University of Bath

PHD

Pole slipping protection for small and medium sized embedded generation

Checksfield, M. J.

Award date:
1997

Awarding institution:
University of Bath

Link to publication

General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners
and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
• You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal ?
Take down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.

Download date: 22. May. 2019

 POLE SLIPPING PROTECTION
FOR SMALL AND MEDIUM
SIZED EMBEDDED GENERATION

Submitted by M J Checksfield,

for the degree of PhD

o f the University o f Bath

1997.

COPYRIGHT

Attention is drawn to the fact that copyright of this thesis rests with its author. This copy
of the thesis has been supplied on condition that anyone who consults it is understood to
recognize that its copyright rests with its author and that no quotation from the thesis and
no information derived from it may be published without the prior written consent of the
author.

This thesis may not be consulted, photocopied or lent to other libraries without the
permission of the author for three years from the date of acceptance of the thesis.

M J Checksfield

M.J.Checksfield/PhD/University of Bath 1996 Page i

 UMI Number: U601737

All rights reserved

INFORMATION TO ALL USERS

The quality of this reproduction is dependent upon the quality of the copy submitted.

In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
a note will indicate the deletion.

Dissertation Publishing

UMI U601737
Published by ProQuest LLC 2013. Copyright in the Dissertation held by the Author.
Microform Edition © ProQuest LLC.
All rights reserved. This work is protected against
unauthorized copying under Title 17, United States Code.

ProQuest LLC
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, Ml 48106-1346
UNIVERSITY OF E r*
 SYNOPSIS

This thesis describes a new power based digital protection algorithm which can be used
to protect a synchronous embedded generation unit against pole slipping. Pole slipping
can cause machine damage and disrupt the quality of supply. The likelihood of embedded
generator instability and hence pole slipping has been studied and it is shown that the
probability of embedded generator pole slipping is high. The new protection algorithm
uses real and reactive power, and rate of change of real power to assess when the
generator operates past the 'point of no return' and pole slipping is unavoidable.

An analysis of the conventional techniques used to detect pole slipping is given, along
with the changes in plant which can be made to reduce the likelihood of pole slipping.
The 'ideal' operating characteristics of a synchronous generator are presented and the
changes to these characteristics which occur in embedded generation sets studied. The
machine characteristics are used to derive the power based pole slipping algorithm, the
basis of which is explained using the Equal Area Criterion method of stability assessment.

The algorithm has been implemented and tested in a commercially available

microprocessor relay and has been shown to work correctly for a wide range of power
system conditions using computer based simulations, laboratory model power systems, and
field trials on three separate generators. The algorithm analysis has included pole slipping,
loss of field, short circuit fault, generator synchronisation, loss of prime mover and other
tests. The performance of the new algorithm has been compared with conventional pole
slipping protection schemes using computer simulation software. The response of standard
embedded generator protection schemes during pole slipping conditions has also been
examined and it has been shown that the standard schemes will rarely trip when pole
slipping occurs.

A summary of the results of the tests performed on the algorithm, conventional techniques
and standard protection schemes has been compiled, and is followed by the conclusions.
The thesis contains a set of appendices which contain the mathematical derivations of
machine and algorithm characteristics, the setting procedures for conventional relays,
embedded generator simulation data, published work and details on the implementation
of the algorithm into a modem microprocessor relay.

M.J.Checksfi el d/PhD /University of Bath 1996 Page ii

 ACKNOWLEDGEMENTS

I would like to express my gratitude to Miles Redfem for his supervision throughout the

course. Thanks is also given to Professor A T Johns for allowing me to use the facilities

of the School of Electronic and Electrical Engineering.

I also wish to thank Martin Balchin, Rod Dunn, Jim Barrett, Jerzy Grzejewski, Alexon

'AAW' Chiwaya, Rich Liewet, David Briggs, Chris Groom, Phil Sapiano Matt Collins and

everyone in the school office and workshop. My special thanks is expressed to James

Hodgeson who had the questionable pleasure of proof reading the thesis.

My thanks also extends to my sponsoring company, GEC Alsthom, T & D, Protection and

Control, Stafford for providing financial and technical support as well as their resources.

Thanks is given specifically to Geoff Weller, Tony Yip, Paul Hindle, Dave Banham, Les

Denning, Peggy Ling, Margaret Leese, Mike Tweed, Andy Ellis, Denise Shed, and Ivor

Scott .

Thanks is also given to Scottish Power, Scottish Hydro, Midlands Electricity and

ECC/Dale Machines. My gratitude is also expressed to Alex Wallis of SWE pic for

providing 'real world' information. My gratitude also extends to the Engineering and

Physical Sciences Research Council and the Royal Academy of Engineers, who both have

provided financial support to enable me to complete this work in addition to attending

international conferences.

M.J.Checksfield/PhD/University of Bath 1996 Page iii

 CONTENTS

Chapter 1
INTRODUCTION

1.1 EMBEDDED GENERATION................................................................................. 1

1.2 PROBLEMS ASSOCIATED WITH EMBEDDED GENERATION................. 1
1.3 POLE SLIPPING .................................................................................................... 3
1.4 THE CAUSES OF POLE SLIPPING ANDASYNCHRONOUS
R U N N IN G ............................................................................................................... 4
1.4.1 The Equal Area C riterion.......................................................................... 5
1.5 DAMAGE CAUSED BY POLE SLIPPIN G ........................................................ 9
1.5.1 Machine Damage Resulting from Pole S lip p in g ................................. 9
1.5.2 The Impact of Pole Slipping on the Restof the Power System .... 11
1.5.3 Legal O blig atio n s..................................................................................... 12
1.6 MICROPROCESSOR RELAYING PLATFORMS ............................. 13
1.7 SUMMARY OF PROPOSED TECHNIQUE FOR DETECTING POLE
S L IP P IN G ............................................................................................................... 14
1.8 RESEARCH OBJECTIVES AND PREVIOUS W O R K .................................... 15
1.9 PRESENTATION FORMAT OF THE THESIS .............................................. 16

Chapter 2
CONVENTIONAL TECHNIQUES FOR DETECTING
AND METHODS FOR PREVENTING POLE SLIPPING

2.1 IMPEDANCE TYPE POLE SLIPPING R E LA Y S.............................................. 22

2.1.1 Loss of Field Scheme ............................................................................. 24
2.1.2 Mho Element S c h e m e ............................................................................. 25
2.1.3 Single Blinder Scheme ........................................................................... 26
2.1.4 Double Blinder S ch em e........................................................................... 28
2.1.5 Lenticular Type Scheme ............................................................................29

MJ.Checksfield/PhD/University of Bath 1996 Page iv

 2.1.6 Double Lens,Triple Lens and Concentric Circle Schemes .................. 30
2.2 POWER BASED POLE SLIPPING PROTECTION ............................................ 31
2.3 MICROPROCESSOR BASED POLE SLIPPING PRO TECTIO N ................... 31
2.3.1 Rate of Change of Apparent Resistance R e la y .........................................31
2.3.2 Multiple Zone Impedance Based Microcomputer Based Protection . . 32
2.3.3 Square Impedance Based Relay Characteristics......................................... 33
2.3.4 Multiple Algorithm Pole Slipping P ro te c tio n ........................................... 34
2.3.5 Artificial Neural Network Pole Slipping Protection..................................35
2.3.6 Pattern Recognition Pole Slipping S c h e m e ..............................................36
2.3.7 Out of Step Voltage Phase Comparison S ch em es.................................... 36
2.3.8 Microcomputer Based Generator Modelling Protection
Techniques..................................................................................................... 37
2.4 EMBEDDED GENERATOR STANDARD PROTECTION SCH EM ES 38
2.4.1 Effect o f Pole Slipping on Over & Under Frequency Relays .... 39
2.4.2 Effect of Pole Slipping on Under & Over Voltage Protection ..............39
2.4.3 Effect of Pole Slipping on Rate of Change of Frequency Relays . . . 40
2.4.4 Effect of Pole Slipping on Voltage DependentOver-current
R e la y s .............................................................................................................41
2.5 TECHNIQUES TO REDUCE THE LIKELIHOOD OFPOLE SLIPPING . . 42
2.5.1 Reducing Fault Clearance T im es................................................................. 42
2.5.2 Resonant L in k s ...............................................................................................43
2.5.3 Fast Acting Voltage Regulators ................................................................. 43
2.5.4 Turbine Fast V alving.....................................................................................44
2.5.5 Fast Governing Systems ..............................................................................44
2.5.6 Braking Resistors ..........................................................................................45
2.5.7 Coordinated Control Schem es...................................................................... 45
2.5.8 Changes in Machine D esig n .........................................................................45
2.5.8.1 Machine Parameter Changes .............................................. 45
2.5.8.2 Novel Machine Designs ......................................................46
2.5.9 Induction Machines ...................................................................................47
2.6 CHAPTER SUM M ARY............................................................................................. 47

M.J.Checksfield/PhD/University of Bath 1996 Page v

 Chapter 3
SYNCHRONOUS GENERATOR CHARACTERISTICS

3.1 IDEAL SYNCHRONOUS POWER CHARACTERISTICS ............................. 55

3.1.1 Round Rotor Machine Real Power Characteristics ............................. 56
3.1.2 Round Rotor Machine Reactive Power Characteristics........................ 57
3.1.3 Salient Pole Generator Real Power Characteristics .............................. 57
3.1.4 Salient Pole Generator Reactive Power Characteristics ....................... 58
3.2 GENERATOR ASYNCHRONOUS POWER CHARACTERISTICS 58
3.3 PRACTICAL SYNCHRONOUS GENERATOR CHARACTERISTICS 59
3.3.1 Non Constant Flux Linkage -
The Direct Axis Open Circuit Transient Time Constant,Tdof ........... 60
3.3.2 Effects of High Stator Resistance on Generator O peration................. 62
3.3.3 Armature Time Constant, Ta .................................................................. 62
3.3.4 Three Phase Short Circuit Torques and Powers ................................ 63
3.3.5 Unbalanced Short Circuit Torques and P o w e rs..................................... 63
3.3.6 The Effect of External Impedance on GeneratorOperating
Characteristics ........................................................................................... 63
3.4 GENERATOR CAPABILITY C H A R T S............................................................. 64
3.4.1 Reactive Power L im its.............................................................................. 64
3.4.2 Saturation E ffe c ts ...................................................................................... 65
3.5 EMBEDDED GENERATOR STABILITY - THE LIKELIHOOD OF
POLE SLIPPING ................................................................................................. 65
3.5.1 Inertia Constant, H & Transient Reactance X d '.................................... 66
3.5.2 Short Circuit Time Constant, Td' ........................................................... 67
3.5.3 Damping Power Provided by Generator and AVR ............................. 68
3.5.4 AVR Field Forcing Action During Short Circuit F a u lts.........................68
3.5.5 Generator G overnors................................................................................. 69
3.5.6 Protection Characteristics of the System Connected to the
G enerator..................................................................................................... 69
3.5.7 Transfer Reactance Between Generator and Infinite B u s ................... 71
3.5.8 Effect of Resistance on S tability............................................................. 71
3.5.9 Affect of Fault Type on Stability .......................................................... 72

M.J.Checksfield/PhD/University of Bath 1996 Page vi

 3.5.10 Affect of Earthing on S tab ility ................................................................ 72
3.5.11 Effects of Embedded Generation System Configuration on
S tab ility ....................................................................................................... 73
3.5.12 Summary of Embedded Generator Stability D iscussion........................ 73
3.6 SYNCHRONOUS GENERATOR POLESLI PPING
CHARACTERISTICS ........................................................................................... 74
3.6.1 Variation in Voltage and Current During Pole S lipping....................... 74
3.6.2 Pole Slipping Explained in Terms of Magnetic F l u x ............................ 76
3.6.3 Direct and Quadrature Axis Flux Variations with Load Angle
In c re a se ........................................................................................................ 76
3.6.4 Real and Reactive Power Pole Slipping C haracteristics....................... 77
3.7 CHAPTER THREE SUMMARY ........................................................................ 78

C hapter 4
THE NEW POW ER BASED POLE SLIPPING ALGORITHM

4.1 BASIS OF THE ALGORITHM .......................................................................... 85

4.2 FILTERING TECHNIQUES.................................................................................. 87
4.2.1 Anti-Aliasing F i l t e r s .................................................................................. 87
4.2.2 Voltage and Current Fourier Full Cycle Cosine Filters ....................... 88
4.2.3 Moving Average F ilte rs ............................................................................. 89
4.2.4 Calculation of the Rate of Change of Power Signal ............................ 90
4.3 THE COMPLETE A LG O R ITH M ........................................................................ 90
4.3.1 The Condition Monitored Real Power Trip Setting, Pt ....................... 91
4.3.2 The Adaptive Rate of Change of Power Trip Setting, (AP/At)min . . . . 92
4.3.3 The Maximum Generator Output Power Monitor, P m a x ..................... 94
4.3.4 Minimum Setting Value for (AP/At)min, Setting (AP/At)fact .................. 94
4.3.5 Implementation of the Power Based Pole Slipping Algorithm into
a Commercial Relay Platform ................................................................ 95
4.4 THEORETICAL OPERATING RANGEOF ALGORITHM ............................. 96
4.4.1 Minimum Value of System Infeed For Algorithm Operation ..............96
4.4.2 Fastest Rate of Pole Slipping Detected by Algorithm ......................... 97
4.4.3 Slowest Rate of Pole Slipping Detected by Algorithm ....................... 98

M.J.Checksfi el d/PhD /University of Bath 1996 Page vii

4.5 OTHER ALGORITHM CONSIDERATIONS ......................................................99
4.5.1 Circuit Breaker Switching Duty ............................................................. 99
4.5.2 Minimum Number of Pole Slips Before T rip p in g .............................. 100
4.5.3 Differentiation Between 'Steady State1 and 'Transient' Pole Slips . . 100
4.6 ADDITIONAL ALGORITHM DEVELOPMENTS - MOTOR POLE
SLIPPING PR O TEC TIO N ................................................................................. 102
4.6.1 Performance for Detecting Loss of Excitation ................................... 103
4.7 ALGORITHM SETTING PR O C ED U R E.......................................................... 104

Chapter 5
ALGORITHM EVALUATION STUDIES

5.1 REAL TIME POWER SYSTEM SIMULATOR (POWSIM) ....................... 113

5.2 THE LABORATORY MODEL POWER SYSTEM 114
5.2.1 Pole Slip T e s t s ........................................................................................ 115
5.2.2 Fault T e s ts ................................................................................................ 116
5.3 PROGRAMMABLE POWER SYSTEM SIMULATOR (PPSS) TESTS . . . 117
5.3.1 Suitability of the ATP Simulation Package for Generator
Simulations ............................................................................................. 121
5.3.2 Simulation of Overhead Lines and C a b le s ........................................... 122
5.3.3 Simulation of T ransform ers................................................................... 123
5.3.4 Simulation of Source Im pedances......................................................... 123
5.4 625 kVA INDUSTRIAL DIESEL GENERATOR FIELD T E S T S ................ 124
5.5 STEAM TURBINE AND HYDRO GENERATOR FIELD T R IA L S 127
5.5.1 Steam Turbine Generator S e t ................................................................. 127
5.5.2 Hydro Turbine Generator S e t ................................................................. 127
5.6 ATP SIMULATIONS OF EMBEDDED GENERATORS............................... 128
5.7 PC BASED SIMULATION OF IMPEDANCE R E LA Y S............................... 130

MJ.Checksfield/PhD/University of Bath 1996 Page viii

 Chapter 6
RESULTS OF ALGORITHM EVALUATION TESTS

6.1 POWSIM TEST RESULTS ................................................................................ 133

6.1.1 Response to Pole Slip T e s t.................................................................... 133
6.1.2 Response to Stable Power Swing T e s t................................................... 134
6.2 LABORATORY MODEL POWER SYSTEM TEST R E S U L T S ................... 135
6.2.1 Weak Field Pole Slip - Test'C ' .......................................................... 135
6.2.2 'Over Torque' Pole Slip - Tests 'E,F,G,H,I,J'....................................... 137
6.2.3 Adjacent Generator Pole Slip and Loss of Excitation - Tests'K,L' . 137
6.2.4 Generator Fault Tests ............................................................................ 138
6.2.5 Observations from the Laboratory Power System Model Tests . . . . 139
6.2.6 Analysis of Pole Slipping Impedance Loci for the 5 kVA
Laboratory Generator ............................................................................ 140
6.3 PPSS TEST RESULTS ........................................................................................ 142
6.3.1 Pole Slipping Algorithm Response to PPSS Pole Slip T e sts 142
6.3.2 Pole Slipping Algorithm Response to PPSS Stable Power Swing
T e s t s .......................................................................................................... 148
6.3.3 Pole Slipping Algorithm Response to PPSS Tests Where the
Generator is Operated in its Dynamic Stability Region - at Load
Angles Greater than 90° ........................................................................ 151
6.3.4 Pole Slipping Algorithm Response to PPSS Loss of Field Tests . . . 152
6.3.5 Pole Slipping Algorithm Response to PPSS Loss of Prime Mover
T e s t s .......................................................................................................... 152
6.3.6 Effect of Altering the Pole Slipping Algorithm 'Slip' S e ttin g 153
6.3.7 Comparison of the Performance of the New Power Based Pole
Slipping Algorithm with Conventional Impedance Based Pole
Slipping S c h e m e s.................................................................................... 154
6.3.7.1 Comparison of Impedance Relays with New Power
Based Protection Algorithm for Pole Slipping
C onditions........................................................................ 156
6.3.7.2 Comparison of Impedance Relays with New Power
Based Protection Algorithm for Stable Power Swing

M.J.Checksfield/PhD/University of Bath 1996 Page ix

 Conditions....................................................................... 159
6.3.7.3 Comparison of Impedance Relays with New Power
Based Protection Algorithm for Other Power System
C onditions.................................................................. 161
6.3.8 Analysis of the Performance of Current, Voltage and Frequency Relays to
the PPSS Pole Slipping C onditions................................................. 163
6.3.8.1 Response of Overcurrent Relay to Pole Slipping . . . 164
6.3.8.2 Response of Under-Voltage Relay to Pole Slipping . 165
6.3.8.3 Response of Overfrequency Relays to Pole
Slipping .......................................................................... 166
6.4 625 KVA DIESEL GENERATOR FIELD TRIAL RESULTS .................... 167
6.4.1 625 kVA Diesel Generator Pole Slip Tests ......................................... 167
6.4.2 625 kVA Diesel Generator Loss of Excitation T e s t s .......................... 169
6.4.3 625 kVA Diesel Generator Steady State and Load T e s ts ................... 170
6.4.4 Comparison of Conventional Impedance Based Pole Slipping Relays
with Power Based Relay for 625 kVA Diesel Generator Pole Slip
T e s t s .................................................................................................... 171
6.5 STEAM TURBINE AND HYDRO GENERATOR FIELD TRIAL
RESULTS ............................................................................................................ 172
6.6 ADDITIONAL ATP SIMULATION TEST RESULTS ................................. 173

Chapter 7
SUMMARY OF RESULTS

7.1 RESPONSE OF THE POWER BASED POLESLIPPING

ALGORITHM 244
7.1.1 Differentiation Between 'Transient' and 'Steady State' Pole Slips . . 244
7.1.2 Adaptive Tripping Times for Minimising Circuit Breaker Operating
Duty ......................................................................................................... 244
7.1.3 Analysis of the Optimum Values for the 'lag' and 'Ptol' Settings . . 245
7.1.4 Algorithm Response to Adjacent Generator Disturbances ............... 245
7.1.5 Algorithm Response to Fault T e s ts .................................................. 245
7.1.6 Algorithm Operational L im its........................................................... 246

M.J.Checksfield/PhD/University of Bath 1996 Page x

 7.1.7 Algorithm Response to Stable Power Swing Tests .......................... 247
7.1.8 The Effect of Using a Lower 'Slip* Setting........................................ 247
7.1.9 Algorithm Performance for Detecting Loss of Field Conditions . . 248
7.1.10 Algorithm Tripping During Synchronous Operation ....................... 248
7.1.11 Algorithm Modification to Detect Motor Pole Slipping ................ 249
7.2 RESPONSE OF THE IMPEDANCE BASED POLE SLIPPING RELAYS 250
7.2.1 The Mho Pole Slipping Protection Scheme ....................................... 250
7.2.2 The Single Blinder Protection Scheme ............................................... 250
7.2.3 The Double Blinder Protection S ch em e................................. 251
7.3 THE EFFECTIVENESS OF STANDARD PROTECTION SCHEMES AT
DETECTING POLE S L IP P IN G .......................................................... 252

Chapter 8
CONCLUSIONS & FURTHER WORK

8.1 CONCLUSIONS....................................................................................... 253

8.2 SUGGESTIONS FOR FUTURE W O R K ............................................. 256

Chapter 9
REFERENCES

REFERENCES .......................................................................................................... 257-270

M.J.Checksfield/PhD/University of Bath 1996 Page xi

 Appendix A
DERIVATION OF THEORETICAL IMPEDANCE,
POW ER VARIATIONS AND ALGORITHM SETTING EQUATIONS

Al l CALCULATION OF THE IMPEDANCE LOCI PRODUCED DURING

POLE SLIPPING ................................................................................................ 271
Al.1.1 The Affect of Resistance on Impedance L o c i.......................... 275
A 1.1.2 The Affect of Variation in the Generator to Source
Impedance R a tio .......................................................................... 275
A2.1 REAL AND REACTIVE POWER TRANSFER CHARACTERISTICS . . 275
A3.1 SYNCHRONOUS GENERATOR REAL AND REACTIVE POWER
CHARACTERISTICS WHEN OPERATING ON AN INFINITE BUS . . . 277
A3.1.1 Transient Power C haracteristics................................................ 279
A3.1.2 Calculation of Synchronous Generator Steady State and
Transient Real and Reactive Power C u r v e s ........................... 281
A3.1.2.1 No Load C alculations..................................................... 281
A3.1.2.2 Full Load Calculations .................................................. 282
A3.1.3 Calculation of Generator Characteristics for a Diesel
G enerator...................................................................................... 282
A4.1 ANALYSIS OF GENERATOR REAL AND REACTIVE POWER
CHARACTERISTICS WHEN OPERATING AGAINST A SIGNIFICANT
SOURCE IMPEDANCE, SALIENCY EFFECTS IN C L U D ED .................... 283
A5.1 THE EFFECT OF MACHINE LOSSES UPON THE POLE SLIPPING
ALGORITHM AND ASSESSMENT OF THE CORRECT VALUE FOR
Pfact ...................................................................................................................... 285
A5.2 THE VALUE OF RATE OF CHANGE OF POWER AT THE CSP _ 287
A5.2.1 Mathematical Analysis of (AP/At)minat the CSP . 288
A5.2.2 Including Transient Saliency Effects in the (AP/At)min
Calculation ................................................................................. 290
A5.2.3 Errors Produced by the Assumptions Used to Derive
(AP/At)min ............................................................................ .. . . 291

M.J.Checksfield/PhD/University of Bath 1996 Page xii

 A5.2.4 Units Used in Equations... ........................................................... 292
A5.3 DERIVATION OF LIMITS OF OPERATION OF ALGORITHM AS THE
SYSTEM INFEED IS V A R IE D ....................................................................... 293

Appendix B
SETTING OF CONVENTIONAL IMPEDANCE RELAYS

B l.l LOSS OF FIELD SCHEME ............................................................................ 303

Bl.1.1 Loss of Field Settings for 200 MVA PPSS Test Generator . 303
B l.l.2 Loss of Field Settings for 5 kVA Laboratory Generator . . . 304
B l.l.3 Loss of Field Settings for 625 kVA Diesel G enerator.. 304
B1.2 MHO ELEMENT POLE SLIPPING SCH EM E............................................... 304
B 1.2.1 Mho Element Settings for 200 MVA PPSS Test Generator . 304
B 1.2.2 Mho Element Settings for the 5 kVA Laboratory
Generator...................................................................................... 305
Bl.2.3 Mho Element Settings for the 625 kVA Industrial Diesel
Generator...................................................................................... 305
B1.3 SINGLE BLINDER POLE SLIPPING PROTECTION SCHEME ............... 306
Bl.3.1 Single Blinder Settings for 200 MVA PPSS Generator . . . . 306
Bl.3.2 Single Blinder Settings for the 5 kVA Laboratory
G enerator...................................................................................... 307
Bl.3.3 Single Blinder Settings for the 625 kVA Diesel Generator . 307
B1.4 DOUBLE BLINDER SCHEME ...................................................................... 308
B 1.4.1 Double Blinder Settings for 200 MVA PPSS Generator . . . 308
B 1.4.2 Double Blinder Settings for the 5 kVA Laboratory
G enerator...................................................................................... 309
B 1.4.3 Double Blinder Settings for the 625 kVA Diesel
G enerator...................................................................................... 310

M.J.Checksfield/PhD/University of Bath 1996 Page xiii

 Appendix C
DATA FOR GENERATORS AND SYSTEMS USED
FOR ALGORITHM TESTS

C l.l PPSS TEST SYSTEM ....................................................................................... 318

C l.1.1 Governor M o d e l........................................................................... 318
C l.1.2 AVR M o d e l.................................................................................. 318
C l. 1.3 Generator Model ................................................................. 319
C l.1.4 Generator InfiniteBus D e ta ils .................................................... 320
C l.1.5 Load & Fault Inform ation.......................................................... 321
C l.2 LABORATORY POWER SYSTEM MODEL ................................................ 321
C l.3 625 KVA INDUSTRIAL DIESEL GENERATOR........................................... 323
C l.4 DATA FOR DIFFERENT SIZED SYNCHRONOUS M A CH IN ES 324

Appendix D
PUBLISHED WORK

Paper 1 - UPEC '93 Conference ................................................................................... 329

Paper 2 - IEEE Transactions on Power Delivery - January 1995 334
Paper 3 - UPEC '94 Conference ................................................................................... 344
Paper 4 - IPEC '95 C onference...................................................................................... 349
Paper 5 - UPEC '95 Conference ................................................................................... 356
Paper 6 - APSCOM '95 Conference............................................................................... 329
Paper 7 - IEEE Transactions on Power Delivery - Ref. PE-486-PWRD-0-11-1996 368
Paper 8 - UPEC '96 Conference ................................................................................... 378
Paper 9 - IEE Colloquium Digest 1996 - Generator Protection .............................. 383
Paper 10 - Western Protection Relay Conference '96, USA ..................................... 393
Paper 11 - CEPSI '96 C o n feren ce................................................................................. 399
Paper 12 - IEE Developments in Power System Protection Conference 1997 . . . . 410
Paper 13 - ERA International Conference 1997 ........................................................... 415

M.J.Checksfield/PhD/University of Bath 1996 Page xiv

 Appendix E
IMPLEMENTATION OF THE POWER BASED POLE SLIPPING
ALGORITHM INTO A COMMERCIAL MICROPROCESSOR RELAY

Appendix E ................................................................................................................. 426-465

Appendix F
EXAMPLES OF COMPUTER FILES USED TO
TEST THE POLE SLIPPING ALGORITHM

F l.l PPSS ATP Simulation - file 'PSL3GA1 ................................................... 466

F I.2 ATP BCTRAN 1.5 MVA Delta-Star Transformer f i l e .................................... 470
F I.3 ATP Fast Diesel Governor - file 'FG2' ................................................... 472
F I.4 ATP Slow Loss of Synchronism DuringLoss of Field .................................. 477
F2.1 'C* code for PC based Simulation of Single Blinder Scheme ....................... 482
F2.2 'C' code for PC based Simulation of Power Based Pole SlippingAlgorithm 485
F2.3 'C' code for PC based Simulation of IDMT Overcurrent Relay ............. 492
F2.4 'C' code for PC based Simulation of Frequency Measuring Algorithm . . . 494

M.J.Checksfield/PhD/University of Bath 1996 Page xv

 LIST OF SYMBOLS AND ABBREVIATIONS

'lag' The number of samples between check points in 'Pm' estimation function
(AP/At) Rate of change of power
(AP/At)min pole slipping algorithm rate of change of power trip setting
(AP/A t)^ Transient saliency rate of change of power setting
(AP/At)fact Minimum magnitude value for (AP/At)min
<|)a Armature flux
<|>d Direct axis flux
<|>f Field flux
<|>q Quadrature axis flux
ANN Artificial Neural Network
ATP Alternative Transients Program
AVR Automatic Voltage Regulator
CCT Critical Clearing Time
CHP Combined Heat and Power
CSP Critical Stability Point
CT Current Transformer
5 Load angle
5c Critical Clearing Angle
DC Direct Current
5m Machine angle (angle between E and Vg)
80 Initial load angle
DSG Dispersed Storage and Generation
E Generator steady state internal voltage
Ed Direct axis voltage
Eg Generator voltage behind impedance
Eg/Es Ratio of source (Es) to generator (Eg) internal voltages
EMF Electro Motive Force
EMTP ElectroMagnetic Transients Program
EPRI Electric Power Research Institute
EPROM Erasable Programmable Read Only Memory
Eq Quadrature axis voltage

M.J.Checksfield/PhD/University of Bath 1996 Page xvi

Eq' Voltage behind transient reactance
Erlr* Apparent power
Er Relay voltage
Es Source voltage behind impedance
f System frequency (Hz)
GPS Global Positioning System
H Generator inertia constant
HP Horse Power
HV High Voltage
I generator stator current (RMS value)
I" Sub-transient fault current

K ,ib »ic instantaneous a,b,c line current

Ic Cosine fourier filter output of signal, i
Id Generator direct axis current
IDMT Inverse Definite Minimum Time
In Relay nominal current
Iq Generator quadrature axis current
Ir Relay current
Is Sine fourier filter output of signal, i
Ld Direct axis synchronous inductance
Lff Field winding inductance
LOE Loss of Excitation
LOF Loss of Field - identical to loss of excitation
m Zg/Ztot
mmf MagnetoMotive Force
NFFO Non Fossil Fuels Obligation
OT Over Torque, a type of pole slip
P Real power

- lag) Sample value of real power 'lag' samples ago

P(n - (2*lag)) Sample value of real power '2 * lag' samples ago

(n) Present sample value of real power

Accelerating power on generator rotor

add-loss Additional losses which occur as a result of generator transient operation

M.J.Checksfi el d/PhD /University of Bath 1996 Page xvii

Pas , Generator asynchronous real power output
PC Personal Computer
Pfact coefficient used to calculate Pt from estimate of Pm
Ploss Losses in generator during steady state operation
Pm Prime mover input power
PPSS Programmable Power System Simulator
PS Pole Slip
PSM Plug Set Multiplier
Pt Pole slipping algorithm real power trip setting
Ptol Real power tolerance band in Pm estimation function
pu Per Unit
0 sin _1 (R/Z) (for generators)
Q Reactive power
Qas Generator asynchronous reactive power output
Qtrip Pole slipping algorithm reactive power trip setting
r External resistance, between generator terminals and infinite bus
R Generator stator resistance
REC Regional Electricity Company
Rf Field winding resistance
RLC Resistance, Inductance, Capacitance model of cable or overhead line
RX Resistance/Reactance (used with reference to impedance diagrams)
S Apparent Power (VA)
s generator rotor slip
SCADA Supervisory Control and Data Acquisition
SF6 Sulphur hexaflouride
Sgen Generator rating
t time
Ta Generator armature time constant
Td' Direct axis short circuit time constant
TdH Direct axis sub-transient short circuit time constant
Tdo' Direct axis open circuit time constant
Tdo" Direct axis sub-transient open circuit time constant
Te Machine air gap torque

M.J.Checksfiel d/PhD /University of Bath 1996 Page xviii

TMS Time Multiplier Setting
Tq" Quadrature axis sub-transient short circuit time constant
Tqo" Quadrature axis sub-transient open circuit time constant
V infinite busbar voltage
v b >vb >v c Instantaneous a,b,c phase voltage
Vg Generator terminal voltage
VT Voltage Transformer
cor Rotor angular speed
cos Synchronous angular speed
X External reactance between generator terminals and infinite bus
XComb Combined value of synchronous reactance plus any external reactance.
Xd Direct axis synchronous reactance
Xd1 Direct axis transient reactance
Xd" Direct axis sub-transient reactance
Xq Quadrature axis synchronous reactance
Xq' Quadrature axis transient reactance
Xq" Quadrature axis sub-transient reactance
Z V (R 2 + Xq2 )
Zd' Generator transient impedance
Zg Generator operational impedance
Zo Zero sequence impedance
Zr 'Apparent' impedance as seen by a protection relay
Zs Source impedance
Ztot combined impedance of generator and source

M.J.Checksfield/PhD/University of Bath 1996 Page xix

 Chapter 1
INTRODUCTION

1.1 EMBEDDED GENERATION

Embedded generation or Dispersed Storage and Generation units (DSG) is that generation
which is connected in parallel with a utility's distribution network. Until recently, the bulk
of electricity was generated in large power stations and transferred to consumers via the
transmission network and utility distribution system. Recent changes in legislation in the
UK[1,2] and the USA[3] have granted permission for local embedded generation to operate
in parallel with utility networks. This has resulted in some of the electricity generation
occurring adjacent to the loads and within the distribution networks.

There are several factors which have contributed to the increased number of small and
medium sized embedded generation units being commissioned. The past two decades have
seen increased economic, political and environmental pressures on the production of
electricity which have increased interest in incorporating embedded generation into public
utility networks [4,5,6]. Combined Heat and Power Schemes improve the overall efficiency
of plant by using waste heat from industrial processes to run generators, or vice versa. The
current trend in alternative energy sources such as landfill gas, waste incineration and
renewables like wind, tidal, solar, mini-hydro and bio-fuel schemes has resulted in a large
number of embedded generators being installed. The Non Fossil Fuel Obligation (NFFO)
acts as an incentive since it entitles alternative energy schemes to receive premium prices
for their electricity.

1.2 PROBLEMS ASSOCIATED W ITH EMBEDDED GENERATION

Connecting a small or medium sized generator to operate in parallel with a utility

distribution network causes problems concerning safety, quality of supply, earthing and
stability[7’8’9’10,1U2>,3]. Safety problems result because in a traditional network, disconnecting
the main source of supply de-energises the network, if embedded generation is connected,
part of the network may remain energised. Quality of supply concerns voltage regulation,

M.J.Checksfield/PhD/University of Bath 1996 Chapter 1, Page 1

frequency regulation, and harmonics. Regional Electricity Companies (REC's) have to
maintain their supplies within pre-defined voltage and frequency limits, and the operation
of an embedded generator can create difficulties since the generator may corrupt the utility
supply if operated incorrectly. Short circuits on the utility distribution system, and faults
in a synchronous embedded generator's control equipment can cause the generator to
become unstable. This instability will result in pole slipping, which may compromise the
quality of utility supply as well as causing plant damage.

Special precautions are therefore required to prevent the embedded generator degrading
the quality of supply to other utility customers. The regulating authorities require the
utilities to ensure that the embedded generators will not detract from the quality of supply
to other customers and have produced guidelines which define the general protection
requirements for protection against all types of faults and abnormal operating conditions.
These guidelines are published as 'G 59'[14] and 'ET113'[15] in the UK and 'ANSI 1001'tl6}
in the USA. It is the embedded generator operator's responsibility to supply the necessary
level of protection and control equipment for the embedded generator. The utility to which
the generator is connected will specify the level of protection that they require. Meeting
these protection specifications117,18’19,20,21’221 can be time consuming and expensive for the
embedded generator operator. There is therefore a need for the availability of in­
expensive relays which offer the required level of protection and do not require
complicated settings.

A widespread complaint from potential operators of embedded generation that the

conventional utility grade relays are too expensive. For a small scheme, the protection
costs can be comparable to that of the generator itself. To make any small generation
scheme economically viable, a low cost multi-functional protection package is therefore
required123,24,25,26,27,28,29,301. Using a microprocessor based relaying platform provides
numerical computing power which can be used to enhance existing protection algorithms
or devise completely new algorithms which were not possible with conventional static or
electro-mechanical relays[31]. The new pole slipping detection algorithm presented in this
thesis is one such algorithm which takes advantage of the computing power available in
a modem microprocessor relay platform.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 1, Page 2

1.3 POLE SLIPPING

A potential difficulty with the operation of an embedded generator is to ensure that the
electrical system can absorb the electrical power produced from the mechanical power
provided to the generator. If too much mechanical power is provided to the generator and
it is unable to deliver that power to the system, instability results. This is referred to as
pole slipping.

Pole slipping occurs when the rotor of a synchronous machine slips one pole pitch with
respect to the stator magnetic flux, which is synchronised to the stator electrical supply.
Two or more machines operating in parallel are required for a pole slip to occur. If there
is just one machine then the rotor speed changes result in a change in the frequency of
the electrical supply produced by the machine. Pole slipping results from the mismatch
in frequencies of two or more machines. For the case of an embedded generator pole
slipping, the other machine can be regarded as the utility infinite bus[15]. If the infeed
capacity at the terminals of the embedded generator is more than 40 to 50 times the rating
of the generator, then the utility infeed can be considered infinite.

The term 'pole slipping' is used only when the field coil on the rotor is excited. Field coil
excitation produces a magneto-motive force (mmf) which drives a flux that links the stator
and rotor. The points where the magnetic flux leaves and enters the rotor structure can
be thought of as magnetic poles, similar magnetic poles will be formed on the stator due
to the three phase alternating current flowing in the stator windings. Figure 1.1a shows
the magnetic poles perfectly aligned in a machine operating at no load. When the
machine is operating as a generator, the generator rotor runs ahead of the rotating flux
field produced by the stator, as shown in Figure 1.1c, and in the phasor diagram of
Figure l.ld . In this state, the magnetic poles move out of alignment and an
electromagnetic torque, Te, results which rotates at speed cos, thus producing power.
During normal stable operation, the 'north' pole of the rotor is attracted to the 'south' pole
of the stator, and vice versa. During pole slipping the generator rotor speed, cor is
different to the electrical supply angular speed, cos, and the rotor magnetic poles therefore
'slip' past the stator poles, and this is the origin for the term 'pole slipping'. This explains
the large fluctuations in torque which result from pole slipping since for the first half of
the pole slip cycle (0° < 8 < 180°), the magnetic poles attract each other producing a
generating torque, whilst for the other half of the pole slip cycle (180° < 5 < 360°), the

M.J.Checksfield/PhD/University of Bath 1996 Chapter 1, Page 3

magnetic poles repel each other, producing a motoring torque.

If the machine's field coil is not excited, the machine operates 'asynchronously'. Since
there is no excitation on the rotor, there can be no rotor magnetic poles and hence no pole
slipping. This mode of operation is referred to as 'asynchronous running' f32].

Technically speaking, pole slipping is classed as asynchronous running. The term

'asynchronous running' will be used in this thesis to refer only to operation out of
synchronism with the rotor winding un-excited. This condition may occur during a loss
of field, or loss of excitation condition, or the generator may even be operated in this
mode as a means of absorbing reactive VArs from the power system [33*34]. In reality, the
rotor will still have weak magnetic poles when there is no excitation, due to residual flux
and the magnetic and electrical asymmetry in the rotor. The difference in the use of the
two words is justified due to the radical difference between the two modes of operation.
With pole slipping, the excitation on the rotor causes severe voltage, power, and reactive
loading surges to occur. Asynchronous operation on the other hand, results in relatively
small fluctuations in the generator voltage and currents in round rotor synchronous
machines. Salient pole machines will produce larger fluctuations in voltage and current
during asynchronous operation, due to the reluctance torque component produced by their
saliency.

1.4 THE CAUSES OF POLE SLIPPING AND ASYNCHRONOUS RUNNING

Pole slipping occurs when there is insufficient electromagnetic torque to hold the rotor in
synchronism with the stator magnetic flux. There are several reasons why the
electromagnetic torque produced by the interaction of stator and rotor flux is not strong
enough to counteract the driving torque of the prime mover and therefore stop the rotor
accelerating to a speed above synchronous speed. Equation 1(1) gives the relationship
between generator output power, P, and machine load angle, 5, for a machine operating
on an infinite bus of voltage, V, during steady state operation.

2
( 1 1 V
P = sin(5) sin(28) 1(1)
Xd {Xq Xd

Where E is the steady state excitation speed voltage, Xd is the direct axis synchronous

M.J.Checksfield/PhD/University of Bath 1996 Chapter 1, Page 4

reactance, and Xq is the quadrature axis synchronous reactance. It can be seen that if E
or V is decreased the maximum amount of power and therefore torque that the machine
is capable of producing decreases. If this maximum amount of electromagnetic torque
falls below the prime mover's mechanical input torque, then a net accelerating torque will
result causing the rotor to rise above synchronous speed and the machine to pole slip.
When the generator rotor travels above synchronous speed, the machine load angle, 5,
increases. Equation 1(1) shows that once the machine load angle exceeds 90°, the
generator output power decreases with an increase in machine load angle. If the machine
load angle has exceeded 90° and the electromagnetic torque is less than the prime mover
input torque, then the generator is operating in an unstable manner and a pole slip will
result.

The discussion so far has only involved steady state operation, most pole slips which
occur result from a transient disturbance on the generator such as a power system short
circuit fault, or line switching operation.

1.4.1 The Equal Area Criterion

The Equal Area Criterion is one of the simplest methods for assessing generator stability
for a given transient disturbance. If the power/machine load angle curves for all states of
power system operation are known, then generator stability and hence the possibility of
pole slipping can be ascertained for a given disturbance.

Consider the case where a transient disturbance results from a three phase fault on one
side of a double circuit line as shown in Figure 1.2a [35]. The fault is removed by
disconnecting the faulty circuit, leaving the healthy side of the line to transfer power at
a reduced capacity. The power/load angle curves and equal area representation of this
scenario is shown in Figure 1.2b.

The pre-fault curve corresponds to the initial operating condition, where both lines are
transferring power. In this condition the embedded generator's power transfer capability
is at its maximum. The fault duration curve represents the generator's power/load angle
characteristic during the fault period. The three phase fault severely reduces system
voltage so that power transfer is impaired. The post fault curve depicts the power transfer
capability once the faulty line has been disconnected, the power transfer capability is
reduced due to an increase in the interconnecting impedance.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 1, Page 5

The generator's prime mover input power is shown by the line Pm. The interception of
the pre-fault output curve and prime mover input curve gives the initial machine load
angle 81. When the fault occurs the generator's output power characteristic drops down
to the fault duration curve. The resulting discrepancy between input and output powers
produces a net accelerating torque on the generator rotor. As the rotor accelerates to a
speed above synchronous speed, the machine load angle increases.

In this example, when the fault is removed, corresponding to a load angle of 53, the
generator's output power jumps up above the line Pm to the post fault curve. This occurs
because the system voltage is restored. This mismatch in generator input and output
powers results in a net retarding force on the generator rotor. Since the rotor is now
travelling above synchronous speed, the initial action of the retarding force is to remove
kinetic energy from the rotor, and return the rotor towards synchronous speed. If the rotor
is returned to synchronous speed before machine load angle 5c is reached, then the
generator will remain stable for that disturbance. Once the generator rotor is at
synchronous speed, any extra retarding force will slow it to below synchronous speed,
causing the machine load angle to decrease back towards its new operating angle, 52.

If the rotor does not reach synchronous speed by machine load angle 5c, then the
generator rotor will again start to accelerate because beyond 5c, the electrical energy out
is less than the mechanical energy in, causing an accelerating torque once more. This is
the unstable situation which results in the rotor travelling above synchronous speed and
the generator pole slipping. If the generator is not able to counteract the accelerating
forces during the pole slipping cycle it will continue to pole slip until the generator
governor decreases the input torque, Pm, or the generator is disconnected from the power
system.

The Equal Area Criterion [36,37] states that the accelerating energy on the rotor due to a
disturbance can be represented by area 'A', whilst the decelerating energy is represented
by area 'B'. For synchronism to be lost, the net accelerating energy for a given
disturbance must exceed the net decelerating energy, i.e a pole slip will occur if area 'A'
is greater than area 'B'. It can be seen from Figure 1.2b that the limiting point for stability
occurs when the generators output power falls below its input power, this point is known
as the Critical Stability Point (CSP) and occurs at the load angle 5c. If the generator
operates past this point, then stability cannot be maintained, and a pole slip will occur.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 1, Page 6

Figure 1.3a shows the same Equal Area Diagram as in 1.2b, and also shows the generator
load angle characteristic for a generator losing synchronism following a sudden transient
shock to the generator [35]. It can be seen that once the generator operates past the CSP,
it accelerates quickly into the pole slip because in addition to the prime mover power
accelerating the rotor, the generator also draws power from the utility supply as it travels
through its motoring region.

The limiting case for stability occurs when the acceleration energy area equals the
deceleration energy area, resulting in zero net accelerating torque for the disturbance. In
this instance, the rotor will swing up to load angle 5c, and then swing back down. The
duration of the three phase fault which causes this condition is referred to as the Critical
Clearing Time (CCT) since any fault whose duration is longer than this will cause
generator instability.

The three phase fault discussed in this example is the most severe fault possible in terms
of generator stability because the generator terminal voltage is reduced to a very low
value, preventing any electrical power transfer from the generator. At its most extreme,
the three phase fault when located close to the generator's terminals will reduce the
generator's voltage to zero and no power transfer out of the generator will be possible.

Other faults can produce instability, however an increased fault duration is necessary
compared with the three phase fault since some load is still maintained, and therefore
there is less accelerating energy going into the rotor. In terms of a fixed fault duration,
one study found [38] that the rotor displacement with a phase to phase fault is one quarter
that of a three phase fault. Similarly, a phase to ground fault only produces one seventh
of the rotor swing, whilst a two phase to earth fault gives five eighths displacement. It
is important to note that these figures do not dictate the stability of a generator to a
disturbance entirely, since stability is also dependent on the post fault impedance.

Pole slipping can also be invoked by a sudden change in transfer impedance between
generator and load, since an increase in transfer impedance will cause a decrease in the
maximum output power of the machine. More commonly, pole slipping of embedded
machines will be caused by long fault clearance times. The fault clearance times are
necessary on distribution systems protected by IDMT type relays to provide adequate
grading [17’39>40]. One study[41] found that the minimum clearing time for industrial feeder

M.J.Checksfield/PhD/University of Bath 1996 Chapter 1, Page 7

short circuits was 0.5 seconds, whilst another [42] stated typical clearing times in the range
0.5 to 1 second.

Faulty excitation systems may also cause the generator to pole slip since they may weaken
the field so that the generator is unable to maintain synchronism. Figure 1.3b shows the
Equal Area Diagram for a pole slip resulting from a loss of steady state stability. This
situation may arise due to insufficient excitation, which takes the peak output power of
the generator below the input power. It could also be caused by increasing the prime
mover input power without a corresponding increase in excitation. Both of these
conditions may result from operator error, or faulty control systems1431.

Complete loss of field is more likely however, since faults can develop in many different
locations in the excitation circuit, such as slip rings, field windings and exciter windings.
Operator error may also cause a loss of field condition due to the field breaker being
accidentally tripped[28,44]. The generator AVR may counteract a partial field fault by
boosting field current, in this condition the generator will not lose synchronism, but may
sustain damage through over-heating or vibration[45]. Complete field failure will eliminate
all synchronous torque except reluctance torque, and asynchronous running will therefore
result. An exception to this occurs with a lightly loaded salient pole generator where the
reluctance torque component may be sufficient to hold the rotor in synchronism.

The effects of asynchronous running are not usually as severe as pole slipping. With no
field on the rotor, the electromagnetic torque pulsations produced as a result of the rotor
and stator magnetic field slipping past one another are dictated only by the electrical and
magnetic asymmetry in the machine producing a reluctance torque. The Equal Area
Diagram for the loss of field condition is shown in Figure 1.3c, note that unlike Figures
1.3a and 1.3b the diagram only contains the first half of the power load angle
characteristic. When a short circuit occurs in the field circuit, the generator flux decays
slowly, resulting in the generator synchronous output power falling to zero. It should be
noted that this diagram does not show the asynchronous component of torque generated
from the rotor travelling above synchronous speed. This can be significant in some
generators, allowing the machine to continue to generate a significant amount of power.

If a generator loses its field from full load, then it is vital that it be disconnected quickly,
since the generator will operate at values of slip as high as 5 %. At this rate of slip high

M.J.Checksfield/PhD/University of Bath 1996 Chapter 1, Page 8

levels of induced currents can flow in the rotor, and the stator current can rise to above
twice its rated value. With such high currents, rotor and stator over-heating can occur
quickly, and the generator will draw excessive reactive power from the system. This may
depress system voltages and thus adversely effect the performance of nearby machines and
generators142,461. The depression in system voltage can cause instability in other machines,
and in extreme cases, cause a cascading loss of synchronism of machines nearby [24,4?1.

1.5 DAMAGE CAUSED BY POLE SLIPPING

The damage caused by pole slipping to the machine is dependent on the amount the
machine has been over designed, if the machine has a sufficient reserve in terms of shaft
rating and thermal capacity, then a pole slipping event may cause no machine damage at
all. However, if the machine does not have sufficient reserves, damage can result.

1.5.1 Machine Damage Resulting from Pole Slipping

The extent of the damage caused to the machine is dependent on the rating of the bus
which the generator is connected to, the speed of the pole slip, and the overcapacity rating
of the machine. When the fault level of the utility system is high in comparison to the
generator rating, the currents experienced during pole slipping will be of a higher
magnitude than three phase fault levels. Since machine windings are only braced for three
phase fault levels, the mechanical forces on the windings produced by these higher current
levels during pole slipping will loosen and deteriorate the windings1421.

In addition to the high stator currents produced as a result of synchronising forces, an

additional component of stator current will be produced as a result of the machine acting
as an induction generator. These currents are highest when the speed of the machine has
risen above the peak of it's torque/slip curve135,481.

Synchronous machines are not designed to be operated as induction generators, the

induced slip frequency currents that flow in the rotor body, damping circuits and
excitation windings can result in serious overheating if prolonged operation is permitted1491.
The rotor body and damper circuits are especially prone, since deep current penetration
due to low frequency currents contributes a high heating component. In one pole slipping
test performed on a 60 MW machine1501, slight blueing of the rotor end rings had occurred

MJ.Checksfield/PhD/University of Bath 1996 Chapter 1, Page 9

after one minute of operation, indicating that the conditions had taken the machine close
to its safe operating limits.

In most machines, stator end region overheating is also likely to result from prolonged
operation, due to an increased axial component of leakage flux1481. Investigations into the
type of end ring material showed that a magnetic end ring will experience a greater
temperature rise than a non-magnetic end ringt50].

Pole slipping can also cause damage to excitation systems. Pole slipping induces large
currents in the field winding, which compensate the high mmf of the stator windings1491.
The magnitude of these currents peaks as the load angle approaches 180°, because the
rotor and stator become magnetically disconnected at this point. The field winding will
try to maintain constant flux linkage and will therefore compensate for the missing flux
contribution from the stator. Once past the 180° point, the currents drop rapidly as the
stator and rotor flux begin to link once more, but in the opposite sense (in a motoring
condition) and the rotor current must decrease rapidly to maintain the mmf balance.
Rectifier excitation systems will not permit the rotor current to reverse, thus ending the
mmf balance and producing a rapid flux change. This high rate of change of flux induces
a large reverse voltage across the rectifiers which can cause reverse breakdown if they are
not sufficiently over rated[51].

Recently, concern has increased over the effect power system disturbances have on
generator shafts, out of phase re-closures and pole slipping can both produce potentially
damaging shaft torques. During pole slipping, if the frequency of the high currents and
pulsating torques produced fall upon a natural frequency of the shaft, then the shaft will
be torsionally excited into a resonance condition, exposing the shaft to oscillatory stress.
Particularly at risk are turbine generator shafts[52] since the bladed rotating parts of the
turbine interact with the couplings and reduced diameter shaft extensions to form a
torsionally resonant system. One study1531, found that the generator-exciter shaft section
experiences the highest stresses. It was found that all of the fatigue life of this shaft
section was used after one pole slip for a shaft over design rating of two. For an over
design rating of five, no significant damage occurred.

The study indicated that the maximum torques and torsional stresses occur during the first
period of oscillation after each torsional impulse, and it is this period where the shaft

M.J.Checksfield/PhD/University of Bath 1996 Chapter 1, Page 10

damage usually occurs. It is therefore advantageous to disconnect the generator from the
system before a pole slipping event occurs in the hope that the torsional impulse will be
prevented. It was also indicated that the cumulative effects of smaller disturbances taken
together, due to less severe transients, might in the long run lead to significant shaft
damage.

1.5.2 The Im pact of Pole Slipping on the Rest of the Power System
The ’system centre’ is an expression often used to describe the point in the power system
where the contribution from the generator equals that from the rest of the system. At this
point the voltage fluctuations produced by pole slipping will be at their most severe, the
voltage may even fall to zero. For the normal generator/infinite bus situation the system
centre appears inside the generator, or its step up transformer. However, if the utility
infeed is low, then the system centre may appear in the load network. The location of the
system centre is not fixed, and it will move around as the effective impedance of the
generator and nearby machines changes. The generator terminal voltage will normally
drop during pole slipping as the load angle approaches 180°. Most generators located on
industrial sites will have loads connected electrically close to them, and these loads will
therefore be susceptible to the voltage disturbances. Since the minimum voltage will
occur when the load angle is 180°, it is desirable to disconnect the generator before this
load angle is reached.

Drives in industrial plants are generally supplied via starter contactors, which have an
instantaneous release time[41] of typically 40 ms. The voltage fluctuations produced can
cause these contactors to open, causing unnecessary loss of plant. If system voltages are
depressed for a sufficiently long time induction motors may stall and this will further
depress system voltage. Synchronous motors may also pole slip in response to a voltage
depression, making them subject to any of the effects described in section 1.5.

The location of the system centre determines how much disturbance a pole slipping
generator will cause to the rest of the system. If the system centre occurs inside the
generator, which occurs when the generator is connected to a 'stiff bus, then the resultant
voltage fluctuations will not be that severe. Unfortunately, the action of AVR's pushes
the system centre away from the generator, therefore moving the voltage depressions away
from the generator terminals into the distribution system. AVR action will tend to raise
machine flux when pole slipping occurs, thus increasing the voltage contribution that the

M.J.Checksfield/PhD/University of Bath 1996 Chapter 1, Page 11

generator makes to the rest of the system1491.

Power system protection may also suffer from the effects of pole slipping, because the
voltage and current variations at the system centre appear the same as a three phase short
circuit[36,54,55,56]. At the system centre, the voltage will be a minimum, whilst the current
will be at its maximum value. The relays most affected are under-voltage, over-current
and impedance type relays.

1.5.3 Legal Obligations

Embedded generator operators have to ensure that their generators do not contravene the
recommendations set by the Electricity Council. Engineering Recommendation G59[14]
provides recommendations on the requirements that should be met for the connection of
private generating plant to an electricity board's distribution system. This states that with
reference to the 1937 Electricity Supply Regulations, 'the private generator must not
operate his plant in such a way that the electricity board is unable to fulfil its obligations
under these regulations'. This makes it unlawful for a private generator to allow their
generator to continuously pole slip or operate asynchronously. In addition to this
statement, section 6.4 states, 'the board requires protective equipment to be provided by
the generator in order to disconnect the generator from the system when a system
abnormality occurs that results in an unacceptable deviation of the voltage or frequency
at the point of supply'. This also demonstrates that if some form of protection is not fitted
which will disconnect the generator when pole slipping or asynchronous running occurs,
then the private generator will be in violation of the electricity boards interconnection
requirements.

In addition to 'G59', Electricity Council Technical Report 113[15] provides notes of

guidance for the production of private generating plant of up to 5 MW for operation in
parallel with an electricity board's distribution network. This report shows the detrimental
effect on system voltage of a pole slipping generator which is not promptly disconnected.
In this instance the system voltage was reduced to an unacceptable mean value of 0.9 p.u.
The minimum protection requirements for generators in the range 250 kVA to 5 MVA are
also given. The minimum protection scheme comprises of Under/Over Voltage/Frequency,
Loss of Mains, Over-current, Earth Fault, Reverse Power, and Neutral Voltage
Displacement. The author of this technical report suggests that this type of scheme would
be effective at tripping the generator for all pole slipping and asynchronous running

M.J.Checksfield/PhD/University of Bath 1996 Chapter 1, Page 12

conditions. Later in this thesis, it will be shown that this is not true.

Technical Report 113 also includes a stability study which demonstrates the effects of
using different interconnection impedances, and governor action. The study showed that
for the 4.5 MVA generator studied, the Critical Clearance Time (CCT) ranged from 84 ms
to 222 ms. Since fault clearance times of up to one second are possible on distribution
circuits, any generators which are not disconnected as a result of the fault are likely to
pole slip.

1.6 MICROPROCESSOR RELAYING PLATFORMS

Nearly all new generator protection is provided by microprocessor based integrated

relaying schemes123,25,27’30,31,57,58,591. A microprocessor relay is comprised of a
microprocessor and memory arrangement, which uses digitally filtered sampled power
system waveform data to execute the protection algorithms stored in its memory.

Microprocessor relays offer many advantages over conventional electromechanical relays.

As well as containing all of the generator protection functions in one unit, the
microprocessor relaying platform allows programmable tripping logic, events recording,
signalling, self-monitoring, test facilities and interface with SCADA systems. A greater
reliability and availability can be achieved as all of the protection functions and
supplementary functions can be factory standardised and housed together in one pre-tested
single unit. Installation is also made easier as only one unit needs connecting to the
instrument transformers, circuit breaker tripping coils and SCADA system. This reduces
the amount of wiring required thereby reducing cost and improving reliability.

The computing power inherent to this type of relay creates numerous opportunities for
developing new protection techniques which were not possible with electromechanical or
static techniques. Pole slipping protection is especially suitable for development in a
digital relaying environment because the phenomenon is relatively slow allowing long
window algorithms to be used. Benefits can also be gained from the calculation of
machine operating conditions. In addition trajectory tracking in conjunction with a-priori
and initial disturbance information allows a faster more secure analysis[31]. The long time
frame also means that low burdens are placed on the microprocessor. An IEEE committee

M.J.Checksfield/PhD/University of Bath 1996 Chapter 1, Page 13

quoted on the subject of digital pole slipping protection "A design challenge is to make
such a sophisticated relay easy for utility engineers to apply and set”. Apart from the
improvement in tripping performance, creating an easy to set pole slipping relay would
greatly reduce the time and effort involved in applying pole slipping protection.

1.7 SUMMARY OF PROPOSED TECHNIQUE FOR DETECTING POLE

SLIPPING

The new pole slipping protection algorithm proposed in this thesis has been designed to
take advantage of the processing power of a modem microprocessor relay and uses power
measurements to ascertain whether the generator will remain stable or not. In designing
a pole slipping algorithm, the protection should trip when it observes conditions which
will produce a pole slip, or correspond to an actual pole slip, but remain stable during
fault conditions and other system abnormalities. Most pole slipping protection schemes
are designed to detect pole slips in the range of 0.1 to 10 % slip[60]. Like pole slipping,
asynchronous running is detrimental to the generator and the system, and in many cases
the distinction between asynchronous running and pole slipping can become vague due
to saliency and residual flux effects. An algorithm which detects both conditions would
therefore be desirable because the generator should be disconnected from the power
system in both cases.

By being able to recognise conditions where a pole slip is inevitable, the algorithm offers
the prospect of disconnecting the generator before a pole slip actually occurs, thereby
disconnecting the machine before it can be damaged, or cause voltage disruption to other
consumers.

The new power based algorithm works by detecting if the generator load angle is still
increasing once the CSP has been exceeded. It uses condition monitoring techniques to
ascertain the generator operating point, and its mode of operation, i.e steady state or
transient, it then adjusts its trip settings based on this information. Generator real and
reactive power are used to check if the generator is operating beyond the CSP. Generator
rate of change of real power is then used to check whether the generator load angle is still
increasing in an unstable manner. If it is, the algorithm issues a trip command for the
generator to be disconnected as it is operating in an unstable manner and a pole slip is

M.J.Checksfi el d/PhD/University of Bath 1996 Chapter 1, Page 14

therefore inevitable. By changing its trip settings according to the generator operating
point, the algorithm is able to use sensitive trip settings to quickly detect slow steady state
pole slips, whilst having more secure settings to avoid nuisance tripping during stable
power swings.

It is important to highlight the distinction between a pole slipping generator and a system
out-of-step or power swing condition161,621. With generator pole slipping, the system centre
appears in, or close by the generator or its step up transformer. This implies that the
generator is 'out of step1with the rest of the system. When a system is out of step, and
a power swing results, then the electrical centre will appear in between groups of
generation, i.e it will appear in a transmission or distribution line. Detection of system
out of step conditions is primarily a task for the transmission line protection relays136,54,55].
This research concentrates on generator pole slipping, it is therefore primarily concerned
with generators operating against infinite buses, or at least buses whose infeed is
significantly greater than the rating of the generator.

The algorithm proposed in this scheme has been designed to detect 'generator' pole slips,
i.e pole slips where the generator rotor speeds up above synchronous speed. It will not
detect 'motor' pole slip i.e pole slips where the generator rotor slows down below
synchronous speed. Since this research work is aimed at generators operating against
buses with a rating larger than their own, only 'generator' pole slips will occur. It is only
when considering system out of step protection, or synchronous motor protection, that
'motor' pole slips also need to be detected. Later in the thesis, it will be shown that the
new algorithm can easily be modified to detect motor pole slipping.

1.8 RESEARCH OBJECTIVES AND PREVIOUS WORK

The main objective of the research was to gain a thorough understanding of the subject
of pole slipping and asynchronous running, in addition to enhancing and testing the pole
slipping algorithm which was first proposed in previous work[63,64]. The previous work had
formed a basic algorithm which had been shown to work using a dynamic computer based
power system simulation package[65], and a laboratory model power system[66].

The computer based power system simulation package did not simulate sub-cycle

M.J.Checksfield/PhD/University of Bath 1996 Chapter 1, Page 15

phenomena and used a reduced order model for its synchronous machines. More detailed
simulations therefore needed to be performed in order to reliably validate the performance
of the algorithm. The laboratory power system model provided useful real life data for
testing the algorithm, however, the generation set up used did not replicate a real life
embedded generation environment exactly1671. More appropriate data from a real life
embedded generation unit was therefore required to test the algorithm more thoroughly.
The results of field tests are reported in this thesis.

The initial algorithm assumed that the generators would be working at fullload, and
therefore one of the main objectives was to develop the algorithm so that it would work
over the whole of the generator's operating range. It was also desirable to design an
algorithm which required minimal setting, since existing pole slipping protection can be
difficult to set reliably.

The final aim of this research project was for a field tested algorithm to be included in
a microprocessor based protection relay.

1.9 PRESENTATION FORMAT OF THE THESIS

The thesis covers the work done in designing a new power based pole slipping algorithm.
Although initially designed for embedded generators, the new technique is equally
applicable to large 'grid' type generators. The causes and consequences of pole slipping,
the presently available protection schemes, and the plant modifications required to reduce
the likelihood of pole slipping are presented. A study into the operating characteristics
of synchronous machines was used as a basis on which to derive the new power based
algorithm. The new algorithm was then comprehensively tested using a range of test
platforms from laboratories through computer simulations to full field trials. The
algorithm's performance was then analyzed and compared with conventional techniques
and conclusions drawn. Six appendices are also provided so that the main body of the
thesis is concise. These contain; theoretical derivations of power system and algorithm
characteristics, the techniques used to set conventional pole slipping relays, data on the
power systems simulated, general embedded generator data, the practical implementation
of the algorithm in a commercial microprocessor relay, the publications produced as a
result of this work, and examples of ATP and 'C' files used in the testing and analysis of

M.J.Checksfield/PhD/University of Bath 1996 Chapter 1, Page 16

the different pole slipping protection schemes.

Chapter 1 provides an introduction to embedded generation. Where commonly available

material is required, numerous references have been given to provide the reader with a
wealth of sources which support the concepts given. The problems associated with pole
slipping and its causes are also discussed. The opportunities which enabled this project
to come to fruition that are provided by modem micro-processor relays are presented and
a summary of the proposed pole slipping protection technique given. The research
objectives are then described.

Chapter 2 gives a summary of the many various approaches to pole slipping protection
giving the benefits and disadvantages of the different schemes with respect to embedded
generation. The standard protection schemes used with embedded generators such as
under voltage and over current protection are presented and their effectiveness at detecting
pole slipping discussed. A study of the changes in plant which can be made to reduce the
likelihood of pole slipping is also included.

Chapter 3 presents the 'ideal* characteristics of synchronous generators, then discusses the
aspects of practical operation which cause distortion. A study into embedded generator
stability has been performed in order to indicate the likelihood of pole slipping. This
study highlights the factors which make embedded generators more likely to suffer from
pole slipping than conventional large 'grid' type machines. .

Chapter 4 presents the new power based pole slipping protection technique. The basis of
the algorithm, the necessary filtering techniques, fault blocking, adaptive trip setting
techniques, limits of operation, and other protection functions provided by the algorithm
are all given.

Chapter 5 contains details on the different test platforms used in validating the operation
of the algorithm. A laboratory model power system, real time dynamic simulator
(POWSIM), Alternative Transients Program (ATP), 625 kVA industrial diesel generator,
26.5 MVA hydro-generator, and 353 MVA turbo-generator have all been used as test beds.
Most of the data used in constructing the test beds has been placed in Appendix C and
Appendix E

M.J.Checksfield/PhD/University of Bath 1996 Chapter 1, Page 17

Chapter 6 discusses the performance of the new algorithm for each of the tests. For many
of the tests, the results and graphs produced are identical. Plots that are virtually identical
have not been repeated, only one example of each plot is given for the test bed being
discussed.

Chapter 7 summarises the results and conclusions of the tests presented in chapter 6. The
reader may wish to read chapter 7 first as it provides a complete and condensed version
of the most extensive chapter of the thesis.

Chapter 8 provides the conclusions, and suggestions for further work. Chapter 9 lists the
references.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 1, Page 18

Stator W inding 60s
Stator

(Or
Rotor
OOOO i v
DCs

Eq

d - axis
Field Winding

Figure 1.1a Figure 1.1b

Diagram Showing Stator and Rotor Field 'Poles' for an Phasor Diagram for an
Underexcited Synchronous Generator at No Load. Underexcited Synchronous
Generator at No Load.

Machine load an g le, I

(Os

Figure 1.1c Figure l . l d

Diagram Showing Stator and Rotor Field 'Poles' for an Phasor Diagram for an
Underexcited Synchronous Generator Operating at a Underexcited Synchronous
M achine Load Angle 8 o f 45 degrees. Generator with 5 = 45
degrees.

NOTE: The stator winding shown on this diagram is a hypothetical coil which
would produce the same resultant m m f distribution as a 3 phase winding.
(Os - angular speed o f rotating flux produced by a three phase winding on the stator.
(Or - angular speed o f generator rotor
Te - electromagnetic air gap torque

M. J. Checksfi el d/PhD/University o f Bath Chapter 1, Fig 1.1, Page 19

 Double Circuit
Interconnecting Utility
Embedded
Line 'Infinite'
Generator
Bus.
-x - -X -

(V-GD Generator
l^ a n a ro f/

Transformer
Fault clearing breakers
<S>
y Fault

Figure 1.2a
Embedded Generator - Utility Connection.

Pre-Fault
Power
Power
i Curve

Critical
Stability
Point
(CSP)
Prime Mover
Input Power
Level, Pm
Post Fault
Power
Curve

Fault-
Duration
Power
Curve

51 52
Machine
Load
Angle

Figure 1.2b
Equal Area Diagram for Clearing of Short Circuit Fault Shown in Figure 1.1a.

M. J. Checksfield/PhD/University of Bath Chapter 1, Fig 1.2, Page 20

 Power
A Machine Load Angle, 5
decelerating energy

999999

—►
90 1 Machine Time
Load Angle
accelerating energy

Figure 1.3a
Equal Area Representation of Loss of Transient Stability Caused by a Fault.

Power
A Machine Load Angle, 5

accelerating energy

Pm

90° for round rotor machine

180 270 360 Machine Time
/ Load Angle

Figure 1.3b
Equal Area Representation of Loss of Steady State Stability Caused by Insufficient Excitation.

Power
A Machine Load Angle, 5

accelerating energy
Pm

Machine
0 90 180 Load Angle Time
Figure 1.3 c
Equal Area Representation of Asynchronous Operation Caused by Loss of Excitation.

M. J. Checksfield/PhD/University of Bath Chapter 1, Fig 1.3, Page 21

 Chapter 2
CONVENTIONAL TECHNIQUES FOR DETECTING AND
METHODS FOR PREVENTING POLE SLIPPING

This chapter discusses the conventional techniques used for detecting pole slipping, as
well as the changes in plant which can reduce the likelihood of pole slipping. The basis
for most of the relays commonly used to detect pole slipping is the measurement of the
apparent impedance at the generator's terminals. Most of the commonly used designs were
conceived around 40 years ago and were based on the work of Edith Clark [185].

Some terms commonly used in this subject area will now be defined. The term 'power
swing' refers to the swings in power output which occur when a generator, or group of
generators undergoes load angle swings. The generators do not need to become unstable
for a power swing to occur. Any oscillation in load angle will cause a power swing. A
'stable' or 'recoverable' power swing therefore occurs when a generator does not lose
stability. A pole slip is therefore a form of power swing, but the term refers to power
swings where a generator has lost stability.

Some pole slipping schemes have two distinct tripping zones so that they can differentiate
between 'stage 1' and 'stage 2' pole slips. A 'stage 1' pole slip is defined as a pole slip
where the system centre occurs inside the generator or its transformer. A 'stage 2' pole
slip is defined as a pole slip where the system centre occurs beyond the HV terminals of
the generator's step up transformer. For most embedded generators, 'stage 1' pole slips
will occur because the system centre appears inside the generator.

2.1 IMPEDANCE TYPE POLE SLIPPING RELAYS

Impedance relays respond to the variations in apparent impedance as seen at the generator
or high voltage transformer terminals during the pole slip and make their trip decision
based on this measurement . Figure 2.1b shows the theoretical variations in impedance
during pole slipping for the system shown in Figure 2.1a. These impedance loci are
derived using the assumptions that the ratio of generator to source electromotive forces,

M.J.Checksfield/PhD/University of Bath 1996 Chapter 2, Page 22

Eg/Es, remains constant. It is also assumed that generator saliency is neglected, generator
damper effects are neglected, transient fault impedance effects have decayed, shunt loads
and shunt capacitance effects are ignored, effects of automatic voltage regulators and
governors are neglected, and the source voltages behind their equivalent impedances are
sinusoidal and at fundamental frequency136,68,69,701. The derivation of these curves is given
in Appendix A.

For the case where the ratio Eg/Es = 1, the impedance locus will be a straight line, PQ,
which is the perpendicular bisector of the system impedance line GS. When a generator
pole slip occurs, the load angle, 5, increases causing the impedance to travel from right
to left across the R-X plane. When the load angle reaches 180°, the loss of synchronism
point has been reached and the apparent impedance will lie on the system impedance line
GS. The point on the system impedance line where the impedance locus appears for 8 =
180° is known as the system centre, at this point the impedance gives the appearance of
a three phase fault. As the pole slip cycle is completed and 5 approaches an in-phase
value of 360° the impedance locus will travel to the left of the system impedance line
before returning to the load area of the R-X plane.

Figure 2.1 also shows impedance loci for the cases where the ratio Eg/Es < 1 and
Eg/Es > 1, which cause the impedance locus to take on a circular arc form. These curves
show that as the generator internal voltage is increased, the system centre is pushed away
from the generator, out into the system. AVR field forcing action will have this effect.

For the majority of situations, the combined impedance of the generator and its
transformer will be larger than the combination of line and equivalent source
impedances162,68,69]. This will result in the system centre appearing inside the generator or
its transformer during pole slipping. If the system centre appears out on a transmission
or distribution line, then power swing detection/blocking relays should be employed to
separate the two systems, since this is not a generator pole slipping problem.

The assumptions made in deriving these theoretical curves do not always apply for
embedded machines. Nearby synchronous machines and induction machines can cause the
impedance trajectory to follow a less than smooth path across the R-X plane which can
make secure installation of relays difficult. Generally the ratio Eg/Es will be in the
range[71] 0.8 to 0.9, causing the impedance locus to travel through the generator or the

M.J.Checksfield/PhD/University of Bath 1996 Chapter 2, Page 23

generator transformer if one is present, i.e the locus would cross the section labelled Zg
in Figure 2.1b. Small embedded machines generally complete their pole slip cycles
relatively quickly due to their low inertias, the high values of slip this produces can cause
difficulty to relays since it can interfere with the fault blocking characteristics required by
some schemes. Careful analysis of the embedded system must be made[42] before deciding
on the appropriate settings for impedance based pole slipping relays, since standard
application procedures developed for large machines are generally inadequate.

It is generally recommended142,62,68,691 that transient stability studies are performed so that

the impedance loci are known and the most appropriate relaying scheme can be selected.
These simulations are necessary to enable correct setting of conventional impedance
relays. It is important to note that there are no industry standards or commonly used
practises for generator pole slipping protection1701.

2.1.1 Loss of Field Scheme

Figure 2.2 shows the loss of field characteristic which sometimes doubles as pole slipping
protection168,691. This type of relay will provide a degree of pole slipping protection, since
it will provide a trip command if the pole slipping impedance locus enters one of the
circular characteristics during the slip cycle. Both of the loss of field characteristics are
time delayed140,42,69,72,731, typically values of 0.5 to 1 second are used.

There are a number of possible flaws in this approach. A pole slipping locus that passes
through the generator transformer will go undetected because the loss of field relay
characteristic only reaches as far as the generator terminals. Due to the time delay
inherent in loss of field relays, tripping will only occur for pole slips whose impedance
locus stays within the locus for a sufficient time. If a time delay of 0.5 seconds is used,
then a generator which is pole slipping a rate of 2 Hz or above will not be tripped. Small
embedded generators are likely to pole slip at high rates1421. If the loss of field relay has
an integrating timer arrangement1741, enabling a delay in relay drop off, then higher rates
of pole slipping can be detected. Several pole slip cycles will be required to make the
relay operate however.

The recommended settings for loss of field usually have an offset equal to Xd'/2 for both
distance relay zones144,46,721, whilst the time delayed zone used for detecting loss of
excitation from low generator loadings has a diameter equal to Xd. The faster operating

M.J.Checksfield/PhD/University of Bath 1996 Chapter 2, Page 24

zone used for detecting loss of field from high initial loadings has a recommended
diameter of 1 pu on the generator base. A delay of 4 to 5 cycles is also used[46,75] in order
to eliminate nuisance trips due to relay vibration.

Small generators will normally have a single Mho unit with an offset of Xd'/2 and
diameter Xd. To ensure nuisance trips do not occur for stable power swings, a time delay
of 0.5 to 0.6 seconds is normally used [46].

2.1.2 Mho Element Scheme

This is the simplest pole slipping protection scheme, its characteristic is shown in Figure
2.3. If the impedance locus enters the circle, a trip signal is issued immediately168,691. This
can be a problem if the relay is too sensitive because a trip signal can be given for a
recoverable power swing. Figure 2.3 shows a recoverable power swing impedance locus,
for such a swing the impedance locus will travel towards the equivalent system impedance
line, then head back towards the load area of the R-X plane as the machine load angle
decreases after reaching its maximum value.

The stable power swing locus[60,62] shown in Figure 2.3 would not cause the Mho scheme
shown to nuisance trip as it does not enter the Mho circle. However, if the relay were
more sensitive, resulting in a larger Mho characteristic, then the power swing locus may
have entered the circle causing a nuisance trip. This type of relay is usually set to a
sensitivity of 120° load angle[68], accepting that this can cause nuisance tripping since
stable swings of up to 155° are possible[38].

One advantage of the scheme is that tripping can occur before the load angle reaches
180°. This is the point where maximum current occurs and therefore the point of
maximum mechanical stress on the generator windings. Another advantage is the scheme's
ability to provide backup protection for multiphase faults occurring in the generator and
a portion of the unit transformer. It also provides inadvertent energisation protection when
properly set.

The main disadvantage of the scheme is that a large characteristic circle is liable to cause
nuisance tripping during stable swings. If a smaller circle is used, it may fail to detect
some pole slips. If it does detect a pole slip, the trip command will be issued to the
generator circuit breaker as the load angle approaches 180°. This subjects the generator

M.J.Checksfi el d/PhD/University of Bath 1996 Chapter 2, Page 25

circuit breaker to a maximum recovery voltage during arc interruption. If a longer time
delay is used to prevent false tripping during stable swings, then the scheme will fail to
detect the fast pole slips. If the scheme is not supervised by an over-current fault detector,
a trip may occur during a voltage transformer (VT) failure[68].

2.1.3 Single Blinder Scheme

It is important to note that there are several possible variations of this scheme[70]. The
basic scheme consists of two impedance elements (referred to as 'blinders' or ohm units)
and a supervisory Mho relay, one example of a commonly used scheme is shown in
Figure 2.4. The relay is supervised by a Mho element, to restrict the operation of the
scheme to swings appearing in the generator or transformer, and to prevent operation for
recoverable power swings that pass through both blinder elements. This also stops
tripping for oscillations in reactive flow after synchronising168,76,771.

The angle of the blinder units can be adjusted so that they are parallel to the equivalent
system impedance. In this scheme[68,69,76,771 for a trip condition to occur, the locus must
originate outside of the mho relay, then enter the Mho characteristic, and cross from one
blinder to the other over a period of around two power system cycles. It is important to
note that the pole slipping locus must enter from one side and leave from the other for a
trip to occur. The scheme is inherently more stable than the mho scheme. Careful setting
of the supervisory mho characteristic is still necessary, because if the mho circle is too
large, nuisance tripping can occur for recoverable swings which cross both of the blinder
elements160,681 or for system out of step conditions where the electrical centre lies out on
a transmission line. An example of a recoverable swing locus is shown in Figure 2.4.

The scheme will issue a trip command either when the impedance locus crosses the left
hand blinder function, or upon reset of the supervisory Mho element. This provides a
means of controlling the load angle at which tripping occurs, the arc interruption duty on
the circuit breaker can therefore be minimised. This scheme will detect both motor and
generator pole slips.

A motor pole slip occurs when a synchronous motor's mechanical load exceeds the
electrical power it is able to draw from the supply. This causes the machines rotor to
decelerate and pole slip relative to the stator flux. The impedance locus will travel across
the RX plane from right to left during a motor pole slip.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 2, Page 26

The single blinder scheme is the most secure of the conventional impedance schemes, the
above scheme does have some limitations however. A complete pole slip cycle is required
for the scheme to operate, the scheme will therefore not trip until the pole slip has
occurred. Although, the scheme is inherently more stable against recoverable power
swings and faults, detailed computer based simulations are still required in order to find
the correct blinder, mho, and timer settings which will guarantee reliable
operation142,60,68,691.

Sometimes an additional offset mho unit is applied to the scheme which reaches out across
the system impedance characteristic160,78,791. With an additional directional relay fitted at
the HV terminals of the generator step up transformer, stage 2 pole slips can also be
detected. The additional mho and directional elements provide a means of differentiating
between ’stage 1' and 'stage 2' pole slips.

A much more secure scheme is suggested by Goody[79] for a pumped storage

motor/generator. Studies had showed that the scheme suggested above could nuisance trip
for unbalanced faults and developing faults as well as circuit breaker pole scatter. It was
also shown that CT saturation could have a detrimental effect to the scheme. The
modified scheme developed to overcome these problems uses the following modifications
to the above scheme. For a trip command to be issued, the supervisory mho element only
needs to be in an operated state when the last ohm element operates. In addition the
current must also be greater than 10 % of rated during the operation of the scheme. These
two measures ensure that the first pole slip cycle is always reliably detected. To stop
incorrect operation during developing faults, more criteria were added. Whilst in the
generating mode, only loci moving from right to left would be detected as a pole slip.
When in motoring mode only the loci moving from left to right would be detected as pole
slipping loci. This logic was implemented using auxiliary switches on phase reversal
disconnecters. The final criteria was that the pick up operation of each ohm element must
be delayed such that its minimum operating time is 20 ms. Note that with this scheme the
ohm elements were set to face the opposite direction to that shown in Figure 2.4. This
new scheme solved many of the above problems, there were still some potential problem
areas however. Developing faults with a development period of around 60 ms may cause
a stage 2 nuisance trip. This could be overcome if stage 2 tripping only occurred after the
second pole slip. CT saturation during certain faults may cause nuisance tripping. This
could be overcome by ensuring that the protection CTs do not saturate during fault

M.J.Checksfi el d/PhD /University of Bath 1996 Chapter 2, Page 27

conditions.

The above scheme therefore provides a trip signal after the first pole slip, and is also
capable of detecting out of step conditions on the system referred to as 'stage 2' pole slips.
This type of detection is necessary in systems with weak transmission, where the system
centre may appear out on transmission lines1801. It is beneficial in this situation, to delay
tripping giving the line relays a chance to separate the system at a more beneficial
location during out of step conditions. It still requires a detailed system analysis to be set
properly however, and does not provide a trip signal until after the pole slip has occurred.
In many situations tripping can only be initiated when the system load angle is much less
than, or much greater than 180°, since the circuit breaking duty on the breaker is too much
at load angles around 180°. The trip command therefore needs to be issued significantly
before 180°, or significantly after. Alternatively, a generator circuit breaker must be
selected which is capable of breaking out of phase fault current levels.

2.1.4 Double Blinder Scheme

A supervisory Mho element is used in this scheme to provide the same security as in the
single blinder scheme. The difference between this approach and the single blinder
scheme is that the blinder logic of the relay functions in a different manner. The
impedance characteristics of the relay are configured as shown in Figure 2.5. A pole slip
is recognised if the impedance remains between the outer and inner blinder elements for
longer than a pre set time.

After the first time constraint is satisfied, a portion of the logic circuitry is 'sealed in' if
the impedance locus enters the area between the inner elements. When the locus leaves
the inner element zone, its traverse time must exceed a preset time before it reaches the
outer characteristic. A trip condition exists when the impedance locus leaves the
supervisory Mho element.

The important distinction between this scheme and the single blinder scheme is that once
the impedance locus has entered the zone made by the inner blinder characteristics and
the Mho element, the impedance locus can leave the inner and outer characteristics in
either direction for successful operation. The recoverable power swing locus shown on
Figure 2.5 would cause the scheme to nuisance trip, the inner blinder elements need
moving closer in to prevent this nuisance trip.

M.J.Checksfi el d/PhD/University of Bath 1996 Chapter 2, Page 28

The inner element setting must therefore be such that it will only pick up for non-
recoverable swings. For this reason the single blinder scheme is inherently more secure
than this scheme. The double blinder scheme is therefore more sensitive to its settings
than the single blinder scheme, computer based simulations are required to provide the
correct blinder, Mho and timer settings.

This scheme has the added disadvantage that if the impedance locus crosses the inner and
outer blinders without first picking up the Mho element, then the scheme will not trip for
that pole slip cycle. This locus is shown in Figure 2.5 and could happen when pole
slipping follows a multiphase power system short circuit fault. During the fault the
impedance locus enters the Mho circle between the inner elements and rests on the system
impedance line. For the time that the fault is applied, the relay will only see the fault’s
effect even if the generator is accelerating towards a pole slip. When the fault is cleared
the generator load angle may have reached a high value, which causes the impedance to
remain between the inner characteristics. The generator will then complete the rest of the
pole slip cycle as normal, however the scheme will not trip, since the first stages of the
scheme logic have not been 'sealed in’. A trip signal will only occur when the complete
second pole slip cycle has occurred, by which time system disruption and machine damage
may have occurred.

2.1.5 Lenticular Type Scheme

Several schemes exist which use lenticular type characteristics rather than blinder and Mho
elements160’68,69,811. The single lens scheme operating characteristic is shown in Figure 2.6.
This scheme has similar operating principles to the single blinder scheme. The scheme
shown in Figure 2.6 can detect both 'stage 1' and 'stage 2' pole slips1601. This scheme was
originally designed for generators where the system centre would often occur out on the
transmission system.

A trip signal is issued when the impedance locus enters from one side, remains inside the
lens for a specified time, and leaves via the opposite side. Typical time settings require
the impedance locus to stay in each side of the lens for at least 25 ms [79,82]. The suggested
generator reach, denoted as Zg in Figure 2.6 is 0.5 Xd [60]. This was chosen to co-ordinate
with the loss of excitation relay, so that pole slips with a ratio of Eg/Es greater than 0.66
p.u. would be detected by the lens scheme. The reach of the HV directional relay is
chosen to detect swings which appear on the generator side of the step up transformer.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 2, Page 29

The other important setting is the lens angle, 5. This is chosen according to the maximum
percentage slip rate, s which the scheme must detect, and is related to the timer setting,
t of the scheme by the following equation t60];

6 - 180 ( 1 - — ) 2(1)
2
The reactance element and HV directional characteristic shown in Figure 2.6 are provided
as a means of discriminating between stage 1 and stage 2 pole slips. This scheme has
similar attributes to the single blinder scheme.

2.1.6 Double Lens,Triple Lens and Concentric Circle Schemes

The operating characteristic for the double lens scheme is shown in Figure 2.7. It
functions in exactly the same manner as the double blinder scheme, it therefore has the
same strengths and weaknesses.

Triple lens schemes have also been devised, which have four steps in detecting a pole
slipping condition. This makes the scheme inherently more secure than the double blinder
scheme. The logic circuitry required to achieve this higher level of reliability is much
more sophisticated than that used in the double lens scheme.

The concentric circle scheme uses two distance elements and operates in the same manner
as the double blinder or double lens schemes.

All of the above schemes suffer from similar problems to the double blinder scheme.
Careful setting of zones is required to ensure that a recoverable swing locus will not enter
the inner zone and cause a nuisance trip. The majority of the impedance based schemes
are able to provide satisfactory pole slipping protection if set correctly. Deriving correct
settings requires a detailed simulation of the generator and power system to be protected.
Collecting sufficient data and adequate models to perform such a simulation can be a time
consuming and expensive process. With the exception of the single Mho scheme, all of
the other schemes trip after the pole slip has occurred, this generally means that the worst
stress point for the generator, and greatest disturbance to the system has already occurred.
A pole slipping detection technique which can disconnect the generator before the first
pole slip has occurred offers the advantage of reducing the ill effects to an absolute
minimum.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 2, Page 30

2.2 POWER BASED POLE SLIPPING PROTECTION

Only one power based pole slipping detection method is to be found in the literature. This
uses the repeated reversal of real power as the generator alternates between a generating
and motoring condition to detect pole slipping136,81,831. The power reversal is indicated by
the alternate closing of the contacts of a duo-directional watt relay. The scheme is
secured against faults by the use of a counter, several pole slip cycles are therefore
required before a trip decision is given. Nuisance tripping due to hunting at light loads is
prevented by the use of an over-current relay in series with the power relay. This
approach's main disadvantage is that several pole slip cycles are required before the
generator is disconnected. It is also possible that full power reversal may not occur for
some pole slips.

2.3 MICROPROCESSOR BASED POLE SLIPPING PROTECTION

The recent pole slipping techniques proposed have all been microprocessor based. All
existing microprocessor based pole slipping protection schemes which have been designed
with a low computational burden in mind are impedance based. The first designs were just
conversions of the conventional electromechanical relay characteristics into a digital
domain. The flexible nature of microprocessor systems has meant that characteristics
other than circles, lines and lens can be implemented easily, and the more recent relays
have used quadrilateral characteristics^4,85,861.

The non-impedance type microcomputer relays at present have a high computational

burden and require information not readily available to most embedded generation
protection engineers. The highly computational and data intensive nature of these schemes
makes them unsuitable for protection of embedded generators because they require a stand
alone relay which makes the cost of the scheme too high. Additionally, complete
generator, control equipment and system data is often difficult to find[87].

2.3.1 Rate of Change of Apparent Resistance Relay

One proposed scheme[88] uses the apparent resistance and its rate of change to predict an
out-of-step condition. This scheme is intended for out of step protection of transmission
lines, but could be used for a generator pole slipping scheme. The scheme initiates

M.J.Checksfi el d/PhD /University of Bath 1996 Chapter 2, Page 31

tripping 'on the way in' to a pole slip, and therefore the circuit breakers used for system
separation in this case need to operate before the angular difference between the systems
is greater than 120°. The rate of separation of the systems is also a factor when
considering circuit breaker duty because of the time lag between relay operation and
circuit breaker clearing.

Apparent resistance is used rather than apparent impedance as this makes the relay
insensitive to the location of the system centre. Use of rate of change of apparent
resistance provides an indiction of how quickly the systems are separating. The relay can
use sensitive settings so that tripping can occur close to the 120° criterion when the rate
of separation is slow, whilst tripping can occur earlier at angles around 100° when the
systems are separating quickly. The relay settings are therefore not restricted by the three
phase fault condition, which produces the highest rate of separation, unlike conventional
relay types. The relay can therefore be set closer to actual stability limits. The scheme
also has an adaptive feature, which enables the overall settings to be changed when an
intertie disturbance occurs. The flag that an intertie disturbance is occurring is provided
by other substation control equipment. This also enhances the relays sensitivity since
worst case considerations do not dictate the relay settings.

The scheme is prevented from operating for power system faults by recording the time it
takes the apparent resistance locus to cross two pre-specified resistance values. If the
locus crosses these values very quickly, scheme tripping is blocked as this indicates a fault
situation. The reactive reach of the scheme is limited by allowing tripping to occur only
if the apparent impedance is less than a set value, in a similar manner to how the Mho
circle limits operation in the blinder schemes.

The main limitation of this scheme is the problem of finding correct settings. This
scheme is devised for a specific 500 kV intertie where the system parameters are well
known. Applying this type of scheme to an embedded generator would involve the same
problems encountered with the blinder pole slipping schemes. Finding the correct
apparent resistance settings for fault blocking requires detailed simulations.

2.3.2 Multiple Zone Impedance Based Microcomputer Based Protection

One scheme developed by Shiwen[89] uses the single blinder principle to produce a relay
that is stable against faults and power swings and operates towards the end of the first

M.J.Checksfield/PhD/University of Bath 1996 Chapter 2, Page 32

pole slip cycle. The relay breaks the R-X plane into five zones, as shown in Figure 2.8,
with each zone's boundary being parallel to the equivalent system impedance line.

A trip signal is given when the impedance locus has passed through each zone in turn, and
remains in each zone for a specified time. This time must be greater than the time the
impedance would stay in a zone for a short circuit condition, but less than the minimum
time it would stay in any zone for a pole slipping condition. The angle at which tripping
occurs can be controlled by selection of the zone reach so that more favourable arc
interruption is obtained. Tripping therefore occurs towards the end of the first pole slip
cycle. In addition, the scheme provides control signals for pole slipping prevention.
These signals are derived when the impedance locus satisfies the trip requirements of the
first two zones. The signals are intended for use with a large generator with a fast valving
or breaking resistor scheme.

The relay offers the advantage over the conventional schemes that it provides preventative
control signals before load angles of 180°, and is inherently more stable due to the
increased number of zones. Problems still exist with choosing the correct timing settings
for the scheme, the scheme may also fail to operate for pole slips where the real power
does not sufficiently reverse, since the impedance locus may fail to enter zone 4.

2.3.3 Square Impedance Based Relay Characteristics

Comparisons between square impedance based characteristics and circular or lens type
schemes suggest that using a square impedance characteristic is more reliable than the
conventional lens, circles or blinder schemes[85,90,91]. The main advantage is that the scheme
is less sensitive to the direction in which the impedance locus travels as illustrated in
Figure 2.9.

An 'ideal' pole slipping locus travels horizontally across the impedance plane, and most
impedance based schemes are set to detect such an impedance locus. If the locus does not
take the 'ideal' trajectory, the single blinder type scheme will fail to detect the pole slip
because the locus would not cross both of the blinder units. The quadrilateral scheme
uses the single blinder principle, in conjunction with the flexibility of having a
quadrilateral detection characteristic to enhance pole slipping detection. The locus must
enter on one side and leave on the other, taking longer than the preset time 'Tl'. The
scheme has the flexibility to detect impedance loci such as those shown in Figure 2.9.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 2, Page 33

Shunt capacitance1361, induction motor loads and other power system equipment^21 can all
change the nature of the impedance locus as seen by the impedance relay. Shunt
capacitors inject current into the network, they therefore cause the out of step behaviour
of a network to deviate from that predicted by the classical two machine model[92].

2.3.4 Multiple Algorithm Pole Slipping Protection

This approach1931 uses three different methods to detect static, dynamic and transient
instability. The static stability monitoring scheme is not impedance based, it monitors the
generator real power and load angle variations, the generator is considered unstable i f :

^ < 0
dS 2(2)
P > K p ,PM„7„
ZD

are both satisfied, where dP/d5 is the synchronising power coefficient, Kp is a coefficient
dependent on the number of generators and lines working, and PMZD is the threshold of
static stability when all generators and lines are operating, and is entered as a relay setting
particular to the system considered. The positive phase sequence values of V and I are
found which are then used to calculate active power. 5 is calculated from the voltages
known at either end of the line and the line impedances. The connecting impedances of
the line are determined within the relay according to switch status. This method therefore
requires data that is not normally available to an embedded generator protection relay, and
would therefore be difficult and expensive to implement.

For the transient stability monitoring, the Equal Area Criterion136,371 is used in a
quantitative manner. If the calculated accelerating area is greater than the decelerating
area then a trip is issued. To calculate these areas, A5 must be calculated and then
applied to the power load angle characteristics. To accurately know all of the power load
angle characteristics requires detailed information on the system configuration as well as
generator and plant data. Such detailed data is often not available, this method is
therefore very susceptible to nuisance tripping due to inaccuracies in the data causing
incorrect calculation of the accelerating and decelerating areas. This method also relies
on the assumption that the mechanical power during steady state is equal to the electrical
output power, and that the governor action is ineffective during the transient period.

The dynamic instability prediction acts as a back up to the previous two schemes, and it

M.J.Checksfield/PhD/University of Bath 1996 Chapter 2, Page 34

detects if an out of step condition is occurring, and whether it is a generator or motor pole
slip. It uses resistance and rate of change of resistance to predict pole slipping. For a
generator pole slip, a trip signal is initiated if the measured resistance lies between two
preset values related to the load angle at the CSP, and the rate of change of resistance is
less than zero. The resistance measurand is a combination of actual apparent resistance
and a predictive term based on the rate of change of resistance. This assumes that the rate
of change of resistance and hence load angle is constant, it therefore trips before the CSP
has been reached. In doing this, the scheme may nuisance trip for recoverable swings
following faults whose duration is just less than the CCT. With such a swing, the rate of
change of load angle decreases as the CSP is reached and it is equal to zero at the CSP.

Both the transient and steady state methods may be effective at providing an early warning
system so that fast valving or breaking resistors can be used. The scheme requires detailed
system data and is therefore impractical for embedded generators.

2.3.5 Artificial Neural Network Pole Slipping Protection

Several pole slipping and loss of excitation schemes have been devised which use an
Artificial Neural Network (ANN) to detect pole slipping or system out of step conditions.
One such approach1941 uses the prime mover input power, kinetic energy deviation and the
average acceleration during the fault as inputs to an ANN. After the network was trained
on a system using 162 samples, it successfully performed for 99 % of the tests conducted.
The problem with this approach is that the inputs used are not readily available, or easy
to derive in a protection relaying environment. The network also needs to be trained 'off
line1using simulations of the power system to be protected. This is too time intensive for
application to embedded generators. These restrictions make this approach impractical for
pole slipping protection of embedded generators.

Another proposed scheme[95] uses the results of a discrete fourier transform on machine
load angle and speed deviation, generator current, voltage, power and admittance and their
cross spectra as inputs to an ANN. It relies on using 'sophisticated system models and
accurate parameters to enhance the speed and accuracy of the relaying scheme'. It also
uses online pattern recognition techniques to enhance its performance. The scheme
successfully differentiated between loss of excitation, and transient loss of synchronism
80 % of the time. This approach has the same problems as the scheme above however,
and is not suitable for protection of embedded generation.

M.J.Checksfi el d/PhD /University of Bath 1996 Chapter 2, Page 35

2.3.6 Pattern Recognition Pole Slipping Scheme
Research has been conducted into using pattern recognition to detect generator out of step
conditions1961. This is the same as using an ANN. The proposed scheme requires generator
output power, and mechanical speed as inputs, it also requires the machine inertia. Using
this input data, it identifies an out of step condition by using these variables to derive the
mechanical input power at the instant of fault initiation, the generator kinetic energy at the
instant of fault clearing, and the average acceleration during the fault. These variables are
then entered into a pattern recognition algorithm which classifies the situation as stable
or unstable. There are several problems in applying this approach to embedded
generation. The inclusion of a speed measuring device would make the scheme too
expensive, in addition, an accurate value of the machine inertia is not always known. This
approach also assumes there is no governor action, no damper windings on the generator,
and that all loads are constant impedance loads. The scheme is only aimed at detecting
a loss of transient stability, it would therefore be unable to detect pole slips resulting from
dynamic or steady state loss of synchronism.

2.3.7 Out of Step Voltage Phase Comparison Schemes

This approach is applied for system out of step problems, where it is vital that the power
system be separated into two individual power islands if an out of step condition is likely
to, or is actually occurring. The advantage of this approach is that unlike impedance
based schemes, the location of the system centre does not effect its operation. The basic
principle is that the phase difference between the voltage phasors at either end of the line
connecting the power system is calculated. If this phase difference is about to, or actually
has exceeded a preset value, then the interconnection is disconnected either at a pre­
determined location, or at the system centre. The aim of separating the systems is to
create two individual 'in step' power systems where load and generation are balanced.

One approach1971 uses sampled voltage information from four locations on the power
system to calculate the phase difference. It then predicts what the phase difference will
be in 200 ms time, by approximating the voltage phase difference oscillation to a sinusoid.
If the predicted value is greater than a preset value, then a trip signal is given to break the
system up at a specific location. This system is very expensive, the cost being justified
in the instance of its application because if the system were to fall out of step, the whole
power system would fail.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 2, Page 36

Another scheme1841 uses central and remote apparatus linked by high speed data
transmission systems to make an out of step detection based on voltage phase comparison.
The internal generator EMF's are calculated from the generator terminal voltage, current
and impedance. The adopted value of impedance used is the average of the synchronous
and transient reactance. The phase difference is calculated and a trip signal issued if the
difference exceeded 180°. This approach may be prone to errors, since the generator
impedance varies between the sub-transient value, Xd" and the synchronous value, Xd.
The method was claimed to have worked when used for phase comparison of several
groups of generation, however, it may become ill-conditioned if applied to an embedded
generator to calculate its load angle.

2.3.8 Microcomputer Based Generator Modelling Protection Techniques

Lyapunov methods have also been used to assess overall system stability for use in power
system contingency planning. Lyapunov methods use the phase plane domain to
determine the critical stability point of the system and assess generator stability. This
approach has been applied to individual synchronous generators1981. The algorithm uses
external circuit information, machine data, accelerating power, load angle and slip are
required as inputs to a complex, computationally intensive algorithm. This type of
algorithm is difficult to instigate on a small integrated protection relay. Two schemes are
proposed, the simple one requires all of the system impedances, and machine and system
inertias and calculates the load angle, slip, acceleration, voltage magnitude, mechanical
input power, and peak transient generator output power. From this information, it
calculates the stability limits for the machine, and trips if the generator operates outside
its stability boundary. The second, more detailed scheme requires additional data on the
generator AVR, governor, saturation, damping and saliency. The scheme also relies on
accurately determining when the generator is disturbed by a fault, and it will therefore not
trip for steady state pole slips. Loss of dynamic stability may also go undetected. The
amount of data required by the scheme is impractical for embedded generation systems.

Another interesting scheme uses time tag, voltage and current phasors and circuit breaker
status information from the rest of the system[86,1901. Sample synchronisation is obtained
with the aid of the Global Positioning System (GPS). Using this information, the system
load angle is derived, and used as a continual update for a simulation of the protected
system performed by the relay. The relay models the protected power system as a two
machine network. If the simulation predicts instability then the relay will initiate a trip

M.J.Checksfield/PhD/University of Bath 1996 Chapter 2, Page 37

signal. This relay is intended for installation on a 500 kV interconnection where the
consequences of out of step operation are dire. The hardware intensive nature of this
scheme is therefore justified. Again, such a scheme can not be justified for smaller
machines.

Other instability predicting algorithms[99,100,1011 developed are based on similar principles.

These can be set to operate before the load angle reaches 180°. However, they are unable
to discriminate between stable and unstable swings under all system conditions.

Algorithms based on an exact calculation of the equal area criterion for the generator
concerned have been developed. One scheme uses a communications link to calculate the
phase angle between points of generation. The areas of the equal area criterion are then
calculated based on estimated machine quantities, and a decision on stability made tl02].

Any scheme which relies on accurate plant data for prediction of instability is unsuitable
for protection of the majority of generators, as the accuracy of the plant data is always
questionable, and changes to the power system will affect their performance. These
changes therefore need taking into account which requires communications capabilities.
The only situation where this approach is justified is out of step protection of important
inter-ties, where accurate data is known, and it is imperative to disconnect the inter-tie
quickly whenever instability is imminent.

2.4 EMBEDDED GENERATOR STANDARD PROTECTION SCHEMES

In the UK, Engineering Recommendation ,G59'[14], and Engineering Technical

Report 113[15] specify the minimum protection requirements for embedded generators. The
USA has similar documents including ANSI 1001[16,21]. Figure 2.10 shows a
comprehensive protection scheme that may be used on embedded generators greater than
250 kVA. On simpler schemes functions 46N, 51V, 40, 87G, 32R, and 67N may be
omitted. Usta[103] and Barrett[104] both provide a comprehensive analysis of these schemes.
This thesis will discuss only the relays' ability to detect pole slipping conditions.

Pole slipping is a balanced phenomenon, and therefore functions which are related to
phase imbalance conditions do not require consideration when discussing pole slipping

M.J.Checksfield/PhD/University of Bath 1996 Chapter 2, Page 38

detection. Pole slipping will therefore have no effect on negative phase sequence, neutral
voltage displacement, current differential, stator earth fault, and directional earth fault
protections.

The basic protection scheme, which does contain the loss of excitation function, is
insufficient for detecting all pole slipping conditions. The main problem is caused by the
oscillatory effect that pole slipping has on most power system variables. The standard
scheme will provide only a limited amount of pole slipping protection cover.

2.4.1 Effect of Pole Slipping on Over & Under Frequency Relays

This form of protection operates when the measured frequency exceeds a preset limit in
a preset time. Engineering recommendation G59 states that the over-frequency limit is
+1 %, and the under-frequency is -4%, both with a time delay of 0.4 seconds. The time
delay requirement comes from the need to detect an abnormal operating condition
following a loss of grid before any auto reclose schemes may reconnect the grid to the
embedded generator. If the generator is not disconnected before the auto-reclose, an out
of phase re-closure may occur which could cause further supply disruption and generator
winding and shaft damage[52,105]. Most re-close schemes have a dead time of 0.5 seconds
or greater before their first re-closure, a time delay of 0.4 seconds was therefore chosen
in order to allow 100 ms for circuit breaker operation.

When a generator pole slips, its real and reactive power output oscillates wildly. This
makes the voltage phase angle at the generator terminals oscillate wildly, which causes
the measured electrical frequency to oscillate also. Pole slipping will not normally cause
under or over-frequency functions to trip, due to the frequency oscillating above and
below the trip settings in a time much below the time delay setting.

2.4.2 Effect of Pole Slipping on Under & Over Voltage Protection

This protection operates when the measured voltage exceeds preset limits for a specified
time. Engineering recommendation G59 states that the over-voltage limit is +10 %, and
the under-voltage limit is -10%, both with a time delay of 0.4 seconds. The time delay
is chosen to ensure co-ordination with auto-reclose schemes (see 2.4.1). If the embedded
generator is connected to a system with no auto-reclose schemes, then a longer time delay
of up to six seconds may be used, which is useful for preventing nuisance tripping of the
embedded generator when system faults are cleared by time delayed protections1151.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 2, Page 39

Depending upon the grid infeed at the terminals of the generator, the under-voltage
protection may operate for a pole slipping condition, since pole slipping can depress
system voltage far below the trip setting of 0.9 pu. A problem can occur if an integrating
timer arrangement is not used on the under-voltage relay because the generator terminal
voltage oscillates from a low value to a normal voltage level during pole slipping. The
period where the voltage is normal may reset the relay, preventing it from tripping during
pole slipping. Under-voltage protection provides a degree of pole slipping protection, but
it will not successfully detect every pole slipping condition.

The 'G59' recommendation for under-voltage relays may cause nuisance tripping following
system faults due to a slow voltage recovery. During the fault, the system voltage is
depressed and the generator load angle increases. This voltage depression can cause the
under-voltage element to pick up. When the fault is cleared, in a time of 350 ms for
example, the reactive power demand of the generator has increased substantially, because
the generator load angle has increased to a high value. This reactive power demand
causes a further depression of system voltage, which may be sufficient to cause the under­
voltage protection to trip, even though the generator may remain stable, and no pole
slipping occurs. There is clearly a co-ordination problem between under-voltage relays
and other system relays. Co-ordination problems also exist with loss of field relays, due
to the 0.5 to 0.6 seconds time delay normally employed[40].

2.4.3 Effect of Pole Slipping on Rate of Change of Frequency Relays

Rate of change of frequency and phase displacement relays[15] will both experience
oscillations in their relaying quantity during pole slipping (section 2.4.1). Engineering
technical report 113[15] quotes a manufacturers recommended setting of 12° over 30 cycles.
This equates to an average rate of change of frequency of 0.055 Hz/s over 0.6 seconds.
This setting is very sensitive and may cause nuisance tripping during disturbances on the
utility grid system. Other sources quote a setting of 0.3 Hz/s with a time delay of 0.3 to
0.5 seconds t,03].

The oscillations in frequency and hence rate of change of frequency may prevent the rate
of change of frequency based relay from tripping during pole slipping, due to the relay
continually resetting before the time delay requirements have been satisfied. This is
dependent upon the time delay used, and the speed of pole slipping. A very sensitive
setting would improve the relays chances of detecting pole slipping, but also increase the

M.J.Checksfield/PhD/University of Bath 1996 Chapter 2, Page 40

chances of nuisance tripping during recoverable power swings.

2.4.4 Effect of Pole Slipping on Voltage Dependent O ver-current Relays

Voltage dependent over-current protection is necessary for synchronous generators because
of the decaying fault current response to a short circuit fault[45]. Two types of voltage
dependent over-current protection are commonly used. Voltage controlled over-current
relays are only permitted to operate as a fault over-current relay if the measured voltage
falls below a set minimum value[24,45], otherwise they function as an over-load relay. The
voltage value used varies between 0.8 and 0.3 per unit depending upon the type of
earthing that the generator has. One type of voltage controlled IDMT over-current relay
multiplies its plug scale multiplier setting by 0.4 if the voltage falls below 60 % [106]. A
typical setting value is a PSM setting of 2*In, with a TMS setting of 0.1 using a standard
inverse IDMT characteristic and a control voltage of 80 % [23A2\ If co-ordination with
short circuit clearance times on adjacent system feeders is required, then higher settings
may be required, resulting in longer fault clearance times[40]. Intertie over-current
protection may be set to a more sensitive level than the generator over-current scheme,
so that when a fault occurs, the generator and site load remain connected together. This
prevents any discontinuity of site load supply [40].

A voltage restrained over-current relay changes its IDMT characteristic according to the
voltage at the machine terminals. It may therefore be regarded as an impedance type relay
with a long dependent time delay1451. The relay therefore operates more or less
independently of the current decrement which occurs in the generator.

Pole slipping may cause an IDMT over-current relay to operate. The magnitude of current,
speed of pole slipping and the relay settings determine the likelihood of such an event.
The current may exceed three phase levels during pole slipping, but only during the period
when the load angle approaches 180°. If the relay does not drop off between pole slip
current maximums, then a trip condition may result after several pole slips due to the
integrating action of the over-current relay[74]. If instantaneous over-current relays are
used, then pole slipping may cause them to operate[36].

Over-current relays have more of a role in preventing pole slipping since they can
disconnect the generator from the utility when a fault occurs whose duration is long
enough to cause pole slipping. If the embedded generator protection scheme philosophy

M.J.Checksfield/PhD/University of Bath 1996 Chapter 2, Page 41

is to disconnect the generator for all external faults, then in theory, pole slipping should
not result from faults. If the over-current relay settings have to be graded with those on
adjacent feeders however[40], then the fault clearance times are likely to be in excess of the
critical clearance time[42] and pole slipping may occur following faults..

2.5 TECHNIQUES TO REDUCE THE LIKELIHOOD OF POLE SLIPPING

While it is possible to fit relays to disconnect generators w'hen they pole slip, there are
numerous other ways of improving power system transient stability which will reduce the
risk of pole slipping. The following sections detail the various modifications that can be
made to the power system to improve transient stability, as well as machine design
changes which alleviate the machine from the phenomenon of pole slipping.

2.5.1 Reducing Fault Clearance Times

Reducing the duration of a fault reduces the accelerating energy which goes into
increasing the machine rotor angle and this therefore decreases the load angle swing for
the fault and increases the likelihood of maintaining stability. This is one of the most
effective methods of increasing transient stability, but reducing the fault clearance time
requires costly changes to the existing utility's protection, in addition to protection changes
within the neighbouring plants[42]. The cost of such changes generally makes this method
impracticable. In order to decrease fault clearance times, the non unit based IDMT
over-current protection predominantly used by utilities has to be changed to unit based
current differential protection1451. Differential protection requires relays at either end of
the feeder, and a means of passing data from one end to the other. This makes the cost
significantly higher than using IDMT over-current type protection, since this only requires
one relay and no communications facility. Improvements could also be made using
distance protection, this has not been favoured by utilities to date for distribution system
protection, due to problems in setting it satisfactorily for short feeders.

Improvements in fault clearance times can also be made by upgrading circuit breakers.
An oil circuit breaker may have a total operating time of 100 ms or more[451, whereas an
SF6 or vacuum circuit breaker may have an operating time of 50 ms[107).

M.J.Checksfield/PhD/University of Bath 1996 Chapter 2, Page 42

2.5.2 Resonant Links
Figure 2.11 shows one form of resonant link, known as the short-circuit limiting
coupler151]. The capacitor and linear series reactor are chosen to be in resonance with each
other at power system frequency. They therefore present almost zero impedance to load
currents.

Under normal operating conditions, the circuit shunted across the capacitor has a high
impedance, making the link appear to have nearly zero impedance. When a fault occurs,
the voltage across the capacitor rises, causing the main saturable reactor to saturate in less
than one system cycle. This results in a large current flowing through the damping
resistor which serves to reduce the fault current asymmetry. The voltage produced by this
current across R, causes the auxiliary reactor to saturate also, diverting the current away
from the resistor and through the auxiliary reactor. The overall combination of inductance
of the saturable reactors and the capacitor adds to the impedance of the linear series
reactor to limit the fault current and improve system stability. The system takes one to
two power system cycles to reset which therefore improves the post fault impedance as
well. The link would normally be installed on the utility/industrial plant intertie, and is
therefore only of use for faults occurring out in the utility system.

2.5.3 Fast Acting Voltage Regulators

Voltage regulators can apply field forcing to the generator for the fault period which
increases the machine's power output when the fault is removed, thereby increasing the
amount of decelerating energy available and enhancing stability for a given fault
duration11081. During a fault, the effect of armature reaction will decrease flux linkage,
especially if the reactive power output of the generator is large[1091. A decrease in machine
flux linkage causes a decrease in output power and a corresponding decrease in machine
stability. A fault duration of a few hundred milliseconds will cause a significant decrease
in flux linkage[110].

Fast AVR's may serve to improve first swing transient stability, but in doing so reduce the
dynamic stability margins for subsequent swings. If the reduction in damping is severe
enough, then the generator may maintain stability for the first swing, and then lose
synchronism in subsequent swings. This dynamic instability requires incorporation of
additional stabilising feedback signals into the AVR to cure the problem11111.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 2, Page 43

A fast AVR will serve to force the levels of excitation during the fault, and therefore
maintain flux levels which would have otherwise decayed. Even though AVR's can
significantly improve the flux levels during faults, their effect on the magnitude of the first
swing in load angle is generally a reduction of just a few degrees. The situation where
they are most beneficial is with long fault clearance times.

Increasing excitation performance also results in more severe pulsations when the machine
actually pole slips, as well as increasing the voltage disturbances to other consumers1501.
The AVR tends to 'push' the system centre away from the generator and out into the
utility system, because it raises the machine internal voltage level[49]. One method of
excitation control proposed known as 'bang-bang' control[112], claims to reduce the impact
of pole slipping. With fast excitation systems, the field forcing action will create a make
or break type situation where the forcing action will either provide enough retardation
energy to maintain synchronism or if it fails in doing this, and the point of no return is
past, the excitation will accelerate the rotor into the pole slip, causing a more severe pole
slip. A bang-bang excitation control system would remove excitation as the pole slip is
entered, and reinstate it after the pole slip at a desirable load angle, thus removing the
strong synchronising forces produced by the interaction of stator and rotor magnetic fields.

2.5.4 Turbine Fast Valving

This method increases the speed of response of steam turbine prime mover systems by
significantly reducing the mechanical input power to the generator during a fault1113]. It
therefore decreases the accelerating power into the rotor and correspondingly increases the
decelerating power. During a fast valving operation, the steam valves are completely shut
in around 80 to 200 milliseconds, thereby decreasing the output power of the steam
turbines1108’114’115’1161. This approach requires a method of detecting rotor acceleration. If a
control signal related to acceleration can be derived, then the method does provide a
significant increase in transient stability for steam turbine generators.

2.5.5 Fast Governing Systems

A fast response governing system can help to increase the critical clearing time, but will
never be able to fully prevent transient instability and hence pole slipping. One stability
study showed[15] that inclusion of the governor into the model improved the critical
clearing time for a gas turbine generator by 40 ms. The time constants of prime mover and
governor systems can vary enormously depending upon the type and sophistication of the

M.J.Checksfield/PhD/University of Bath 1996 Chapter 2, Page 44

system being considered. Although governors can not completely prevent pole slipping,
they are the single most important factor when considering the re-synchronisation of a unit
once pole slipping has occurred. A fast governing system will cause a pole slipping
generator to re-synchronise after one pole slip cycle if it is not first disconnected.

2.5.6 Braking Resistors

Braking resistors act as a substitute to the load lost due to a fault on the power system
therefore 'braking' the generator over the fault period, and decreasing the degree of load
angle advance. The resistors are switched off when there is a rotor deceleration1114,116,1171.
One problem with braking resistors is determining the instant of switching in and out of
the resistors so that transients are controlled most effectively. A sophisticated control
scheme is required to enable optimum switching.

When considering embedded generation, the cost of installing a braking resistor and the
necessary control schemes would be prohibitive. Braking resistors are normally used with
large grid generators in order to meet design stability requirements11181

2.5.7 Coordinated Control Schemes

Fast AVR's, fast valving, and braking resistors can all be employed at the same time. If
they are all controlled by one integrated device, then the transient and steady state stability
limits can both be significantly improved1113,1161.

2.5.8 Changes in Machine Design

Changes can be made to the parameters of the machine at the design stage which will
improve the machine's transient stability. Several novel machine designs have been
proposed which produce synchronous machine like output characteristics, but alleviate the
transient stability problem.

2.5.8.1 Machine Parameter Changes

The main changes that can be made are to the machine inertia constant and the transient
reactance. An increase in the machine inertia constant results in a 'stiffer' machine since
more energy is stored in the rotating mass. More acceleration energy will therefore be
required to obtain a given deviation in load angle. A higher inertia constant requires a
heavier and therefore more expensive machine rotor.

M.J.Checksfiel d/PhD/University of Bath 1996 Chapter 2, Page 45

A lower transient reactance will increase the power output of the machine during and after
transient disturbances such as faults, providing more deceleration energy. The transient
saliency produced by the difference in transient reactances in the direct and quadrature
axes causes a further increase in machine output power, as well as shifting the load angle
at which the peak occurs to a value greater than 90° [1191. A low value of transient
reactance will increase rotor length and therefore the size and cost of the generator11081.

Increased damping will also improve stability, since the resultant asynchronous torque
produced by the damper windings and rotor iron will provide a higher real power output
for increasing load angles[120]. High damper performance may cause other problems
however, since the machines fault current contribution may increase to unacceptable
levels, or poor voltage regulation may result due to the damper circuit limiting the rate of
change of flux[108].

Embedded machines can sometimes have high values of stator resistance, due to their
small size. This causes a decrease in inherent stability11211, low per unit stator resistance
is therefore also a virtue. Section 3.5.3 discusses this in more detail.

2.5.8.2 Novel Machine Designs

One development11121 allows the machine load angle to be kept constant irrespective of
load power factor, allowing the machine to absorb reactive power up to the thermal limit
of the stator without compromising transient stability. It achieves this by using a divided-
winding rotor, one half of which is responsible for providing the torque necessary to hold
the rotor in synchronism. The other half is responsible for the reactive power control of
the machine, applying positive or negative excitation depending on the power factor of the
load supplied. The machine load angle can be kept at a constant value of 40°. Transient
stability is improved because when load angle increase occurs during a fault, the position
of the resultant rotor mmf can be held constant by altering the reactive excitation level.
The down side of the scheme is the additional cost of the excitation system, since thermal
differential expansion and contraction problems result due to the separate rotor windings
carrying different levels of excitation current.

Another machine development removes the problems of synchronous stability and hence
pole slipping by relieving the rotor magnetic field from being fixed to the rotor11221. A
polyphase excitation winding is used on the rotor, which is excited with polyphase slip

M.J.Checksfi el d/PhD /University of Bath 1996 Chapter 2, Page 46

frequency voltages of proper sequence and phase. In doing this the desirable features of
synchronous and asynchronous machines can be realised. The machine’s output power
characteristics are similar to those of a conventional generator, yet the asynchronised
synchronous machine is capable of operating with slip. Machine stability is related to the
performance of the excitation system, and the benefits of the new scheme are accounted
for in the increased cost of the excitation system required.

2.5.9 Induction Machines

An obvious way of eliminating pole slipping is to use induction machines instead of
synchronous machines. An induction machine rotor does not rotate in synchronism with
the stator electromagnetic field and therefore can not lose synchronism. Although the
initial cost of an induction machine may be less than that of a synchronous machine,
induction generators require additional design considerations that synchronous generators
do not[7]. The savings in equipment purchases may easily be swamped by more expensive
operating costs if the situation is not fully analyzed. Induction machines only operate
efficiently at full load. At part load exciting VAr's are required from the system,
worsening the operator's power factor and increasing the tariff set by the utility. Induction
machines can not be used to operate an independent power island, whereas synchronous
machines can. Synchronous generators can also be used to offset a site's reactive and real
power demand.

2.6 CHAPTER SUMMARY

All of the suggestions given in this chapter for improving transient stability are expensive
and are more applicable to large 'grid' generators rather than small and medium sized
embedded machines. All such methods improve transient stability, and therefore reduce
the risk of pole slipping, but they do not eliminate it completely however.

The alternative is to install relays that will disconnect the generator from the system when
a threatening situation occurs, preventing the generator from the damage caused by pole
slipping and the possible disturbance to the power system. The established techniques
generally offer a level of pole slipping protection, the single blinder scheme being the
most reliable. This scheme requires a complete pole slip cycle to operate however, it is
not able to disconnect a generator before a pole slip occurs. Another problem with the

M.J.Checksfield/PhD/University of Bath 1996 Chapter 2, Page 47

impedance based schemes is that they are difficult to set so that they will operate correctly
for all conditions. A detailed computer based simulation study is necessary in order to
set the relays with a high degree of confidence.

The majority of new pole slipping protection schemes proposed rely on data being
available on the generator and the power system to which it is connected. Data
acquisition can be difficult, time consuming and therefore expensive. Even if all of the
data is available, techniques which use a quantitative approach to predicting generator
instability may fail due to inaccuracies in the data, or in the measurements of the
generator's state of operation.

The design challenge is to devise a pole slipping relay which can detect all asynchronous
conditions, restrain for all other operating conditions, is easy to set, and inexpensive to
realise.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 2, Page 48

 utility
rrmnaratnr generator transformer line source . „ /
yciiciaiui . - t . i nlfinit
finite
impedance impedance im pedance impedance
bus

E g /.8 E s /,0

Figure 2.1a
Two Machine Representation o f a Power System.

KEY

Zsys, System impedance (Zl+Zs)

Zt, Transformer impedance
Zg, Generator impedance (Xd')
Zl, Line impedance
Zs, Source impedance

Eg/Es > 1

999999999999999999999999999941

Eg/Es = 1

Eg/Es < 1

pole slipping loci

Figure 2.1b
Classical Pole Slipping Impedance Loci.

M. J. Checksfield/PhD/University o f Bath Chapter 2, Fig 2.1, Page 49

 supervisory directional
Zsys control
relay chanfcteristic

Pole
Slipping
Locus

Loss of Field
Characteristic

Loss of Field
Characteristic

Figure 2.2
Loss o f Field Relaying Scheme also used for Pole Slipping Protection.

KEY
Zsys, System impedance
Zt, Transformer Impedance
Zg, Generator impedance (Xd')

Mho relay trip Zsys

characteristic

Zt
Pole Slipping
Locus

Recoverable
Power Swing
Locus

Figure 2.3
Mho Element Pole Slipping Protection Scheme

M. J. Checksfield/PhD/University of Bath Chapter 2, Fig 2.2/2.3, Page 50

 For operation
T2 - T1 > 2 cycles
Left Zsys I Right
Blinder ^ I Blinder

pole
slipping
T2 T1
**locus
Supervisory
Mho Element
stable power
Left Right swing locus
element element
pickup pickup

Figure 2.4
Single Blinder Pole Slipping Detection Scheme

KEY
Zsys, System impedance
Zt, Transformer impedance
Zg, Generator impedance (Xd') pole slipping locus following
/ a multiphase fault

left inner blinder right inner blinder

left outer blinder right outer blinder

Zsys

pole slipping locus

recoverable
power swing
Supervisory Mho locus
Element

Figure 2.5
Double Blinder Pole Slipping Protection Scheme.

M. J. Checksfield/PhD/University of Bath Chapter 2, Fig 2.4/2.5, Page 51

 Impedance element
(Ohm line or blinder)

Zsys
reactance element
HV directional relay
^ operate
operate

pole
slipping
Lens ^
locus
Characteristic
lens angle, 8

Figure 2.6
Lenticulare Type of Pole Slipping Detection Scheme

KEY
Zsys, System impedance
Zt, Transformer impedance
Zg, Generator impedance (0.5Xd)

Zsys

Z t,

pole
slipping
locus
Lens , "
Characteristics

Figure 2.7
Double Lens Type of Pole Slipping Detection Scheme

M. J. Checksfiel d/PhD/University of Bath Chapter 1, Fig 1.4, Page 52

 jxA

Zsys
Reactance relay
characteristic

pole
Zg< slipping
locus

zone 4 zone 3 zone 2 zone 1 zone 0

Figure 2.8
Multiple section Microprocessor Based Pole Slipping Protection.

___________KEY____________
Zsys, System impedance
Zt, Transformer impedance
Zg, Generator impedance (Xd1)

unconventional
pole slip locus
Zsys

unconventional 'ideal' pole slip

pole slip locus locus

Figure 2.9
Quadrilateral Impedance Relay Characteristic.

M. J. Checksfield/PhD/University of Bath Chapter 2, Fig 2.8/2.9, Page 53

 810 Over Frequency Relay.
Site <5> 81U Under Frequency Relay.
Load 59 Over Voltage Relay.
27 Under Voltage Relay.
25 Check Sync Relay.
46N* N.P.S. Overcurrent.
51V* Voltage Dependent
Overcurrent.
46N A 5IV K 40 K 32RnLOG 59N Neutral Displacement.
87G
40* Loss of Excitation.
87G* Differential Relay.
32R* Reverse Power.
67N 5 IN
5 IN Stator Earth Fault.
67N* Directional Earth Fault.
59N
LOG Loss of Grid Protection.
52 Circuit Breaker.
* these functions may be ommitted

Figure 2.10
Comprehensive Embedded Generator Protection Scheme.

Capacitor

/YYYYYY
Linear Series
Reactor Damping
Resistor
R
Main
Saturable
Reactor - r /W r r L

Auxiliary
Saturable Reactor

Figure 2.11
Resonant Link Used for Improving Stability.

M. J. Checksfield/PhD/University of Bath Chapter 2, Fig 2.10/2.11, Page 54

 Chapter 3
SYNCHRONOUS GENERATOR CHARACTERISTICS

3.1 IDEAL SYNCHRONOUS POW ER CHARACTERISTICS

The characteristic response of synchronous machines is dominated by the interaction
between the field winding flux and the stator flux. This results in a sinusoidal power/load
angle characteristic. For a machine connected to an infinite bus, supplying a balanced set
of voltages, the steady state real and reactive power load angle characteristics are given
by;

\
1 'X d -Xq 3(1)
Pss = V 2( sin(25) - R +EVZ sin (8 +0)
2 +XdXq . { 2 J

1 Xd -Xq Xd+Xq 3(2)

Qss = EVZ cos (5 +0) +V cos(25) -
R 2 +XdXq

where Z = V (R2 + Xq2 ) and 0 = sin 1 (R/Z). Appendix A covers the derivation of the
effective generator voltage, E for various machine loadings.

Due to the small field resistance, the field flux linkage will generally remain constant
following a change in operating condition11231. When a sudden change in stator current
occurs, due to a short circuit for example, the resulting current transient can be broken
down into two components[36124,125 126]. The first component is referred to as the sub­
transient component, the second the transient component. The sub-transient component
occurs due to circuits other than the field winding on the machine rotor. In salient pole
machines, these circuits are provided by damper windings, whilst in round rotor machines,
the solid steel rotor provides the circuits for induced eddy currents to flow.

The circuits formed by the field winding and the damper windings form two coupled
circuits. These are at rest with respect to each other, but both rotate with respect to the
stator. These two circuits have two time constants, a short sub transient time constant, and
a longer transient time constant. Both of these time constants are effected by the
impedance of the stator. If the stator is short circuited, the time constants are given by
their short circuit values Td' and Td", whilst they take on their open circuit values, Tdo'

M.J.Checksfield/PhD/University of Bath 1996 Chapter 3, Page 55

and Tdo" if the stator is open circuited.

The magnitude of the stator current during the sub transient period is determined by the
direct axis sub transient reactance, Xd", whilst the transient magnitude is determined by
the direct axis transient reactance, Xd'. Typical values for Tdo" and Td" are 0.125 and
0.035 secondstll0]. Since pole slipping has a considerably longer time period than these
values, the sub transient period is of less interest than the transient period. The sub
transient period is important for the short circuit characteristics of the machine when
setting protection relays or calculating plant ratings. Typical values for Tdo' and Td' are
6 and 1.5 seconds11101, the transient period is therefore of major interest when considering
pole slipping characteristics. The transient real and reactive power machine angle
characteristics are given by;

1 {(Xd> - X q \ 3(3)
Ptr = V 2 sin(28) - / f j + £ 9 'FZ sin (5 +0)
2 +Xd'Xq {[ 2 J

rX d f +XqV 3(4)
Qtr = E q 'V Z cos (5 +0) +V —^?-jcos(25) -
R 2 +Xd'Xq

Appendix A contains details on how to calculate the voltage behind transient reactance,
Eq', for various machine loadings. Equations 3(1,2,3,4) are for a general synchronous
machine.

In general two types of synchronous machine need to be considered. The round rotor and
the salient pole type. Round rotor design is typically used for high speed, two pole
generators driven by steam turbines. Salient pole designs are used for lower speed multiple
pole pair generators such as hydro generators. The real and reactive power characteristics
for both types of generators will be discussed in the following sections.

3.1.1 Round Rotor Machine Real Power Characteristics

Figure 3.1a shows the real and reactive power rotor angle plots for a 200 MVA round
rotor machine operating at no load and full load in steady state and transient modes. The
steady state curves show the fundamental component, produced by the interaction between
the field winding and the stator mmfs. This is the sinusoidal characteristics on the no
load curves. The real power reaches its peak at a rotor angle of 90°, and stable steady
state generator operation can therefore only be maintained between 0 and 90°.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 3, Page 56

The transient curves have an additional 25 component, due to transient saliency. This
occurs due to the absence of a field winding on the quadrature axis. During transients,
the direct axis reactance changes from its synchronous value, Xd, to its transient value,
Xd', whilst the quadrature axis reactance stays at its synchronous value. The effect of this
25 component is to shift the peak of the real power curve past 90°. The peak of the real
power curve increases during transient conditions, and this helps the machine to maintain
stability following power system disturbances such as short circuits faults. The additional
power output capability during transient conditions enables the generator to produce more
decelerating energy, which helps retard the rotor towards synchronous speed.

When a pole slip occurs, the machine real power output follows the characteristic given
in Figure 3.1a. This shows that at the 120° rotor angle point, the machine is generating
five times its rated power, whilst at 240°, it is importing five times its rated power. Note
that the difference between the no load and full load curves is small. The main difference
is that during full load, there is more mechanical input power into the generator, which
promotes faster, more violent pole slips. A pole slip is unlikely to occur from no load
because there is virtually no mechanical input power to accelerate the rotor into a pole
slip.

3.1.2 Round Rotor Machine Reactive Power Characteristics

Figure 3.1b shows the reactive power/rotor angle curves for steady state and transient, no
load and full load conditions. This is also sinusoidal with the difference between this and
the real power plot being a phase shift of 90°. The transient curves have been raised from
the steady state curves for load angles between 40 to 90° and 270 to 320°, whilst they are
much lower than the steady state curves between 90 and 270°. At a value of 90°, all of
the curves converge. Equations 3(2) and 3(4) show that at 90°, the equation simplifies

t0; - 1
Q = ----------------- feFZsin(0) + V 2Xd) 3(5)
R 2 +XdXq

This is correct for steady state conditions, during transient conditions, Xd must be replaced
by Xd', and E by Eq'.

3.1.3 Salient Pole Generator Real Power Characteristics

Figure 3.2a shows the real power rotor angle characteristics for the 625 kVA salient pole
diesel generator described in appendix A, section A3.1.3. The most obvious difference

M.J.Checksfield/PhD/University of Bath 1996 Chapter 3, Page 57

between the round rotor and salient pole generator curves is the rotor angle at which the
peak in steady state real output power occurs. The 25 term in equation 3(1) causes the
peak to occur at load angles below 90°. The transient peaks also occur at lower load
angles for salient pole machines, but will still occur at load angles greater than 90°.

3.1.4 Salient Pole Generator Reactive Power Characteristics

Figure 3.2b shows reactive power rotor angle characteristics for the 625 kVA salient pole
generator. The main difference between these curves and the round rotor generator curves
are that the steady state curves flatten off as they approach rotor angles of 180°. The
common point where all of the curves converge is the same for salient pole and round
rotor generators, and is given by 3(5).

3.2 GENERATOR ASYNCHRONOUS POWER CHARACTERISTICS

When the generator rotor travels at speeds other than synchronous speeds, real and
reactive power is also produced due to induction machine action. The output power
characteristics are therefore not only dependent upon the rotor angle, but also on the time
rate of change of rotor angle. The asynchronous component of real power, PM, depends
upon the asynchronous constants of the generator, the applied voltage, rotor angle, and its
time rate of change, referred to as the slip, s. This is defined as;

</5 3(6)
s =-
dt
Slip is negative for speeds above synchronous speed. The asynchronous power, PMwhich
is produced during steady asynchronous operation, is given by;

P =

v
_ v 2 X d-X d' sT d' (
(1 + f I + (sTd' ) 2 ). sin [25 -ta n -f 1 1-1st
2 XdXd' 1 + (sTd') 2 , /; 3(7)
_ v 2 X d '-X d " sTd"
(1 - f 1+ (sTd n) 2 ). sin 25 -ta n - (_ !_ W
2 X d 'X d " 1 + (sTd ") 2 U Td") .

_ v 2 Xq-X q" sTq "

(1 - I + (sTq u) 2 ). sin [25 - tan-1 (—?— -2 s/|
2 XqXq " n\ 2
1 + (sTq ") I I ° \sT q ") JJ

Where 50 is the initial rotor angle, and the values of the tan'1 terms lie between 0 and
180° [35]. The first term is produced as a result of the direct axis field winding, whilst the

M.J.Checksfield/PhD/University of Bath 1996 Chapter 3, Page 58

second and third terms are produced as a result of the damper windings. If the generator
is operating on a non infinite bus, than the reactances of the tie line can be added to the
leakage reactance of the generator when calculating the generator reactances and time
constants. The asynchronous reactive power, QMis given by;

Q
^ as =

V 2 X d-X d' /\2

(sTd') sTd'
cos | 2st - tan 1 [—5—
~ XdXd' [l +( ^ r f\7(sfdY { { sT d ' //

3(8)
V 2 X d-X d " (sTd" )2 sTd"
.cos 12si - t a n '1
2 XdXd" l +( s T d " ) 2 •/! + (sTd ")2 i sTd

_ V 2 X q-X q" (sTq " ) 2 sTq"

-.cos 2st - t a n '1
~2~ X q X q " (sT q " J

Both the real and reactive power asynchronous equations consists of a steady component
plus a pulsating component. It is important to note the units used in these equations.
Units of slip are in absolute values so that a slip of 1 % corresponds to 0.01. The
reactances are all in per unit, whilst the time constants and time are in radians, where
t[rad] = W0 t[sec]. C0o is the electrical angular frequency of the generator [,27].

The expressions for synchronous and asynchronous power can be added to give the
complete output power of the generator, noting that effects such as saturation have not
been taken into account. The equations are useful for gaining an understanding of
synchronous machine characteristics. If more accurate characteristics are required a
computer based simulation should be used because the mathematics can become too
complicated for easy use.

3.3 PRACTICAL SYNCHRONOUS GENERATOR CHARACTERISTICS

The characteristics discussed in sections 3.1 and 3.2 are ideal. For the operation of
practical machines, other phenomena need to be considered. The discussions in this
chapter relate mainly to generator power characteristics because the algorithms developed
use power as their basis. References [110,123,128,129,130] all contain good discussions
of generator operating current and voltage characteristics. There are three additional
characteristics which are caused by non constant flux linkage, and trapped flux in the

M.J.Checksfield/PhD/University of Bath 1996 Chapter 3, Page 59

stator. Stator resistance also effects the characteristics, although to a lesser extent.

3.3.1 Non Constant Flux Linkage -

The Direct Axis Open Circuit Transient Time Constant, Tdo’

In section 3.1, it was assumed that the field flux linkage remained constant following a
transient^23,126’1311. For a sudden application of a step change in field voltage, the time
constant of the field during stator open circuit conditions is given by Tdo'. Since the open
circuit stator terminal voltage is proportional to the field current in an unsaturated
machine, the stator voltage will also change with this time constant. This time constant
is given by [11C1;

Tdo' = M 3(9)
Rf
Where Lff is the field inductance, and R f is the field resistance. Most discussions
involving synchronous generators are related to large generators which have ratings of
hundreds of megawatts. These will have a very high field inductance to resistance ratio
and so have a long field time constant, the assumption of constant field flux linkage is
therefore valid for at least the first second of a transient. Small embedded generators have
much higher per unit resistances, and consequently have much shorter field time constants,
the assumption of constant flux linkage is therefore not valid. Tables Cl and C2 of
Appendix C contain values of Tdo* for generators ranging from 5 kVA up to 384 MVA.
For typical embedded generators, rated below 3 MVA, Tdo* ranges between 1 and 2
seconds, whilst for the larger grid type machines with ratings above 10 MVA, Tdo' ranges
between 4 and 7 seconds. Small and medium sized embedded generators generally have
shorter time constants than grid connected generators.

Tdo1is applicable to a generator with an open circuited stator, a similar measure known
as Tdf, the short circuit time constant, is applicable when the stator is short circuited. Td*
is related to Tdo' by the following equation;

Td' = — Tdo' 3(10)

Xd
Tables Cl and C2 show that for small embedded generators, Td' is typically between 100
and 200 ms, whilst for the larger 'grid' type generators, Td1 is typically above 1 second.
When the generator is connected to balanced external impedances, the field time constant
will lie between Tdo' and Td'. Td' is useful when considering generator stability following

M.J.Checksfield/PhD/University of Bath 1996 Chapter 3, Page 60

a short circuit fault, since it indicates how long the field flux will remain constant, and
therefore how long the generator reactance will remain at Xd* and the internal emf at Eq'.
When the flux linkage starts to decay, the generator operating characteristic will move
from that defined by equation 3(3) to that of 3(1). Having short values of Td' therefore
has implications for generator stability, it also affects the performance of conventional
impedance based protection schemes. Figure 3.3a shows the real power against time
output for three simulations of a diesel driven generator following a close up short circuit
fault11321. In two of the simulations, the generator time constant, Tdo' is equal to 2 and 6
seconds and the generator AVR and governor have been disabled. For the other
simulation, Tdo' was set to 2 seconds with a slow acting AVR enabled. The generator
parameters are given in Appendix C. For all three simulations, the fault duration was such
that stability was just maintained, i.e the generator swung up to the critical stability point
(CSP)

The generator is initially operating at full power, which is 80 % of rating. Referring to

Figure 3.3a, when the fault begins a small oscillation occurs before the power falls to
almost zero. This occurs due to trapped flux in the stator and will be discussed in
section 3.3.3. The load angle then begins to increase until it reaches a value of around
135°, the fault is then cleared and the power jumps up to 3 pu. The peaks of the curves
A l, B l, and Cl are all at different heights. The highest is the Tdo' = 6.0 s curve, then the
Tdo' = 2 s, AVR included case, and then the Tdo' = 2 seconds with no AVR. The
Tdo' = 6 s curve jumps the highest because the field flux has decayed the least during the
fault. The simulation with the AVR included out performs the manual AVR simulation
due to the field forcing action of the AVR during the fault. This has the effect of
increasing machine flux. The peaks of the curves did not reach the level predicted in
Figure 3.2a due to external circuit reactance and field flux linkage decay. Kimbark[1101
gives details on how to calculate the exact change in field flux and states that for
generators with a value of Tdo' of around 5 seconds, the assumption of constant flux
linkage is a reasonably accurate one unless the clearing time for severe faults is greater
than 0.2 sec. A large amount of faults that an embedded generator experiences will be
longer than 200 ms, due to the IDMT type over-current protection used on distribution
feeders. This coupled with the fact that Tdo' for embedded generators is much less than
5 seconds means that a significant flux linkage decay will occur for faults. This has
implications for stability and the operation of conventional pole slipping schemes.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 3, Page 61

Referring back to Figure 3.3a, the curves then continue towards the CSP, which is roughly
at the input power level of 0.8 pu. The CSP's for each simulation are not exactly at 0.8
pu because the fault clearance time was adjusted in one millisecond steps to determine
the CSP. After the CSP, the curves head back towards their original steady state operating
points. Again the curves all attain different maximum swings in output power, A2, B2,
and C2 for the same reason as above. The A2 curve flattens off much earlier than the
others because of the decaying flux, this reduces the synchronising coefficient, making the
system less oscillatory. It therefore can not be assumed that the flux linkage is constant
for embedded generators.

3.3.2 Effects of High Stator Resistance on Generator Operation

Equations 3(1,2,3,4) show the effect of high stator resistance, R, on generator operation.
A high stator resistance has two effects on the real power characteristics. It shifts the
peak of the power rotor angle curve back by a value, 0, so that it is less than 90°. It also
shifts the whole curve downwards by a value R/(R2+XdXq). In general, unless the
generator is less than 50 kVA, the stator resistance will have little effect on the steady
state operating characteristics. Tables Cl and C2 of Appendix C give data for a range of
synchronous machines. The stator resistance is typically 0.05 pu in small machines, which
is negligible compared to a value of Xd between 1 and 2 pu. The greatest effect which
resistance has on generator operation is with regard to stability. This will be discussed
in section 3.5.8.

3.3.3 Armature Time Constant, Ta

When there is a sudden change in stator current following a fault or large load change,
there exists dc, fundamental, and second harmonic components in the phase current. The
dc and second harmonic components arise from flux trapped in the stator circuits at the
instant of the fault. This flux decays to zero in a time governed by the armature time
constant, Ta[110123]. The stationary trapped stator flux induces a fundamental frequency
current in the rotor, which then induces a second harmonic flux back into the stator. The
armature time constant is dependent upon the ratio of stator short circuit inductance to
resistance. Typical values range between 0.04 and 0.4 seconds[109].

M.J.Checksfield/PhD/University of Bath 1996 Chapter 3, Page 62

3.3.4 Three Phase Short Circuit Torques and Powers
During and after a three phase short circuit, the generator torque and therefore real power,
can be broken down into three components. A dc component due to i2r losses in the stator
and rotor, a fundamental frequency component due to trapped stator flux, and a double
frequency component dependent upon the sub transient saliency[123]. The double frequency
terms during balanced faults are usually negligible compared to those generated during
unbalanced conditions.

As well as the above components, the components due to equation 3(3) also occur at a
reduced level because the fault reduces the generator terminal voltage. The fundamental
frequency component decays according to the value of armature time constant, Ta. It is
important that the existence of such oscillations in torque are known when designing
power based protection algorithms.

3.3.5 Unbalanced Short Circuit Torques and Powers

Unbalanced short circuit faults will cause all of the components in torque and power
discussed above, the main difference is that the second harmonic component is much
greater due to the unbalance. This introduces a 'saliency' in the stator which in addition
to any rotor saliency results in double frequency terms in generator real power. An
unbalanced load will also produce twice power system frequency terms in power.

3.3.6 The Effect of External Impedance on Generator Operating Characteristics

The equations presented above apply to a generator operating against an infinite bus. The
external reactance between generator terminals and bus is by definition, zero. This does
not occur practically, however in most cases the external impedance is negligible
compared to the generator impedances and the generator can therefore be regarded as
operating on an infinite bus. In situations where this is not the case, the real and reactive
power equations and time constants can be modified to take the external impedance into
account by adding the external impedance to the stator leakage value[35,127]. This approach
has been dealt with in Appendix A4.1, for a general synchronous machine.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 3, Page 63

3.4 GENERATOR CAPABILITY CHARTS

The generator capability, or operating chart is used by generator operators to show the
limits of operation of the generator1120’129,130,133,1341. It uses real and reactive power as its
two axis. Figure 3.4a shows a round rotor generator operating chart, whilst figure 3.4b
shows the under-excited section of a salient pole generator operating chart.

Figure 3.4a and 3.4b shows that the theoretical stability limit is determined by Xq. Figure
3.4a has been labelled Xd since this parameter is most commonly used when referring to
round rotor machines. In a round rotor machine Xd and Xq are approximately equal. The
practical stability limit line on both diagrams shows that, the generator is always operated
to the right of the theoretical stability limit. This allows for fluctuations in load and
improves the transient stability margin following power system short circuit faults11351.

3.4.1 Reactive Power Limits

Figure 3.4a shows that with a fast, high performance AVR fitted, it is possible for the
generator to operate beyond the theoretical stability limit. Depending upon the impedance
of the intertie connecting the generator to the system, it is possible to operate the
generator at load angles of up to 130° [1201. This improves the generator's reactive power
absorbing capability beyond the theoretical stability limit level of -V2/Xd. If the AVR is
capable of supplying negative field current the theoretical no load level becomes -
V2/Xq [,23]. However, operation beyond the stability limit does not occur in practice
because the generator is likely to become unstable following power system short circuit
faults. Some generators are not required to have any reactive power absorbing
capability1131, and it is common practice to fit minimum excitation limiter circuitry11361 to
embedded generator AVRs to prevent absorption of high amounts of reactive power during
high system voltages. This prevents the generator from pulling out of step and thermal
overloading during high grid voltages.

Large generators are also fitted with minimum excitation limiters, which are set so that
the machine can not operate past the practical stability line[137]. In practice the reactive
power absorbing capability of the generator may be less than predicted by theory due to
shorted rotor field turns. Many synchronous generators are shipped with one or more
shorted tumstl3?1.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 3, Page 64

Figure 3.4b shows an excitation margin, at the low power region of the capability chart.
This is required because the excitation system on some generators is not capable of
reducing the field voltage to zero, they therefore have circuitry fitted which limits the field
voltage to 5 % of the full load value[133].

3.4.2 Saturation Effects

At low values of excitation, saturation effects are at their lowest, and the synchronous
reactance remains constant. The synchronous reactances will therefore be at the
manufacturer’s values at low excitations, because the open circuit characteristic almost
coincides with the air gap line. The unsaturated value of synchronous reactance is
normally quoted by manufacturers. As field current is increased, saturation decreases the
value of synchronous reactance[130]. Saturation effects are most dominant in the lagging
power factor region. In the leading VAr operating area, the effect of saturation is to
flatten the normal power angle curve and at the same time, push the maximum power
angle above 90°. This drift of the stability limit increases with excitation; at low
excitations which correspond to low output powers close to the theoretical stability limit,
the effect is negligible. The extent of the drift of the theoretical stability limit depends
on the degree of saturation, but even in extreme cases, the overall effect is quite small and
shifts the maximum power angle up by 5 to 10° at full load[32]. The effect of stator
resistance is the opposite to that of saturation, and the two effects will therefore cancel
each other to some extent.

3.5 EMBEDDED GENERATOR STABILITY - THE LIKELIHOOD OF

POLE SLIPPING

Embedded generators which operate in parallel with utility distribution systems are more
likely to suffer transient instability than the larger grid type generating sets which are
connected to transmission circuits. This is due to their low inertias, low time constants and
the long fault clearance times associated with utility distribution systems.

Transient stability is the ability of the generator to produce forces which act to restore the
generator back to a state of equilibrium following a transient disturbance. Transient
disturbances can be produced by load changes, switching operations and electrical faults.
Section 1.4 explains the process of losing stability following a transient disturbance to the

M.J.Checksfiel d/PhD /University of Bath 1996 Chapter 3, Page 65

generator using the Equal Area Criterion

The time taken for the load angle to swing from its initial value to the critical clearing
angle for a 3 phase short circuit fault, is referred to as the Critical Clearing Time (CCT).
Faults must therefore be cleared within this time in order to maintain stability. In most
cases of transient stability assessment, if the machine does not lose stability during the
first swing in load angle, it will maintain stability for that disturbance. If the generator
loses stability during subsequent swings in load angle, this is referred to as dynamic or
small signal instability, because the loss of stability is a result of insufficient damping.

For most embedded generators, the point of connection to the REC's distribution system
can be thought of as a connection to an infinite bus, that is a bus whose frequency and
voltage remain unaffected by the embedded generator’s operation. This simplifies the
discussion of stability because the one machine-infinite bus model is the most basic
system for stability analysis. If the REC's bus is not infinite, then the system can be
represented as a two machine system, which can also be analyzed without too much
difficulty.

The factors which affect synchronous generator stability can be split into two different
categories, those which are determined by the generator and its control equipment, and
those which are dependent upon the configuration and operation of the system to which
the generator is connected.

The two main generator parameters which have the most affect on stability are the inertia
of the generator, and the direct axis transient reactance. The most useful measure of
generator inertia is the inertia constant, H in units of kWs/kVA. The main advantage of
using H compared with other measures of rotational inertia is that its value is fairly
constant for a wide range of generator sizes and types, it therefore makes comparisons
between machines easier.

3.5.1 Inertia Constant, H & Transient Reactance Xd’

From a stability viewpoint the higher the generator's value of H, the more stable it will
be. For a given fault clearing time, the machine load angle at which the fault is cleared
decreases with an increase in inertia. The machine requires more acceleration energy to
make its rotor advance and therefore appears less responsive and more able to withstand

M.J.Checksfi el d/PhD /University of Bath 1996 Chapter 3, Page 66

a mismatch between generator input and output powers.

The direct axis transient reactance, Xd' is also a vital measure of transient stability because
it determines the amplitude of the generator's transient power load angle characteristic.
Equation 3(3) shows that the lower the transient reactance, the more output power the
generator can produce. This serves to provide a greater decelerating force on the
generator rotor following a transient disturbance, assisting stability.

Tables Cl and C2 of appendix C contain values of H for a range of generator sizes. For
generator sizes above 9 MVA, the inertia constant is above 2.6 seconds, whilst for those
generators below 9 MVA, the inertia constant is below 1.8 seconds. Embedded generators
are generally more 'twitchy' than large grid generators, and therefore accelerate faster for
a given fault duration. The tables also show Xd' which is similar regardless of the size of
the generator. The duration for which the generator reactance is equal to Xd' following a
fault does vary however .

3.5.2 Short Circuit Time Constant, Td'

Section 3.3.2 discusses the effect that the value of Tdo' and Td' have on the generator
power characteristics. When considering short circuit faults, the parameter Td' is more
useful, since this time constant is appropriate for short circuit conditions. Equation 3(10)
gives the relationship between Tdo' and Td'.

Td' determines the time taken for the reactance of the machine to change from the
transient value Xd' back to the synchronous value Xd following a transient. With this
change comes a corresponding decrease in the output power capabilities of the generator
and hence its potential to produce decelerating forces in response to a disturbance. It
follows from equation 3(9) that the higher the resistance compared with the inductance of
the field, the shorter the time constant, and the less time the generator has for producing
high output powers. If the generator is exposed to a severe fault, which is cleared in a
time of around a few hundred milliseconds the constant flux theorem does not hold
because of the decrease in flux caused by the fault[110]. Embedded generators have higher
per unit resistances than the larger grid type generators, this results in them having smaller
time constants.

Tables Cl and C2 of Appendix C show that for generator sizes above 9 MVA, the time

M.J.Checksfield/PhD/University of Bath 1996 Chapter 3, Page 67

constant Td1 is typically greater than 1 second, whilst for the generators below 9 MVA,
Td' is typically in the 100 to 200 ms range. This shows that as well as having lower
inertias, embedded generators are unable to maintain the high transient output powers
which provide stabilising decelerating forces. In terms of the Equal Area Criterion, this
reduces the decelerating energy and increases the accelerating energy for a given
disturbance, resulting in a shorter CCT and therefore a higher likelihood of instability and
hence pole slipping.

3.5.3 Damping Power Provided by Generator and AVR

Another factor which can influence stability is the damping power produced by the
generator. This is dependent mainly upon the type of damper winding in the machine, the
magnitude of armature/tie-line resistance and the AVR parameters. A positive damping
coefficient will aid stability by producing a decelerating force on the rotor when it is
accelerating.

The effect that the combined value of armature and tie line resistance has on the damping
coefficient is dependent upon the AVR used[138]. If the AVR is not in service, then an
increase in resistance will result in a decrease in the damping coefficient, to the point
where the machine can become negatively damped11251. This damping is the natural
damping of the machine. If the AVR is in service, then an increase in the resistance
increases the damping resulting from the AVR. The damping provided by the AVR is
dominant over the generator natural damping, and therefore a high resistance can improve
damping overall.

However, a poorly tuned AVR can make the damping worse. If the loop gain of the AVR
is too high, the damping can become negative. If the AVR does not have a suitable value
of stabilising derivative feedback signal1371, then the high gain necessary to produce a small
steady state voltage error will remain high during transient disturbances. This can result
in rotor oscillations which increase in magnitude until the generator becomes unstable and
pole slipping occurs. This is known as dynamic instability and additional stabilising
feedback signals often need to be incorporated into the AVR to prevent ittlll].

3.5.4 AVR Field Forcing Action During Short Circuit Faults

The effect of high fault current and therefore armature reaction during a fault tends to
decrease flux linkage, especially if the reactive power output of the generator is large[109].

M.J.Checksfield/PhD/University of Bath 1996 Chapter 3, Page 68

If the generator has an AVR in operation, the action of the AVR tends to force the
excitation system to boost the exciter's mmf and therefore maintain flux levels.
Anderson11091, states that the dominant factor with AVRs in maintaining stability is the
ability to offset the effects of armature reaction during faults. Even though AVRs can
significantly improve the flux levels during faults, their effect on the magnitude of the first
swing in load angle is generally only a reduction of a few degrees. They are most
beneficial with long clearance time faults. It is therefore advantageous to have a high
performance AVR. Unfortunately embedded generator AVRs are often basic due to the
size of the generator and the application. High performance from an AVR is not seen as
a pre-requisite since embedded generators are too small to affect system voltages. AVRs
are normally used on embedded generators to provide constant power factor operation in
order to meet the site real and reactive power demand[7,12,71].

3.5.5 Generator Governors

Depending upon the type of governor/prime mover set used, the governor may offer a
significant improvement in the CCT. If the governor and prime mover system has a small
overall time lag, the CCT can be increased by governor action. In many transient stability
texts, the effect of governor action is neglected on the grounds that the governor will not
respond in the one second period of the transient under consideration. This may be true
for a three stage high, medium, and low pressure steam turbine combination, but the
assumption is invalid for fast valved steam turbines, gas turbines, and diesel generators.
Hydro prime movers are different because their characteristics produce a power increase
before a power decrease during an over-speed condition.

Studies conducted1151 on a 4.51 MVA gas turbine generator show that the governor can
increase the critical clearance time by approximately 40 ms. The critical clearance times
found with this study were in the region of 200 ms. If the governor and prime mover
system has a long time lag, it's time constants will render it practically in-active during
the first swing of a transient disturbance, and it will therefore have little effect on the
c c t [139].

3.5.6 Protection Characteristics of the System Connected to the Generator

The two main external factors which affect the transient stability margins of a generator
are the effective transfer impedance between generator and infinite bus and the protection
clearance times for faults.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 3, Page 69

Protection clearance time is the most important factor in maintaining stability, because the
longer the fault exists on the system, the longer the power transfer capabilities of the
generator are disrupted. The protection clearance times normally associated with high
power transfer transmission circuits are usually in the region of several power system
cycles. This means that only a small amount of accelerating energy is produced as a
consequence of the fault. The generator rotor should therefore not accelerate much.

The protection associated with distribution feeders can have clearance time in excess of
one second where graded IDMT over-current relays are employed121,41,42,1401. With this long
clearance time, the load angle increases so much that the likelihood of the generator
maintaining stability is low. Embedded generator CCT's are generally between 200 and
400 ms. Since distribution circuit breakers can take up to 100 ms to operate, this leaves
only 100 to 300 ms for fault detection. This figure is hard to achieve with IDMT type
over-current relaying .

Engineering recommendation G59[14,1411 states in section 6.4 that the generator should be
disconnected from the system "when a system abnormality occurs that results in an
unacceptable deviation of the voltage or frequency at the point of supply". Power system
faults can produce unacceptable deviations at the point of supply, G59 therefore requires
that the generator be disconnected for all short circuit faults. If this is achieved, then pole
slipping should not occur following short circuit faults. However, in practice the generator
is not disconnected for all external short circuit faults. Pole slipping may therefore occur.
Additionally, the above definition also demands that a pole slipping generator be
disconnected due to the voltage or frequency disruptions which may occur.
«

Engineering Technical Report 113[15] also specifies that over-current protection should be
fitted to the generator and gives methods for deriving the fault current contribution from
the generator. The amount of fault current contribution from the generator depends upon
the generator reactances, time constants and AVR parameters. A low value of sub
transient and transient reactance will result in high fault currents, whilst the longer the
time constants, the longer the transient components of fault current will last. Because
embedded generators generally have short time constants, these transient currents will
therefore be short-lived. The AVR ceiling voltage determines the steady state fault
current. If this ceiling limit is low, then the generator fault current contribution may not
be sufficient to operate an IDMT type over-current relay. Generator manufacturers supply

M.J.Checksfield/PhD/University of Bath 1996 Chapter 3, Page 70

as standard, generators fitted with static excitation systems without terminal current
boost[42]. The fault current in this type of generator will quickly fall to zero because the
excitation system will not supply any field voltage during the fault.

Fielding[140] states that voltage controlled IDMT over-current protection is used on

embedded generators. The generator protection coordination is such that the generator
relays grade with the utility distribution relays, the generator would therefore not be
disconnected for faults on the utility feeders. Such faults would be cleared in around
500 ms. If the clearance of the fault does not result in the disconnection of the generator
network from the utility grid supply, then pole slipping will occur.

3.5.7 Transfer Reactance Between G enerator and Infinite Bus

The value of connecting impedance between the generator and the infinite bus has a
crucial affect on generator stability. The study conducted on the 4.5 MVA gas turbine
generator in reference1151 revealed that the CCT rapidly decreases once the utility source
impedance reaches the sub-transient infeed. With source impedances below the sub
transient level, the CCT ranged from 200 to 220 ms, for impedances above the sub
transient level, the CCT dropped to 150 ms.

Any reactance between the terminals of the generator and the infinite bus can be added
to the reactance of the generator, including that of a step up transformer. It follows that
a large external impedance reduces the synchronising power transfer capability. The
generator therefore cannot produce the decelerating forces necessary to maintain
synchronism. As a general rule[15], the critical clearing time is related to inertia and inter
tie or external reactance in the following manner;

CCT a ----------- 3<n )

N (Xd' + X) (Xd + X)

Where X is the external or 'transfer* reactance. The situation may be worse following the
clearance of a fault because one of several parallel interconnections would be disconnected
to clear the fault, resulting in a higher post fault transfer impedance.

3.5.8 Effect of Resistance on Stability

Section 3.5.3 stated that high stator and inter-tie resistances may reduce the natural
damping capabilities of the generator, lessening the stability margin. The greatest concern

M.J.Checksfield/PhD/University of Bath 1996 Chapter 3, Page 71

with high resistance values in an embedded generation environment is the effect on
dynamic stability. A high value of resistance found in a long length of cable, combined
with a poorly damped machine, low performance AVR and poorly damped governor could
result in a negative value of damping1111,121,142,1431. This could cause the machine to
oscillate with the amplitudes of the oscillations increasing until the generator loses
stability.

Increased resistance can also be beneficial to stability for close up short circuit faults
because the power dissipated in the resistance provides some load for the generator,
reducing the accelerating power going into the rotor.

3.5.9 Affect of Fault Type on Stability

A decrease in fault impedance reduces the power transfer from the generator, causing a
lower stability limit for a given fault duration. In order of fault severity, the most severe
fault is the 3 phase fault. This is followed by 2 phase to earth, two phase, and single
phase to earth faults. Kimbark[139] states that the difference in severity of faults becomes
smaller as the fault duration is decreased, however, with a fault of 200 ms there is a
significant difference.

3.5.10 Affect of Earthing on Stability

The grounding of the system affects it's zero sequence impedance, Zo which in turn
affects the impedance of unbalanced faults and therefore the level of power mismatch
caused by the fault. As Zo increases with reference to the fault location, the severity of
an unbalanced fault decreases. If the extreme of an infinite value of Zo is considered, it
can be seen that a two phase to earth fault would transform into a two phase fault. It is
therefore beneficial from a stability viewpoint to ground through impedances rather than
using solid grounding techniques. Grounding resistors close to generators are more
beneficial than reactors as they provide an additional braking effect to the generator during
unbalanced faults.

Embedded generators are normally un-earthed when operating in parallel with a utility
supply. This therefore improves stability with respect to unbalanced faults.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 3, Page 72

3.5.11 Effects of Embedded Generation System Configuration on Stability
Some embedded configurations have multiple embedded generators located at one site.
If the generators are connected to the same bus, they can be grouped together, their
behaviour appearing as one larger generator. From a stability viewpoint, adding extra
generation to a site will reduce the ratio of infinite bus to machine ratios. If the ratio
becomes too low, the synchronising power following disturbances will decrease and the
generators will be more prone to instability.

Step up transformers also affect stability, the impedance of the transformer can be added
to that of the generator's stator circuit. If the transformer has a high impedance, the
interchange of synchronising power between generator and grid will be reduced, causing
a decrease in the stability margin. The transformer may improve stability for close up
terminal faults because the transformer will partially load the generator.

3.5.12 Summary of Embedded Generator Stability Discussion

Embedded generators connected to distribution systems using over-current protection are
much more prone to instability than large 'grid' style generators connected to transmission
systems. The long fault clearance times associated with utility IDMT style over-current
protection, coupled with low generator inertias and short time constants substantially
decreases their stability margins.

Prolonged disturbance to the surrounding power system may result following a fault, due
to the effects of a synchronous generator operating in an unstable manner. These
problems can only be remedied by either reducing fault clearance times so that the
generator remains stable, or installing protective relays to disconnect the generator when
it becomes unstable.

If the generator has over-current protection set to operate for any external short circuit
fault, then pole slipping due to faults will not occur because the fault which would cause
the pole slip will cause the generator to be disconnected. If the over-current protection
is set to grade with protection on adjacent feeders, then pole slipping is likely to occur
following faults.

The Critical Clearance Time (CCT) for small and medium sized embedded generation has
been found in most studies to be between 200 and 300 ms[12»15144 145] When compared with

M.J.Checksfield/PhD/University of Bath 1996 Chapter 3, Page 73

the 0.5 to 1 second fault clearance times typically produced by over-current relays, it can
be concluded that there is a high likelihood that embedded generators will suffer
instability and pole slip.

3.6 SYNCHRONOUS GENERATOR POLE SLIPPING CHARACTERISTICS

This section presents the voltage, current, flux and power characteristics for pole slipping
generators. Impedance is not discussed because it has already been discussed in chapter
2. Appendix A gives full derivations of impedance and power variations during pole
slipping for a range of power system models.

3.6.1 Variation in Voltage and Current During Pole Slipping

The voltage and current variations are most easily explained if the synchronous
machine/utility connection is represented as the classical one machine/infinite bus[146], as
shown in Figure 2.1a. The machine and utility are represented by a voltage behind a
reactance, and are connected via the distribution line impedance.

As a generator falls out of step, the phase difference between the two voltages increases,
and when this difference is 180°, the voltages are in direct opposition. This produces a
point between the two machines where the voltage will fall to almost zero. This point is
known as the system centre. It's location is fixed by the relative impedances of the two
machines and the interconnecting equipment between. If the combination of generator and
transformer impedance equals the utility source impedance in Figure 2.1a, the system
centre would occur at the geometric centre of the connecting line.

If the generator rating is decreased, its impedance Zg would increase, and the system
centre will move towards the generator. For embedded machines, the rating of the
machine is small in relation to the rating of the system it is connected to. The system
centre will therefore normally occur inside the machine or its transformer unit. With a
system centre inside the generator, local loads will be affected by the voltage fluctuations
produced by pole slipping. If the generator voltage, Eg, were increased using a higher
excitation level, the system centre, and therefore the point of worst voltage fluctuations
would move away from the generator, causing more of a voltage disturbance to other
consumers.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 3, Page 74

The current produced when the voltages are 180° in opposition is high, giving the
appearance of a short circuit at the system centre. Figure 3.5 shows generator terminal
voltage and current plots in relation to load angle for two consecutive pole slips, followed
by re-synchronisation. These plots were produced by POWSIM,[107] using the power
system model described in Chapter 5. The 588 MVA generator modelled had a 220 ms
three phase fault placed at its terminals which resulted in two pole slip cycles. The source
capacity at the terminals of the generator was 4.5 times that of the generator. The source
impedance was equal to 72 % of Xd'. This resulted in the system centre appearing just
inside the generator.

Figure 3.5b shows the voltage fluctuations for the pole slips, which at a time of
1.1 seconds fall to 0.4 pu at the machine terminals. This voltage minimum coincides with
the point where the load angle is at 180° and the machine voltages are in complete phase
opposition. As the load angle increases past 180°, the two voltages oppose each other less
until they are back in phase again, resulting in normal system voltages and currents.

The current plot shows the current maximum produced during pole slipping at a time of
1.1 seconds to be higher than the sub transient fault level which occurs at the fault on
point. This demonstrates the strain which pole slipping can place on the generator
windings. The pole slipping current is at the sub transient fault level for much longer than
the sub transient time constant. Consider a generator11471 with typical values of reactance
such that Xd' = 0.3 and Xd" = 0.2. The maximum level of sub-transient current for a solid
three phase terminal fault is given by;

I" = — = -L = 5 pu 3(12)
Xd" 0.2

For pole slipping, the worst case is where the generator is connected directly to an infinite
bus. The source impedance is therefore negligible when compared to the transient
reactance. The current level when the voltages are 180° apart is then given by;

/ . , ,., = 2V = — -— = 6.67pu 3(13)

M p) X d '+ X 0 .3 + 0

The situation will be worse if generator AVR field forcing raises the generator EMF to
above 1 p.u. This shows that the currents experienced during pole slipping may be
substantially higher than three phase fault levels.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 3, Page 75

3.6.2 Pole Slipping Explained in Terms of Magnetic Flux
If all of the magnetic effects due to induced currents in the field, rotor iron and damper
windings are neglected, then a simpler explanation of the process of losing synchronism
is possible. This assumption is valid when pole slipping results from a slow loss of
synchronism, such as a loss of steady state stability. In this situation, the whole pole slip
can take one second or more, and at this rate of slip, the induced effects are negligible.

As the load angle starts to increase from its steady state value, the air gap flux will
decrease due to the demagnetising effect of armature reaction[45,148]. The demagnetising
effect results from currents flowing in the stator, which create an mmf that opposes the
rotor mmf, thus weakening the resultant air gap mmf. With further load angle increase,
the flux linking the stator and the rotor together will begin to unlink, taking leakage flux
paths instead. This causes a weakening of the magnetic coupling between the stator and
rotor. The point where the magnetic coupling between the stator and the rotor becomes
zero is defined as the loss of synchronism point[148] and at this point the air gap flux will
be at its minimum and the resultant restraining torque on the rotor is negligible. In an
’ideal' machine this point would correspond to a load angle of 180°.

With no magnetic coupling, the rotor does not oppose the driving force of the prime
mover, and the load angle will increase rapidly, as the machine moves from generating
to motoring. The flux between rotor and stator will again start to link. However, it will
link with the rotor pole next to and behind of the pole it used to link with, and the
machine will be in a motoring condition. With the machine in a motoring condition,
power is drawn from the electrical supply, as well as the prime mover. Most of this power
is used to accelerate the rotor. The rotor will therefore travel quickly through the
motoring region until once more, the rotor and stator flux link in a generating condition.
The machine will only re-synchronise if the strength of the magnetic coupling between
stator and rotor is sufficient to overcome the extra kinetic energy in the rotor produced by
it travelling above synchronous speed. If the machine is to re-synchronise, all of the
excess kinetic energy must be removed before the CSP for that slip cycle is reached.

3.6.3 Direct and Q uadrature Axis Flux Variations with Load Angle Increase
For low load angles ( « 90°), the major proportion of the air gap flux is comprised of
the direct axis component, with the quadrature axis contributing very little. As the load
angle increases, the flux produced by the stator, <|)a begins to lag further behind <|)f , the

M.J.Checksfield/PhD/University of Bath 1996 Chapter 3, Page 76

field flux. As the phase angle between <|)a and <t>f increases, a corresponding increase in
armature current occurs, causing an increase in braking torque, and thus balancing the
accelerating torque that produced the load angle increase. The magnitude of the
quadrature component of flux, <{>q, changes with variations in load angle. The upper
limit[149] on <|)q is set by V/Xcomb, this peak occurs at a load angle of 90°, beyond this point,
it decreases again. V is the infinite bus voltage, and Xcomb is the combined value of
synchronous reactance plus any external reactance. The direct axis component of flux, <|>d,
will reach its minimum at a load angle of 180°. For the process of losing synchronism,
the direct axis component will gradually decrease as the rotor is accelerated, accompanied
by an increase in quadrature axis component. The quadrature axis component provides
an increasing amount of braking torque as the load angle approaches 90°. Once the load
angle exceeds 90°, <|>q will begin to decrease along with <|)d, resulting in an unstable
decrease in braking torque with load angle increase. When the load angle approaches
180°, <|>q will vanish and <|)d will be at its lowest level. This causes the lowest magnetic
coupling between stator and rotor, and with no flux interaction to provide a braking
torque, the rotor will accelerate further and the machine will pole slip into a motoring
condition.

References [148] and [149] provide a finite element analysis of the process of losing
synchronism, and contain figures showing the change in flux paths as the machine loses
synchronism.

3.6.4 Real and Reactive Power Pole Slipping Characteristics

The main real and reactive power characteristic during pole slipping are determined by the
sinusoidal terms in equations 3(3) and 3(4). If the slip rate during the pole slip is above
0.33 %, then the generator reactance will be less than twice the value of Xd'[45]. The
transient saliency terms will therefore contribute a significant amount to the fundamental
sinusoidal characteristics resulting from the interaction of stator and rotor flux.

As the load angle completes a 360° cycle, one sinusoidal variation in real and reactive
power will occur. Any asynchronous power generated during the pole slip will shift the
sinusoidal variation in power up. A detailed analysis of the power characteristics during
pole slipping are given in Chapter 6.

M.J. Checksfiel d/PhD /University of Bath 1996 Chapter 3, Page 77

3.7 CHAPTER THREE SUMMARY

This chapter presents the steady state and transient power characteristics of a generator.
These have a sinusoidal power load angle characteristic, with a second harmonic
component caused by saliency. Embedded generators are far from 'ideal' however and the
constant flux linkage assumption normally used for large 'grid' type generators can not
be used. Embedded machines have short time constants which for long duration faults,
cause a significant decay in the field flux. This distorts the sinusoidal variations in power,
because the reactance of the generator changes from Xd' to Xd as the flux decays. Field
flux decay can cause a decrease in embedded generator transient stability margins.

Power system frequency oscillations and second harmonics can also occur in the
generator's real power during and after short circuit faults, due to trapped flux in the
generator stator. These oscillations decay at a rate determined by the armature time
constant.

When the generator rotor travels above synchronous speed, the induced slip frequency
currents in the field structure enable the generator to generate asynchronous power in the
same way as an induction machine. Reactive power is absorbed by the generator during
asynchronous operation as this provides the necessary magnetising VArs.

A fast acting AVR enables the operation of a generator beyond its steady state stability
limit, at machine rotor angles greater than 90°. Although this type of operation is
possible, it is never practised, because the reserve in hand for transient stability is
generally too low.

Embedded generators are more likely to pole slip than 'grid' type machines because they
have short time constants, low inertias, and are exposed to long fault clearance times.
Their CCT is typically between 200 and 400 ms, whereas the fault clearance time
exhibited by the typical over-current protection used on distribution feeders is between 0.5
and 1.0 seconds.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 3, Page 78

 6
full load transient
5
no load transient
4

2
Real Power (pu)

1
0
1 no load steady state
2 full load steady state
3

6
0 30 60 90 120 150 180 210 240 270 300 330 360
Machine Rotor Angle (degrees)

Figure 3.1a
£L

Power Load Angle Characteristic for a 200 MVA Round Rotor Steam Turbine Generator.

no load steady state

0

■2
Q = -V /Xq Q = -V /Xq
Reactive Power (pu)

3
full load steady state

5
no load transient
6

8
full load transient
9
0 30 60 90 120 150 180 210 240 270 300 330 360
Machine Rotor Angle (degrees)

Figure 3.1b
Reactive Power Load Angle Characteristic for a 200 MVA Round Rotor Steam Turbine Generator

M. J. Checksfield/PhD/University o f Bath Chapter 3, Fig 3.1, Page 79

 6
full load transient
5
no load transient
4

2
Real Power (pu)

1
0
1 no load steady state
■2 full load steady state
3

6
0 30 60 90 120 150 180 210 240 270 300 330 360
Machine Rotor Angle (degrees)

Figure 3.2a
Real Power Load Angle Characteristic for 625 kVA Salient Pole Diesel Generator.

1
no load steady state
0

•2
= -V /Xq Q = -V /Xq
Reactive Power (pu)

■4 full load steady state

•5
no load transient
6

9 full load transient

-10
0 30 60 90 120 150 180 210 240 270 300 330 360
Machine Rotor Angle (degrees)

Figure 3.2b
Reactive Power Load Angle Characteristic for 625 kVA Salient Pole Diesel Generator.

M. J. Checksfield/PhD/University o f Bath Chapter 3, Fig 3.2, Page 80

 A B C
KEY Tdo' = 2s, (no avr) Tdo' = 2s, (avr included) -Tdo' = 6s, (no avr)
L . - . - *- . v A l
The curves shown in Figure 3.3b have
K \ —7 v\
not been shown completely on this \ N*\
graph for reasons o f clarity.

Steady state
pre-fault starting
/ point

r
M achine angle increasing during fault p en od Rotor swings
up to CSP

80 100 120 140| 160

Machine Angle (degrees) Fault Clearance
Points
Figure 3.3 a
Real Power / Machine Angle Plot for Three Phase Fault Simulations with
Various Values o f Tdo' on a Diesel Generator, Governor Disabled.

KEY T d o '= 2s, (no avr) T d o '= 2s, (avr included) -Tdo'= 6s, (no avr)
3

2 C2

Fault on
-B 2

1 -A 2

0
Oscillations in power due
Fault off to stator time constant, Ta

1
0.0 0.25 0.5 0.75 1.0 1.25 1.5
Time (seconds)

Figure 3.3b
Real Power/Time Plot for Three Phase Fault Simulations
on a Diesel Generator, Governor Disabled.
M. J. Checksfi el d/PhD/University o f Bath Chapter 3, Fig 3.3, Page 81
 Pi MW limit
MVA
to limit
Stator
end
heating
limit
Excitation
•-D
\ limit
practical stability limit
is usually based on
10% safety margin

Leading Vars Lagging Vars

Figure 3.4a
Round Rotor Generator Operating Chart.

Locus of
Synchronous
Operation x>
Operating
for If = 0
u K Point

5 % Excitation Margin

Figure 3.4b
Under-Excited Section of Salient Pole Generator Operating Chart

M. J. Checksfield/PhD/University of Bath Chapter 3, Fig 3.4, Page 82

 a) Load Angle.
200
Generator Re-synchronised
Load Angle (degrees)

100

-100

-200
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (s)
b) Terminal Voltage.
Fault on Voltages in phase opposition
O
8 = 180, voltage at minimum value
Voltage (p.u.)

0.8
Fault off

0.4

0.2 -

0.2 0.4 0.6 0.8

Time (s)

Voltages in phase opposition

Sub-Transient c) Terminal Current.
5 = 180, current at maximum value
3.5 Fault Current

3
2.5
Current (p.u.)

2
1.5
1
0.5
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (s)

Figure 3.5
Voltage and Current Variation During Pole Slipping for a 588 MVA Generator with A
Terminal Source Capacity 4.5 Times its Rating. (Source Impedance = 12% Xd' )

M. J. Checksfield/PhD/University of Bath Chapter 3, Fig 3.5, Page 83

 Chapter 4
THE NEW POWER BASED POLE SLIPPING ALGORITHM

In designing a pole slipping algorithm, it must be able to predict when a generator is

committed to a pole slip or actually is pole slipping. It must also remain stable for all
other power system conditions, including faults and power swings. The algorithm has
been designed for use in a multi-function microprocessor based relaying
environment123’27’28’29,30,31,57,91,1501. A limited amount of computer processing power is
therefore available. An IEEE committee report on 'Experience with Generator Protection
and Prospects for Improvements Using Digital Computers' reported that respondents from
a utility engineer survey stated that "generator protection can be improved by providing
out-of-step relays that detect loss of synchronism for conditions that do not cause the
apparent impedance to move through the conventional blinder characteristics". This
confirms that conventional pole slipping relays can be difficult to set and suggests a need
for improved pole slipping relay design.

The paper reports that out-of-step relaying can benefit most from micro-processor based
relaying since sophisticated algorithms can be developed which take advantage of the
computing power and long time frames involved with pole slipping. The paper also stated
that "A design challenge is to make such a sophisticated relay easy for utility engineers
to apply and set". This point has been given much consideration when designing the new
power based pole slipping algorithm, since there is little point in designing an algorithm
which is too complicated for utility protection engineers to easily set.

The protection functions found in a micro-processor relay are generally independent; there
is therefore little difficulty in using a comm on hardware system. An entire suite of
protection functions necessary for protection of a synchronous generator can therefore be
included into one microprocessor relay, thus providing the economy required to meet
modem requirements. This is especially tme for embedded generators because they are

M.J.Checksfield/PhD/University of Bath 1996 Chapter 4, Page 84

relatively inexpensive. They therefore require inexpensive protection schemes. Other
protection functions intended for embedded generators1103,104’151,1521 also require real and
reactive power calculations. The software overhead for this task can therefore be shared,
minimising the demand on the processor made by the algorithm.

Using power in this type of environment has an advantage over reactance and resistance
measurement, since exact point on wave synchronisation is not required and the voltage
and current samples do not have to be resolved into direct and quadrature components.
Generator real power, P, and reactive power, Q, can be calculated from instantaneous
voltage and current samples using1153];

p = va */' a + vb.*/.
b
+ vc * i c 4(1)

Q = (v„*('«-'») + v»* 0 , - 0 + K *('rO ) 4(2)

&

where va ,vb ,vc are the sampled values of phase to earth voltages, and ia ,ib ,ic are the
sampled values of line current. Alternatively, if the relay only measures two line to line
voltages then the 'two wattmeter' method can be used;

P = V**!'. + vbc* ic 4(3)

Q = (v„t *(',"'») + 4(4)

ft

Note that the sign convention used for reactive power is that reactive power will be
negative at the terminals of an under-excited generator.

4.1 BASIS OF THE ALGORITHM

The algorithm predicts pole slipping by detecting if the generator's load angle is still
increasing once the Critical Stability Point (CSP) has been exceeded. Section 1.4.1
explains the Equal Area Criterion and Figure 1.2b shows the Equal Area Diagram and the
CSP for the clearance of a short circuit fault. It is important to note that the new pole
slipping algorithm does not use the equal area criterion in a quantitative manner. The
Equal Area Criterion is used to explain the concept of the CSP.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 4, Page 85

Inspection of Figure 1.2b shows that at, or beyond the CSP, two conditions apply. The
load angle is greater than 90°, and the generator real power output, P is equal to, or less
than the prime mover input power, Pm. Section 3.1.2 states that the reactive power
characteristics for a generator are such that they always take on the same value at 90° and
270°. For load angles between 90° and 270°, the reactive power will be less than the
value at 90°, and for load angles from 270° to 90°, the reactive power will always be
greater. If resistance is neglected, then from equation 3(5) the reactive power at 90° is
given by;

= Qtrip 4(5)

This quantity is referred to as Qtrip, and is used to identify operation at or beyond the
CSP. The unsaturated value of Xq is used because saturation effects are negligible when
a generator's reactive power output is close to Qtrip. The manufacturer normally supplies
unsaturated values of reactances. Figure 1.2b shows that the generator electrical power
output can also be less than the prime mover input power, Pm, for load angles of less than
90°. This could occur at load angle 51 in Figure 1.2b. Generator reactive power output
is therefore used to reveal when the generator is operating in the potentially unstable area
when the load angle is greater than 90°. The real power measurement is then used to
determine if the generator is operating at, or beyond the CSP. Operation at, or beyond
the CSP therefore occurs when; ^ _ .
Q < Qtnp
* 4(6)
P < Pt

Where Pt is the algorithm real power trip setting, that is based on the prime mover input
power, Pm.

The criteria of equation 4(6) could predict that pole slipping will occur, since pole slipping
is inevitable if the generator operates past the CSP. However, Chapter 3 shows that
synchronous generators do not behave in an ideal manner. Effects such as asynchronous
power generation, prime mover dynamics and measurement errors would result in too
many false trip conditions if only equation 4(6) was used to predict pole slipping. To
overcome this problem, generator rate of change of power is used to determine whether
or not the load angle is still increasing once the CSP has been exceeded. If the load angle
is still increasing when the generator operates beyond the CSP, then pole slipping is
inevitable. If the generator just maintains stability after reaching the CSP, then the load
angle will be temporarily static, or decreasing.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 4, Page 86

Figure 1.2b shows that if the load angle is increasing at the CSP, then the generator real
power output will be decreasing, and therefore the time rate of change of real power will
be negative. Load angle increase at the CSP is therefore identified when;

(APIAt) < (AP/Ar)min 4(7)

Where (AP/At)min is a margin for error, allowed for in this expression to ensure algorithm
mal-operation due to measurement errors and non-ideal machine effects does not occur.
The theoretical form of the power based algorithm therefore takes the form;

Q < Qtrip
TRIP IF: and
P < Pt 4(8)
and
(AP/At) < (AP/At)mm

4.2 FILTERING TECHNIQUES

The algorithm has been implemented and tested in a commercial micro-processor based
relaying platform[741. Several different filtering techniques have been used to remove
noise, or enhance algorithm security. This section explains the different techniques and
the reasons for their use.

4.2.1 Anti-Aliasing Filters

The sampling theorem states[154,155] that the sampling frequency must be greater than twice
the highest frequency to be sampled. If this is disobeyed, then aliasing effects will distort
the sampled waveform, since the higher frequencies will impersonate the lower ones and
cause distortion.

The pole slipping algorithm has been designed to work for frequencies 10 Hz above or
below the nominal 50/60 Hz system frequency, i.e from 40 to 70 Hz. To avoid the effects
of aliasing, a first order passive low pass filter with a cut of frequency of 132 Hz is used.
This ensures that any frequencies above half the maximum sampling frequency of 840 Hz
are filtered out before they reach the analogue to digital converters. This ensures that only
the 50 Hz power system frequency and its second and third harmonics are used by the
relay. Note that a 'gentle' first order filter cut off has been used to minimise the effects
of group delay, thus speeding up relay response for functions which require very fast

M.J.Checksfield/PhD/University of Bath 1996 Chapter 4, Page 87

response times1154].

4.2.2 Voltage and Current Fourier Full Cycle Cosine Filters

The relay uses a sampling rate of 12 samples per power system cycle, this is achieved by
the use of a frequency tracking algorithm1741. The Fourier filter extracts the fundamental
frequency component of the voltage and currents. The pole slipping algorithm should use
only balanced three phase fundamental frequency components of voltage and current, since
pole slipping is a balanced phenomenon.

A Fourier cosine filter provides the in phase component of a signal, a sine filter provides
the quadrature phase component. The cosine Fourier filter signal, Ic of a sampled signal,
i takes the form;
N - 1
4(9)
Ic = + £ /wcos(cottA?)
N n=1

whilst the sine Fourier filter signal, Is of a sampled signal, i takes the form;

N -1
4(10)
in sin(co«At)
/!=1

where N is the number of samples per power system cycle, i„ is the instantaneous value
of voltage or current sampled at time nAt, co is the system angular frequency, i0 is the
instantaneous value of signal, i, sampled at time 0, and iN is the instantaneous value of
signal, i, sampled at time NAt [156]. Equations 4(3) and 4(4) use only the cosine filter
components to calculate real and reactive power. The sine component is required for
sample error compensation. The relay has only one analogue to digital converter, the
voltage and current samples are therefore each delayed by 20 fis. The fourier sine and
cosine filter components are used with pre-determined multiplicative constants to 'rotate'
the voltage and current vectors so that they are all effectively sampled at the same time.

The frequency response characteristics of the 1 cycle fourier filter are shown in
Figure 4.1a. The response of the filter to the 11th harmonic is the same as the response
to the fundamental components, due to the effects of aliasing. However, the anti-alias
filter attenuates these harmonics significantly, and the combined response of the anti-alias
and full cycle fourier filters is shown in figure 4.1b.

Figure 4.1b shows that the fundamental frequency component passes through almost

M.J.Checksfield/PhD/University of Bath 1996 Chapter 4, Page 88

unchanged, whilst all harmonics up to the 10th harmonic are removed. This removes
distortion in the calculated power signals due to harmonics. Embedded generators are
exposed to harmonics because they are located at distribution levels where high levels of
harmonics are more likely to exist.

DC components are also removed. This is desirable since it reduces the power system
frequency oscillations which occur in the real power signal during faults. These are
produced as a result of DC current components interacting with power frequency voltage
components. These oscillations are discussed in section 3.3.4.

4.2.3 Moving Average Filters

After filtering the voltage and current signals, the algorithm processing rate is reduced
from 12 to 4 samples per power system cycle. This significantly reduces the demand
which the algorithm places on the relay's processor. Processing at a rate of 4 samples per
cycle also has some advantages from a filtering viewpoint.

Power system unbalance produces double power frequency components in the calculated
real and reactive power signals[152,157]. This can interfere with the operation of the
algorithm, because the algorithm may use the top or bottom of the unbalanced signal, as
shown in Figure 4.2a. Using the top or bottom of the unbalanced signal could result in a
significant error. Applying a two point, half cycle moving average filter removes this
problem completely. A half cycle moving average filter removes the second harmonic,
and also attenuates the fundamental, whilst allowing DC through.

Figure 4.2b shows the effect that the moving average filter has on the power signal for
an unbalanced fault. The level of unbalance on the 'raw' value of instantaneous power can
be seen to increase to a maximum over one power system cycle. The moving average
filtered signal suffers a small oscillation, then settles on the average value of power within
a half cycle of the unbalance reaching a steady state. The mathematical equation for a
moving average filter is given below;
N- 1
v
y {ri)
= J_
M
v X(» -0 4(11)
r «0

Where y(n) and x(n) are the n* output and input respectively, and N is the window size.
With a processing rate of 4 samples per cycle, the equation used for a half cycle moving
average filter becomes;

M.J.Checksfield/PhD/University of Bath 1996 Chapter 4, Page 89

This places a small burden on the relay's processor because with the correct internal
scaling, the calculation can be reduced to a simple addition.

4.2.4 Calculation of the Rate of Change of Power Signal

The rate of change of power signal, (AP/At), is calculated using the following equation;

API A t = —^ P{- 2) 4( 13)

2 At

Note that this method uses the previous but one sample to calculate (AP/At), instead of the
previous sample. This technique was originally proposed by Barrett[104] as a method of
removing unbalance. However, the half cycle moving average filter applied to the real
power signal achieves this. The effect of using this technique is shown in Figure 4.2c.
This figure shows the half cycle moving average filtered power of Figure 4.2b, and the
two rate of change of power signals. The advantage of the alternate sample method is that
less disruption is caused to the (AP/At) signal. The alternate sample method only causes
two 'spikes' in the (AP/At) signal when a change in the level of unbalance occurs.

4.3 THE COM PLETE ALGORITHM

The combined criteria specified in equation 4(8) will successfully predict and detect pole
slipping. However, they can also occasionally be satisfied during power system short
circuit faults. The faults which have been found to cause problems are three phase and
two phase to earth faults. The problem arises at the fault off point. Additional fault
blocking constraints are therefore required to stop algorithm operation for faults. Section
3.3.4 discusses the short circuit torques which occur during faults. The discussion also
applies to electrical output power since at synchronous speed, torque equals power. This
section states that double power frequency terms will occur in the instantaneous power
signal during unbalanced faults, whilst power frequency terms will occur in the power
signal during balanced faults. These power frequency terms will be reduced by the dc
rejection characteristics of the fourier filters, although a small amount will still occur since
the fourier filters still allow low frequencies through.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 4, Page 90

Section 4.2 showed that unbalanced faults will cause a positive and a negative peak in
(AP/At) at their fault on and fault off points.

Empirical analysis has shown that if a time delay of one and a quarter power system
cycles is used on the algorithm trip criteria of equation 4(8), the algorithm successfully
restrains for all power system faults. All of the faults tested did not cause the criteria of
4(8) to be satisfied for more than one power system cycle. Choosing one and a quarter
cycles therefore introduces an extra safety margin. The complete pole slipping algorithm
is therefore;
Q < Qtrip Satisfied

and Continuously
For
P < Pt 4(14)
1.25 Power
and
System
(AP/At) < (AP/At) .
Cycles

4.3.1 The Condition Monitored Real Power Trip Setting, Pt

The real power trip setting, Pt is based on the value of prime mover input power, Pm, as
shown in Figure 1.2b. In an embedded generation relaying environment, the only signals
available are the voltage and current signals from the VT's and CT's. Information such
as the governor position or prime mover output power are not available. The value of Pm
is therefore estimated from an on-line analysis of the generator electrical power output,
P. The input power, Pm is estimated by observing the output power of the generator to
see whether it remains constant for a specific amount of time. If the output remains
constant it can be concluded that the generator is in a steady state and the output power
will approximately equal the input power, except for losses. The function which is used
by the relay to do this is given below;

V l ^ ( W) “ P ( r ,- la g ) I a n d l^ („ ) ~ P ( „ - (2 * lag)) I < P t°^ •

4(15)
then Pt = P^n)*Pfact

where P(n) is the present real power sample value, P(n . jag) is the sample value of 'lag1
samples ago, P(n. (2*iag» ls the sample value of '2*lag' samples ago, and Ptol is a tolerance
band, chosen as nominally 5% of machine rating. Pfact is a coefficient to take into

M.J.Checksfield/PhD/University of Bath 1996 Chapter 4, Page 91

account the losses and inaccuracies inherent in measuring machine input power from
output power. The algorithm processing rate is 4 times per power system cycle, the
nominal value of ’lag' typically chosen is 200 which equates to 1 second, Pfact is
nominally set to 0.9. This results in a safety margin of 10 % to allow for losses and
prime mover response times. The effects of synchronous machine losses on the algorithm
are discussed in Appendix A. Figure 4.3 shows the flowchart for the function in the
algorithm. Note that Pt can never take on a negative value, i.e. the function will not work
for motoring conditions, it has only been designed as a generator pole slipping protection.
Section 4.6 states how the algorithm can be adapted for synchronous motor pole slipping
protection.

The algorithm implemented in the commercially available hardware platform had several
settings which were included for algorithm development purposes only. These enabled
'fine tuning' of the algorithm during field trials, without having to make costly
modifications to the software placed into the relay's EPROM. Section 4.7 contains the
settings used, their ranges, and the settings necessary if the algorithm were released
commercially.

4.3.2 The Adaptive Rate of Change of Power Trip Setting, (AP/At)min

The rate of change of power trip setting, (AP/At)min is calculated by the algorithm
according to the external relay setting, slip, and two internal real power parameters, Pt and
Pmax. The method for determining Pmax, the maximum generator output power is
presented in section 4.3.3.

The margin for error, (AP/At)min used in the (AP/At) criterion is based on the theoretical
value of (AP/At) at the CSP, it is given by;

(AP/At)min = -s *Pmax * cos(8 c) 4(16)

where s is the slip and 5Cis given by;

8 = 180° - sin -'f—^ - 1 4(17)

\Pm ax )

the derivation of (AP/At)min is given in appendix A5.2. The value of slip chosen will not
be the exact value at which the relay trips, due to effects not taken into account in the
derivation of 4(16), such as damping power. Equation 4(16) is used by the algorithm to
continuously update the trip setting (AP/At)min according to what the variables Pmax and

M.J.Checksfield/PhD/University of Bath 1996 Chapter 4, Page 92

Pt dictate.

Pmax is a measure of the maximum output power that the generator has produced in the
last second. It enables the algorithm to adjust its setting according to whether transient or
steady state conditions are prevailing. With 'steady state' pole slips, a gradual loss of
synchronism occurs, and the resulting power output waveforms change gently to begin
with, as shown in Figure 1.3b. This produces low values of (AP/At), a sensitive trip setting
is therefore required. If the generator has been transiently disturbed, its output during the
first swing in load angle, be it stable or unstable, will be much greater than its nominal
output, because the power load angle characteristics change to the transient curves. This
results in a greater magnitude in (AP/At) at the CSP. The algorithm therefore needs to
adjust its setting (AP/At)min accordingly. The algorithm adjusts the setting, (AP/At)min
according to 4(18);
Pt
If ——— < 0.6
Pmax 4(18)
then (AP/At)am = (A P /A t)^ * (AP/At):rm

where (A P/A t)^ is an external relay setting based on the degree of transient saliency in
the generator. Transient saliency arises due to differences in the values of Xd' and Xq'.
In a salient pole generator, the changing stator flux produced by a transient can pass
sideways through the field coils without linking them and inducing current. Consequently,
for salient pole machines, Xq' is equal to Xq[110]. In a solid round rotor machine, a value
of Xq' does exist due to eddy currents in the solid rotor structure. Typically the value of
Xq' is 2 to 4 times larger than Xd', and has a time constant, Tqo' which is typically 10
times shorter than the field time constant, Tdo' [109]. For the purpose of deriving the
setting, (AP/At)^, it is assumed that Xq' is equal to Xq. (AP/At)^ is given by;

(AP/AtY = ( _L -_L ) *Sgen 4(19)

V K Xq X d '
This approach has the advantage that algorithm security against power swings is inherently
increased because a greater magnitude of (AP/At) is required before tripping will occur.
For a 'steady state' pole slip, no peak will occur in the power output before the pole slip.
The algorithm trip setting should therefore be at its most sensitive level to detect the pole
slip because low values of (AP/At) will result.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 4, Page 93

4.3.3 The Maximum Generator Output Power Monitor, Pmax
The value of Pmax used by equations 4(16), 4(17) and 4(18) is nominally set to 1.4 * Pt,
so that the ratio Pt/Pmax equals 0.71. Equation 4(18) shows that (AP/At)^ is not included
in the calculation of (AP/At)min during steady state conditions. If the generator's power
output rises above the existing value of Pmax for longer than one power system cycle, the
value of Pmax is updated to the new maximum for a duration of one second. The one
second limit is used so that the new value of Pmax is only used for the period where a
pole slip or recoverable power swing associated with that value is likely to occur. The
one cycle constraint is necessary to prevent updates in Pmax for spurious spikes in the
power waveform. Figure 4.4 shows the Pmax function flowchart.

4.3.4 Minimum Setting Value for (AP/At)mln, Setting (AP/At)fjlct

If the generator was operating at a low value of output power, then it would be possible
that the setting (AP/At)min calculated by equation 4(16) could almost be zero. This may
reduce algorithm security because if a fault causes a stable power swing, the (AP/At)min
setting may be too sensitive. The setting (AP/At)fact is used so that if the magnitude of
(AP/At)rain calculated by 4(16) falls below this setting, then (AP/At)min is set to (AP/At)fact.
At this development stage, this setting can be adjusted externally. If the algorithm were
released commercially, it would not be provided as an external setting, a value would be
chosen which would be embedded into the algorithm code. The nominal value chosen is
given by;

(AP/Ar)/ac, = -0.25 *Sgen 4(20)

i.e 25 % of the generator rating. Using equation A(57) from appendix A and
equation 4(20), the theoretical operating power at which (AP/At)fact is used for the value
of (AP/At)min is given by;
m -0.255 *Sgen 4(21)
(fit which (AP/AO^ is used) SHp(%) * K

The factor of pi is necessary to convert slip from units of percent to absolute units which
are in the time frame of the generator. This equation has been calculated assuming a
power system operating frequency of 50 Hz. With the default setting of slip of -0.5 %,
(AP/At)fact is used when Pt falls below 16 % of generator rating.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 4, Page 94

4.3.5 Implementation of the Power Based Pole Slipping Algorithm into a
Commercial Relay Platform

The algorithm has been implemented in a commercial micro-processor relay[74]. This

enabled the testing of the algorithm in several ways;
a) It confirmed that the algorithm was suitable for inclusion into a
microprocessor based relaying platform.
b) It made the testing of the algorithm in an industrial environment easier
because the disturbance recorder function of the relay could be used to
store the algorithm measurands and trip levels.
c) It permitted continuous field trials to be conducted on the algorithm since
the relay could be interrogated remotely with a modem. This served to
increase the confidence in the algorithm's ability to restrain for non-pole
slipping conditions since the algorithm was installed on two generators
which were regularly taken on and off load. One of these generators was
a salient pole hydro set, the other a round rotor steam turbine set. The
algorithm was therefore tested with both types of synchronous generator.

Figure 4.5 shows a flowchart for the complete algorithm. The flowchart also shows how
relay output chatter has been prevented. Relay output chatter occurs when an algorithm
measurand continuously crosses back and forth across its trip setting. This is undesirable
as it causes excessive output contact wear. The problem is usually overcome by including
some hysteresis in the trip level, so that after it is first satisfied, the trip level is reduced
to prevent the output contact chatter.

Relay output chatter in the pole slipping algorithm was prevented by keeping the
algorithm in a 'tripped' state until the reactive power criterion was no longer satisfied.
This results in only one trip output per pole slip cycle.

Appendix E contains the document written for the practical implementation of the
algorithm in a commercial relay hardware platform[74]. This document contains practical
information such as simplification of algorithm equations in order to reduce processor
power, numerical overflow considerations, setting ranges, default settings and disturbance
recorder operation.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 4, Page 95

4.4 THEORETICAL OPERATING RANGE OF ALGORITHM

The theoretical operating range of the algorithm is dictated by;

a) The minimum value of system infeed at the generator terminals which
permits the reactive power criterion to operate correctly and the algorithm
to function successfully.
b) The fastest rate of pole slipping which can be detected.
c) The slowest rate of pole slipping which can be detected.

4.4.1 Minimum Value of System Infeed For Algorithm Operation

If the system infeed at the generator terminals is too low, the reactive power criterion will
not be satisfied. The algorithm will therefore fail to detect poleslips. Put another way, if
the system centre does not appear sufficiently inside the generator, the algorithm will fail
to work. For most embedded generators, the infeed is high enough for the algorithm to
operate correctly. For situations where the system centre appears out in the
transmission/distribution network, the algorithm will fail to operate. This is not considered
a problem however, because this is generally thought of as a job for the transmission line
relays. To determine the point where the algorithm will no longer operate, a detailed
analysis was performed using the equations for real and reactive power developed in
Appendix A, section 4.1. These equations are too complicated for a simple equation to
be derived that shows how the limit for system infeed changes with generator parameters.
The mathematics software package, 'Mathcad' was therefore used to derive graphs which
can be applied to any synchronous machine, showing the limits of operation. The method
used to produce the graphs is given in Appendix A, section A5.3.

Four graphs were derived, one showing the effect of varying the external resistance, r, the
others showing the effect of varying Xd' for different values of Xq and E. Figure 4.6a
shows the setting graphs for different values of tie line resistance. This graph shows that
the effect of tie line resistance is small and a high tie-line resistance aids algorithm
operation.

Figures 4.6b, 4.7a and 4.7b show the setting graphs for varying values of Xq, Xd1, and E.
To use the graphs the maximum expected value of internal generator voltage, E is chosen.
This determines which graph is used. For most applications the highest value of E
is 1.25 pu[79]. The graph of Figure 4.6b should therefore be used. In communications with

M.J.Checksfield/PhD/University of Bath 1996 Chapter 4, Page 96

a private protection consultant, it was stated that the practical values for Eg/Es range
between 1.1 pu and 0.8 pu. In rare cases, the maximum value of E can be 1.5 pu, and if
an extremely fast, high ceiling AVR is used, then E may reach 1.75 pu. Once the graph
has been selected, Xd' is calculated as a fraction of Xq. The appropriate curve on the
graph is then selected, and the maximum value of external reactance, X, looked up for the
corresponding value of Xq.

If the value of X taken from the graph is greater than the actual value of external
reactance for the system to be protected, the algorithm will work correctly. The value of
external reactance used should be the largest value possible, i.e the situation of lowest
capacity infeed. The external reactance should include the source reactance, transformers,
and inter-connecting lines. The setting lines were calculated for the case where the
external resistance, r, was equal to 10 % of the external reactance, X. This introduces a
margin for error because it is a pessimistic outlook.

If the relay’s VT's are located on the system side of the generator step up transformer, the
transformer reactance should not be included in the external reactance value. Instead, the
transformer reactance should be added to the values of Xd1and Xq of the generator. This
enables the algorithm to detect pole slips that it could not otherwise detect because the
system centre does not lie sufficiently inside the generator.

4.4.2 Fastest Rate of Pole Slipping Detected by Algorithm

Most pole slip schemes specify that slip rates of up to 10 % should be detected160,79,1471.
The one and a quarter cycle tripping constraint used in the algorithm sets the limit on the
fastest pole slip which the algorithm can detect. It is assumed that the reactive power
criterion, Qtrip is satisfied for machine angles between 90° and 270°. For the fastest rate
of pole slipping to occur, the generator must be operating at full power, the real power trip
setting will therefore equal 0.9 pu if the default value of Pfact of 0.9 is used. The peak
in real power, Pmax is assumed to occur at a load angle of 90°, and equal two times rated
power. This assumption is based on empirical values. Using a value for Pt of 0.9 pu, and
Pmax of 2 pu, and assuming a sinusoidal power/load angle curve, the CSP when the real
power criterion will be satisfied will occur at an angle, 8Cgiven by (equation 4(17));

8 c =180 - sin_1(0.9/2) = 153° 4(22)

The real power criterion will therefore be satisfied between the load angles of 153 and

MJ.Checksfield/PhD/University of Bath 1996 Chapter 4, Page 97

297°. Using the default value for (AP/At)^ of -3.0 pu (see table 4-1), a value of 8Cof
153°, and the nominal slip setting of -0.5%, equations 4(16) and 4(18) give a value for
the rate of change of power trip setting, (AP/At)min of -5.72 pu. Assuming a slip rate of
10 %, and using equations A(54) and A(55) of Appendix A, the equation for rate of
change of power is given by;

(AP/At) =10% * n * — (2 * sin(8 )) = 62.8 cos(8 ) 4(23)

dt

Using the value for (AP/At)min and equation 4(23), the load angles at which the rate of
change of power trip setting is satisfied are 95 and 265°.

The final assumption made is that the 90 to 270° section of the pole slip occurs at twice
the speed of the 270 to 90° section. This assumption is made because the machine
accelerates using power from the prime mover and power from the utility during this
period.

All three trip criteria are satisfied between the load angles of 153 and 265°, i.e they are
satisfied for 112°. The algorithm requires 25 ms to trip. The fastest pole slipping rate
which the algorithm can detect is therefore calculated as follows;
1) the 90 to 270° section of the slip cycle requires 25 ms * 180/112 = 40 ms
2) the mean time to complete 360° is therefore 40 ms * 3 = 120 ms
3) A period of 120 ms equates to 8.3 Hz, i.e 16 % slip at 50 Hz.
This is greater than the 10 % specification which most other pole slipping schemes use,
and is therefore sufficient.

4.4.3 Slowest Rate of Pole Slipping Detected by Algorithm

Very slow pole slips occur rarely. They can occur during times of low excitation, when
a generator is at low load. It is possible that one pole slip cycle can take a minute to
complete1511. Other pole slipping schemes are specified to operate down to a pole slipping
rate of 0.1 % [60’79]. At a system frequency of 50 Hz, this translates to a single pole slip
cycle taking 20 seconds. It is difficult to give a theoretical basis showing the slowest pole
slip rate which the algorithm could detect because it is dependent upon the generator
characteristics. The setting which determines the limit for detection of slow pole slips is
(AP/At)min. At very slow pole slip rates, the rate of change of power may not be sufficient
to operate the relay. However, although the whole pole slip may take 20 seconds, the part
of the pole slip where the generator passes the CSP, and the algorithm operates will occur

MJ.Checksfield/PhD/University of Bath 1996 Chapter 4, Page 98

at a speed much faster than the mean rate of pole slipping indicates because the generator
rotor accelerates at this point. The algorithm should therefore trip. A simulation has been
performed to confirm this theory, details of which can be found in Chapter 5.

4.5 OTHER ALGORITHM CONSIDERATIONS

This section discusses considerations which are secondary to the theoretical basis of the
algorithm, such as circuit breaker operating duty, and post fault analysis.

4.5.1 Circuit Breaker Switching Duty

Conventional pole slipping schemes such as the single and double blinder impedance
schemes are generally set to trip at a system load angle of 270°. This reduces the
switching duty placed on the generator circuit breaker, since it is exposed to a maximum
voltage across its poles at a load angle of 180°. The application engineer must decide
which is more important, minimising the damage to the generator and disturbance to the
system, or minimising the switching duty on the generator circuit breaker. This is
dependent upon the over-rating of the generating plant and the rating of the local system.

Flurscheim[158] states that an important requirement for inter-connectors between generating

stations is the ability of the circuit breaker to open satisfactorily under asynchronous
conditions. In this case the recovery voltage across an opening circuit breaker may be
much higher than under short circuit conditions and the duty may be more severe in some
respects than the short circuit duty. If no special provision is made to relate the opening
with phase displacement, it must be assumed that a circuit breaker may open at any phase
angle. For a solidly earthed system, the recovery voltage has a maximum value of twice
the phase to neutral voltage at a phase angle of 180°. IEC Document 56 calls for an
asynchronous switching capability of interrupting 25 % rated short circuit current at 2.0
times normal voltage. However, the asynchronous switching duty is decreased because
the severity of the transient recovery voltage is small compared with that arising during
short circuit switching. This is due to the damping provided by the impedances either side
of the circuit breaker. Westinghouse[61] state that if the circuit breaker has insufficient
dielectric strength to withstand the asynchronous recovery voltage, re-ignitions will
continue until a more favourable load angle is reached.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 4, Page 99

The micro-processor relaying environment enables several additional facilities which can
overcome the switching duty problem. The theoretical point where the algorithm trip
signal is reset occurs when the reactive power criterion is no longer satisfied. This results
in only one trip output per pole slip cycle, and for an ideal system, this occurs at a load
angle of 270°. A facility can therefore be provided whereby the tripping is delayed until
this point is reached, thus minimising the duty on the circuit breaker. This can be
provided within the relay as a setting option so that immediate tripping, or delayed
tripping can be chosen. It is possible during loss of excitation conditions, that the reactive
power criterion may not be reset, a timer must therefore be included so that if, after a
specified time, the reactive power criterion had not been set, the trip command would be
issued anyway. A suitable time would be 100 ms after the algorithm originally tripped.
This would allow 90° of load angle increase for a mean pole slipping rate of 2.5 Hz. This
ensures that the algorithm would always trip once it had detected a pole slipping
condition.

4.5.2 Minimum Number of Pole Slips Before Tripping

For some applications, it is desirable to trip the generator only if continued pole slipping
occurs. A facility should therefore be provided within the relay which delays circuit
breaker tripping until a specified amount of pole slips have occurred. For instance, the
decision to trip could be delayed until a second pole slip is detected. This may be useful
for situations where it is possible for the generator to re-synchronise after the first pole
slip. There would be little point in dis-connecting the generator if it is about to re-
synchronise and commence normal stable operation. This decision can therefore be left
to the applications engineer by providing an V pole slips until trip setting.

4.5.3 Differentiation Between 'Steady State' and 'Transient' Pole Slips

The causes of ’steady state' and 'transient' pole slips are quite different, the protection
engineer may therefore find it useful to know what type of pole slip caused the algorithm
to trip. In the case of a 'transient' pole slip, the pole slip is normally due to a short circuit
fault close to the generator. A 'transient' pole slip does not occur due to a failing in the
generator or its control system. The generator can therefore be re-synchronised as soon
as the disturbance which caused the pole slip has been removed. A 'steady state' poleslip
may be caused by a defect in the generator's AVR or governor. It is therefore useful to
know which type of pole slip occurred so that any corrective work required can be
conducted.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 4, Page 100

Section 4.6 states that the power based pole slipping algorithm also detects some loss of
excitation conditions. There is a grey area where generator pole slipping and loss of
excitation overlap because during loss of excitation, the generator can commence
asynchronous operation before the field flux has fully decayed. The remaining field flux
gives the generator a degree of excitation. It can therefore have the appearance of a
generator which has pole slipped due to insufficient excitation. Loss of excitations
occurring from a high initial generator power output can therefore have the same
appearance as a steady state pole slip.

The generator maximum power output monitor provides one means of differentiation
between the types of pole slips. A steady state pole slip or loss of excitation will not
cause a peak in real power output. The Pmax measurement can therefore be used to
differentiate between the two. If Pmax has been updated to greater than the nominal value
of 1.4 * Pt, then the pole slip would be a transient pole slip. The simplest way of
implementing this in the algorithm code would be to check the ratio Pt/Pmax at the instant
of tripping. If this is below the default level of 0.7, then Pmax is not at its nominal level.

Monitoring the magnitude of (AP/At) also provides a complimentary method. Loss of

excitation or steady state pole slips do not cause an abrupt change in real power before
the trip command is issued. If the magnitude of (AP/At) exceeds a specified level before
the algorithm trips, it can be inferred that a transient disturbance has occurred on the
network, and that the pole slip is a 'transient' type. Checking if a variable exceeds a
specified level requires little additional processor time. Such checks therefore do not
impose a large computational burden. A (AP/At) level of five times the rating of the
generator would be suitable for detecting a transient disturbance. If the algorithm is
installed alongside fault detection algorithms, then information from their starter elements
could also be used to improve reliability.

The steady state or transient pole slip information may also be used to implement an
adaptive tripping scheme within the algorithm. Pole slips caused by faults will generally
accelerate quickly into the pole slip. The algorithm typically produces a trip signal at a
load angle of approximately 160° because the CSP does not occur until 140°. This
inevitably means that the circuit breaker is forced to open at load angles of 180°, the
worst place for circuit breaker opening duty. This problem can be overcome by delaying
tripping as described in section 4.5.1 for 'transient' pole slips.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 4, Page 101

For steady state pole slips, the algorithm generally produces a trip signal at much lower
load angles, typically 110°. Since the process of losing synchronism is much slower
during steady state pole slips, the load angle does not increase as quickly as for transient
pole slips. Circuit breaker tripping can therefore be initiated the instant the algorithm has
predicted the 'steady state' pole slip, i.e pole slips which the algorithm detected when
Pmax was at nominal value and (AP/At) had not exceeded its preset level.

4.6 ADDITIONAL ALGORITHM DEVELOPMENTS - MOTOR POLE

SLIPPING PROTECTION

The algorithm has been specifically developed for embedded generators. However, it is
simple to extend its operation to motor pole slipping protection. This is especially
important for hydro generator pumped storage schemes, as they are particularly prone to
pole slipping when operating in the pumping mode[60,80,82,147]. One stability analysis11451
showed that the CCT was 120 ms whilst in the pumping mode. It was also stated that
"generator manufacturers usually don't want to guarantee the asynchronous operation
period of their synchronous machines when they are excited. The machine protection
should consequently detect and prevent such kind of operation".

Synchronous motor protection is based on either impedance measurement, with a time

delayed offset mho characteristic, or a power factor relay, which operates if a very low
power factor occurs with a high current145,1591. The single and double blinder schemes can
also provide motor pole slipping protection.

To adapt the algorithm to enable motor pull out protection, the signs of the real power and
rate of change of power trip settings need reversing. To provide motor and generator pole
slipping protection in one algorithm, the algorithm would need to ascertain whether the
machine was in a motoring or generating condition. This is easily done by monitoring the
sign of the real power trip criterion, Pt. If this is negative, then the machine is motoring
and the (AP/At)min calculation needs changing so that the sign of (AP/At)min is positive.
The maximum power output measuring function would also need changing so that
maximum negative powers are measured. The motor pole slipping algorithm would take
the following form;

M.J.Checksfield/PhD/University of Bath 1996 Chapter 4, Page 102

 >
Q < Qtrip
} continuously
and
} for 1.25 power
P > Pt 4(24)
} system
and } cycles
(AP/At) > (AP/At)min }

where Qtrip remains the same, Pt will be negative instead of positive, and (AP/At)min will
be positive instead of negative.

4.6.1 Performance for Detecting Loss of Excitation

The algorithm will also detect the most damaging of loss of excitation conditions because
essentially, the algorithm is an asynchronous operation detection unit. The algorithm will
successfully detect loss of excitations from high initial powers because the field flux has
not fully decayed at the onset of asynchronous operation. The necessary fluctuations in
real power required to operate the real power and rate of change of power portions of the
algorithm are therefore produced.

In round rotor machines, the difference in Xd* and Xq' results in a transient saliency which
also produces oscillations in power causing the algorithm to trip. At very low initial
output powers. There is not enough variation in real power and hence (AP/At) to satisfy
the (AP/At) trip criterion. To offer full loss of excitation cover from zero to full power,
the algorithm would need supplementing by an additional algorithm which would provide
detection of the low initial power loss of excitations[160].

With salient pole machines, a different problem can occur. The (25) term in equation 3(1)
shows that a significant reluctance torque can be generated, due to the differences in Xd
and Xq. Assuming Xq to equal one half of Xd, a quarter of rated power can be generated
by a salient pole generator with no excitation, whilst still maintaining synchronism. A
large amount of embedded generators have salient pole construction because they can be
operated at much slower speeds, and are easier to manufacture. Even if a salient pole
generator operates synchronously without a field, it still requires immediate disconnection
because it will be consuming a large amount of reactive power which may depress system
voltage. The power based algorithm will not detect this condition, because a loss of
synchronism has not occurred. The algorithm described in reference [160] could be used

M.J.Checksfield/PhD/University of Bath 1996 Chapter 4, Page 103

to supplement the power based algorithm.

If a salient pole generator does operate asynchronously, the power based algorithm will
detect this condition as large oscillations in real power will occur due to the saliency.

4.7 ALGORITHM SETTING PROCEDURE

Table 4-1 provides the equations used for setting the algorithm. It is important to note
that if the algorithm were released commercially, it is envisaged that the only settings
required by the algorithm would be Xq, Xd1and the generator rating. A setting program
within the relay would then be used to calculate the internal settings. More settings are
used at present since at the development stage, greater control of the settings is required
to fine tune the algorithm if required. If settings such as slip were 'hard coded' into the
algorithm code, then any modification would require modifications to the processor board
within the field trial relays, which can not be achieved remotely.

Table 4-1 Summary of Settings for Pole Slipping Algorithm

Default Minimum Maximum

Setting Formula
Value Setting Setting

Qtrip -1/Xq * Sgen Xq = 1.5 pu Xq = 2.5 pu Xq = 0.3 pu

Ptol x % /l 00 * Sgen x= 5% x = 1% x = 20 %

Pfact none 0.9 0.7 1.0

(AP/At)** y %/100 * Sgen y = -25 % -10 % -50 %

Xq = 2.5 pu Xq = 2 pu
(A P/A tU (1/Xq - 1/Xd') * Sgen -3 * Sgen
Xd' = 0.1 pu Xd' = 2 pu

slip none -0.5 % -0.1% -0.5%

lag none 1s 0.1 s 10 s

M.J.Checksfield/PhD/University of Bath 1996 Chapter 4, Page 104

In some instances, the only data available on the generator may be Xd, Xd' and its
rating. If the generator has round rotor construction, then a value of Xq can be

- 0.97 4(25)

This value is an average value, based on data from 18 round rotor generators11091. If
the generator has salient poles, then Xq can be estimated using;

- 0 6 * x d 4(26)

M.J. Checksfi el d/PhD /University of Bath 1996 Chapter 4, Page 105

 c
3
Ui
O
+-> 0.6

0.2

0 1 2 3 4 5 6 7 8 9 10 11 12
Multiples of 50 Hz

Figure 4.1a
Frequency Response of 12 Point, One Cycle Fourier Cosine Filter [74].

anti-aliasing filter response

0.8 combined response of anti-aliasing and fourier cosine filters

c
3
t-c
<D
+-<

0.2

2 4 6 8 10 12
Multiples of 50 Hz

Figure 4.1b
Frequency Response of Anti-Aliasing Filter and Fourier Cosine Filter [74].

M. J. Checksfield/PhD/University of Bath Chapter 4, Fig 4.1, Page 106

K EY

'raw* instantaneous power ---------------- (A P /A t) , (P(n> - P (n - 2 )) method

0 ------------- 9 I ” ^ ( * » > - * - > ) method

a) Real Pow er - Steady State

Algorithm may use incorrect values at top or bottom of
unbalanced signal if moving average filter is not used

10 15 20 25 30 35 40 45 50 55 60 65 70
Time (ms)

b) Real Pow er during unbalanced fault

240
Fau t on Fault off
220
3 200
^

| 180
s ' 160
I 140
Pm
120
100

0 10 20 30 40 50 60 70 80 90 100
Time (ms)

Real Pow er & Rate o f Change o f P ow er

Rate of change of power (W /s)

240 ( P ( n) - P (n -2 )) 2,000
220 (A P /A t)

200
( P ( n) - P ( n - l) )
£ 180 - 2,000

160
-4,000
140
- 6,000
100
Tim e (ms)

Figure 4.2
Graphs Showing Two Point Moving Average Filter Effect on Power Signals,
and Effect o f Two Point Rate o f Change o f Power Calculation M ethod

M. J. Checksfield/PhD/University o f Bath Chapter 4, Fig 4.2, Page 107

 Pt Calculation Function.

^ (check generator has

been operating in the
steady state for the last
"2 * 5 ms* lag” seconds)
(Update Pt setting to
New value)

P t = Pfact * P

Pi = P2
P2 = P (Ensure Pt and therefore
Pmax is not negative)
(Store previous values)

NOTE
1) The Pt calculation function is called every '5 ms * lag' seconds, which for a 'lag1
setting of 200, equates to 1 second. The most recent value of moving average
filtered real power calculated at the time the function is called is set to P'. This
function is separate from the rest of the algorithm and is performed
asynchronously, all of the other algorithm functions are called synchronously
every quarter of a power system cycle. This serves to reduce processor demand.

2) Pi is the value of P used '2*5 ms * lag1seconds ago, whilst P2 is the value of P used
'5ms * lag' seconds ago.

3) Pfact is the scaling factor used to introduce a safety margin in the Pt estimation.
Nominally, it is set to 0.9 in absolute terms. It can be externally set from 0.7 to 1.0
in 0.1 increments.

4) Ptol is the tolerance band which the generator output power must stay within over
the time period' 2* 5ms* lag' seconds for the value of Pt to be updated.

Figure 4.3
The Generator Input Power Estimation Function (Calculates Trip Setting, Pt.)

M. J. Checksfi el d/PhD/University of Bath Chapter 4, Fig 4.3, Page 108

 Pmax Calculation Function.

Start

Yes Pmax = P
Yes Increment 'count Pmax'
Is P > Pmax >= 4 'count Pmax' = 1
'count Pmax'
v (1 cycle)iV 'count Pmax off = 1
No
No
'count Pmax' = 0

' This part of the function sets Pmax to values greater than
nominal if the power output of the generator rises above
V 1.4 * Pt for more than 1 power system cycle.
\ /
This part resets Pmax to its nominal value of 1.4 * Pt
if no update has occurred in the last second

Yes
Increment
'count Pmax off
count Pmax off
V NO /

No
Yes
'count Pmax off
Pmax = 1.4* Pt count Pmax off = 0 > = 200
S. (1 second) V

No

NOTE
1) The function tracks any slow changes in power output from the generator which are greater
than the nominal value of 1.4 * Pt. It will reject any spikes produced by disturbances such
as faults, this is why the moving average filtered power, P has to be greater than Pmax for
one power system cycle before an update occurs. Once an update occurs, the updated
value is 'held' for one second, Pmax is then reset to its nominal value.

2) The variables 'count Pmax' and 'count Pmax off are counters used in the c code to count the
necessary time constraints. Since the function i s called every quarter power system cycle,
a count of 4 is equivalent to one power system cycle.

Figure 4.4
Generator Maximum Output Power Monitor (Pmax Measuring Function).

M. J. Checksfield/PhD/University of Bath Chapter 4, Fig 4.4, Page 109

 The Power Based Pole Slipping Algorithm.

start

NOTE
Cosine fourier filter V & I's at 12 samples
/ cycle. Reduce to 4 samples/cycle
The pole slipping function is
synchronously locked to the power
Calculate real power & apply half cycle system frequency and processed 4
moving average filter. Store present and times per power system cycle.
previous two values of filtered power for
(AP/At) calculation. Store 'raw* value of Pt, the real power trip level is
power for next moving average calculation. calculated Asynchronously every
’5ms * lag' seconds.
Calculate reactive power & apply half cycle
moving average filter. Store 'raw* value of The algorithm only resets once the
power for next moving average calculation. reactive power criterion is NOT
satisfied. This is to prevent
Calculate Pmax 'chatter1of the relay outputs.

Calculate ( A P /A t) m in
Calculate (AP/At)

yes Are P & yes

Is Qtrip
Criterion (AP/At) trip Increment
Criteria Trip Count
Satisfied ?
^Satisfied “L

no no

Is 'TRIP' yes
Reset Trip
Reset Trip Count signal NOT Count
TRUE?

no
Trip Count = 6

yes Trip Count = 6

Trip Count 'TRIP' = TRUE
>=6 ? TRIP SCHEME LOGIC

no
TRIP'= FALSE
BLOCK SCHEME
LOGIC

Figure 4.5
Flowchart Showing Operation of The Pole Slipping Algorithm

M. J. Checksfield/PhD/University of Bath Chapter 4, Fig 4.5, Page 110

 KEY r = 0.1 * X r= X

4.6a - Setting Limits for Pole Slipping Algorithm (Xd' = 0. lXq)

0.16

0.14 power based pole slipping algorithm may

fail to operate for external reactance values
External Reactance, X (p.u)

on this side o f the setting lines.

0.08

0.06

0.04

0.02 _
0.5 0.75 1.0 1.25 1.5 1.75 2.0 2.25 2.5
Quadrature Axis Synchronous Reactance, Xq (p.u.)

4.6b - Setting Limits for Pole Slipping Algorithm ( E = 1.25 pu, r = 0. IX)

0.5 power based pole slipping algorithm may

fail to operate for external reactance values
on this side o f the setting lines.
External Reactance, X (p.u)

0.4

0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1 2.3 2.5
Quadrature Axis Synchronous Reactance, Xq (p.u.)

Figure 4.6
Graphs Showing Maximum Amount o f External Reactance,
for which the Pole Slipping Algorithm Can Still Operate.

M. J. Checksfield/PhD/University o f Bath Chapter 4, Fig 4.6, Page 111

 4.7a - Setting Limits for Pole Slipping Algorithm ( E = 1.50 pu, r = 0. IX)

power based pole slipping algorithm may

fail to operate for external reactance values
on this side o f the setting lines.
External Reactance, X (p.u)

0.2

0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1 2.3 2.5
Quadrature Axis Synchronous Reactance, Xq (p.u.)

4.7b - Setting Limits for Pole Slipping Algorithm ( E = 1.75 pu, r = 0. IX)

power based pole slipping algorithm may

fail to operate for external reactance values
on this side o f the setting lines.
External Reactance, X (p.u)

0.4

0.2

0.5 0.7 0.9 1. 1 1.3 1.5 1.7 1.9 2.1 2.3 2.5
Quadrature Axis Synchronous Reactance, Xq (p.u.)

Figure 4.7
Graphs Showing Maximum Amount o f External Reactance,
for which the Pole Slipping Algorithm Can Still Operate for E = 1.5 and 1.75.

M. J. Checksfield/PhD/University o f Bath Chapter 4, Fig 4.7, Page 112

 Chapter 5
ALGORITHM EVALUATION STUDIES

A variety of methods were used to verify the operation of the algorithm. Initially the
algorithm was proved off-line using data from computer based simulations performed
using POWSIM[161], and data taken from tests performed on a laboratory power system
model. The algorithm was then implemented in a microprocessor relay[162] and tested
using the laboratory power system model. The algorithm was then implemented into a
commercially available relay platform[74] and tested using a Programmable Power System
Simulator (PPSS) [45]. The data files used by the PPSS were created using the PC based
Alternative Transients Program (ATP) version of the ElectroMagnetic Transients
Program (EMTP). An off-line software version of the algorithm was also used to test the
algorithm using results directly from the ATP. This was particularly useful for testing
various power system operating conditions because the only equipment required to
perform the tests was a PC.

The algorithm correctly operated within the commercial relay for all of the tests
conducted. To test the algorithm's operation in an industrial environment, the commercial
relay containing the power based pole slipping algorithm was installed on a 625 kVA
industrial diesel generator. In addition to this, two relays were installed in power stations.
Both relays were installed with modems so that the relays could be interrogated remotely.
These relays were commissioned in October 1996, one on a 26.5 MVA salient pole
generator and the other on a 353 MVA turbo-generator. The field trials were intended to
test the algorithm to ensure that it did not nuisance trip.

5.1 REAL TIM E POW ER SYSTEM SIMULATOR (POWSIM)

POW SIM165,161] is a real-time electro-mechanical transient power system simulator. It uses

a partitioned approach to solve the sets of equations which describe the behaviour of the
machines, control systems and transmission network. The differential equations for each
machine group are solved independently to calculate the current injection into the
transmission network. The algebraic network equations are then solved and the whole
calculation repeated until the required accuracy is obtained.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 5, Page 113

The machine model used is based on Park's transformation^69,1701. Damper winding and
eddy current effects are represented by a short circuited winding on both the direct and
quadrature axis. POWSIM is not suitable for simulating asynchronous operation, since
it assumes that speed changes are negligible. The machine flux linkages are therefore not
altered by a change in speed. The simulations will therefore be acceptable until
synchronism is lost. The algorithm should operate before the loss of synchronism point.
The simulations were therefore suitable for testing the algorithm. POWSIM also cannot
simulate the effects of unbalance because it is a three phase simulation package. However,
it is useful in testing and illustrating the basic theory behind the new algorithm, since the
results it produces are of an 'ideal' nature and contain the fundamental characteristics of
pole slips and power swings.

The classical single machine infinite bus model shown in Figure 5.1 was used. This is
similar to most embedded generation sets because their rating is small compared with the
system infeed. The rating of the 'infinite bus' used was one hundred times the rating of
the machine.

Pole slipping was triggered by applying a three phase fault on the generator transformer
busbar with a fault duration just greater than the critical clearance time (CCT). This is
the maximum fault duration allowable before instability occurs. This technique was also
used to trigger a power swing but for this the fault duration was chosen to be just less
than the CCT. This will produce the largest possible load angle swing for a given
operating condition and therefore provide the most testing conditions for the algorithm.

Both the pole slip and power swing tests were preceded by a reduction in generator
loading from full to half power, this was done in order to test the condition monitoring
section of the algorithm.

5.2 THE LABORATORY MODEL POW ER SYSTEM

Section C l.2 of appendix C contains a schematic diagram of the laboratory power system
model and data on the generator, prime mover and utility infeed. A variety of tests were
conducted to test the algorithm's ability to detect pole slipping and restrain for faults.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 5, Page 114

5.2.1 Pole Slip Tests
For health and safety reasons, it was not possible to apply solid three phase faults at the
terminals of the generator. Pole slipping was therefore induced using other methods.

In one of the pole slipping tests conducted, pole slipping was induced by switching a
resistor in parallel with the field winding. This reduced the field's strength and hence the
synchronising torque which holds the rotor in synchronism with the stator magnetic flux.
The generator excitation was initially set to its full load value, with the generator
operating at a low initial power. The generator's power input was then increased by
increasing the power output of the DC machine. The instant the generator was operating
at full power, the resistor was switched in parallel with the field, causing the generator to
pole slip. Apart from the fluctuations in ammeter readings, a strobe directed at a marker
on the generator's rotor was used to verify that the generator was pole slipping. The test
was performed with a local load of 4 kW. These tests served to verify the operation of the
condition monitored real power trip setting, Pt as well as the algorithm. Due to the
limited length of the disturbance recording equipment, the time setting, 'lag', in the
algorithm's prime mover power estimation function was set to 0.5 seconds instead of the
normal 1 second. This was necessary so that the algorithm could be seen to operate
correctly over the time span of the disturbance record.

Pole slipping was also induced by a sudden increase in the input torque to the generator.
This type of poleslip will be referred to as an 'over torque' pole slip. Such a pole slip
could happen if a microprocessor prime mover control system failed or due to operator
error. A variety of tests were performed, at varying levels of excitation and local load.
In addition tests were performed with another generator operating on the same bus. Pole
slipping and loss of excitation tests were performed on this generator to check whether
the algorithm remained stable during disturbances occurring on adjacent generators. Table
5-1 contains details of the tests conducted.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 5, Page 115

 Table 5-1 Pole Slipping Tests Performed on 5 kVA Generator.

OT - Over Torque, PS - Pole Slip, LOE - Loss of Excitation.

Test type of local Excitation

letter test load

C Weak 4 kW PS caused by inserting a resistor

Field PS in parallel with field

E OT PS 4 kW low ( 3 Amps)

F OT PS 4 kW medium ( 6 Amps)

G OT PS 4 kW high ( 9 Amps)

H OT PS none low ( 3 Amps)

I OT PS none med ( 6 Amps)

J OT PS none high ( 9 Amps)

K Adjacent 4 kW high ( 9 Amps), protected

OT PS machine in steady state.

L Adjacent 4 kW high ( 9 Amps), protected

LOE machine in steady state

5.2.2 Fault Tests

The data for the fault tests was taken from tests which Barrett[104] performed when
developing a micro-processor based loss of grid relay. To limit the fault current, a 'fault
resistance' was used at the fault point. Three phase to earth, three phase, two phase to
earth, two phase, and phase to earth faults were applied at position CB2 in Figure C4 of
Appendix C. The tests were performed with and without the generator neutral solidly
earthed. Embedded generators generally operate with an isolated neutral, this is a
requirement of the REC's single point earthing regulation.

One three phase fault was of particular interest as it caused the generator to pole slip,
giving data which tested the algorithm's ability to restrain for faults and trip for a pole slip
shortly after. Table 5-2 contains details on the fault tests.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 5, Page 116

 Table 5-2 Fault Tests Performed on 5 kVA Generator.

Test Fault Type Generator Neutral Generator Initial

Connection Operating Point

1A A-E Earthed P = 3.5 kW, Q = 250 VAr

IB A-E Isolated P = 3.75 kW, Q = 250 VAr

2A A-B Isolated P = 3.6 kW, Q = 500 VAr

3A A-B-E Earthed P = 3.5 kW, Q = 0 VAr

3B A-B-E Isolated P = 3.7 kW, Q = 400 VAr

4A A-B-C Isolated P = 4 kW, Q = 0 VAr

5A A-B-C-E Earthed P = 3.9 kW, Q = 0 VAr

5B A-B-C-E Isolated P = 3.8 kW, Q = 0 VAr

5.3 PROGRAMMABLE POW ER SYSTEM SIMULATOR (PPSS) TESTS

The power system model used to generate the PPSS test data was taken from an EPRI
transient stability model[163] of a 200 MVA generator. The commercial relay
implementation of the algorithm was used for the tests. Appendix E contains details on
the techniques used to install the algorithm in a microprocessor relaying environment.
Appendix C, section Cl contains details of the power system, AVR and governor models
used in the simulations. The ATP was used to simulate pole slips, power swings, faults,
circuit breaker pole scatter, operation at machine load angles in excess of 90°, loss of
excitation, and loss of prime mover conditions. Details of all of the tests conducted can
be found in Tables 5-3, 5-4, and 5-5.

The performance of the conventional pole slipping schemes described in chapter 2 was
compared with the power based pole slipping algorithm. The impedance based relays
were modelled using a PC. Section 5.7 contains details of the techniques used to model
the impedance relays.

M.J.Checksfi el d/PhD/University of Bath 1996 Chapter 5, Page 117

 Table 5-3 Details of PPSS Pole Slipping Simulation Tests

Test Name
Description of Pole Slipping Test
(Code)

PSL1NN Pole Slip (following a 940 ms, 3 Phase to Earth Fault), no governor or
(Al) AVR modelled, 20 % initial load, Fault on point staggered by 10 ms for
each phase.

PSL2NN Pole Slip (following a 650 ms, 3 Phase to Earth Fault), no governor or
(A2) AVR modelled, 50 % initial load.

PSL3NN Pole Slip (following a 350 ms, 3 Phase to Earth Fault), no governor or
(A3) AVR modelled, 100 % initial load

PSL1GN Pole Slip (following a 1180 ms, 3 Phase to Earth Fault), governor
(A4) modelled, AVR not modelled, 20 % initial load

PSL2GN Pole Slip (following a 950 ms, 2 Phase to Earth Fault), governor
(A5) modelled, AVR not modelled, 50 % initial load

PSL3GN Pole Slip (following a 950 ms, Phase to Phase Fault), governor
(A6) modelled, AVR not modelled, 100 % initial load

PSL1NA Pole Slip (following a 1170 ms, 3 Phase to Earth Fault), no governor
(A7) modelled , AVR modelled, 20 % initial load

PSL2NA Pole Slip (following a 750 ms, 3 Phase to Earth Fault), no governor
(A8) modelled, AVR modelled, 50 % initial load

PSL3NA Pole Slip (following a 290 ms, 3 Phase to Earth Fault), no governor
(A9) modelled, AVR modelled, 100 % initial load

PSL1GA Pole Slip (following a 1170 ms, 3 Phase to Earth Fault), governor &
(Bl) AVR modelled, 20 % initial load

PSL2GA Pole Slip (following a 650 ms, 3 Phase to Earth Fault), governor &
(B2) AVR modelled, 50 % initial load

PSL3GA Pole Slip (following a 330 ms, 2 Phase to Earth Fault), governor &
(B3) AVR modelled, 100 % initial load

PSL1GAF Pole Slip (due to faulty avr action - Vf reduced by 95 %), governor &
(B4) AVR modelled, 20 % initial load

PSL2GAF Pole Slip (due to faulty avr action - Vf reduced by 50 %), governor &
(B5) AVR modelled, 50 % initial load

PSL3GAF Pole Slip (due to faulty avr action - Vf reduced by 20 %), governor &
(B6) AVR modelled, 100 % initial load

M.J.Checksfield/PhD/University of Bath 1996 Chapter 5, Page 118

 Table 5-4 Details of PPSS Stable Power Swing Simulation Tests

Test Name
Description of Stable Power Swing Test
(Code)

PSW1NN Stable Power Swing (following a 900 ms, 3 Phase to Earth Fault), no
(B7) governor or AVR modelled, 20 % initial load.Fault on phases staggered
by 10 ms, fault off by 20 ms.

PSW2NN Stable Power Swing (following a 610 ms, 3 Phase to Earth Fault), no
(B8) governor or AVR modelled, 50 % initial load

PSW3NN Stable Power Swing (following a 263 ms, 3 Phase to Earth Fault), no
(B9) governor or AVR modelled, 100 % initial load

PSW1GN Stable Power Swing (following a 1155 ms, 3 Phase to Earth Fault),
(Cl) governor modelled, AVR not modelled, 20 % initial load

PSW2GN Stable Power Swing (following a 810 ms, 2 Phase to Earth Fault),
(C2) governor modelled, AVR not modelled, 50 % initial load. Fault off
phases staggered by 5 ms.

PSW3GN Stable Power Swing (following a 820 ms, 2 Phase Fault), governor
(C3) modelled, AVR not modelled, 100 % initial load. Fault off phases
staggered by 10 ms.

PSW1NA Stable Power Swing (following a 1150 ms, 3 Phase to Earth Fault), no
(C4) governor modelled, AVR modelled, 20 % initial load

PSW2NA Stable Power Swing (following a 630 ms, 3 Phase to Earth Fault), no
(C5) governor modelled, AVR modelled, 50 % initial load

PSW3NA Stable Power Swing (following a 280 ms, 3 Phase Fault), no governor
(C6) modelled, AVR modelled, 100 % initial load

PSW1GA Stable Power Swing (following a 1150 ms, 3 Phase to Earth Fault),
(C7) governor & AVR modelled, 20 % initial load

PSW2GA Stable Power Swing (following a 630 ms, 3 Phase to Earth Fault),
(C8) governor & AVR modelled, 50 % initial load

PSW3GA Stable Power Swing (following a 320 ms, 2 Phase to Earth Fault),
(C9) governor & AVR modelled, 100 % initial load

M.J.Checksfield/PhDAJniversity of Bath 1996 Chapter 5, Page 119

 Table 5-5
Details of PPSS AVR Assisted High Load Angle, Loss of Field and
Loss of Prime Mover Simulation Tests

Test Name Description of Loss of Excitation or Loss of Prime Mover Test

(Code)

AVR4NA Operation at 8 > 90°, no governor modelled , AVR modelled, 87.5

(Dl) to 44 % initial load. Tdo' reduced from 6.2 to 4.2 s, AVR gain
reduced from 400 to 200.

AVR5NA Operation at 8 > 90°, no governor modelled , AVR modelled, 50

(D2) to 0 % initial load. Tdo' reduced from 6.2 to 4.2 s, AVR gain
reduced from 400 to 200.

AVR6NA Operation at 8 > 90°, no governor modelled , AVR modelled, 87.5

(D3) to 70 % initial load. Tdo' reduced from 6.2 to 4.2 s, AVR gain
reduced from 400 to 200.

LOF1GA Loss of Field , governor modelled, Field voltage reduced to 0,

(D4) 20 % initial load. Tdo' reduced from 6.2 to 2.2 s.

LOF2GA Loss of Field , governor modelled, Field voltage reduced to 0,

(D5) 50 % initial load. Tdo' reduced from 6.2 to 3.8 s.

LOF3GA Loss of Field , governor modelled, Field voltage reduced to 0.

(D6) 100 % initial load.

LOP1NA Loss of Prime Mover , AVR modelled, Input power reduced to

(D7) -3 % rated power, 20 % initial load

LOP2NA Loss of Prime Mover , AVR modelled, Input power reduced to

(D8) -3 % rated power, 50 % initial load

LOP3NA Loss of Prime Mover , AVR modelled, Input power reduced to

(D9) -3 % rated power, 100 % initial load

Note: For the tests involving operation at load angles in excess of 90°, the forward path
gain of the AVR had to be reduced from 400 to 200 in order to improve the
damping. In addition, the generator open circuit time constant, Tdo' was changed

M.J.Checksfield/PhD/University of Bath 1996 Chapter 5, Page 120

 from 6.2 to 4.2 seconds. This was necessary to improve the speed with which the
AVR was able to cause changes in the generator's field excitation.
Note: The maximum time which the PPSS can simulate is 6 seconds. The time constant,
Tdo' was therefore adjusted for the low initial power loss of excitation tests so that
the generator lost synchronism within the 6 second time period.

The PPSS uses the data from ATP simulations to re-create the voltage and current
waveforms which would normally come from the generator VTs and CTs. These
waveforms are then injected into the relay to test the algorithm. Reference [45] contains
details on the PPSS which was used for the tests.

5.3.1 Suitability of the ATP Simulation Package for Generator Simulations

This section discusses the suitability of the ATP for performing simulations which will
test the power based algorithm. There are many different methods and models available
for simulating synchronous machines, and it is important to choose an approach which
will model the synchronous generator accurately as it pole slips. It is also important to
model the effects of the generator control systems. In many studies, it is assumed that the
effect o f the generator governor can be neglected because the time constants involved are
several seconds[,08]. This assumption is invalid. The stator, field, damper windings, AVR
and governor all need including in the model. This section will not give a detailed
discussion of synchronous machine simulation, this subject has been well covered in many
other texts[37,103,109,123’164,165]. This section will discuss the synchronous machine model used,
and the reasons for its use.

The model used[166,167,168] assumes that the electrical part of the machine has three armature
windings, connected 120° apart, one field winding, which produces flux in the direct axis
when excited, one hypothetical winding on the direct axis which represents damper bar
effects, one hypothetical winding on the quadrature axis which represents damper bar
effects, and one hypothetical winding on the quadrature axis which represents slowly
changing fluxes in the quadrature axis, which are produced by deeply flowing eddy
currents (negligible in salient pole machines). The model is in the form of the general
synchronous machine equations derived by Park[169,170]. These hold for any variation of
the speed of the machine, and any kind of transient change in currents and voltages11031.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 5, Page 121

The ATP model was originally developed to investigate sub-synchronous resonance (SSR).
The electrical and mechanical shaft torsional dynamics are therefore represented. One
paper11681 states with reference to the application of the model that, ’’Other typical
applications include transmission line re-closure, independent pole switching, load
rejection, loss of synchronism and multi-machine interaction problems”.

To simulate a pole slipping machine, it is necessary to take into account the effects of
changes in rotor speed. Changes in rotor speed cause two main effects. Firstly a change
in rotor speed will result in a change in the flux linkage, this change therefore needs
including in the model. Secondly, as the rotor speed departs from synchronous speed,
induction machine effects also need modelling because induced currents will flow in the
field and damper windings. Additionally, flux cannot be assumed constant, since the
simulations are of a long duration and embedded generators can have short time constants.
The model used by the ATP includes all of these effects, as well as including saturation,
and modelling of AVR's and governors. It is therefore suitable for performing simulations
which can be used to test the algorithm.

Examples of the ATP simulation files can be found in Appendix F.

5.3.2 Simulation of Overhead Lines and Cables

Another modelling consideration is the way in which lines, transformers and sources are
simulated. As shown in Figure 4.1b, the relay is sensitive to frequencies below 200 Hz.
Effects such as travelling waves do not require representation in the power system model,
since they will not be seen by the relay. In addition, embedded generators are generally
connected to short cables or overhead lines. The lumped parameter mutually coupled pi
model of a transmission line can therefore be used[37,164,166]. It is impractical to use a
distributed parameter model for such short lines because the travel time of such a line
would require an unsuitably small simulation time-step. The important consideration when
modelling lines is to adequately represent the positive, negative and zero phase sequence
components so that balanced and unbalanced faults can be simulated. The ATP type 1,2,3
mutually coupled RLC pi model[166] is suitable for representing unbalanced faults. The
parameters can be entered manually, or derived by the ATP cable or line constants
program.

M.J.Checksfi el d/PhD/University of Bath 1996 Chapter 5, Page 122

5.3.3 Simulation of Transformers
Transformers have been represented in two different ways. In the PPSS simulations, the
transformer model was taken directly from an EPRI transient stability study model[163].
This uses the ATP internal saturable transformer model called TRANSFORMER, which
models transformers using the classical model shown in Figure 5.2. This uses leakage
reactance and resistances on both the primary and secondary side, an ideal transformer,
and a resistor in parallel with a saturable non-linear reactor representing the magnetisation
circuit. This model will simulate the losses found in a transformer as well as the zero
sequence trap which occurs in delta connected windings. It is therefore suitable for
simulation of unbalanced faults.

The second technique uses the ATP supporting routine 'BCTRAN' to derive a six by six
impedance matrix which represents the transformer. This uses standard transformer test
data and is therefore a convenient way of converting manufacturer's data into a usable
format. This is the recommended approach for simulating the zero sequence capabilities
of transformers^631. Examples of BCTRAN simulation files can be found in Appendix F.

5.3.4 Simulation of Source Impedances

It is impractical to represent in detail, the entire power system. It is necessary to derive
a thevenin equivalent of the system at some point and it is important to properly represent
the source impedance which represents the 'utility' when performing embedded generation
simulations11711. If the three phase and phase to earth fault currents for that and only that
part of the system are known, then a thevenin equivalent impedance can be derived as
follows[1711;
Z. = !----- 5(1)
-

1 V a u lte d

where Z x is the per unit positive sequence impedance, and Ifault(3ph) is the per unit three
phase fault current o f the section of power system to be represented. The per unit zero
sequence thevenin impedance, Zc is given by;

Z„ = ------ - 5(2)
1 ph -E ) 3 ph)

where Ifault(1 ph_E) is the per unit single phase to earth fault current of the section to be
represented. This approach can be used only if there is no significant capacitance or
transmission lines at the bus which is being converted to a thevenin equivalent.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 5, Page 123

5.4 625 kVA INDUSTRIAL DIESEL GENERATOR FIELD TESTS

The 625 kVA diesel generator used to perform field tests is a salient pole machine.
Appendix C contains the data available on the generator and utility connection. The
commercial relay implementation of the algorithm was installed for the tests. An
additional relay was installed in parallel with the pole slipping relay so that raw samples
of voltage and current could be taken. This enabled comparison between conventional
pole slipping schemes and the new approach. Pole slipping, loss of excitation, steady state,
and local load change tests were performed, the details of which are given in tables 5-6,
5-7, and 5-8.

Pole slipping could not be induced by placing short circuit faults close to the generator,
as this was not permitted. Instead, pole slipping was induced in two ways. One method
involved weakening the field excitation until the generator lost synchronism, the other
involved suddenly increasing diesel engine output torque until a pole slip occurred. This
type of pole slip could occur due to control system mal-function, or operator error. There
was also a 593 kVA diesel generator next to the test machine, this was forced to pole slip
and to lose excitation in order to test the algorithm's ability to restrain for nearby
disturbances.

Loss of excitation was invoked by disconnecting the field supply. This effectively short
circuited the field winding. The largest local load which could be switched on and off at
one discrete point was a compressor. This caused an abrupt load change when switched
on, but not when switched off. Large local load disconnections therefore could not be
performed. The compressor rating was 120 HP.

The relay was operational throughout the whole test period. Generator run-up and
synchronisation tests were effectively performed, though no disturbance records were
taken.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 5, Page 124

 Table 5-6
Pole Slipping Tests Conducted on 625 kVA Diesel Generator.

Test Protected Adjacent Notes

Generator Generator
(625 kVA) (593 kVA)

Al Pole Slip Not in Operation Tried to induce pole slip by

(100 kW) weakening the field. But field
would not weaken enough.

A2 Pole Slip Not in Operation. Tried to induce pole slip by

(250 kW) weakening the field. But field
would not weaken enough.

A3 Pole Slip Not in Operation. Induced pole slip by weakening

(500 kW) the field.

A4 Steady State. Pole Slip Induced pole slip by weakening

(100 kW), (474 kW). the field.
(-200 kVAr)

A6 Pole Slip . Not in Operation. Tried to induce pole slip by

(100 kW), suddenly increasing input power.
(-200 kVAr)

Al Pole Slip Not in Operation. Induce pole slip by suddenly

(0 kW) increasing input power.
(- 200 kVAr)

Note: The figures given in brackets are the initial generator output powers. The first
figure is the real power output, the second the reactive power output. The reason
pole slips did not occur in tests A l, A2 and A6 is that the control which adjusted
the field excitation could only be decreased to a certain level. An open circuit,
un-synchronised test was performed to see what this level corresponded to. In the
middle position, the generator open circuit terminal voltage was 440 volts. This
is above the no load value of excitation, which should produce a voltage of 415
volts. With the excitation set to a minimum, the open circuit terminal voltage was

M.J.Checksfiel d/PhD/University of Bath 1996 Chapter 5, Page 125

 380 volts, i.e 8 % less than the no load value. This limit combined with the effect
of the generator AVR meant that pole slipping could not be induced for low levels
of generator loading
Table 5-7
Loss of Excitation Tests Conducted on 625 kVA Diesel Generator.

LOE - Loss of Excitation

Test Protected Generator

B1 LOE from 20% load. (100 kW), under-excited ( -200 kVAr)

B2 LOE from 20% load. (100 kW), over-excited ( 200 kVAr)

B3 LOE from 50% load. (250 kW), under-excited ( -160 kVAr)

B4 LOE from 50% load. (250 kW), over-excited ( 200 kVAr)

B5 LOE from 100% load. (500 kW), under-excited ( -130 kVAr)

B6 LOE from 100% load. (500 kW), over-excited ( 200 kVAr)

Table 5-8
Steady State and Local Load change Tests Conducted on 625 kVA Diesel Generator.

Test Protected Generator (625 kVA) Notes

Cl Steady State. 30 % load Steady State

(150 kW), under-excited (-200 kVAr) Operation.

C2 Steady State. 30 % load Steady State

(150 kW), level-excited (0 kVAr) Operation.

C3 Steady State. 30 % load Steady State

(150 kW), over-excited (200 kVAr) Operation.

C4 Steady State, 100 % load (500 kW), Switch compressor

under excited (-130 kVAr). on.

C6 Steady State, 20 % load (100 kW), Switch compressor

under excited (-200 kVAr) on.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 5, Page 126

5.5 STEAM TURBINE AND HYDRO GENERATOR FIELD TRIALS

Two relays were also installed in power stations in a purely supervisory role in order to
test whether the algorithm would operate correctly. Capturing a pole slip on one of these
generators would be a considerable bonus because the trials were primarily conducted to
ensure that the algorithm did not nuisance trip for power system conditions other than pole
slipping or loss of excitation.

The relays had been installed for three months at the time of writing this thesis, it is
intended that they will remain connected to the generator for a period of one year. Both
of the relays had modems connected to their serial communications ports so that they
could be interrogated remotely. As well as the pole slipping algorithm, a voltage
controlled over-current protection function1741 withsensitive settings wasenabledso that
the relays disturbance recorder would be triggered for any power system faults. The
disturbance recorder was also set to trigger if the algorithm tripped, or if the reactive
power criterion was satisfied.

5.5.1 Steam Turbine Generator Set

The turbo-generator used for the field trials is a coal fired 353 MVA machine. It
generated at 17 kV giving a full load current of 12000 Amps. The generator was a round
rotor type, with the following parameters;

Xd = 2.52 pu
Xq = 2.40 pu
Xd' = 0.268 pu.

5.5.2 Hydro Turbine Generator Set

The hydro-generator used for the field trials is a 26.5 MVA machine. It generated at
11 kV giving a full load current of 1390 Amps. The generator was a salient pole type.
The only data on the machine available was measured at 50 % rated volts;

Xd = 1.148 pu
Xq = 0.700 pu
Xd* = 0.358 pu.

M.J.Checksfi el d/PhD /University of Bath 1996 Chapter 5, Page 127

The reactances at 100 % voltage will be lower than at 50 %, since synchronous machines
are designed to operate in the saturated region. The amount by which these reactances
change is largely dependent upon the design of the machine. In one machine, the
reactances were typically 10 % larger at 50 % volts compared to rated volts[130]. Other
studies show that this figure is typically between 10 % and 20 % [129]. Assuming a higher
percentage change will result in a more secure reactive power trip level because the
magnitude of Qtrip becomes larger as Xq decreases.

A 10 % drop was assumed for this particular machine, the resultant figures used for
deriving the algorithm settings were therefore;

Xd = 1.03 pu
Xq = 0.63 pu
Xd' = 0.322 pu.

The algorithm settings were derived for both machines as defined by table 4-1 of
Chapter 4.

5.6 ATP SIMULATIONS OF EMBEDDED GENERATORS

Some additional simulations were required so that aspects of generator operation which
had not been addressed by any of the other tests were investigated. Section 3.3.2 of
Chapter 3 showed that the 'ideal' sinusoidal power-load angle characteristics predicted if
a constant field flux linkage was assumed may not occur with embedded generators, due
to a very low value of time constant, Tdo1. Stable power swing and pole slip simulations
with low values of Tdo1 were therefore performed in order to verify that the algorithm
operated correctly. The 750 kVA generator described in table Cl of Appendix C was used
for these simulations. A slow AVR model was used because a fast one would have offset
the effect o f a short value of Tdo'. No governor was modelled in these tests, it was
assumed that a slow acting type was used. A slow acting governor would not respond in
the short time window of the simulation.

A fast diesel governor was also simulated to ensure that the algorithm did not nuisance
trip. This test was important because the algorithm assumes that the generator input power

M.J.Checksfi el d/PhD/University of Bath 1996 Chapter 5, Page 128

remains constant during a disturbance. A 4 % droop governor was represented using a first
order lag, with a time constant of 40 ms. The diesel engine was modelled as a 15 ms time
delay1172’173,174,175'176’177’178,179,1801. This is the fastest prime mover/governor combination likely
to occur in practice, and is appreciably faster than steam turbines, gas turbines, and hydro
turbines1180,181,182,163]. It therefore presents the most onerous conditions for testing the
algorithm. Appendix F contains an example of the ATP file used in the simulation.

Section 4.6 of Chapter 4 stated that theoretically, the algorithm should detect synchronous
motor pole slipping with little modification. A hydro-generator in pumping mode was
therefore simulated to see if the modified algorithm of equation 4(22) would work. A fast
hydro governor was simulated11631, using the same 750 kVA generator model as in the
above tests. A stable power swing and a pole slip were simulated. The CCT for such a
generator was 75 ms, which shows the vulnerability to instability that hydro-generators
have when they are in their pumping mode.

The AVR was disabled from the hydro-generator model, and another pole slip simulated
by decreasing the machines terminal voltage by 80 %. This is a steady state motor pole
slip.

A simulation was also performed using the PPSS test system. The 200 MVA generator
was initially operating under-excited, with an initial output power of 20 %. The
generators excitation was then reduced so that a pole slipping rate of approximately two
pole slips/minute was achieved. This tested the algorithm against slow pole slips, and
conformed the theory of section 4.4.3.

Table 5-9 contains details of the tests conducted.

M.J.Checksfiel d/PhD/University of Bath 1996 Chapter 5, Page 129

 Table 5-9
Details of the Additional ATP Simulations Performed to
Test the Power Based Pole Slipping Algorithm.

Test Type Governor AVR Generator Details /

Type of Disturbance

LT1 Stable Swing None Slow Tdo' set to 1.0 s / 170 ms 3 Ph. fault

LT2 Stable Swing None Slow Tdo' set to 2.0 s / 200 ms 3 Ph. fault

LT3 Stable Swing None Slow Tdo' set to 3.0 s / 210 ms 3 Ph. fault

LTP1 Pole Slip None Slow Tdo' set to 1.0 s / 200 ms 3 Ph. fault

LTP2 Pole Slip None Slow Tdo' set to 2.0 s / 230 ms 3 Ph. fault

LTP3 Pole Slip None Slow Tdo' set to 3.0 s / 250 ms 3 Ph. fault

FG1 Stable Swing Fast Slow Tdo1 set to 1.0 s / 260 ms 3 Ph. fault

FG2 Stable Swing Fast Slow Tdo' set to 3.0 s / 370 ms 3 Ph. fault

MOT1 Stable Swing Hydro Slow Tdo' set to 3.0 s / 70 ms 3 Ph. fault

MOT2 Pole Slip Hydro Slow Tdo' set to 3.0 s / 80 ms 3 Ph. fault

MOT3 Pole Slip Hydro None Tdo' set to 3.0 s /

Infinite bus voltage reduced to 80 %.

SLW Pole Slip PPSS PPSS generator at 20 % initial load / pole

slipping rate of 2 cycles/minute.

5.7 PC BASED SIMULATION OF IMPEDANCE RELAYS

The apparent impedance at the generator terminals was calculated using half cycle moving
average filtered, three phase values of real power, reactive power and voltage. Before the
variables were calculated, the voltages and currents were pre-filtered using a full cycle
fourier cosine filter, as described by equation 4(9). For the analysis of the 200 MVA
generator PPSS tests, the processing rate used was 1000 Hz. For the 625 kVA diesel

M.J.Checksfield/PhD/University of Bath 1996 Chapter 5, Page 130

generator tests, a processing rate of 12 samples per power system cycle was used. The
'three phase' voltage was calculated using;

■ fta ~ + V b ~ + V c* 5(3)

The real and reactive power was calculated using the three phase technique described in
Chapter 4. All three signals were then half cycle moving average filtered as described in
Chapter 4, to remove unbalance and smooth the calculated value of impedance. Apparent
resistance, R, and reactance, X were then calculated as follows;

* = P -*L L 5(4)
P 2 +Q 2

Q *V2
X =
P 2 +Q

The proof of these equations is simple, the denominator is equal to the apparent power
squared , S2. The term V2/S gives the impedance, Z, whilst the term P/S is equal to
cos(0), and Q/S is equal to sin(0).

The mho, loss of field, single blinder and double blinder relays were all simulated using
the logic described in Chapter 2. An example of the C code used to simulate the
commonly used single blinder scheme is given in Appendix F.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 5, Page 131

 Generator NOTE: Generator per unit on Generator Base
All other values in per unit are on a 100 MVA base
588 MVA
Xd = 2.77 pu Generator
Xq = 2.63 pu Transformer
Xd' = 0.305 pu r = 0.038 %pu CEGB
H = 4.46 s X = 2.7 %pu
AVR
Generator Transmission
IEEE Type 1
Busbar Ka = 200
Busbar
AVR Ta = 0.5
IEEE Type 1 Ke= 1
Ka = 30 Te = 0 s
Ta = 0.3 i K f= 0.05
r = 0.135 %pu a
Ke= 1 J Tf =0.5 s
X = 1.752 %pu a
Te = 0 s B = 73.5 %pu 35
o
Governor
a\
Kf = 0 ( split between E 4 % droop
ir j
both ends of Time constant = 0.5
Governor line) f— i

4 % droop Governor
Time constant = 0.3 s 4 % droop
Time constant = 0.5 s
Transmission

Busbar
CEGB Infinite Bus
100 * 588 MVA Generators

Figure 5.1
Classical One M achine/ Infinite Bus Power System Model Used in POWSIM

ideal

R2 L2
winding 1 winding 2

Figure 5.2
Classical Transformer Model used by ATP's 'TRANSFORMER' Component.

M. J. Checksfield/PhD/University o f Bath Chapter 5, Fig 5.1, Page 132

 Chapter 6
RESULTS OF ALGORITHM EVALUATION TESTS

In the tests conducted, the algorithm successfully operated for all pole slipping conditions,
and restrained for all other power system conditions apart from loss of excitation. The
algorithm provided limited loss of excitation protection, the amount of cover depending
upon the initial operating power and the type of generator. The concept of tripping time
cannot be applied to pole slipping in the same way as it is for faults, since the point at
which the algorithm trips depends upon the load angle. Tripping times are therefore
expressed in terms of load angle, where load angle information is available, or generator
real power output. When the real power reverses sign during a pole slip, the load angle
will equal 180° if asynchronous effects are neglected.

6.1 POWSIM TEST RESULTS

The algorithm successfully detected the pole slip, also successfully categorising it as a
'transient' pole slip. If delayed tripping had been required due to circuit breaker operating
duty, then tripping would have been delayed until the load angle was approaching the
optimum switching point.

The algorithm successfully restrained for the stable power swing test.

6.1.1 Response to Pole Slip Test

Figure 6-1 shows the algorithm's response to the simulated pole slip. The algorithm
predicted that a pole slip would occur at a load angle of 168°, i.e. just before the machine
actually slipped. The reactive power criterion was satisfied first, followed by the rate of
change of power criterion, then the real power criterion. 25 ms after all three criteria had
been satisfied, a trip signal was issued.

The reactive criterion was satisfied the instant the fault was cleared, this indicates that the
load angle was greater than 90°, the load angle plot confirms this. The fact that the rate
of change of power criterion was satisfied next indicates that the generator rotor was

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 133

travelling above synchronous speed, i.e the load angle was increasing. This is also
confirmed by the load angle plot, which wraps around at load angles of +/- 180°. Finally
with both the Qtrip and (AP/At)min criterions satisfied, the real power falls below its
criterion, Pt, indicating that the CSP had been reached and no further decelerating energy
was available. Instability was therefore unavoidable and the algorithm tripped.

The 'Q criterion satisfied' curve shows the point at which the trip signal would have been
reset, this is at 1 second on the record. The maximum output power measuring function,
Pmax was greater than its nominal value at the time of tripping, the ratio Pt/Pmax was
less than 0.6 and the pole slip was classified as a 'transient' pole slip. The rate of change
of power signal also reached a magnitude of 50,000 MW/s before the trip output, this
served as another indicator that the pole slip was a 'transient' type.

If tripping was delayed until reset of the reactive power criterion, due to concern over the
generator circuit breaker operation duty, then the 'trip circuit breaker' command would
have been issued at a load angle of -100°, i.e as the generator was approaching the 'in
phase' position. This would have been the ideal time for the circuit breaker to commence
arc interruption, and would have placed minimum duty on the circuit breaker.

The algorithm successfully detected the second pole slip, at a time of 1.4 seconds.
However, by this time, the speed of pole slipping had considerably increased and the
algorithm tripped at a load angle of 188°.

6.1.2 Response to Stable Power Swing Test

The simulated power swing which produced the largest possible stable load angle swing
of 152° is shown in Figure 6-2. For the time period between 0.3 and 0.8 seconds the
reactive power is less than Qtrip, indicating that the load angle is greater than 90°, as the
load angle plot confirms. However, during this time, the real power does not satisfy its
trip setting Pt, the algorithm therefore restrained from operation. This condition indicates
that decelerating energy is still available, and that the CSP has not yet been reached. The
section marked AB represents the generator swinging up to its maximum load angle. The
rate of change of power curve is zero at point 'B', showing that the rotor is once again at
synchronous speed. This is confirmed by the load angle plot which is at a maximum at
this point. The section BC occurs when the generator rotor backswings towards a stable
operating point. In doing so, the peak of the real power load angle curve is crossed once

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 134

more, resulting in a rise in the power output of the machine between points B and C. As
the load angle swings back from its maximum value, the reactive power rises above Qtrip.
This causes the algorithm to restrain and ensures that no nuisance tripping occurs. The
algorithm therefore restrained during the recoverable power swing.

6.2 LABORATORY MODEL POWER SYSTEM TEST RESULTS

The algorithm successfully operated for all of the pole slipping tests, whilst restraining for
all of the fault tests.

6.2.1 Weak Field Pole Slip - Test ’C’

Figure 6.3 shows the pole slipping algorithm response to the pole slip caused by insertion
of a resistor in parallel with the generator field winding. The time limit on the
disturbance recorder meant that the algorithm time setting, lag, which is used in the prime
mover output power measuring function was set to 0.5 seconds rather than the 1.0 second
recommended in table 4-1 of chapter 4. This was necessary so that the algorithm would
update its settings over the duration of the disturbance record.

Plots a and b of Figure 6.3 show that the generator was initially operating overexcited at
low power. The (AP/At)min curve in plot c shows that the (AP/At)min setting was initially
determined by (AP/At)fact, because of the low real output power of the generator. This
ensures that there is always an adequate margin for error in the rate of change of power
trip criterion. For the 5 kVA generator, this value was set to 25 % of the generator rating,
as defined in table 4-1.

At a time of 1 second on the record the power input to the generator was increased from
150 W to 2.75 kW. The maximum output power monitor, Pmax tracks this increase in
output power, causing a corresponding increase in (AP/At)min. The trip setting (AP/At)min
gradually increased in magnitude as Pmax increased because a high value of generator
output power results in high values of (AP/At), a less sensitive trip level should therefore
be used. This change in Pmax also caused the ratio Pt/Pmax to fall below 0.6, the
transient saliency factor (AP/At)^ was therefore included in the (AP/At)min calculation, de­
sensitising the algorithm further.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 135

At 3 seconds, the real power trip setting, Pt is updated to the new prime mover input
power. Shortly after this, Pmax is updated to its default level of 1.4 * Pt. This causes the
(AP/At)min level to become more sensitive, because the transient saliency scaling factor,
(AP/At),^ is no longer included in the calculation because the generator is in a steady
state.

At 3.2 seconds, the resistor was inserted in parallel with the field winding, causing the
generator to pole slip. The reactive power plot shows that this criterion was satisfied at
3.75 s, indicating that the load angle had increased to a value above 90°. A further update
in Pt occurs at 4 seconds, this would not have occurred if the setting, lag, had been set
to its default value of 1.0 second. The effect of this update was to decrease the real
power trip setting sensitivity, but increase the rate of change of power setting sensitivity.
The setting Ptol limits the extreme to which this can occur. The default setting ensures
that Pt can only update to within 5 % of the generator rating when compared to the
previous value.

At 4.4 seconds, the real power trip criterion is satisfied, leaving just the (AP/At) criterion
to be satisfied. The (AP/At) criterion is satisfied at a time of 4.85 seconds, and therefore
all three criteria were satisfied for the one and a quarter cycles required. The algorithm
therefore tripped at this point. The trip signal is included on the real power plot. The trip
command was issued a significant time before the generator real power went negative. The
algorithm therefore tripped before the pole slip. The usual point where a pole slip is said
to have occurred is when the machine goes from generator to motor action, i.e when the
power output of the machine goes negative. Unfortunately, the whole pole slip could not
be recorded, due to limitations in the data acquisition system.

The pole slip was classified as a 'steady state' pole slip because the ratio Pt/Pmax was at
its nominal value at the time of tripping. There was however, a large deviation in (AP/At)
of -37,500 W/s before the algorithm tripped which resulted from the switching of the field
resistor. This caused conflicting information. If fault starter element information was also
available to the algorithm, then there would be no doubt that this was a 'steady state' pole
slip because the starter elements would not have picked up. The protection engineer
would therefore know that the pole slip was caused by a problem with the generator.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 136

6.2.2 ’Over Torque’ Pole Slip - Tests *E,F,G,H,I,J*
The results to low, medium, and high excitation 'over torque' pole slips, with local load
connected are given in Figures 6.4, 6.5, & 6.6 respectively. The results to tests H, I, &
J were nearly identical. Local load has little effect on pole slipping characteristics.

The algorithm successfully detected all of the pole slips, producing one trip output per
pole slip cycle. Two pole slips occur in Figure 6.4, the First being a much slower pole
slip than the second. The high levels of noise in the (AP/At) are due to problems with the
data acquisition system. These problems did not occur in the algorithm implemented in
the commercial micro-processor relay.

The main difference between the three graphs is the fluctuations in algorithm measurands.
As the generator excitation is increased, the magnitude of the fluctuations also increases.
This is due to a higher level of effective generator internal voltage. In all cases, the
algorithm tripped before the pole slip occurred. In Figures 6.5 and 6.6 an update in Pt
occurs after the prime mover input torque was increased. This was due to the 0.5 second
'lag' setting. These updates would not have occurred if the nominal value for 'lag' of 1.0
second had been used. This information is useful because it shows the effect of using a
smaller 'lag' setting. The update produces a more sensitive real power trip setting,
resulting in an earlier algorithm trip. It must be emphasised however, that this is not
desirable as the algorithm may incorrectly update during a power swing. It would therefore
use an incorrect value for the prime mover input power. The 'over torque' pole slip tests
do not produce typical pole slipping characteristics. The tests were repeated with the 'lag'
setting set to 1 second, and the algorithm still successfully detected all of the pole slips.
These tests show that having a short 'lag' setting can make the algorithm prone to incorrect
trip level updates, which may result in nuisance tripping. It is therefore recommended that
the default level of 1.0 second for lag is used.

6.2.3 Adjacent Generator Pole Slip and Loss of Excitation - Tests ’K, L’
Figure 6.7 shows the algorithm measurands for the test where a generator connected to
the same bus as the protected generator was forced into 'over-torque' pole slipping. It pole
slipped four times at a rate 2 Hz before it was re-synchronised. The algorithm remained
stable for this disturbance. The reactive power output actually increases and moves
further away from the trip setting, Qtrip, when the adjacent generator begins to pole slip.
The algorithm was therefore less likely to operate. This occurs because of the inherent

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 137

regulation characteristics of the machine. The adjacent generator consumed a large
amount of reactive power during its pole slipping, thus depressing system voltage. The
protected generator therefore increased its reactive power output in an attempt to maintain
its terminal voltage. This test shows that the algorithm is stable when operating alongside
a pole slipping generator.

Figure 6.8 shows the algorithm response to the adjacent machine loss of excitation test.
The algorithm successfully restrained for this test, for the same reasons as with the
adjacent generator pole slipping test. None of the three criteria were satisfied for the one
and a quarter power system cycles required for the algorithm to operate.

6.2.4 Generator Fault Tests

The algorithm did not nuisance trip for any of the fault tests. Graphs for the isolated
generator neutral tests have been included. The graphs for the solidly earthed tests are
similar and have therefore not been included.

Figure 6.9 contains the algorithm response to the single phase to earth fault test (IB). The
fault causes the real and reactive power to rapidly fluctuate at the fault on and off points,
but causes no significant change during the fault period. Neither of the real and reactive
power trip criterions were satisfied during the fault test. It can be concluded that the
algorithm remains very secure for single phase to earth faults.

Figure 6.10 contains the phase to phase fault test results (Test 2A). A significant amount
of distortion occurs in the algorithm measurands, this was due to an increase in the
harmonic content of the voltage and currents. The reactive power criterion was not
satisfied during the test, the algorithm therefore restrained. The reactive power changed
to a negative value after removal of the fault. This occurred because the generator load
angle increased during the fault. After the initial change in reactive power at the fault off
point, the reactive power returns to its original value as the generator returns to its original
operating position. The harmonic distortion in the algorithm measurands shows that the
algorithm in unaffected by this type of interference. This aspect of the laboratory model
power system is useful for testing protection algorithms. The voltage and current
waveforms are far from ideal. If the algorithm can operate successfully under these
conditions, then it should operate for any other conditions.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 138

Figure 6.11 shows the algorithm response to the two phase to earth fault, test '3B'. The
reactive power criterion was satisfied at the fault off point for this test. This indicates that
the generator load angle increased to a value above 90° during the fault. However, at the
instant the reactive power criterion is satisfied, the real power jumps above its trip level,
blocking algorithm operation. This provides decelerating energy to the generator rotor,
indicating that the CSP had not been reached. The algorithm therefore restrained from
tripping.

The three phase fault test (test 4A) results are given in Figure 6.12. Once more, the fault
off point is the most testing part of the fault test for the algorithm, due to the reactive
power criterion being satisfied. This shows that the generator load angle increased to
above 90° during the fault. When the fault was cleared the real power jumps above its
trip level, showing that the CSP has not yet been reached. This prevented the algorithm
from tripping. Additionally, the (AP/At) criterion was only satisfied for half a power
system cycle.

After the fault has been cleared, the generator real and reactive power continue to
decrease, as the generator loses stability. The sign of the (AP/At) signal is continuously
negative after the fault has been cleared, indicating that load angle increase is occurring.
The real power starts to decrease more rapidly as the generator accelerates into a pole slip,
and the algorithm trips. This occurs at an output power of 2 kW. The algorithm therefore
tripped before the pole slip. The pole slip was diagnosed as 'transient' in nature because
of the large fluctuations in (AP/At) produced by the fault. If the algorithm's slip setting
had been lower, the algorithm would have tripped much earlier. A slip setting of -0.25 %
rather than the nominal -0.5 % would have resulted in an algorithm trip at a real power
value o f 3 kW.

The algorithm response to the three phase to earth fault (test 5B) is shown in Figure 6.13.
The algorithm did not trip for the same reasons as for the other faults. The fundamental
basis of the algorithm, which ensures that tripping does not occur until the CSP has been
reached, blocks algorithm operation during power system faults.

6.2.5 Observations from the Laboratory Power System Model Tests

The main observation is that the algorithm detected three different types of pole slips,
namely 'weak field', 'over-torque' and 'fault induced'. The algorithm was secure for all

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 139

possible fault types. The 'over-torque' tests showed that reducing the 'lag' timer setting
below 1.0 second can result in updates in the algorithm trip levels which increase
algorithm sensitivity. Although the updates in Pt during the over-torque pole slip tests
were correct, because the prime mover input power had been increased, it is safer to use
the 1.0 second 'lag' setting so that incorrect updates do not occur during stable power
swings, where the prime mover output power has not increased. Most generator prime
movers increase their torque outputs at a much slower rate than the rates used in these
tests. This is one of the inaccuracies of the laboratory power system model. Prime movers
are normally constant torque devices, whereas the DC machine used as the prime mover
is a constant speed device. The laboratory tests do however provide proof that the
algorithm operates satisfactorily in a real power system environment, and are therefore
useful. The parameters of the 5 kVA generator are also atypical, due to the very high
value of transient reactance. This in conjunction with all of the other tests performed
strengthens the value of the tests, as it proves that the algorithm works for a wide variety
of machines.

6.2.6 Analysis of Pole Slipping Impedance Loci for the 5 kVA Laboratory
Generator

Inspection of the real power plots for the pole slipping tests also shows that the real power
does not go as negative as theory suggests. This is mainly due to the asynchronous power
generated during the pole slip, which shifts the whole real power characteristic up. This
effects the path of the apparent impedance locus and therefore it affects impedance based
pole slipping relays. Using the single blinder characteristic derived in Appendix B for the
5 kVA generator, the response of the scheme to the pole slip tests was analyzed. The
scheme operated for two of the 6 'over torque' pole slipping tests. Figure 6.14 shows
several impedance loci for differing tests, along with the single blinder relay characteristic.

Figure 6.14a is for the pole slip which occurred following the three phase fault test. The
power based pole slipping algorithm response to this test is shown in Figure 6.12. If the
most commonly used single blinder scheme is chosen, as described in chapter 3, the
scheme would have failed to detect the pole slip. This scheme requires the supervisory
mho to have picked up before the impedance crosses the blinder element. Figure 6.14a
shows that the impedance locus crosses the blinder elements before the mho element picks
up. This problem can be remedied by increasing the size of the mho characteristic. This

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 140

highlights the intricacies required in setting the single blinder scheme. The mho pole
slipping scheme would have successfully detected this pole slip. The double blinder
scheme would not have detected it. Details on this and the mho scheme can be found in
Appendix B.

Figure 6.14b shows the response to the low excitation over-torque pole slip test. The
power based algorithm response for this test is given in Figure 6.4. The single blinder
scheme did not operate for this pole slip, and could not be modified so that it would. One
of the quadrilateral impedance based pole slipping protection characteristics described in
section 2.3.3 could have detected this condition however. The real power does not reverse
sufficiently in this test to make the impedance iocus traverse from right to left across the
impedance plane. Consequently the locus does not cross both blinder elements, and the
scheme therefore failed to trip. The mho scheme would have detected this pole slip, as
would a loss of field scheme. The double blinder scheme described in Appendix B would
not have detected the pole slip. If the size of the supervisory mho characteristic used in
the double blinder scheme had been doubled it would have detected this slip. Increasing
the size of the supervisory mho characteristic must be performed cautiously however, since
too large a mho characteristic may result in nuisance tripping during stable power swings.

Figure 6.14c shows the impedance locus for the high excitation over torque pole slipping
test. The single blinder scheme successfully detected this pole slip, the double blinder
scheme did not however, because the impedance locus left the supervisory Mho circle
before crossing the blinder. Again a larger supervisory mho characteristic would have
remedied this problem.

The above analysis highlights some of the problems with conventional impedance based
pole slipping schemes. It is not intended as a thorough analysis of these schemes. A more
thorough analysis will be given when discussing the PPSS and 625 kVA diesel generator
tests.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 141

6.3 PPSS TEST RESULTS

The commercial relay implementation of the algorithm successfully detected all pole
slipping conditions, and restrained for all of the recoverable power swing tests. It also
detected all of the loss of field conditions, and restrained for the loss of prime mover
simulations. The algorithm did operate for the tests where the generator was operated at
load angles above 90° with the aid of a fast AVR. This was expected and as stated in
chapter 4, does not cause a problem because generators are never practically operated in
this manner.

As well as using the ATP simulations to run the PPSS, a comparison with the
conventional impedance based protection techniques was performed using PC models of
the relays. Details of these models can be found in appendix F.

6.3.1 Pole Slipping Algorithm Response to PPSS Pole Slip Tests

Test 'PSL1GA'
Figure 6.15 contains the algorithm response to the pole slipping test 'PSL1GA'. In this
test, pole slipping was caused by a 1170 ms, 3 Phase fault. A long fault duration was
required because the generator was initially operating at 20 % load. The algorithm
successfully restrained for the duration of the fault, and tripped at a time of 1.5 seconds
on the record. The pole slip was diagnosed as a 'transient' pole slip. The generator slipped
one pair of poles before re-synchronising. The ATP machine angle plot is not a plot of
the machine load angle, or the system load angle, the angle is measured relative to a
reference point and is given here to show the overall increase of 360° which occurs during
a pole slip. These curves were produced using a PC based simulation of the algorithm
because this file was unavailable when the PPSS tests were conducted.

Pole slipping caused by faults from such a low initial generator output power is unlikely
in practice, since the fault would be cleared much faster than 1170 ms, unless a stuck
circuit breaker situation occurred. The test is still required because the algorithm must be
tested for all operating conditions.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 142

Test 'PSL2GA'
Figure 6.16 shows the variables down-loaded from the relay’s disturbance record for PPSS
test PSL2GA. All of the graphs are given in terms of relay secondary quantities. The
generator terminal voltage on the secondary side of the VT is 110 Volts (phase to phase
rms), whilst nominal or full load current is 1 Amp (rms). The relay secondary rating for
the generator is therefore 190.5 VA. For this test, the generator was initially at 50 %
load. It pole slipped twice before re-synchronisation. The algorithm detected both pole
slips, which were correctly diagnosed as 'transient' pole slips. The reactive power plot
shows that Qtrip was satisfied three times during the test, and the real power plot shows
that during the third time, the generator developed enough decelerating power to re-
synchronise. The generator therefore stopped short of the CSP after the second pole slip.
This is shown as point A on the real power plot.

Test 'PSL3GA'
Figure 6.17 shows the algorithm response to test PSL3GA. The generator failed to re-
synchronise for the duration of the test because it was initially operating at full load. The
algorithm detected the first pole slip, but failed to detect subsequent pole slips because of
AVR field forcing action pushing the system centre away from the generator. The effects
of field forcing can be observed in the real and reactive power plots. In the reactive power
plots, the reactive power criterion is satisfied for the first pole slip, but not for any
subsequent slips because the mean level of reactive power drifted up above the reactive
power trip level, Qtrip. In the real power plots, the magnitude of the positive and negative
peaks which occur during every slip cycle increases. The peaks reach a value of 650 W,
which is almost 3.5 times the generator rating.

Section 4.4.1 of Chapter 4 provided graphs which show the theoretical operating range of
the algorithm. For the PPSS test generator, the value of external reactance, X given in
Appendix C is 0.133 pu, whilst Xq = 1.64 pu, and Xd' = 0.238 pu. Expressed as a
fraction of Xq, Xd' = 0.145*Xq. If these figures are applied to the setting graphs 4.6b and
4.7a then the graph for an effective generator voltage, E of 1.25 pu shows that the
algorithm should just work; the maximum value of external reactance from this graph is
0.145 pu, whilst the value from the E = 1.5 pu graph is 0.125 pu. This simulation
therefore confirms that the setting graphs are correct. The algorithm detected the first pole
slip, before the reactive power criterion was no longer satisfied, and would therefore have
disconnected the generator avoiding damage. This system is close to the operational limits

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 143

of the algorithm, and was chosen to confirm the theoretical setting limits.

The PPSS has a pre-fault cycling function which enables you to continuously cycle the
first power system cycle of the simulation file. If the first cycle is at a steady state, then
the initial steady state period at the start of the simulation can be extended almost
indefinitely. This function was used because the maximum PPSS simulation time was 6
seconds. The algorithm requires at least 3 seconds before the real power trip level, Pt is
updated. A pre-fault cycling period of 3 seconds was therefore used with all of the PPSS
tests. The update in Pt in this case occurs just before 500 ms. This implies that at the
instant the PPSS simulation started, the Pt function must have been approximately half
way between its sample check points, which are 'lag' seconds apart. The default value of
'lag' is 1 second, the update before 500 ms therefore confirms that Pt is updating correctly.

The Pmax function shown in Figure 6.17b was reset to its nominal value just before the
update in Pt occurred. Since Pt at this point was 0, Pmax was reset to zero for one power
system cycle before tracking back onto the power waveform. This produces a temporary
drop in the magnitude of (AP/At)min. This does not cause a significant increase in
algorithm sensitivity however, because the trip level is changed for a period of less than
the 1.25 cycles required by the algorithm to trip.

Note that once Pt is updated, the (AP/At)min trip level does not reset to its lower 'steady
state' level. This is a result of computer code 'bomb-proofing'. If the ratio Pt/Pmax is
greater than 0.7, or less than 0, when (AP/At)min is calculated, the algorithm automatically
uses the 'transient' value (AP/At)^ in the calculation of (AP/At)min. Pmax does not update
to its nominal value of 1.4*Pt the instant Pt is updated because once an update in Pmax
occurs, Pmax is not updated for another second, unless an even higher value of real power
occurs. This provides an additional amount of algorithm security.

Test 'PSL3GN'
Figure 6.18 shows the algorithm response to test 'PSL3GN'. This is identical to the above
test, except that the AVR was on 'manual' control. No field forcing action therefore
occurred and the algorithm successfully detected all of the pole slips. The reactive power
shows that with no field forcing action, the reactive power criterion was satisfied for every
slip cycle. Note that the peaks in real power during pole slipping are much less severe
with the AVR on manual control. This shows that once a generator is pole slipping, AVR

M.J.Checksfield/PhDAJniversity of Bath 1996 Chapter 6, Page 144

action generally increases the severity of the pole slips, and does nothing to aid the re­
synchronisation process. The single most important factor for re-synchronisation is the
speed of the governor and prime mover system.

Test 'PSL3NN'
Figure 6.19 shows the results to test 'PSL3NN'. In this test the effects of AVR and
governor action were not modelled The prime mover input power and generator excitation
therefore stay at their initial steady state values through the entire test. The algorithm
successfully detected the first 7 pole slips, and then failed to detect the next 3. This was
a result of the high speed at which the generator was pole slipping. The algorithm trip
criterions are not satisfied for a sufficient duration. After failing to detect the three pole
slips, the algorithm trips again, and remains tripped. This occurred because the reactive
power criterion was continually satisfied.

Figure 6.20 contains an expanded section of the real power and rate of change of power
plots, showing why the algorithm failed to detect the later pole slips. The real power and
rate of change of power are only satisfied for five consecutive sample points. Six
consecutive sample points are required for the algorithm to trip.

This confirms the performance predicted in section 4.4.2. If the time between consecutive
peaks in real power is measured, it gives a mean pole slipping rate of 8 Hz. Section 4.4.2
predicted that the algorithm would detect a maximum pole slip rate of 8.3 Hz.

A glitch in the reactive power plot of Figure 6.19 occurs just before 2000 ms. The glitch
is a jump in reactive power to zero exactly. It is therefore thought to be a result of a
problem in the computer code. It is impossible to say if the glitch occurred when
converting the 32 bit integer variables used by the algorithm to 16 bit variable for storage
in the disturbance record, or if it occurred in the calculation of reactive power by the
algorithm. This was the only glitch seen in over 100 disturbance records which have been
down-loaded from the commercial relay.

Test 'PSLIGAF'
Figure 6.21 shows the algorithm measurands for the 'PSLIGAF' test. This pole slip was
caused by a 'fault' in the generator AVR, which reduced the AVR's reference voltage from
unity down to 0.05 per unit. The generator was initially operating at 20 % load, the entire

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 145

6 seconds of the PPSS test was therefore required for the generator to lose synchronism.
If the reference voltage had been reduced less, the generator would not have lost
synchronism in the 6 second limit. The incredibly low AVR reference voltage practically
makes this a loss of field test. The algorithm tripped before the pole slip occurred, then
shortly after, the simulation ended. The pole slip was correctly diagnosed as a 'steady
state' pole slip. The graph shows the levels of noise found on the (AP/At) signal. This was
higher than expected, and is power system frequency noise. This suggests that the source
of the noise is due to an offset in either the voltage or currents. A DC offset in one of the
signals, when multiplied by the power system frequency alternating component of the
other, superimposes a power system frequency oscillation onto the mean level of the
calculated power signal

Without performing more steady state tests with varying amounts of real and reactive
power, the exact cause of the noise can not be found. There was no opportunity to do
this, and since the noise levels do not effect the algorithm operation, no further
investigations were conducted. The noise may be due to timing errors in the analogue to
digital converters. It is not due to a drift in the analogue component tolerances, since
several different data acquisition boards have been used in the relay, all of which produce
the same level of noise.

The noise caused a small time delay in algorithm tripping in this test because the 1.25
cycle trip criterion was not satisfied until the mean level of (AP/At) fell below the
(AP/At)min trip setting by the peak amplitude of the noise. This pole slip was the most
demanding for the algorithm to detect. The low levels of (AP/At) produced as a result of
the generator pole slipping slowly meant that the (AP/At)min criterion was barely satisfied.
Neglecting the effect of the minium setting value, (AP/At)fact, if the slip setting had been
reduced from -0.5 % to -0.25 %, then the algorithm would have tripped at a time of
1724 ms rather than 1844 ms. It would therefore have tripped 120 ms earlier. The -
0.25 % slip trip point is shown as the dotted line in Figure 6.21.

Test 'PSL2GAF'
Figure 6.22 shows the curves for the 'PSL2GAF' test. For this test, the AVR reference
voltage was set to 0.5 per unit, the generator was initially operating at 50 % load. The
higher initial load resulted in much higher values of (AP/At). The noise in the signal
therefore had much less effect on the tripping time than with the 20 % initial generator

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 146

loading case. The algorithm tripped before the pole slip occurred, and the pole slip was
diagnosed as a 'steady state' pole slip. The pole slip shown took 4.0 seconds to occur, the
next pole slip would have occurred in another 5 to 6 seconds. The mean pole slipping
rate was therefore 0.2 Hz. This shows that the algorithm can detect slow pole slips. With
slow pole slips, the increase in load angles to 100° takes a long time, as does the period
when swinging from 270° to 100°. The part of the pole slip cycle where the generator
load angle moves from 90 to 180° occurs quickly, producing sufficient rates of change of
power for the algorithm to trip. The mean rate of pole slipping may therefore be low, and
the algorithm will still detect the pole slip, because of the fast rate at which part of the
pole slip occurs.

If the slip setting had been reduced from -0.5 % to -0.25 %, the algorithm would have
tripped earlier. The lower slip setting trip point is shown on Figure 6.22 as the dotted trip
line. The algorithm tripped at a time of 2004 ms, this was 60 ms before the -0.5 % slip
setting trip point.

Test 'PSL3GAF'
Figure 6.23 shows the algorithm response to test 'PSL3GAF'. The AVR reference voltage
was reduced to 0.8 per unit to induce the pole slip, which occurred quickly because the
generator was initially operating at 100 % full load. The algorithm tripped before the
generator pole slipped. The pole slip was correctly diagnosed as a 'steady state' pole slip.
This would inform the protection engineer that a fault in the generator or its controls was
the cause of the pole slip.

The speed at which the pole slipping took place meant that additional reactive power was
drawn from the system to supply exciting VArs for asynchronous induction generation.
This resulted in the reactive power remaining below Qtrip for the first few pole slips, and
consequently the algorithm produced one long trip signal for several pole slip cycles. The
real power also shows the amount of asynchronous power generated, since the positive
peaks in real power are much greater than the negative peaks. The mean value of real
power generation was shifted up by 100 Watts. The generator was therefore generating
approximately 50 % of rated power asynchronously.

After three or four pole slips, the AVR responded to the voltage depression caused by the
excessive reactive power demand, and increased excitation. This pushes the system centre

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 147

away from the generator and causes the reactive power to oscillate either side of Qtrip.

If the slip setting had been reduced from -0.5 % to -0.25 %, then the algorithm would
have tripped 35 ms earlier. This shows that the gain in tripping time decreases as the
generator loading increases, due to higher values of (AP/At).

The graphs for the other tests have not been included because they are practically identical
to the graphs already presented.

6.3.2 Pole Slipping Algorithm Response to PPSS Stable Power Swing Tests
The power based pole slipping algorithm successfully restrained for all of the stable power
swing tests.

Test 'PSWIGA'
Figure 6.24 shows the pole slipping algorithm response to test 'PSWIGA'. This was a
severe stable power swing, induced by a 1150 ms three phase fault. The generator was
initially operating at 20 % output power. The algorithm remained very secure during the
fault because the reactive power criterion was not satisfied. It is rare for the reactive
power criterion to be satisfied during a fault because faults are normally inductive. The
instant the fault was cleared, the reactive power criterion was satisfied because the
generator load angle had risen above 90°. For the duration that Q was less than Qtrip, the
generator's real output power was greater than the trip setting, Pt. This indicated that there
was decelerating power available. The algorithm therefore restrained from tripping.

After this initial period, the reactive power criterion was no longer satisfied because the
generator load angle had swung back to a stable operating point below 90°. The
algorithm therefore remained stable for the duration of the entire stable power swing.

Test 'PSWL2GA'
Figure 6.25 shows the pole slipping algorithm response to test 'PSW2GA'. This was a
severe stable power swing, induced by a 630 ms three phase fault. The generator was
initially operating at 50 % output power. The algorithm did not trip for the above reasons.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 148

Test 'PSW3GA'
Figure 6.26 shows the response to test 'PSW3GA'. This was a severe stable power swing,
induced by a 320 ms two phase to earth fault. The generator was initially operating at
100 % output power. Again the algorithm did not trip for above reasons.

Test 'PSWIGN'
In the above three tests, the generator AVR was in service, Figure 6.27 shows the
algorithm response to test 'PSWIGN', when the AVR was set to manual control. The
stable power swing was caused by a 1155 ms three phase fault. The real and reactive plots
show that without the AVR in service, more dramatic fluctuations occur in the algorithm
measurands. Without the AVR in service, the overall damping of the rotor oscillations is
decreased, because of the high value of impedance between the generator and infinite bus.
This condition is made worse by the fault clearance. This disconnected the 235 MW load
which also assisted damping. In addition, the initial generator excitation was low because
the generator was initially operating at a low output power. This results in low
synchronising torques, which can be over-shadowed by the asynchronous torques produced
during the swing. The operating characteristic of the generator is no longer dominated by
the sinusoidal load angle variations produced as a result of field excitation. The flux due
to field excitation will also be weakened by the long fault duration, due to the de­
magnetising effects of armature reaction. The end result is that the generator rotor swings
more violently compared with the case when the AVR is in service. The reactive power
criterion is therefore satisfied four times rather than the one time which occurred with test
'PSWIGA'. This increases the number of occasions where algorithm operation may occur.
The real power plot shows the points where the reactive power criterion was satisfied, and
the rate of change of power plot shows when both the real and reactive power criteria
were satisfied.

The purpose of the transient saliency scaling factor, (AP/At)^ can be seen. This de­
sensitises the algorithm during severe power swings so that it does not operate. The non-
sinusoidal variation in real and reactive power results in the real and reactive power
criterions being satisfied more times than if the components of power were only produced
due to synchronising torques. A de-sensitised rate of change of power trip criterion is
therefore required to improve algorithm security. The additional component in (AP/At)min
caused by (AP/At)^ ensures that the rate of change of power criterion is not satisfied, and
that the algorithm restrains. If one sensitive setting had been used for (AP/At)min which

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 149

detected the slow 'steady state' pole slips, the algorithm would have nuisance tripped for
this stable power swing test.

Test 'PSW2GN'
Figure 6.28 shows the algorithm measurands for test 'PSW2GN'. The AVR was also set
to manual control on this test. This resulted in the reactive power criterion being satisfied
twice. The first time, the real power was above Pt, showing that decelerating power was
available. The second time that the reactive power fell below Qtrip, the real power
criterion was also satisfied. This occurred when the machine angle was at a minimum. It
was therefore a result of asynchronous induction generation effects. Inspection of the rate
of change of power plot shows that the algorithm was prevented from operating due to the
large magnitude of (AP/At)min. This test illustrates the advantages of having an adaptive
rate of change of power setting.

Test 'PSW3GN'
Figure 6.29 shows the results to test 'PSW3GN'. The AVR was also set to manual control
on this test. In this test the reactive power criterion is only satisfied during the period of
the swing when the load angle is above 90°. The machine characteristics are dominated
by the effects of field excitation, because the initial value of excitation was much higher.
The algorithm measurands therefore behaved in the 'ideal' way predicted by basic
synchronous generator theory, and the algorithm restrained from tripping with relative ease
when compared to the 'PSW2GN' and 'PSWIGN' tests.

The 'PSWIGN' and 'PSW2GN' tests are very severe power swing tests. They would
happen rarely in practice, since they require a fault duration of at least 800 ms, coupled
with a generator AVR set to manual, and an initial low value of generator operating
power. They are however, possible and are more likely to happen in an embedded
generation environment, where long IDMT fault clearance times exist. There are several
reasons why an embedded generator may not be operated at full load. The generator may
be used to offset a site load, or may use excess steam from an industrial process in a CHP
scheme. It is also possible that the AVR may be set to manual. The tests are useful
because they provide the severest power swing test conditions possible. If the pole
slipping algorithm does not operate for these stable power swings, it is unlikely that it will
operate for any others.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 150

The graphs for the other tests have not been included because they are practically identical
to the graphs already presented.

6.3.3 Pole Slipping Algorithm Response to PPSS Tests Where the Generator is
Operated in its Dynamic Stability Region - at Load Angles Greater than 90°

The algorithm operated for all three of these tests, as predicted in Chapter 4. When a fast
AVR is used to operate the generator at load angles above 90°, the generator is effectively
being operated beyond its conventional CSP. This 'fools' the algorithm when a sudden
input power decrease occurs because the reactive power criterion is satisfied due to the
load angle being above 90°. The real and rate of change of power criteria are also satisfied
as a result of the load decrease.

However, as stated in chapter 4, this is not considered a problem, since generators are
never practically operated in this area because the reserve in hand for transient stability
is too low. These simulations have been included for completeness.

Test 'AVR5NA'
Figure 6.30 shows the algorithm variables for test 'AVR5NA'. The curves were produced
by the PC based version of the algorithm code, in order to see the full 16 second duration
of the test. The generator is initially operated in the over-excited region. The AVR voltage
reference is then reduced, which results in the generator moving over into the under­
excited region. This causes the load angle to increase above 90°, and results in the
reactive power criterion being satisfied. A period of steady state operation is then
simulated to allow the generator to settle. At 10 seconds, the generator input power is
reduced from 0.5 pu to 0 pu. This causes the load angle to decrease and the reactive
power to increase. Before the reactive power rises above its trip level, the real and rate
of change of power trip levels are satisfied, and the algorithm nuisance trips.

It is important to note that this is a worst case simulation. In practice, the output power
of a generator prime mover is not instantaneously reduced or increased. A slower change
in output power will produce lower values of (AP/At). The algorithm may therefore not
nuisance trip for some forms of generation. For example, hydro generators are normally
loaded up or down over a period of 30 seconds. At this rate of change the algorithm real
power trip level, Pt may track the change in output power, and the algorithm would not

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 151

nuisance trip.

6.3.4 Pole Slipping Algorithm Response to PPSS Loss of Field Tests

The algorithm tripped for all of the loss of excitation tests performed.

Test 'LOFIGA'
The algorithm response to test 'LOFIGA' is shown in Figure 6.31. This was a generator
loss of field test from 20 % initial load. The algorithm successfully tripped before the
generator lost synchronism. The trip was diagnosed as a 'steady state' pole slip.
Comparison of these test results with those of test 'PSLIGAF' shown in Figure 6.21 shows
the two tests to be almost identical. This occurred because the field voltage was reduced
to such a low value that the generator would lose synchronism in the 6 second time
window of the PPSS.

Test 'LOF2GA'
The algorithm measurands for test 'LOF2GA' are shown in Figure 6.32.. This was a
generator loss of field test from 50 % initial load. The algorithm tripped before
synchronism was lost. Comparison between these results and those produced by test
'PSL2GAF' in Figure 6.22 shows the similarity between the two tests. The AVR reference
voltage in test 'PSL2GAF' was reduced to 50 % per unit, whilst the field voltage in the
loss of field test was reduced to zero. The similarity occurs because the generator AVR
reduces the field voltage to zero in both tests.

Test 'LOF3GA'
Figure 6.33 shows the results to test 'LOF3GA', a loss of field from an initial load of
100 %. The algorithm tripped before the generator lost synchronism. The trip was
diagnosed as a 'steady state' trip. These results are different to those of test 'PSL3GAF'
shown in Figure 6.23, but only after the initial algorithm trip. The effects of AVR action
are apparent in the latter test, as the peaks in real power increase in magnitude once
synchronism is lost. With the loss of field case, the peaks become smaller due to a
decaying generator flux.

6.3.5 Pole Slipping Algorithm Response to PPSS Loss of Prime Mover Tests
Figure 6.34 shows the algorithm measurands for the loss of prime mover test 'LOP3NA'.
The algorithm successfully restrained for this operating condition. The algorithm

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 152

remained very secure.

6.3.6 Effect of Altering the Pole Slipping Algorithm 'Slip' Setting

The effect of altering the algorithm's 'slip' setting has been discussed in some of the above
tests. A reduction in the slip setting results in earlier tripping times during 'steady state'
pole slips. These pole slips produce very low values of (AP/At) as the generator loses
synchronism. The (AP/At) criterion is the last criterion to be satisfied. It therefore
determines the point at which the algorithm will trip. With 'transient* pole slips, the real
power criterion determines the tripping point because the (AP/At) and reactive pov/er
criteria are satisfied as the real power begins to decrease from its maximum value, Pmax.

The power system frequency noise experienced in the commercial relay implementation
of the algorithm further delayed the tripping point because the mean value of (AP/At) must
exceed the trip level (AP/At)min by the peak value of the noise. The noise in the PPSS tests
was typically +/- 50 W, which for a generator secondary rating of 190.5 VA, is
approximately equal to 25 % of the generator rating. It is therefore feasible that the slip
setting can be reduced. The noise will effectively counteract the increase in sensitivity
which occurs by decreasing the magnitude of the trip level, (AP/At)min. The PPSS ATP
tests were run through the PC based simulation of the algorithm, with a slip setting of -
0.25 %. No nuisance trips occurred. The only tests where the trip time was improved
were the 'PSL7GAF' , and the 'LOF7GA' tests because all of the pole slips in these tests
were 'steady state' pole slips.

The PPSS 'steady state' pole slip test disturbance records were also analyzed with a slip
setting reduced from -0.5 % to -0.25 %. The analysis showed that a lower slip setting
would improve the steady state pole slip trip times, without compromising algorithm
security.

Table 6-1 contains the improvements in trip times for a reduction in the 'slip' setting. The
figures in brackets are where (AP/At)fact , the minimum set value for (AP/At)min was
reduced from 25 % of generator rating to 10 %. This occurred because it was inhibiting
any effect that a reduction in the slip setting might have on (AP/At)min at low levels of Pt.

The PC based simulations show the theoretical gain, whilst the PPSS disturbance record
tests show the practical gain. In the 50 % and 100 % initial load tests, denoted by the

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 153

tests with a 2 or 3 in the title, the theoretical gain is greater than the practical gain,
because of the noise in the practical system. For the 20 % initial load cases, the real
power criterion Pt inhibits tripping because an update in Pt occurs just before the real
power begins to fall as the generator loses synchronism. This effect can be seen on
Figure 6.31. Figure 6.35 shows the change in (AP/At)min and trip points for test
'PSLIGAF.

This analysis shows that an appreciable gain in 'steady state' pole slip tripping times can
be achieved by reducing the slip setting to -0.25 %. This must also be accompanied by
a decrease in the setting (AP/At)fact so that the new slip setting can take effect at low
values of Pt.
Table 6-1
Improvement in trip times for a reduction in the 'slip' setting from -0.5 % to -0.25 %.
(figures in brackets are where (AP/At)ftct has also been reduced from 25 % to 10 % o f generator rating.)

Test PC Based PPSS Disturbance

Algorithm (ms) Record (ms)

PSLIGAF 0 (60) 0 (120)

PSL2GAF 80 60

PSL3GAF 55 35

LOFIGA 15 (110) 0 (140)

LOF2GA 85 40

LOF3GA 70 60

The change in slip setting does not effect the algorithm security for 'transient' pole slips
because the setting (AP/At)^ compensates for this.

6.3.7 Comparison of the Performance of the New Power Based Pole Slipping
Algorithm with Conventional Impedance Based Pole Slipping Schemes

The performance of the new and conventional approaches to pole slipping protection was
analyzed using the PPSS test simulation files and PC based models of the relays. The

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 154

tripping times for the power based algorithm are for when the ATP generator voltage and
current output files were passed through the PC 'relay simulator' code. They are not related
to the PPSS disturbance records given in the previous sections. Tables 6-2, 6-3 and 6-4
contain the trip times for all of the tests.

Table 6-2
Results of Comparisons Between Conventional and New Pole Slipping Protection
Methods for PPSS Pole Slipping Simulation Tests.

Test Loss of Mho Single Double Power

Field Trip Scheme Blinder Blinder Algorithm
(s) Trip (s) Trip (s) Trip (s) Trip (s)

PSL1NN NT 1.227 NT* NT* 1.305

PSL2NN NT 0.929 1.357 1.364 0.995

PSL3NN NT 0.633 0.882 0.887 0.655

PSL1GN NT 1.456 NT* NT* 1.560

PSL2GN NT 1.119 1.608 1.617 1.565

PSL3GN NT 1.158 1.503 1.253 1.255

PSL1NA NT 1.451 NT* NT* 1.545

PSL2NA NT 1.290 1.377 1.386 1.330

PSL3NA NT 0.635 0.978 0.982 0.670

PSL1GA NT 1.451 NT* NT* 1.540

PSL2GA NT 0.932 1.592 1.602 1.030

PSL3GA NT 0.664 1.031 1.036 0.705

PSL1GAF 3.232 NT* NT* NT* 5.225

PSL2GAF 2.589 NT* NT* NT* 3.115

PSL3GAF 1.677 2.000 2.253 2.255 1.465

* - signifies that the relay did not operate as required.

M.J.Checksfi el d/PhD /University of Bath 1996 Chapter 6, Page 155

6.3.7.1 Com parison o f Im pedance Relays w ith N ew Pow er Based Protection
Algorithm for Pole Slipping Conditions

Table 6-2 shows that the new power based pole slipping algorithm tripped for all of the
pole slipping conditions, whilst some of the conventional schemes failed to trip. The Mho
scheme failed to detect all of the 'faulty AVR' pole slips, whilst the single and double
blinder schemes failed to detect some of the pole slips which occurred from 20 % initial
generator load. The setting procedure used to set the impedance based relays is given in
Appendix B. The loss of field scheme operated for all of the 'faulty AVR pole slips'.

Test 'PSL3GN'
Figure 6.36 contains the single blinder scheme characteristic and the impedance trajectory
for test PSL3GN. The single blinder scheme failed to detect the first pole slip for this
test, but correctly detected the subsequent pole slip cycles. For the scheme to operate, the
impedance locus must first enter the supervisory mho characteristic, cross one of the
blinders, remain in between the blinders for 40 ms, then cross the opposite blinder. The
mho element must remain energised for this entire sequence. During the fault, the
impedance locus crossed the right hand blinder with the mho element dropped out. This
resulted in the scheme failing to trip for the first pole slip cycle when the impedance locus
crossed the left hand blinder. If the right hand blinder had been placed a fraction more to
the left, the scheme would have detected the first pole slip. The scheme successfully
detected the subsequent pole slip cycles because the impedance locus took the classical
trajectory.

The double blinder scheme detected the first pole slip cycle because its supervisory Mho
circle was located higher up the R-X plane.

Test 'PSL1GA'
Figure 6.37 shows the single blinder scheme and the pole slip impedance locus for test
'PSL1GA'. The single and double blinder schemes failed to detect the pole slip because
the impedance locus crossed the right hand blinder unit during the fault when the
supervisory Mho unit was dropped out. The generator was operating at 20 % initial load,
and therefore re-synchronised after the first pole slip. A full pole slip impedance locus,
with no alteration due to fault effects was therefore not produced. It is sometimes argued
that pole slipping schemes should not trip if only one pole slip occurs. However, the

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 156

author believes it is desirable to detect the first pole slip cycle, even if tripping does not
occur until the second cycle. The decision of when a pole slipping scheme should trip can
then be left to the protection engineer.

The common single blinder scheme described in chapter 2 fails because the supervisory
Mho unit must have picked up before the impedance locus crosses the blinder units. An
enhanced scheme used for a pumped storage generator, which is also described in
chapter 2 would not have suffered these problems, but would still suffer the same setting
dilemmas as the commonly used scheme.

A stable swing impedance locus can also be seen on Figure 6.37. This occurred after the
generator re-synchronised. This shows that care must be taken when setting the Mho and
double blinder schemes; if the mho characteristics, and blinder locations are set so that
they reach too far out to the right of the R-X plane, then they can nuisance trip for stable
swings.

Test PSL3GA'
Figure 6.38 shows the mho scheme characteristic and the pole slipping loci produced by
test 'PSL3GA'. The Mho scheme tripped for the first three pole slip cycles. It failed to
trip for the subsequent cycles because the pole slipping impedance locus did not remain
inside the Mho characteristic for the 20 ms delay necessary for it to trip. After the sixth
pole slip cycle, the impedance locus was pushed above and outside of the Mho
characteristic by the field forcing action of the AVR.

The single blinder scheme only detected the second and third pole slip cycles. It did not
detect the first cycle and the fourth and subsequent cycles because the impedance locus
crossed the right hand blinder unit before entering the supervisory mho characteristic.

The double blinder scheme did not detect the first pole slip cycle for the same reasons as
the single blinder scheme. It did detect all subsequent cycles because its supervisory Mho
characteristic was located higher up the R-X plane compared to the single blinder scheme.

Test 'PSL3NN'
Section 6.3.1 stated that the power based algorithm detected the first seven pole slips,
failed to detect the next three, then detected the subsequent pole slips. Analysis of the

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 157

Mho scheme's response to this test showed that it detected the first eight pole slip cycles,
then failed to detect any further pole slips.

The single blinder scheme operated for only the second and third pole slip cycles. The
double blinder scheme operated for the second, third, and fourth cycles. All of the
impedance based approaches failed to operate for the fast pole slips because the
impedance locus travelled through their trip characteristic too quickly.

'Test PSL2GAF
Figure 6.39 contains the Mho pole slipping protection scheme characteristic, the loss of
field characteristic, and the pole slipping locus for test 'PSL2GAF'. The Mho, single and
double blinder schemes failed to detect the pole slip because the impedance locus passed
underneath their trip characteristics. This occurred because the generator dynamic
impedance during the pole slip was greater than Xd', and the generator internal voltage
was less than unity. The loss of field protection did trip for the pole slip however. The
times when the impedance entered and left the loss of field characteristic are shown in
Figure 6.39. The impedance locus entered the loss of field characteristic for a duration
of 1.48 s. This satisfied the 0.6 second time delay used to stop nuisance tripping during
stable power swings. The power based algorithm also tripped for this simulation,
approximately 0.5 seconds after the loss of field scheme.

'Test PSL3GAF'
Figure 6.40 contains the Mho pole slipping protection scheme characteristic, the loss of
field characteristic, and the pole slipping loci for test 'PSL3GAF1. The Mho, single and
double blinder schemes failed to detect the first pole slip because the impedance locus
passed under their trip characteristics. The mho, single and double blinder schemes
successfully detected the second and subsequent pole slips, which are labelled 2,3...7 on
Figure 6.40.

The loss of field scheme operated for the first pole slip cycle, labelled '1' on Figure 6.40.
It failed to trip for the subsequent cycles, because a portion of the impedance locus
occurred outside the loss of field characteristic. This is shown as point 'A' on Figure 6.40.
This portion of the locus prevented the 0.6 s loss of field time delay from being satisfied.
If a 'delay on reset' function was included in the loss of field scheme, as used in some
modem relays[74], the relay would have detected the subsequent pole slips.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 158

6.3.7.2 Com parison o f Im pedance Relays w ith N ew Pow er Based Protection
A lgorithm for Stable Pow er Swing Conditions

Table 6-3 shows the results for the recoverable power swing simulations. The power based
pole slipping scheme and single blinder scheme successfully restrained for all of the stable
power swing simulations. The Mho scheme nuisance tripped for 8 of the 12 simulations.
The double blinder scheme nuisance tripped for 1 of the 12 simulations. The loss of field
scheme remained stable for all of the simulations, showing that the 0.6 s time delay
blocked relay operation during such conditions.

Table 6-3
Results of Comparisons Between Conventional and New Pole Slipping Protection
Methods for PPSS Stable Power Swing Simulation Tests

Test Loss of Mho Pole Single Double Power

Field slipping Blinder Blinder Based
Algorithm

PS W INN NT 1.188* NT NT NT

PSW2NN NT NT NT NT NT

PSW3NN NT NT NT NT NT

PSW1GN NT 1.431* NT NT NT

PSW2GN NT 1.094* NT NT NT

PSW3GN NT 1.176* NT NT NT

PSW1NA NT 1.431* NT NT NT

PSW2NA NT NT NT NT NT

PSW3NA NT 0.670* NT 0.846* NT

PSW1GA NT 1.431* NT NT NT

PSW2GA NT NT NT NT NT

PSW3GA NT 0.702* NT NT NT

* - signifies that the relay did not operate as required.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 159

Test PSW3GA'
Figure 6.41 shows the Mho characteristic and the stable power swing impedance locus for
test PSW3GA. The impedance locus entered the Mho characteristic at a time of 0.683 s,
and remained inside until 0.745 s. It was therefore inside the characteristic for 62 ms. This
was ample time to satisfy the 20 ms time delay applied to the scheme to enhance its
security towards stable power swings. If the time delay had been increased to 65 ms to
prevent the nuisance trip, the scheme would fail to detect fast pole slips because the
impedance locus would pass through the characteristic in under 65 ms. If the Mho circle
size is reduced to prevent the nuisance trip, some pole slips may go undetected.

The Mho scheme nuisance tripped in the same way for 7 of the other stable power swing
tests. This confirms the weaknesses of the Mho scheme stated in Section 2.1.2.

Test 'PSW3NA'
A section of the double blinder characteristic, and the impedance locus for test 'PSW3NA'
are shown in Figure 6.42. The double blinder scheme nuisance tripped for this simulation.
After fault clearance at a time of 0.52 s, the impedance trajectory travels from inside the
two inner elements to outside the right outer blinder element. At 0.57 s, the supervisory
Mho element is picked up, but the two right hand blinder units are not. The locus then
heads back to the left and crosses both of the blinders with a time delay of approximately
0.13 s. This 'seals in' the first stage of the double blinder logic. The locus then stays in
between the inner blinder elements for long enough to seal the next stage of the scheme's
logic. It then crosses back over the two right hand blinder units in 0.12 s, satisfying the
final stage of the scheme logic, and issues a trip signal at a time of 0.85 s.

The small 'oval' in the impedance trajectory is produced as the generator swings up to the
CSP. After the fault clearance, the generator swings up to the CSP, and in doing so passes
over the peak of its real power load angle curve. It then swings back towards a stable
operating point, and again passes over the peak of its power output curve. In terms of
time, the CSP is reached at 0.70 s on the record.

Although the double blinder scheme only nuisance tripped once for the 12 stable power
swing tests conducted, this nuisance trip highlights one of the problems with the double
blinder scheme. Careful setting is required to ensure that this kind of nuisance trip does
not occur. The 'oval' which occurred in the impedance locus often occurs during severe

M.J.Checksfi el d/PhD /University of Bath 1996 Chapter 6, Page 160

stable power swings. The inherent security of the double blinder scheme is much lower
than for the single blinder scheme, because the double blinder scheme does not require
the impedance locus to enter from one side and leave from the other.

6.3.7.3 Comparison of Impedance Relays with New Power Based Protection

Algorithm for Other Power System Conditions

Table 6-4 contains the results of the performance comparison tests for the loss of field,
loss of prime mover, and operation of the generator in the dynamic stability region tests.

The loss of field and power based relays both operated for all of the loss of field tests.
The single and double blinder schemes did not operate. The Mho scheme operated for the
loss of field from 100 % initial load test.

The loss of field scheme and the new power based algorithm nuisance tripped for the
simulations where the generator was operated at load angles of greater than 90°. This is
not considered a problem because generators are never operated in this way in practice
because the reserve in hand for transient stability is too low.

All of the protection relays remained stable during the loss of prime mover tests.

Test 'LOF3GA’
Figure 6.43 shows the impedance locus and loss of field relay operating characteristic for
test 'LOF3GA'. The impedance locus stays within the operating characteristic almost
continuously for this test. The circle which occurs in the impedance locus is due to the
effects of magnetic saliency in the generator rotor structure.

The loss of field relay operated before the power based relay because it only required the
generator to import large amounts of reactive power to trip. The power based algorithm
on the other hand, requires the onset of asynchronous operation before it will trip. The
difference in tripping times is most pronounced at low initial operating powers. If the
generator was operating at zero initial operating power, the power based algorithm would
not trip because the generator would not lose synchronism. An additional power based
algorithm, requiring almost no additional relay processor time could be used alongside the
power based pole slipping algorithm to provide full loss of excitation protection11601.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 161

Test 'AVR5NA1
Figure 6.44 contains the impedance locus for test 'AVR5NA1. The generator was initially
operating at point 'A'. The AVR reference voltage was then reduced, causing a reduction
in generator excitation in an attempt to regulate the terminal voltage to the new reference
point. This resulted in the generator importing reactive power, which moved the
impedance locus to point 'B\ The loss of field relay therefore tripped because point ’B'
is located inside the relay trip area. At a time of 10 seconds into the simulation, the
generator prime mover input power was reduced from 0.5 pu to zero. The change in power
moved the impedance locus from point 'B' to point *C\ Point 'C' was still inside the loss
of field relay trip zone. After several seconds of operation at point 'Cf, the AVR reference
voltage was returned to unity, causing the impedance locus to return to the positive
reactive power region.
Table 6-4
Results of Comparisons Between Conventional and New Pole Slipping Protection
Methods for PPSS AVR Assisted High Load Angle, Loss of Field and Loss of Prime
Mover Simulation Tests.

Test Loss of Mho Pole Single Double Power

Field slipping Blinder Blinder Based
Algorithm

AVR4NA 1.796* NT NT NT 10.345*

AVR5NA 2.406* NT NT NT 10.330*

AVR6NA 1.796* NT NT NT 10.385*

LOF1GA 3.488 NT NT NT 6.825

LOF2GA 3.716 NT NT NT 4.870

LOF3GA 2.910 3.384 NT NT 3.030

LOP1NA NT NT NT NT NT

LOP2NA NT NT NT NT NT

LOP3NA NT NT NT NT NT

* - signifies that the relay did not operate as required.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 162

6.3.8 Analysis of the Performance of Current, Voltage and Frequency Relays to the
PPSS Pole Slipping Conditions

The performance of under and over-voltage, under and over frequency, and over-current
relays for detecting pole slipping conditions was analyzed using the PPSS test simulation
files and PC based models of the relays. Table 6-5 contains the results for all of the PPSS
pole slip tests. For all of the tests, the effects of the fault on the relays which caused the
pole slip have been neglected, since a large majority of the faults used to induce pole
slipping caused the relays to operate. The analyses performed in this section will solely
concentrate on the effects that a pole slipping generator will have on these relays.

Table 6-5
Response of Current, Voltage and Frequency Relays to
Pole Slipping Conditions in PPSS Test Files

Test Over-Current trip Under-Voltage trip Over-Frequency trip

PSL1NN no trip no trip no trip

P2SL2NN no trip no trip no trip

PSL3NN TRIP TRIP no trip

PSL31GN no trip no trip no trip

PSL2GN no trip no trip no trip

PSL3GN no trip no trip no trip

PSL1NA no trip no trip no trip

PSL2NA no trip no trip no trip

PSL3NA TRIP no trip no trip

PSL1GA no trip no trip no trip

PSL2GA no trip no trip no trip

PSL3GA no trip TRIP no trip

PSL1GAF no trip TRIP no trip

PSL2GAF no trip TRIP no trip

PSL3GAF no trip TRIP no trip

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 163

The analysis shows that these relays do not provide protection against pole slipping
conditions, as suggested by Engineering Technical Report 113[15]. Although this study
system has not focused specifically on embedded generators, the results apply equally to
embedded machines. The main difference between the two environments is that pole
slipping is more likely to occur with embedded generators.

6.3.8.1 Response of Over-Current Relay to Pole Slipping Conditions

The PC based 'C' code model of the Voltage Controlled IDMT over-current relay1451 used
in the analysis can be found in Appendix F. The over-current relay operated for 2 of the
15 pole slipping tests. The following settings were used;

Plug Set Multiplier Setting (PSM) = 1.5 * Generator Full Load Current
Time Multiplier Setting (TMS) = 0.1
Voltage Control Setting = 70 % Nominal Voltage
IDMT Curve Type = Standard Inverse.

These settings are sensitive, yet the IDMT relay still failed to detect most of the pole
slipping conditions. Figure 6.45 contains the Phase A RMS currents and IDMT trip points
for tests PSL3NN, PSL2GA, and PSL3GA. The phase B and C currents will be identical
for the pole slipping section of the plot because pole slipping is a balanced phenomenon.

Figure 6.45a shows the current for test 'PSL3NN'. The first trip occurs as a result of the
fault, the second occurs due to the pole slipping. A total of 17 pole slip cycles were
required to operate the relay. Although the mean level of current was 2.75 pu, and the
maximum level around 4 pu, the IDMT relay did not trip until the minimum current level
approached the relay pick-up current setting of 1.5 pu. The minimum current is the
important quantity when considering the performance of IDMT relays at detecting pole
slipping conditions. The relay simulated used a 5 cycle exponential reset function[57]. This
meant that for the first 10 pole slip cycles, the relay reset after every cycle. One pole slip
cycle is not sufficient to operate the relay, the relay therefore failed to trip until a reset
did not occur at the end of each cycle.

Figure 6.45b shows the current for test TSL2GA'. The over-current relay failed to trip
for the three pole slip cycles that occurred, for the same reasons as above. The peak
current during the pole slip cycle lasted for much longer in this test, but was still not

M.J.Checksfield/PhDAJniversity of Bath 1996 Chapter 6, Page 164

sufficient to operate the over-current relay. A current level of 4 pu, in conjunction with
the relay settings used would require the current to remain at this level for 700 ms before
the relay will trip. If the voltage control element operates, the PMS setting would be
adjusted by a factor of 0.4, changing it from 1.5 to 0.6*In. With the new PMS setting of
0.6*In, a fault duration of 362 ms is still required. This was just longer than the current
duration that occurred during the test.

Figure 6.45c shows the current for test 'PSL3GA1. The overcurrent relay did not operate
because the IDMT relay was reset after each pole slip cycle.

The above analysis shows that unless very sensitive settings are used, IDMT type over-
current relays do not provide effective pole slipping protection because the relay is reset
between pole slip cycles. A long delay on reset function would alleviate this problem..

6.3.8.2 Response of Under-Voltage Relay to Pole Slipping Conditions

The under-voltage relay operated for 5 of the 15 pole slipping tests conducted. The relay
trip setting was 90 % of rated volts, with a time delay of 0.5 seconds. The time delay
enables the relay to 'ride through' most short circuit faults. Figure 6.46 shows the phase
'A' RMS voltage for tests 'PSL3NN', 'PSL2GA', and 'PSL3GAF'. The 200 MVA generator
used in the PPSS model is connected to an infinite bus through a relatively weak tie line.
The system centre is therefore close to the terminals of the generator, rather than far inside
the generator. This results in large fluctuations in voltage during pole slipping, or when
excessive amounts of reactive power are drawn from the system.

Figure 6.46a shows the voltage, under-voltage trip, and trip level for test 'PSL3NN'. The
under-voltage relay tripped after 25 pole slip cycles. As with the over-current relay, the
under-voltage relay is inhibited from tripping because it is reset during every pole slip
cycle. For this test, the reactive power consumption becomes so large that eventually no
resets occur, and the relay trips. Figure 6.19a shows the reactive power for part of this
test, which decreases as the pole slipping continues. If the generator had been connected
to a 'stiffer' bus, the under-voltage relay may not have tripped because it would have reset
during the 'in phase' part of each pole slip cycle.

Figure 6.46b shows the response for test 'PSL2GA'. This test has a slower rate of pole
slipping, and shows the extent to which pole slipping can disturb the system voltage.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 165

However, the disturbances are still insufficient to cause the relay to trip for the pole slip.
It did trip during the fault however.

Figure 6.46c shows the response for test 'PSL3GAF1. This pole slip was caused by a
faulty AVR. As the reactive power absorbed by the generator increases, the terminal
voltage decreases. Figure 6.23a shows a plot of the reactive power for this test. The
decreasing terminal voltage causes the under-voltage relay to trip before the generator
starts to operate asynchronously. The relay remains tripped for the first 4 pole slip cycles,
and is then reset. The voltage fluctuations increase towards the end of the test as a result
of the field forcing action of the AVR. Although faulty, the AVR still responds to the
voltage depression.

6.3.8.3 Response of Over-Frequency Relays to Pole Slipping Conditions

The over-frequency relay did not operate for any of the 15 pole slipping tests conducted.
The relay was set to trip at 1 % above the nominal frequency of 50 Hz, with a time delay
of 0.4 seconds. The PC based 'C' code model of the frequency measuring device[182] used
in the analysis can be found in Appendix F. As with the under-voltage and over-current
relays, the over-frequency relay was inhibited from tripping because of the cyclic
fluctuations in measured frequency which occur during pole slipping.

Figure 6.47a shows the measured frequency for test 'PSL3NN'. A glitch occurs in the
measured frequency during the fault on and fault off points. This occurs because the
position of the voltage waveform's zero crossing changes rapidly. The frequency
measurement soon returns to its correct value. The frequency can be seen to rise during
the fault because the generator rotor accelerates. Once the fault has been cleared, the
frequency oscillates wildly at the rate of pole slipping. Although the actual generator
rotor is travelling at least 1 % above synchronous speed, the measured frequency does not
show this. The fluctuations in frequency are produced because of the fluctuations in
voltage phasor which occur due to the large fluctuations in real and reactive power, as
shown in Figure 6.19. If the input to the frequency relay was derived from the generator
rotor speed, the relay would have tripped.

Figure 6.47b shows the result for test 'PSL2GA'. The frequency can be seen to rise during
the fault period, then fluctuate at the rate of pole slipping afterwards. The fluctuations are
less severe than in test 'PSL3NN' because the fluctuations in real and reactive power are

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 166

not as severe. Note that even after the generator has re-synchronised, at a time of
2 seconds on the record, small fluctuations in frequency still occur. Figure 6.16 shows
the real and reactive power fluctuations for this test.

Figure 6.47c shows the frequency plot for test 'PSL3GAF'. There is no fault in this test.
The fluctuations observed in frequency are therefore solely due to pole slipping. The
frequency remains at 1 pu until the generator loses synchronism. It then begins to fluctuate
as the oscillations in real and reactive power build up, as shown in Figure 6.23. The
magnitude of the frequency fluctuations increases with the magnitude of the real power
fluctuations. The fluctuations make the over-frequency relay reset very often, the 0.4
second time delay is therefore never satisfied.

6.4 625 KVA DIESEL GENERATOR FIELD TRIAL RESULTS

The power based pole slipping algorithm successfully tripped for all of the pole slipping
conditions. For the loss of excitation tests, the algorithm detected the loss of excitations
from high initial operating powers, but failed to detect those which occurred from 50 %
and 20 % load because the generator did not lose synchronism over the duration of the
test period. The algorithm restrained for all other operating conditions, such as
synchronisation, local load changes, adjacent generator pole slipping, and all generator
load and excitation changes.

6.4.1 625 kVA Diesel Generator Pole Slip Tests

Figure 6.48 shows the algorithm response to test 'A2\ An attempt was made to induce
pole slipping from 50 % initial load by weakening the field excitation. However, the
excitation could not be sufficiently weakened to do this. The generator therefore did not
pole slip, and the algorithm did not trip. Figure 6.48a shows that the reactive power
criterion was satisfied, but the real and rate of change of power trip criteria were not.

4 Hz oscillations occurred in the real power for this test. Their magnitude was 10 kW
peak. This is equal to 1.6 % of generator rating, the setting Ptol is nominally set to 5 %.
The oscillations did not cause any problems to the adaptive real power trip setting section
of the algorithm. Oscillations in real power are common in diesel generators, and take
two forms, either forced or natural. Forced oscillations occur as a result of the periodic

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 167

torque produced by the diesel engine. Generator-diesel units are normally designed to
minimise the effects of such oscillations11831. Natural oscillations can occur for a variety
of reasons, such as poor AVR or governor performance. An adequate explanation for the
oscillations in real power for this 625 kVA diesel generator has not been put forward.
Anderson[184] suggests that these oscillations may be due to poor quality fuel.

It is important for the algorithm to track the output power, regardless of the oscillations.
It is equally important that the algorithm does not track the oscillations caused by power
swings. The field trial tests therefore provide data on this subject.

Figure 6.49 shows the power based pole slipping algorithm measurands for test fA3\ The
plots are made up of two disturbance records which were taken consecutively. A pole slip
was induced by decreasing field excitation. The algorithm tripped before the generator
pole slipped, at a real power of 306 kW.

The algorithm also tripped during the oscillations in real power which occurred after the
pole slip. These do not normally cause a problem. However, the reactive power plot
shows that the load angle was increasing and a pole slip was about to occur. The increase
in load angle resulted in the reactive power criterion being satisfied. The oscillations in
real power satisfied the remaining two criteria. This trip is not seen as problematic, since
the generator was close to pole slipping again. The problem could easily be overcome by
increasing the duration of time for which Pmax is held before being reset to its nominal
value. This would increase algorithm security, without effecting sensitivity towards
detecting steady state pole slips. A period of two seconds would ensure that this type of
problem would not occur because the damping in most synchronous machines is such that
most oscillations have abated in this period.

Figure 6.50 shows the disturbance records for test 'A4', where an adjacent generator was
forced to pole slip. The algorithm remained stable for this test. The reactive power plot
has the reactive power demands of the adjacent pole slipping generator translated onto it.
As the demand for reactive power by the pole slipping machine increases, the protected
generator supplies some of the increased demand. This takes the reactive power output
further away from the trip level, Qtrip. This test demonstrates that the algorithm is not
liable to trip for adjacent generators which are pole slipping.

M.J.Checksfield/PhDAJniversity of Bath 1996 Chapter 6, Page 168

The real power plot contains the 4 Hz oscillations, which appear to be triggered by the
disturbances from the adjacent generator. This suggests that the oscillations are triggered
via some form of voltage feedback mechanism, since the generators were completely
independent apart from being connected to the same busbar. The peak to peak amplitude
of the oscillations was 100 kW, which equates to 16 % of generator rating. The algorithm
still successfully updated Pt, the setting for a value of Ptol of 5 %. This setting is
therefore adequate. It is unlikely that oscillations more severe than this would occur.

Figure 6.51 shows the combined disturbance records for test 'A7\ Pole slipping was
induced by increasing the diesel engine's output power. The increase in real power from
250 to 575 kW over 3 seconds was the fastest rate possible; the servo-mechanism in the
governor limited it to this rate. The algorithm correctly updated its Pt setting just after
the maximum power output had been reached. The algorithm then tripped at a power
level o f 428 kW. It therefore tripped before the pole slip occurred. After the first pole
slip, the machine's output power rose to 2.6 MW, which is equivalent to five times the
rating of the generator. Allowing for a circuit breaker interrupting time of 100 ms, the
algorithm would have disconnected the generator before the power reversed sign, the
algorithm would therefore have avoided the potentially damaging variation in output
power.

The generator slipped another pair of pole before re-synchronising. The algorithm tripped
for this pole slip also.

6.4.2 625 kVA Diesel Generator Loss of Excitation Tests

The algorithm tripped for both of the loss of excitation from 100 % initial load tests, but
did not trip for the 25 % and 50 % initial load cases. It did not trip for the low initial
load tests because the generator did not lose synchronism. The saliency in its rotor
structure provided enough reluctance torque to maintain synchronism. An additional power
based algorithm, using a rectangular setting characteristic in the PQ plane tripped for the
lower power loss of excitations*160]. If this type of algorithm is used to compliment the
power based pole slipping algorithm, then full loss of excitation protection, which covers
the generator's entire operating range, can be achieved.

Figure 6.52 shows the algorithm measurands for test 'B2', loss of excitation from 20 %
initial load. The reactive power does not satisfy the Qtrip criterion because the load angle

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 169

has not increased to 90°. The real power criterion was not satisfied either. The algorithm
therefore did not trip.

Figure 6.53 shows the results to test 'B3', loss of excitation from 50 % initial load. The
reactive power criterion was satisfied, indicating that the load angle had advanced beyond
90°. The real power only dropped by 10 kW however, which was insufficient to satisfy
the criterion, the algorithm therefore did not trip. It is possible for the generator to
temporarily maintain a level of output power above the level predicted from the steady
state saliency term of equation 3(1). Inspection of equation 3(3) shows that the transient
saliency term may be sufficient to temporarily hold the rotor in synchronism. A
simulation was performed to confirm this, the results of which are shown in Figure 6.54.
The ATP file used for the simulation can be found in Appendix F. The power remained
constant for 12 seconds, and then the generator lost synchronism. The field trial loss of
excitation tests were run for 8 seconds or so. The generator was disconnected after this
time. The generator may therefore have lost synchronism if the test had been continued
for longer. Figure 6.54d shows the effect that the excessive reactive power demand can
have on generator terminal voltage.

Figure 6.55 shows the algorithm response to test 'B6', loss of excitation from 100 % initial
load. The algorithm successfully tripped for this test because the generator lost
synchronism before the generator was disconnected. Note that even during loss of field
conditions, the power pulsations produced still reach 1.5 MW. The loss of excitation was
correctly diagnosed as a 'steady state' trip.

6.4.3 625 kVA Diesel G enerator Steady State and Load Tests
The algorithm remained stable for all of the steady state and load change tests. Figure 6.56
shows the algorithm measurands for steady state test 'C3'. The reactive power plot shows
the quantisation levels as a result of the disturbance recorder storage mechanism. The 'C'
code for the algorithm uses 32 bit integers for its variables such as the reactive power, but
to optimize storage space, 16 bit integers are used for the disturbance records. Details can
be found in Appendix E.

Figure 6.57 shows the measurands for load change test 'C6f. The load change can be
seen in the reactive power output, which step changes by 5 kVAr. The load change did
not cause the algorithm to come even close to tripping.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 170

6.4.4 Comparison of Conventional Impedance Based Pole Slipping Relays with
Power Based Relay for 625 kVA Diesel Generator Pole Slip Tests

The relay models described in section 6.3.6 were used for this analysis. Details on the
settings used can be found in Appendix B. The mho, single blinder, and double blinder
pole slipping schemes all failed to detect either of the pole slips which occurred in field
trial tests 'A3f, and 'A7'.

Figure 6.58a shows the single blinder pole slipping scheme and the impedance locus for
test 'A3'. The pole slip occurs between the time of 0 and 0.2 on the RX plot. The
impedance locus shows that the single blinder scheme failed to detect the pole slip
because the impedance locus passed underneath the scheme’s characteristic. Inspection of
Figures BIO, B l l , and B12 of Appendix B, which contain the mho, single and double
blinder schemes trip characteristics, shows that they all occupy approximately the same
area on the R-X plane. This is the area bounded by the generator impedance, Xd\ This
confirms that the mho and single blinder schemes would also fail to operate for this pole
slip, because the impedance locus passed beneath their operating characteristics.
Comparison between this figure and Figure 6.49, which shows the power based pole
slipping algorithm response for the same test provides a means of correlation between the
power, current and voltage produced during pole slipping.

Figure 6.58b and 6.58c show the generator terminal voltage and current for the same test.
The effect of pole slipping on the terminal voltage can clearly be seen. The voltage is
almost depressed to 90 % of its nominal value. If the generator were to continually pole
slip, the voltage depressions would continuously remain at this level. The current plot
shows the high current levels which occur during pole slipping, which in this case reached
2.5 pu.

Figure 6.59a shows the single blinder characteristic, and impedance locus for test pole
slip 'A7\ All of the conventional impedance based pole slipping schemes failed to detect
either of the pole slips for the same reason as in test 'A3'. The impedance locus passed
underneath their operating characteristic. The first pole slip occurs between 0 and 0.2
seconds, whilst the second occurs between 0.4 and 0.6 seconds.

M.J.Checksfi el d/PhD/University of Bath 1996 Chapter 6, Page 171

The voltage and current for this test are shown in Figures 6.59b and 6.59c. The voltage
depression which occurs leading up to and during the pole is a 10 % depression, whilst
the current approaches 3 pu. After the pole slips occurred, the peak in current reaches
5 pu. This could cause cumulative damage to the machine windings.

6.5 STEAM TURBINE AND HYDRO GENERATOR FIELD TRIAL RESULTS

At the time of writing, the two field trial relays had been in place for 10 weeks. Neither
relay nuisance tripped during this period showing that the power based algorithm is stable.
No pole slips have been reported for either generator, the algorithms ability to trip has
therefore not been demonstrated.

Three disturbance records have been triggered by the over-current element on the turbo­
generator, one of these records is shown in Figure 6.60. The over-current element showed
this to be a phase to phase fault, which was remote from the generator because it caused
little disturbance to the generator's output power. The reactive power did not fall even near
to the trip level, Qtrip. The algorithm therefore remained very secure for this fault. The
disturbance records for the other two faults were virtually identical, and were therefore not
included.

Figure 6.61 shows a disturbance record taken when the turbo-generator was in steady
state. Figure 6.62 shows a steady state disturbance record taken from the hydro-generator
relay. These plots show that the oscillations experienced during the diesel generator field
trials did not occur for either type of machine. Since the Pt section of the algorithm
worked with the diesel generator, it has no problem whatsoever with these types of
generator.

The noise levels in the (AP/At) signal were +/- 6 MW, which equates to the same
magnitudes experienced with the diesel generator and PPSS tests; the level was 23 % of
generator rating.

M.J.Checksfi el d/PhD /University of Bath 1996 Chapter 6, Page 172

 6.6 ADDITIO NAL ATP SIM ULATION TEST RESULTS

The pole slipping algorithm tripped for all the generator and motor pole slipping tests, and
restrained for all the stable power swing simulations.

Figure 6.63 shows the algorithm measurands for test 'LT11, a simulated power swing with
a generator open circuit time constant, Tdo' of 1.0 second. This is very low, and distorts
the 'ideal' sinusoidal power load angle characteristics produced if a constant field flux
linkage is assumed. The distortion is most evident in the real power plot. The algorithm
remained stable for this simulation because the reactive power criterion was only satisfied
when the real power was above its trip level, Pt.

Figures 6.64 and 6.65 shows the plots for the stable swing tests where Tdo' was set to
2 and 3 seconds. The algorithm remained secure for these tests for the same reasons as
above.

Figure 6.66 shows the power plots for test 'LTP1', pole slipping with the value of Tdo' set
to 1 second. The generator pole slipped one pair of poles before re-synchronising. The
algorithm successfully detected this pole slip, which would have been diagnosed as a
'transient' pole slip due to the large fluctuations in (AP/At) caused by the fault which
induced pole slipping.

Figures 6.67 and 6.68 show the results for tests 'LTP2' and 'LTP3'. The algorithm
successfully tripped during these tests, and diagnosed the pole slips as 'transient'. Note that
the higher the value of Tdo', the more pole slips occurred before the generator re­
synchronised. Equation 3(7) of chapter 3 shows that as the time constant, Tdo' increases,
the amount of asynchronous power which is generated decreases. A shorter time constant
therefore enables the generator to generate more asynchronous power. This provides the
necessary additional retardation energy required for the generator to regain synchronism.

Figures 6.69 and 6.70 show the results for the fast governor simulations. The algorithm
did not trip for either of these simulations because the reactive power criterion reset before
the real power criterion was satisfied. These tests show that the algorithm will work with
fast governors. It is unlikely that governors with a faster response will be used in practice.

M.J.Checksfi el d/PhD/University of Bath 1996 Chapter 6, Page 173

Figure 6.71 shows the modified 'motor' power based pole slipping algorithm measurands
for the stable power swing for a hydro-generator in pumping mode test. The reactive
power criterion plot is identical to that produced by the 'generator' pole slipping algorithm.
The real and rate of change of power plots are different however. They both have been
reflected across the x-axis so that Pt and Pmax are now negative numbers, and (AP/At)min
is a positive number. The algorithm did not trip for this test because when the reactive
power criterion was satisfied, indicating that the motor's load angle had reached 90°, the
real power was less than its trip setting Pt. This indicated that there was still accelerating
power left, and that the motor's CSP had not been reached.

Figure 6.72 shows the modified 'motor' pole slipping algorithm's response for the hydro­
machine pole slip from pumping mode. The algorithm tripped for the pole slip, and
diagnosed it as the 'transient' variety. The algorithm did not reset between pole slip cycles
due to the reactive power being constantly below the trip level Qtrip,

Figure 6.73 shows the plots for test 'MOT3', the pole slipping of a hydro-generator in
pumping mode due to a 20 % decrease in terminal voltage. The algorithm tripped for all
of the 11 pole slip cycles which occurred before the machine governor reduced the motor
load torque so that the motor could re-synchronise. The pole slip was diagnosed as 'steady
state', which in conjunction with information about the voltage depression that occurred
before the pole slip, would have enabled the protection engineer to ascertain the cause.

Figure 6.74 shows the algorithm response to the slow pole slipping rate test 'SLW'. The
generator pole slipped at a mean rate of 2 to 3 pole slips per minute. A pole slip rate of
3 cycles/minute equates to a mean slip of 0.1 %. The algorithm successfully tripped for
the slow pole slip, and diagnosed it as a 'steady state' type. This simulation shows that
the algorithm meets the 0.1 % slip specification used by most other pole slipping schemes.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 6, Page 174

 a) Load Angle
200
Load angle (degrees)

100 TRIP TRIP

-100

-200
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Fault on Time (s)
b) Generator Reactive Power
600 Fault off
Reactive power (MVAr)

400
200

-200
-400
-600
-800 Q criterion satisfied f
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Time (s)
c) Generator Real Power
Pmax
1,000
Real power (MW)

500

0
TRIP TRIP
-500
P criterion satisfied
- 1,000
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Time (s)
d) Generator Rate of Change of Power (RCP)
5.000
RCP (MW/s)

-5,000 (AP/At)min
10.000
15.000 -50.000 MW/s
20,000
I (AP/At) criterion satisfied \
0 0.2 0.4 0.6 0.8
Graph limited Time (s)
for accuracy
Figure 6.1
Pole Slipping Algorithm Response to Simulated Pole Slip Using POWSIM.

M. J. Checksfi el d/PhD/University of Bath Chapter 6, Fig 6.1, Page 175

 a) Load Angle
160 NO TRIP
Load angle (degrees)

140
120
100 90 degrees

60
40

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8

Fault on Time (s)
b) Generator Reactive Power
600
Reactive power (MVAr)

400
200
0
Qtrip
-200
-400
-600
-800
\i1 Q.....
criterion satisfied
"I---------1—
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
Time (s)
c) Generator Real Power
Pmax
1,000
800
Real power (MW)

600
400
200 NO TRIP

-200
criterion satisfied
-400
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
Time (s)
d) Generator Rate of Change of Power (RCP)
4,000
2.000
RCP (MW/s)

2.000 (AP/At)min

4.000
6,000
8,000 50,000 MW/s J \ (AP/At) criterion satisfied J ^
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
Graph limited Time (s)
for accuracy
Figure 6.2
Pole Slipping Algorithm Response to Simulated Stable Power Swing Using POWSIM.

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.2, Page 176

 KEY Trip Level Pmax

a) Generator Reactive Power

1000
^ Parallel field
resistor inserted
Reactive Power (VAr)

-1000 ■■

-2000 ■■

-3000 ■■

-4000 ■■

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
Time (s)

b) Generator Real Power

3000 ■■
Real Power (W)

2000

1000 ■■

TRIP

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
Time (s)

c) Generator Rate of Change of Power (RCP)

4000

2000
RCP (W/s)

-2000

-4000

-6000
-37,500 W/s
-8000
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
Time (s)

Figure 6.3
Pole Slipping Algorithm Response to a W eak Field' Pole Slip on
the Laboratory Model Power System 5 kVA Generator - Test 'C'

M. J. Checksfield/PhD/University o f Bath Chapter 6, Fig 6.3, Page 177

 KEY Trip Level Pmax

a) Generator Reactive Power

1000 f
Reactive Power (VAr)

-1000 ■■

-2000 ■ ■'

-3000 ■■

-4000 ■■

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
Time (s)

b) Generator Real Power

5000

4000
Real Power (W)

3000

2000

1000
TRIP TRIP

-1000
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
Time (s)

c) Generator Rate of Change of Power (RCP)

4000

2000
RCP (W/s)

-2000

-4000

-6000

-8000
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
Time (s)

Figure 6.4
Pole Slipping Algorithm Response to an 'Over Torque' Pole Slip on the Laboratory
M odel Power System 5 kVA Generator - Test 'e' (low excitation, 4 kW o f local load)

M. J. Checksfield/PhD/University o f Bath Chapter 6, Fig 6.4, Page 178

 KEY Trip Level Pmax

a) Generator Reactive Power

2000
Reactive Power (VAr)

-2000

-4000

-6000

-8000
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
Time (s)

b) Generator Real Power

8000

6000 ■■
Real Power (W)

4000

2000 ■■

TRIP

-2000
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
Time (s)

c) Generator Rate of Change of Power (RCP)

RCP (W/s)

-5000 ■ ■

-10000 ■■

-15000
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
Time (s)

Figure 6.5
Pole Slipping Algorithm Response to an 'Over Torque' Pole Slip on the Laboratory
Model Power System 5 kVA Generator - Test 'F' (medium excitation, 4 kW o f local load)

M. J. Checksfield/PhD/University o f Bath Chapter 6, Fig 6.5, Page 179

 KEY Trip Level Pmax

a) Generator Reactive Power

4000

2000 ■■

-2000 ■

•3 -4000 ■■

-6000 ■■

-8000
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
Time (s)

b) Generator Real Power

6 ■■
¥

I
C L,

cC TRIP

-2 ■■

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
Time (s)

c) Generator Rate of Change of Power (RCP)

-5000 ■ ■

—
i
-10000 ■■

-15000 ■■

-20000 ■■

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
Time (s)

Figure 6.6
Pole Slipping Algorithm Response to an 'Over Torque' Pole Slip on the Laboratory
Model Power System 5 kVA Generator - Test 'G' (high excitation, 4 kW o f Local Load)

M. J. Checksfield/PhD/University o f Bath Chapter 6, Fig 6.6, Page 180

 KEY Trip Level Pmax

a) Generator Reactive Power

1500

1000 ■■
Reactive Power (VAr)

-1000 ■■

period o f adjacent generator pole slipping

-1500 ■■

-2000 ■■

-2500
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
Time (s)

b) Generator Real Power

4000

3500

3000
Real Power (W)

2500

2000

1500

1000

500 NO TRIP

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
Time (s)

c) Generator Rate of Change of Power (RCP)

RCP (W/s)

-5000 ■

10000 ■■

-15000
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
Time (s)

Figure 6.7
Pole Slipping Algorithm Response to an Adjacent 5 kVA Generator Suffering an 'Over
Torque' Pole Slip on the Laboratory Power System Model - Test 'K'

M. J. Checksfield/PhD/University o f Bath Chapter 6, Fig 6.7, Page 181

 KEY Trip Level ----------------------- Pmax

a) Generator Reactive Power

2000

1000

<5
I-
Q
> -1000 ■■
Adjacent generator loses excitation
I“ -2000 ■■

-3000
o.oo 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
Time (s)

b) Generator Real Power

3000

2800

^ 2600

2400

2200

J 2000

1800
NO TRIP
1600

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
Time (s)

c) Generator Rate of Change of Power (RCP)

2000

Cu
y
& -2000

-4000

--------------1--------------1-------------- 1------------- 1--------------1 I---------------1--------------1--------------1—

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
Time (s)

Figure 6.8
Pole Slipping Algorithm Response to an Adjacent 5 kVA Generator Loss o f Excitation on
the Laboratory Model Power System 5 kVA Generator - Test 'L'

M. J. Checksfield/PhD/University o f Bath Chapter 6, Fig 6.8, Page 182

 KEY Trip Level ----------------------- Pmax

a) Generator Reactive Power

1000
500 ■■
Reactive Power (VAr)

-500 ■■

-1000 ■■

-1500 ■■

-2000 ■■

-2500
0.00 0.38 0.75 1.12 1.50 1.88 2.25 2.62 3.00 3.38
Time (s)

b) Generator Real Power

4500 ■■
Fault on Fault off
Real Power (W)

4000

3500 ■■

NO TRIP
3000
0.00 0.38 0.75 1.12 1.50 1.88 2.25 2.62 3.00 3.38
Time (s)

c) Generator Rate of Change of Power (RCP)

-5000
RCP (W/s)

-10000

-15000

-20000
0.00 0.38 0.75 1.12 1.50 1.88 2.25 2.62 3.00 3.38
Time (s)

Figure 6.9
Pole Slipping Algorithm Response to a Single Phase to Earth Fault on the Laboratory
Model Power System 5 kVA Generator, Generator Neutral Isolated - Test 'IB '

M. J. Checksfield/PhD/University o f Bath Chapter 6, Fig 6.9, Page 183

 KEY Trip Level Pmax

a) Generator Reactive Power

3000

2000
Reactive Power (VAr)

1000 ■■

-1000 ■■

-2000 ■■

-3000
0.00 0.38 0.75 1.12 1.50 1.88
Time (s)

b) Generator Real Power

5000
Fault on
Fault off
4000
Real Power (W)

3000 ■■

2000 ■■
NO TRIP

0.00 0.38 0.75 1.12 1.50 1.88

Time (s)

c) Generator Rate of Change of Power (RCP)

o ■■
= h f ------------------------
-20000 ■•

-40000 ■■
RCP (W/s)

-60000 ■■

-80000 ■■

•100000 ■■
•120000 ■■

-140000 ■'
0.00 0.38 0.75 1.12 1.50 1.88
Time (s)

Figure 6.10
Pole Slipping Algorithm Response to a Phase to Phase Fault on the Laboratory
Model Power System 5 kVA Generator, Generator Neutral Isolated - Test *2A'

M. J. Checksfi el d/PhD/University o f Bath Chapter 6, Fig 6.10, Page 184

 KEY Trip Level Pmax

a) Generator Reactive Power

3000

2000 ■■
Reactive Power (VAr)

1000 ■■

-1000 ■■

-3000 ■■

-4000 ■■

-5000
0.00 0.38 0.75 1.12 1.50 1.88 2.25 2.62 3.00 3.38
Time (s)

b) Generator Real Power

6000

Fault on
5000 ■■
Real Power (W)

4000

3000 ■■

2000
Fault off
1000
NO TRIP

0.00 0.38 0.75 1.12 1.50 1.88 2.25 2.62 3.00 3.38
Time (s)

c) Generator Rate of Change of Power (RCP)

RCP (W/s)

-5000 ■■,

10000 ■■

-15000
0.00 0.38 0.75 1.12 1.50 1.88 2.25 2.62 3.00 3.38
Time (s)

Figure 6.11
Pole Slipping Algorithm Response to a Two Phase to Earth Fault on the Laboratory
Model Power System 5 kVA Generator, Generator Neutral Isolated - Test '3B'

M. J. Checksfield/PhD/University o f Bath Chapter 6, Fig 6.11, Page 185

 EY Trip Level Pmax
a) Generator Reactive Power
4000

2000
Reactive Power (VAr)

0
-2000

-4000

-6000

-8000

-10000
0.00 0.38 0.75 1.12 1.50 1.88 2.25 2.62 3.00 3.38
Time (s)

b) Generator Real Power

8000
Fault on Fault off
6000
Real Power (W)

4000

2000 v NO TRIP
DURING FAULT TRIP
0
BEFORE
-2000 POLESLIP

-4000
0.00 0.38 0.75 1.12 1.50 1.88 2.25 2.62 3.00 3.38
Time (s)

c) Generator Rate of Change of Power (RCP)

o
RCP (W/s)

-5000

-10000

-150,000 W/s : 75,000 W/s

-15000
0.00 0.38 0.75 1.12 1.50 1.88 2.25 2.62 3.00 3.38
Time (s)

Figure 6.12
Pole Slipping Algorithm Response to a Three Phase Fault on the Laboratory Model
Power System 5 kVA Generator, Generator Neutral Solidly Earthed - Test '4A',
Fault Resulted in the Generator Pole Slipping.

M. J. Checksfield/PhD/University o f Bath Chapter 6, Fig 6.12, Page 186

 KEY Trip Level Pmax

a) Generator Reactive Power

4000

2000 ■■
Reactive Power (VAr)

-2000 ■

-4000 ■■

-6000 ■■

-8000
0.00 0.38 0.75 1.12 1.50 1.88 2.25 2.62 3.00 3.38
Time (s)

b) Generator Real Power

6000

5000 ■■ Fault on
Real Power (W)

4000 ■■.

3000 ■■
NO TRIP
2000 ■■

iooo ■■ Fault off

o.oo 0.38 0.75 1.12 1.50 1.88 2.25 2.62 3.00 3.38
Time (s)

c) Generator Rate of Change of Power (RCP)

RCP (W/s)

-5000 ■■r-

10000 ■■

-15000
0.00 0.38 0.75 1.12 1.50 1.88 2.25 2.62 3.00 3.38
Time (s)

Figure 6.13
Pole Slipping Algorithm Response to a Three Phase to Earth Fault on the Laboratory
Model Power System 5 kVA Generator, Generator Neutral Isolated - Test '5B'

M. J. Checksfield/PhD/University o f Bath Chapter 6, Fig 6.13, Page 187

 Mho! I Fault o ff
Characteristic
Fault on
0
Q.
6.14a
Impedance Diagram Showing 5 kVA
-1
Generator Single Blinder Scheme and
Pole Slipping Loci for Test 4A - Pole
Slipping After a Three Phase Fault.
■2
TRIP Left Right
Blinder Blinder
•3
•2 1 0 1 2
Resistance, R (pu)

1
system & generator
^ impedance line
0
6.14b Oh
Impedance Diagram Showing 5 kVA 1
Generator Single Blinder Scheme and
Pole Slipping Loci for Test E - Over
2
Torque Pole Slip (low excitation).

•3
TRIP
4
1 0 1 2 3
Resistance, R (pu)

0
6.14c Cl
Impedance Diagram Showing 5 kVA
Generator Single Blinder Scheme and 1
Pole Slipping Loci for Test G - Over
Torque Pole Slip (high excitation).
■2
TRIP

•3
■2 -1 0 1 2
Resistance, R (pu)

Figure 6.14
Single Blinder Pole Slipping Protection Scheme Characteristic for
_______ Laboratory 5 kVA Generator Pole Slipping Tests._______
M. J. Checksfield/PhD/University o f Bath Chapter 6, Fig 6.14, Pagel88
 KEY Trip Level Pmax
a) ATP Machine Angle (degrees)
200
Angle (D egrees)

100

■100
■200,
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00
b) Generator Reactive Power

Fault on
Fault off
Reactive Power (pu)

0.5

- 0.5

- 1.5
o.oo 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00
Time (s)
c) Generator Real Power

TRIP
2“
Real Power (pu)

-2
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00
Time (s)
d) Generator Rate o f Change o f Power (RCP)

-10"
RCP (pu/s)

-20 ■■

-30 ■■

-40 4—
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00
Time (s)

Figure 6.15
Pole Slipping Algorithm Response to Test 'PSL1GA' - Pole Slipping of a 200 MVA
Generator Induced by a Fault from 20% Initial Load, AYR and Governor Modelled
M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.15, Page 189
 KEY Trip Level Pmax

a) Generator Reactive Power (Relaying Secondary Values)

300
Fault on
200 ■■
Reactive Power (VAr)

Fault off
100 ■■

-100 ■■

-200 ■■

-300
0.000 499.862 999.301 1495.402 1997.679 2497.718 3000.115 3498.884
Time (ms)

b) Generator Real Power (Relaying Secondary Values)

600

400 ■■
TRIP TRIP
Real Power (W)

200 ■■

-200 ■■

-400 "

-600
0.000 499.862 999.301 1495.402 1997.679 2497.718 3000.115 3498.884
Time (ms)

c) Generator Rate of Change of Power (RCP) (Relaying Secondary Values)

2000

-2000 ■■
RCP (W/s)

-4000 ■■

-6000 ■■

-8000 ■■

-10000
0.000 499.862 999.301 1495.402 1997.679 2497.718 3000.115 3498.884
Time (ms)

Figure 6.16
Pole Slipping Algorithm Response to PPSS Test' PSL2GA' - Pole Slipping of a 200 MVA
Generator, Induced by a 650 ms Three Phase Fault, Governor and AVR Modelled.

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.16, Page 190

 KEY Trip Level Pmax

a) Generator Reactive Power (Relaying Secondary Values)

300

200 ■■
Reactive Power (VAr)

100 ■■

-100 ■■

-200 —
0.000 499.857 999.706 1495.338 2001.982 2502.969 2998.431 3490.173
Time (ms)

b) Generator Real Power (Relaying Secondary Values)

600 ■■Pmax Pt
update update TRIP
400 ■■ II
Real Power (W)

200 "-

-200 ■■

-400 ••

-600
0.000 499.857 999.706 1495.338 2001.982 2502.969 2998.431 3490.173
Time (ms)

c) Generator Rate of Change of Power (RCP) (Relaying Secondary Values)

-5000 ■■
t
change in
RCP (W/s)

-10000 + (AP/At)min due to

temporary reset of
isooo + Pmax to nominal
value
-20000 +

-25000 + + + +
0.000 499.857 999.706 1495.338 2001.982 2502.969 2998.431 3490.173
Time (ms)

Figure 6.17
Pole Slipping Algorithm Response to PPSS Test' PSL3GA' - Pole Slipping of a 200 MVA
Generator, Induced by a 330 ms Two Phase to Earth Fault, Governor and AVR Modelled.

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.17, Page 191

 KEY Trip Level Pmax

a) Generator Reactive Power (Relaying Secondary Values)

400

300 ■■
Reactive Power (VAr)

200 ■■
100 ■■

-100

-200 ■■
-300
0.000 499.843 996.291 1495.712 1999.990 2500.492 3001.338 3502.162
Time (ms)

b) Generator Real Power (Relaying Secondary Values)

400 ■■
TRIP
Real Power (W)

200 •:

-200 ■■

0.000 499.843 996.291 1495.712 1999.990 2500.492 3001.338 3502.162

Time (ms)

c) Generator Rate of Change of Power (RCP) (Relaying Secondary Values)

-5000 ■■
RCP (W/s)

-10000 ■■

-15000 ■■

-20000
0.000 499.843 996.291 1495.712 1999.990 2500.492 3001.338 3502.162
Time (ms)

Figure 6.18
Pole Slipping Algorithm Response to PPSS Test' PSL3GN1- Pole Slipping of a 200 MVA
Generator, Induced by a 950 ms Two Phase Fault, AVR not Modelled, Governor Modelled.

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.18, Page 192

 KEY Trip Level Pmax

a) Generator Reactive Power (Relaying Secondary Values)

300

200 ■■
Reactive Power (VAr)

100

-100 ■■

-200 ■■

-300
0.000 519.849 1019.688 1533.915 2051.156 2551.785 3072.411 3591.013
Time (ms)

b) Generator Real Power (Relaying Secondary Values)

600

400 ■■ TRIP
Real Power (W)

200 *-
31

-200 ■■

-400
0.000 519.849 1019.688 1533.915 2051.156 2551.785 3072.411 3591.013
Time (ms)

c) Generator Rate of Change of Power (RCP) (Relaying Secondary Values)

-5000 ■■
RCP (W/s)

-10000 ■■

-15000 ■■

-20000 ■ ■

-25000 + + +
0.000 519.849 1019.688 1533.915 2051.156 2551.785 3072.411 3591.013
Time (ms)

Figure 6.19
Pole Slipping Algorithm Response to PPSS Test ’ PSL3NN1- Pole Slipping of a 200 MVA
Generator, Induced by a 350 ms Three Phase Fault, No Governor or AVR Modelled.

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.19, Page 193

 a) Generator Real Power

Average Rate o f Pole Slipping = 8 Hz

400

-200 -
5 sample points satisfied the criterion,
6 are needed for the algorithm to trip

2502.434 2720.592 2822.336 2921.820 3021.336 3122.805

2551.785 26j71.420 2771.124 2873.520 2971.092 3072.411 3171.393
Time (ms)
b)Rate of Change of Power (RCP)

-5000 ■

-10000 •

-15000 ■

-20000 ■

-25000
2502.434 2622.631 2720.592 2822.336 2921.820 3021.336 3122.805
2551.785 2671.420 2771.124 2873.520 2971.092 3072.411 3171.393
Time (ms)

Figure 6.20
Expaned Graphs from PPSS T est' PSL3NN1 Showing That the Algorithm Failed to Detect
Some o f the Later Pole Slips Due to a Pole Slipping Rate o f 8.0 Hz.

M. J. Checksfield/PhD/University o f Bath Chapter 6, Fig 6.20, Page 194

 KEY Trip Level ----------------------- Pmax

a) Generator Reactive Power (Relaying Secondary Values)

-20
Reactive Power (VAr)

-40

-60

-80

-100
End o f Simulation
-120
-140 —
0.000 499.838 999.676 1499.503 1999.454 2499.412 2999.599 3499.519
Time (ms)

b) Generator Real Power (Relaying Secondary Values)

140
TRIP
120 "
(-0.5% slip)
100 ■■
Real Power (W)

TRIP
(-0.25% slip)

-20
0.000 499.838 999.676 1499.503 1999.454 2499.412 2999.599 3499.519
Time (ms)

c) Generator Rate of Change of Power (RCP) (Relaying Secondary Values)

200

-200
RCP (W/s)

-600

-800

-1000 —
0.000 499.838 999.676 1499.503 1999.454 2499.412 2999.599 3499.519
Time (ms)

Figure 6.21
Pole Slipping Algorithm Response to PPSS T e st' PSL1GAF' - Pole Slipping o f a 200 MVA
Generator, Induced by Setting the AVR Reference Voltage to 5% pu, Governor & AVR Modelled.

M. J. Checksfield/PhD/University o f Bath Chapter 6, Fig 6.21, Page 195

 KEY Trip Level Pmax

a) Generator Reactive Power (Relaying Secondary Values)

s
1 -100
S.
d>>

,
12 -150

-200
0.000 499.831 999.652 1499.486 1999.315 2498.700 2999.217 3499.240
Time (ms)

b) Generator Real Power (Relaying Secondary Values)

400
TRIP
(-0.5% slip)
300 ■■
TRIP
200 ■■ (-0.25% slip)

I 100 - k
1

-100 + + + +
0.000 499.831 999.652 1499.486 1999.315 2498.700 2999.217 3499.240
Time (ms)

c) Generator Rate of Change of Power (RCP) (Relaying Secondary Values)

500

-500 ■■
CO

-2000 ■■

-2500 ■■

0.000 499.831 999.652 1499.486 1999.315 2498.700 2999.217 3499.240

Time (ms)

Figure 6.22
Pole Slipping Algorithm Response to PPSS Test' PSL2GAF - Pole Slipping of a 200 MVA
Generator, Induced by Setting AVR Reference Voltage to 50 % Rated, Governor and AVR Modelled.

M. J. Checksfi el d/PhD/Uni versity of Bath Chapter 6, Fig 6.22, Page 196

 KEY Trip Level Pmax

a) Generator Reactive Power (Relaying Secondary Values)

-50 ■■

-100

-150 ■■

-200 ■■

-250
0.000 499.873 999.640 1499.367 1999.255 2499.218 2998.587 3499.854
Time (ms)

b) Generator Real Power (Relaying Secondary Values)

400 ■■

TRIP

-200

0.000 499.873 999.640 1499.367 1999.255 2499.218 2998.587 3499.854

Time (ms)

c) Generator Rate of Change of Power (RCP) (Relaying Secondary Values)

1
.1 .

-5000 ■■

Pm
-10000 +

-15000 ■■

+ + + + + +
0.000 499.873 999.640 1499.367 1999.255 2499.218 2998.587 3499.854
Time (ms)

Figure 6.23
Pole Slipping Algorithm Response to PPSS Test1PSL3GAF - Pole Slipping of a 200 MVA
Generator, Induced by Setting AVR Reference Voltage to 80 % pu, Governor and AVR Modelled.

M. J. Checksfi el d/PhD/University of Bath Chapter 6, Fig 6.23, Page 197

 KEY Trip Level Pmax

a) Generator Reactive Power (Relaying Secondary Values)

300
Fault on

200 ■■
Reactive Power (VAr)

Fault off
too ■■

-100 ■■

-200
0.000 499.780 1002.704 1495.221 1997.866 2496.589 2995.855 3496.108
Time (ms)

b) Generator Real Power (Relaying Secondary Values)

400 ■■

200 ■■ NO TRIP
Real Power (W)

-200 ■■

-400 f
0.000 499.780 1002.704 1495.221 1997.866 2496.589 2995.855 3496.108
Time (ms)

c) Generator Rate of Change of Power (RCP) (Relaying Secondary Values)

4fMte*i40UMNflK59t9CMl

-2000 ■■
RCP (W/s)

-4000 ■■

-6000 ■■

-8000
0.000 499.780 1002.704 1495.221 1997.866 2496.589 2995.855 3496.108
Time (ms)

Figure 6.24
Pole Slipping Algorithm Response to PPSS Test' PSW1GA' - Stable Power Swinging of a 200
MVA Generator, Induced by a 1150 ms Three Phase Fault, Governor and AVR Modelled.

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.24, Page 198

 KEY Trip Level Pmax

a) Generator Reactive Power (Relaying Secondary Values)

300

100 ■■

-100 ■■

-200
0.000 499.845 998.575 1495.682 2000.023 2499.297 2998.425 3498.408
Time (ms)

b) Generator Real Power (Relaying Secondary Values)

400 ■■

NO TRIP
200 ■■

-200 ■■

0.000 499.845 998.575 1495.682 2000.023 2499.297 2998.425 3498.408

Time (ms)

c) Generator Rate of Change of Power (RCP) (Relaying Secondary Values)

2000

CO -2000 ■■
r ' -4ooo ■■
Oh
3 -6000 ■■

-8000 ■■

-10000
0.000 499.845 998.575 1495.682 2000.023 2499.297 2998.425 3498.408
Time (ms)

Figure 6.25
Pole Slipping Algorithm Response to PPSS Test' PSW2GA1- Stable Power Swinging of a 200
MVA Generator, Induced by a 630 ms Three Phase Fault, Governor and AVR Modelled.

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.25, Page 199

 KEY Trip Level Pmax

a) Generator Reactive Power (Relaying Secondary Values)

300
Fault on
200 ■■
Reactive Power (VAr)

Fault off
too ■■

-100 ■■

-200 +—
0.000 499.857 999.711 1495.495 1998.783 2498.152 2998.267 3498.253
Time (ms)

b) Generator Real Power (Relaying Secondary Values)

500

400

NO TRIP
Real Power (W)

300

200

100

-100
0.000 499.857 999.711 1495.495 1998.783 2498.152 2998.267 3498.253
Time (ms)

c) Generator Rate of Change of Power (RCP) (Relaying Secondary Values)

2000

-2000 ■■
RCP (W/s)

-4000 ■■

-6000 ■■

-8000 ■■

-10000
0.000 499.857 999.711 1495.495 1998.783 2498.152 2998.267 3498.253
Time (ms)

Figure 6.26
Pole Slipping Algorithm Response to PPSS Test1PSW3GA' - Stable Power Swinging of a 200
MVA Generator, Induced by a 320 ms Two Phase to Earth Fault, Governor and AVR Modelled.

M. J. Checksfiel d/PhD/University of Bath Chapter 6, Fig 6.26, Page 200

 KEY Trip Level Pmax

a) Generator Reactive Power (Relaying Secondary Values)

300
Fault on
200 ■■
Reactive Power (VAr)

Fault off
100 ■■

-loo ■■

-200 ■■

-300
0.000 499.660 1002.489 1496.649 1995.672 2495.328 2995.406 3494.870
Time (ms)

b) Generator Real Power (Relaying Secondary Values)

NO TRIP ■
Real Power (W)

— A / \ / \ L
1u . J _ J

Qtnpsatisfied,— , , | i n .
0.000 499.660 1002.489 1496.649 1995.672 2495.328 2995.406 3494.870
Time (ms)

c) Generator Rate of Change of Power (RCP) (Relaying Secondary Values)

2000
RCP (W/s)

-2000 ■■
NO TRIP

-6000 ■■ Qtrip & Pt satisified

-8000
0.000 499.660 1002.489 1496.649 1995.672 2495.328 2995.406 3494.870
Time (ms)

Figure 6.27
Pole Slipping Algorithm Response to PPSS Test' PSW1GN1- Stable Power Swinging of a
200 MVA Generator, Induced by a 1155 ms Three Phase Fault, AVR Out of Service.

M. J. Checksfi el d/PhD/University of Bath Chapter 6, Fig 6.27, Page 201

 KEY Trip Level Pmax

a) Generator Reactive Power (Relaying Secondary Values)

Fault on

Fault off

100
200 ■■

-300
0.000 499.862 1002.039 1499.757 1999.548 2498.682 2998.788 3498.702
Time (ms)

b) Generator Real Power (Relaying Secondary Values)

400

300 ■■

-100 ■■

-200
0.000 499.862 1002.039 1499.757 1999.548 2498.682 2998.788 3498.702
Time (ms)

c) Generator Rate of Change of Power (RCP) (Relaying Secondary Values)

-2000 ■■

-4000 ■■

-6000 ■■

-8000
0.000 499.862 1002.039 1499.757 1999.548 2498.682 2998.788 3498.702
Time (ms)

Figure 6.28
Pole Slipping Algorithm Response to PPSS Test' PSW2GN1- Stable Power Swinging of a
200 MVA Generator, Induced by a 810 ms Two Phase to Earth Fault, AVR Out of Service.

M. J. Checksfi el d/PhD/University of Bath Chapter 6, Fig 6.28, Page 202

 KEY Trip Level -------------------- Pmax

a) Generator Reactive Power (Relaying Secondary Values)

400
Fault on
300 ”

200 ■■ Fault off

% 100"

S -loo -

-200 ■■
-300
0.000 499.800 996.460 1495.375 1999.447 2498.968 2998.742 3498.554
Time (ms)

b) Generator Real Power (Relaying Secondary Values)

400

300 -

-100
0.000 499.800 996.460 1495.375 1999.447 2498.968 2998.742 3498.554
Time (ms)

c) Generator Rate of Change of Power (RCP) (Relaying Secondary Values)

2000

-2000

T
Pi -4000 ■■
u
-6000 ■■

-8000 ■■

-10000
0.000 499.800 996.460 1495.375 1999.447 2498.968 2998.742 3498.554
Time (ms)

Figure 6.29
Pole Slipping Algorithm Response to PPSS Test1PSW3GN1- Stable Power Swinging of a
200 MVA Generator, Induced by a 820 ms Two Phase Fault, AVR Out of Service.

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.29, Page 203

 a) ATP Machine Angle (degrees)
Angle (Degrees)

0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 11.00 12.00 13.00 14.00 15.00

b) Generator Reactive Power (PU)

l
Generator moves into AVR reference voltage restored
under-excited operation
0.5 ^ ---------------------- ► Steady state operation
Power (pu)

-0.5
jfy r
-l
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 11.0012.00 13.00 14.00 15.00

c) Generator Real Power (PU)

Input power reduced from 0.5 to 0 pu

0.5
Power (pu)

-0.5
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 11.0012.00 13.00 14.00 15.00

d) Generator Rate of Change of Real Power (PU)

2

0
RCP(pu/s)

6
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.0011.0012.0013.0014.0015.00
T im e (s)_

Figure 6.30
Pole Slipping Algorithm Response to Test D2 (AVR5NA) Generator Operation
Absorbing System VArs Using A Fast AVR.

M. J. Checksfield/PhD/University o f Bath Chapter 6, Figure 6.30, Page 204

 KEY Trip Level Pmax

a) Generator Reactive Power (Relaying Secondary Values)

o
-20
Reactive Power (VAr)

-40

-60

-80

-120
-140
0.000 499.761 999.561 1499.323 1999.160 2498.952 2998.599 3498.344
Time (ms)

b) Generator Real Power (Relaying Secondary Values)

140

120 ■■
100 ■■
Real Power (W)

80 ■■

40 ■■

20 ■■
TRIP

-20
0.000 499.761 999.561 1499.323 1999.160 2498.952 2998.599 3498.344
Time (ms)

c) Generator Rate of Change of Power (RCP) (Relaying Secondary Values)

200

-200
RCP (W/s)

-400

-600

-800

-1000
0.000 499.761 999.561 1499.323 1999.160 2498.952 2998.599 3498.344
Time (ms)

Figure 6.31
Pole Slipping Algorithm Response to PPSS Test' LOF1GA'
Loss of Field on a 200 MVA Generator, Governor Modelled.

M. J. Checksfi el d/PhD/University of Bath Chapter 6, Fig 6.31, Page 205

 KEY Trip Level Pmax

a) Generator Reactive Power (Relaying Secondary Values)

o
Reactive Power (VAr)

-50

-100

-150
End of simulation

-200 —
o.ooo 499.838 999.652 1499.515 1999.315 2499.451 2999.688 3499.483
Time (ms)

b) Generator Real Power (Relaying Secondary Values)

400

300 ■■
Real Power (W)

TRIP

-100
0.000 499.838 999.652 1499.515 1999.315 2499.451 2999.688 3499.483
Time (ms)

c) Generator Rate of Change of Power (RCP) (Relaying Secondary Values)

1000

500 ■■
RCP (W/s)

-1000 ■■

-1500 ■■

-2000 ■■

-2500
0.000 499.83 8 999.652 1499.515 1999.315 2499.451 2999.688 3499.483
Time (ms)
Figure 6.32
Pole Slipping Algorithm Response to PPSS Test' LOF2GA'
Loss of Field on a 200 MVA Generator, Governor Modelled.

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.32, Page 206

 KEY Trip Level Pmax

a) Generator Reactive Power (Relaying Secondary Values)

0

-50
Reactive Power (VAr)

-100

-150

-200

-250
0.000 499.785 999.606 1499.367 1999.129 2498.522 2998.603 3498.459
Time (ms)

b) Generator Real Power (Relaying Secondary Values)

300

200 ■■
Real Power (W)

100 ■■

TRIP

-100 ■■
0.000 499.785 999.606 1499.367 1999.129 2498.522 2998.603 3498.459
Time (ms)

c) Generator Rate of Change of Power (RCP) (Relaying Secondary Values)

2000

-2000 ■■
RCP (W/s)

-4000 ■■

-6000 ■■

-8000 - ■

-10000 ■■
0.000 499.785 999.606 1499.367 1999.129 2498.522 2998.603 3498.459
Time (ms)

Figure 6.33
Pole Slipping Algorithm Response to PPSS Test' LOF3GA'
Loss of Field on a 200 MVA Generator, Governor Modelled.

M. J. Checksfi el d/PhD/University of Bath Chapter 6, Fig 6.33, Page 207

 KEY Trip Level Pmax

a) Generator Reactive Power (Relaying Secondary Values)

150

100 ■■
Reactive Power (VAr)

50 ■■

-50 ■■

-100 ■■

-150
0.000 499.857 999.715 1499.572 1999.430 2499.283 2999.865 3499.344
Time (ms)

b) Generator Real Power (Relaying Secondaiy Values)

300

200 ■■
Real Power (W)

NO TRIP

-100 ■■

-200
0.000 499.857 999.715 1499.572 1999.430 2499.283 2999.865 3499.344
Time (ms)

c) Generator Rate of Change of Power (RCP) (Relaying Secondary Values)

1000

500 ■■
RCP (W/s)

-500 ■■

-1000

-1500 ■■

-2000
0.000 499.857 999.715 1499.572 1999.430 2499.283 2999.865 3499.344
Time (ms)

Figure 6.34
Pole Slipping Algorithm Response to PPSS Test' LOP3NA'
Loss of Prime Mover of a 200 MVA Generator, AVR Modelled.

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.34, Page 208

 KEY_______________________________________________________________
...............(AP/At)mjnfo r-0.25% slip ------------- (AP/At)min fo r-0.5% slip

a) Rate of Change of Power (RCP) in Per Unit for PC Based Algorithm Simulation

0.2 --
sample points which satsify trip criterion
RCP (pu/s)

-0.2

Pt criterion inhibits
-0.4 tripping until here

-0.6 -0.25% slip -0.5% slip

i—i tr*p

4.96 5.04 5.12 5.20 5.28 5.36 5.44

Time (s)

b) Rate of Change of Power in Relaying Secondary Values for PPSS Simulation

sample points which satsify trip criterion

RCP (W/s)

\/
-50

\/
'\
-100

-0.5% slip
-0.25% slip
-150

1414.530 1494.504 1574.485 1654.459 1734.446 1814.439 1894.446

Time (ms)

Figure 6.35
Graphs Showing Rate of Change of Power Criterion Trip Points for
test 'PSL1GAF for PPSS and PC Based Algorithm Test Systems.

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.35, Page 209

 Left Right The 'fault on'
0.4
Blinder Blinder point occurs
at a per unit
impedance
value of
0.2 -
Fault off R = 0.985 pu
Transformer Mho jX = 0.121 pu
3
reactance Characteristic
3
*
8c

- 0.2

-0.4
-0.4 -0.2 0 0.2 0.4
Resistance, R (pu)
Figure 6.36
Impedance Diagram Showing 200 MVA Generator Single Blinder Pole
Slipping Protection Scheme and Pole Slipping Loci for Test 'PSL3GN1

0.4 Right
The 'fault on'
Blinder Blinder
point occurs
at a per unit
impedance
0.2 Fault off value o f
Transformer R = 2.11 pu
reactance jX = 2.47 pu
3
3
x
8e
Stable
Power
$
- 0.2 Locus

\ Mho
Characteristic
-0.4 i 1
-0.4 - 0.2 0 0.2 0.4
Resistance, R (pu)
Figure 6.37
Impedance Diagram Showing 200 MVA Generator Single Blinder Pole
Slipping Protection Scheme and Pole Slipping Locus for Test 'PSL1GA'

M. J. Checksfield/PhD/University o f Bath Chapter 6, Fig 6.36,6.37 Page 210

 The 'fault on'
system point occurs
equivalent at a per unit
impedance impedance
line value of
R = 0.985 pu
jX = 0.121 pu

generator
equivalent
impedance,
i Xd'

\M h o
Characteristic

- 0.1 0 0.1 0.2

Resistance, R (pu)
Figure 6.38
Impedance Diagram Showing 200 MVA Generator M ho Pole Slipping Protection Scheme
Characteristic and Pole Slipping Loci for Test 'PSL3GA'

tl = 1.99 s
| M ho I Loss o f Field
Characteristic t2 = 3.47 s
Characteristic
0.5
t3 = 3.79 s
t4 = 4.02 s
0 t5 = 4.44 s
t6 = 4.48 s
3
a, 0.5
* tl to t2
represents
§ 1 the
I
poleslip
I
1.5
t4 ' locus

t2 to t6
■2 respresent
stable
power
-1 -0.5 0 0.5 1 1.5 2 2.5 swings
Resistance, R (pu)
Figure 6.39
Impedance Diagram Showing 200 MVA Generator M ho Pole Slipping Protection Scheme
Characteristic and Pole Slipping Locus for Test 'PSL2GAF'
M. J. Checksfield/PhD/University o f Bath Chapter 6, Fig 6.38,6.39 Page 211
 Mho Characteristic

-0.5

3
Oj -1
X
i
-1.5
i

-2 -
Loss of Field
Characteristic
-2.5
-1.5 -1 -0.5 0 0.5 1
Resistance, R (pu)
Figure 6.40
Impedance Diagram Showing 200 MVA Generator Loss o f Field Protection Scheme
Characteristic and Pole Slipping Loci for Test 'PSL3GAF'

The 'fault on'

01 t ( system I........................f..........1 point occurs
jequi valent Fault off at a per unit
impedance impedance
line value of
R = 0.985 pu
^ ’ jX = 0.121 pu
cu
X
<u
8
Stable power
I -0-1 generator swing locus
equivalent causes M ho
impedance, scheme to trip
1 Xd' at this point
-0.2
^ \M h o |
Characteristic

-0.1 0 0.1 0.2

Resistance, R (pu)
Figure 6.41
Impedance Diagram Showing 200 MVA Generator M ho Pole Slipping Protection Scheme
Characteristic and Power Swing Locus for Test 'PSW3GA'
M. J. Checksfield/PhD/University o f Bath Chapter 6, Fig 6.40,6 .41 Page 212
 0.2

Right Right The 'fault on'

0.15 Inner Outer point occurs
Blinder Blinder at a per unit
impedance
0.1 value o f
3 M ho R = 0.895 pu
Q. Characteristic jX = 0.121 pu
X 0.05
8
fi /
1
I

-0.05

-0.1
-0.05 0 0.05 0.1 0.15 0.2 0.25
Resistance, R (pu)
Figure 6.42
Impedance Diagram Showing 200 MVA Generator Double Blinder
Scheme and Power Swing Loci for Test 'PSW3NA'

Mho

-0.5

X
8c

I
&

Loss o f Field
Characteristic
-2.5
1.5 1 -0.5 0 0.5 1
Resistance, R (pu)
Figure 6.43
Impedance Diagram Showing 200 MVA Generator Loss o f Field Protection Scheme
Characteristic and Loss o f Field Loci for Test 'LOF3GA'
M. J.Checksftel d/PhD/University o f Bath Chapter 6, Fig 6.42, 6.43, Page 213
 stable operating point
P = 0.5 pu, Q = -0.67
o.

Loss of Field
Characteristic

_ lo ad change from
P = 0.5 pu to P = 0 pu AVR reference voltage
reset - causing locus to
swing round in a large
clockwise arc to the
positive reactance zone
Resistance, R (pu) (not shown)
Figure 6.44
Impedance Diagram Showing 200 MVA Generator Loss o f Field Protection Scheme
Characteristic and Impedance Locus for Test AVR5NA'

Figure 6.45a - A Phase RMS Current and IDMT trip for Test PSL3NN
Fault fa u lt off
on \ 1

TRIP TRIP!
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.0C
Time(s)

Figure 6.45b - A Phase RMS Current and IDMT trip for Test PSL2GA
Fault on Fault o ff
NO TRIP during pole slip

PSM setting
T R IP f
0.00 0.50 1.00 1.50 Time(s)
2.00 2.50 3.00 3.50 4.00

Figure 6.45c - A Phase RMS Current and IDMT trip for Test PSL3GA
6 Fault o ff NO TRIP during pole
Fault
I 45 on
e
3
2
u 1 -
TRIP
0 1—
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00
Time(s)

Figure 6.45 - Phase A RMS Current for PPSS Tests PSL3NN, PSL2GA and PSL3GA

M. J. Checksfi el d/PhD/University o f Bath Chapter 6, Fig 6.44,6.45 Page 214

 NOTE: A time delay of 0.5 seconds was used with the Under-Voltage relay

Figure 6.46a - A Phase RMS Voltage and Under Voltage Trip for Test 'PSL3NN1
1.2
Fault on Fault off
1
Voltage (pu)

TRIP
0
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00
Time(s)

Figure 6.46b - A Phase RMS Voltage and Under Voltage Trip for Test 'PSL2GA1

Fault on
Voltage (pu)

0.8 Under-Voltage Trip Level

0.6 Fault off

0.4

0.2 TRIP

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00

Time(s)

Figure 6.46c - A Phase RMS Voltage and Under Voltage Trip for Test 'PSL3GAF
Voltage (pu)

0.4

0.2
TRIP

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00

Time(s)

Figure 46
Under-Voltage Relay Response to Tests 'PSL3NN1, 'PSL2GA', and 'PSL3GAF'

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.46, Page 215

 NOTE: A time delay of 0.4 seconds was used with the Over-Frequency Relay

Figure 6.47a - Measured Frequency at Generator Terminals,

and Over Frequency Trip for Test 'PSL3NN1

1.05

Fault on
Fault off
0.9
NO TRIP I
0.85
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00
Time(s)

Figure 6.47b - Measured Frequency at Generator Terminals,

and Over Frequency Trip for Test 'PSL2GA'
1.06
Fault off N 0 TRIP
1.04
? 1.02
w> 1
| 0.98
Fault on
0.96
0.94
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00
Time(s)

Figure 6.47c - Measured Frequency at Generator Terminals,

and Over Frequency Trip for Test ’PSL3GAF
1.05
NO TRIP

&0.95

0.85
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00
Time(s)

Figure 47
Over-Frequency Relay Response to Tests 'PSL3NN1, 'PSL2GA', and ’PSL3GAF'

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.47, Page 216

 KEY Trip Level Pmax

a) Generator Reactive Power

-100,000

-200,000

-300,000
Power (VAr)

-400,000

-500,000

-600,000

-700,000
0 499.92 999.84 1499.714 1999.514 2499.314 2999.114 3498.914

b) Generator Real Power

320,000

300,000 ■■

280,000 ■■
Power (W)

260,000 ■■
NO TRIP
240,000 ■■

220,000 ■■

200,000
0 499.92 999.84 1499.714 1999.514 2499.314 2999.114 3498.914

c) Generator Rate of Change of Power (RCP)

400.000

200.000
RCP (W/s)

-200,000

-400,000

-600,000
0 499.92 999.84 1499.714 1999.514 2499.314 2999.114 3498.914
Time (ms)
Figure 6.48
Pole Slipping Algorithm Response to 625 kVA Diesel Generator Field Trial Test 'A2'
Unsuccessful Pole Slip Attempt by Weakening the Field Excitation.

M. J. Checksfield/PhD/University o f Bath Chapter 6, Fig 6.48, Page 217

 KEY Trip Level Pmax

a) Generator Reactive Power

-200,000 ■■
Power (VAr)

-600,000 ■■

-800,000 ■■

■1,000,000
-1,200,000 ■■
-1,400,000 ■■

-1,600,000
0 999.58 1999.161 2998.742 3998.34 4998.043 5997.674

b) Generator Real Power

2,000,000

1,500,000 ■■
Power (W)

500,000 ■■

TRIP
-500,000
0 999.58 1999.161 2998.742 3998.34 4998.043 5997.674

c) Generator Rate of Change o f Power (RCP)

-5,000,000 ■■

-10,000,000 ■■
RCP (W/s)

-15,000,000 '■

-20,000,000 ■■

-25,000,000 ■■

-30,000,000
0 999.58 1999.161 2998.742 3998.34 4998.043 5997.674
Time (ms)
Figure 6.49
Pole Slipping Algorithm Response to 625 kVA Diesel Generator Field Trial Test 'A3'
Pole Slip Induced by Weakening the Field Excitation.

M. J. Checksfield/PhD/University o f Bath Chapter 6, Fig 6.49, Page 218

 KEY Trip Level Pmax

a) Generator Reactive Power

-100,000

-200,000
Power (VAr)

-300,000

-400,000

-500,000

-600,000 H----------- 1----------- 1----------- 1----------- 1----------- 1----------- 1----------- 1----

0 997.42 1994.822 2992.456 3990.004 4987.603 5985.168 6982.685

b) Generator Real Power

180,000

160,000 ■■

140,000 ■■
Power (W)

120,000 ■■

100,000 ■■

80,000 ■■

60,000 ■■
NO TRIP
40,000
0 997.42 1994.822 2992.456 3990.004 4987.603 5985.168 6982.685

c) Generator Rate of Change o f Power (RCP)

1,000,000
RCP (W/s)

-1,000,000

-2,000,000 ■■

-3,000,000
0 997.42 1994.822 2992.456 3990.004 4987.603 5985.168 6982.685
Time (ms)
Figure 6.50
Pole Slipping Algorithm Response to 625 kVA Diesel Generator Field Trial Test 'A4'
Adjacent Generator Pole Slip Induced by Weakening the Field Excitation.

M. J. Checksfi el d/PhD/University o f Bath Chapter 6, Fig 6.50, Page 219

 KEY Trip Level Pmax
a) Generator Reactive Power

-500,000

-1,000,000

glitch produced
-1,500,000 due to change over period
in disturbance records

-2,000,000
1001.215 2002.502 3003.866 4005.306 5006.79 6008.276 7009.578
b) Generator Real Power
3.000.000

2,500,000 ■■

2.000.000 ■■

^ 1,500,000 +

| 1,000,000 ■■
pu,
500,000 ■■

-500,000 t ™ ECU
1001.215 2002.502 3003.866 4005.306 5006.79 6008.276 7009.578

c) Generator Rate of Change of Power (RCP)

-10,000,000 ■■

-20,000,000 ■■

-30,000,000 "

%
-40,000,000"

-50,000,000 "

-60,000,000
0 1001.215 2002.502 3003.866 4005.306 5006.79 6008.276 7009.578
Time (ms)
Figure 6.51
Pole Slipping Algorithm Response to 625 kVA Diesel Generator Field Trial Test ’A7'
Pole Slip Induced by Suddenly Increasing Prime Mover Output Power.

M. J. Checksfi el d/PhD/University of Bath Chapter 6, Fig 6.51, Page 220

 KEY Trip Level Pmax

a) Generator Reactive Power

200,000

0
Power (VAr)

-200,000

-400,000

-600,000

-800,000
0 998.754 1997.528 2996.283

b) Generator Real Power

130,000

120,000
Power (W)

90,000

NO TRIP
80,000
0 998.754 1997.528 2996.283

c) Generator Rate of Change o f Power (RCP)

400,000

200,000 ■■
RCP (W/s)

-200,000 ■■

-400,000 ■*-+
998.754 1997.528 2996.283
Time (ms)
Figure 6.52
Pole Slipping Algorithm Response to 625 kVA Diesel Generator Field Trial Test 'B2'
Loss o f Excitation from 20% Initial Load.

M. J. Checksfield/PhD/University o f Bath Chapter 6, Fig 6.52, Page 221

 KEY Trip Level Pmax

a) Generator Reactive Power

-100,000

-200,000 ■■

-300,000 ■■
Power (VAr)

-400,000 ”

-500,000 ■■

-600,000 "

-700,000
998.692 1997.376 2996.068

b) Generator Real Power

340,000

320,000

300,000
Power (W)

280,000

260,000

240,000

220,000
NO TRIP
200,000
0 998.692 1997.376 2996.068

c) Generator Rate of Change o f Power (RCP)

600.000

400,000 ■■

200,000 ■■
RCP (W/s)

-200,000 "
-400,000 ■■

-600,000 ■■

-800,000 -H- 1---- 1-----

0 998.692 1997.376 2996.068
Time (ms)
Figure 6.53
Pole Slipping Algorithm Response to 625 kVA Diesel Generator Field Trial Test 'B3'
Loss o f Excitation from 50 % Initial Load.

M. J. Checksfi el d/PhD/University o f Bath Chapter 6, Fig 6.53, Page 222

 a) Machine Rotor Angle
180
160
Angle (Degrees)

140
120
100

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
b) Machine Reactive Power

-200000
Power (VAr)

-400000
-600000
-800000 ■
-1000000 ■
-1200000

c) Machine Real Power

500000
400000
Power (W)

300000
Power remains constant for a long duration
200000

-100000

d) Machine Terminal Voltage

415
Voltage (Volts)

410

405

400

395

e) Machine Current
1600
1400
Current (Amps)

1200
1000
800
600
400
200
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time (s)

Figure 6.54
Generator Variables for ATP Simulation of a 625 kVA Diesel
Generator Loss of Excitation from 45 % Initial Load.

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.54, Page 223

 KEY Trip Level Pmax
a) Generator Reactive Power
500,000
generator disconnected

-500,000 ■■

5 -1,000,000 ■■
Q-i

-1,500,000 ■■

-2,000,000
0 999.272 1998.565 2997.838 3997.151

b) Generator Real Power

2,000,000

1,500,000 ■■

^ 1,000,000 ■■
tH
<D
I
Oh 500,000

TRIP
-500,000
0 999.272 1998.565 2997.838 3997.151

c) Generator Rate of Change of Power (RCP)

-5,000,000 ■■

^ -10,000,000 ■■

^ -15,000,000 ■■
U
&
-20,000,000"

-25,000,000 ■■

-30,000,000 1— +
0 999.272 1998.565 2997.838 3997.151
Time (ms)
Figure 6.55
Pole Slipping Algorithm Response to 625 kVA Diesel Generator Field Trial Test 136'
Loss of Excitation from 100 % Initial Load.

M. J. Checksfi el d/PhD/University of Bath Chapter 6, Fig 6.55, Page 224

 KEY Trip Level Pmax

a) Generator Reactive Power

214,000

213,000

212,000
Power (VAr)

210,000

209,000

208,000

997.92 1995.84 2993.76

b) Generator Real Power

150,000

140,000
Power (W)

110,000

100,000

90,000
NO TRIP
80,000
0 997.92 1995.84 2993.76

c) Generator Rate of Change o f Power (RCP)

400,000

200,000 ■■
RCP (W/s)

-200,000 ”

-400,000 -L+
997.92 1995.84 2993.76
Time (ms)
Figure 6.56
Pole Slipping Algorithm Response to 625 kVA Diesel Generator Field Trial Test 'C3'
Steady State Operation at 30 % Load, Over-Excited Exitation.

M. J. Checksfield/PhD/University o f Bath Chapter 6, Fig 6.56, Page 225

 KEY Trip Level — Pmax

a) Generator Reactive Power

-222,000
Change in reactive loading
-224,000
due to compressor being
switched on
Power (VAr)

-226,000

-228,000

-230,000

-232,000
0 1000.099 2000.179 3000.259

b) Generator Real Power

150.000

140.000
......... i'
130.000 ■■
Power (W)

120.000
NO TRIP
110,000

100,000

90,000
0 1000.099 2000.179 3000.259

c) Generator Rate of Change o f Power (RCP)

300,000 ■■

200,000 ■■
100,000"
RCP (W/s)

-100,000 ■■
-200,000 ■■
-300,000 ■■

-400,000 -H- 1----- 1-----

0 1000.099 2000.179 3000.259
Time (ms)
Figure 6.57
Pole Slipping Algorithm Response to 625 kVA Diesel Generator Field Trial Test 'C6'
Compressor Energisation Test at 20% Initial Generator Loading.

M. J. Checksfield/PhD/University o f Bath Chapter 6, Fig 6.57, Page 226

 Left Right
Blinder Blinder
Mho
Characteristic 0.8
0.2

-0.5 -

cx
NO TRIP

0.6

1 -0.5 0 0.5 1
Resistance, R (pu)
Figure 6.58a
Impedance Diagram Showing 625 kVA Generator Single Blinder Pole
Slipping Protection Scheme and Pole Slipping Loci for Test 'A3'

Figure 6.58b - Machine Terminal Voltage for Test 'A3'

RMS Voltage (PU)

0.98

o.%
0.94

0.92

0.9
0.00 0.20 0.40 0.60 0.80 1.00 1.20
Time (s)
Figure 6.58c - Machine Terminal Current for Test 'A3'
RMS Current (PU)

2.5

0.5
0.00 0.20 0.40 0.60 0.80 1.00 1.20
Time (s)

Figure 6.58
Plots Showing Generator Impedance, Voltage and Current Variations for
625 kVA Diesel Generator Field Trial Pole Slip Test 'A3'
M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.58 Page 227
 0.5

Left Right
Blinder Blinder
or
Mho
Characteristic 0.2
-0.5

0.6

1 -0.5 0 0.5 1
Resistance, R (pu)
Figure 6.59
Impedance Diagram Showing 625 kVA Generator Single Blinder Pole
Slipping Protection Scheme and Pole Slipping Loci for Test 'A7'

Figure 6.59b - Machine Terminal Voltage for Test 'AT

RMS Wtage (PU)

0.98
0.96
0.94 ■■
0 .9 2 "
0.9
0.88
0.00 0.20 0.40 0.60 0.80 1.00 1.20
Time (s)
Figure 6.59c - Machine Terminal Current for Test 'A7f
RMS Current (PU)

4-

0.00 0.20 0.40 0.60 0.80 1.00 1.20

Time (s)

Figure 6.59
Plots Showing Generator Impedance, Voltage and Current Variations for
625 kVA Diesel Generator Field Trial Pole Slip Test 'A7'
M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.59 Page 228
 KEY Trip Level Pmax

a) Generator Reactive Power

200
Reactive Power (VAr)

100 ■■
Millions

-100 ■■

-200
0.000 498.555 997.075 1495.555 1993.507 2492.377 2991.110 3489.590
Time (s)

b) Generator Real Power

340

320

300 ■■
Real Power (W)
Millions

280

260 ■■ NO TRIP
240 ■■

220 ■“
200
0.000 498.555 997.075 1495.555 1993.507 2492.377 2991.110 3489.590
Time (s)

c) Generator Rate of Change of Power (RCP)

0.00 ■

-0.50
RCP (W/s)

-1.00 ■■
Billions

-1.50 ■■

-2.00 ■■

-2.50 ■■

-3.00 +—
0.000 498.555 997.075 1495.555 1993.507 2492.377 2991.110 3489.590
Time (s)

Figure 6.60
Pole Slipping Algorithm Response to a Fault During the
Field Trials on a 353 MVA Turbo-Generator

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.60, Page 229

 KEY Trip Level Pmax

Reactive Power (VAr) a) Generator Reactive Power

Millions

0.000 1000.608 2001.216 3001.824

Time (s)

b) Generator Real Power

70

65
Real Power (W)
Millions

50

0.000 500.299 1000.608 1500.916 2001.216 2501.515 3001.824 3502.152

Time (s)

c) Generator Rate of Change of Power (RCP)

100
RCP (W/s)
Millions

-100 1------------1----------- 1----------- 1----------- 1------------1-----------1--------

0.000 500.299 1000.608 1500.916 2001.216 2501.515 3001.824 3502.152
Time (s)

Figure 6.61
Pole Slipping Algorithm Measurands for Generator Steady State Operation During the
Field Trials on a 353 MVA Turbo-Generator

M. J. Checksfi eld/PhD/University o f Bath Chapter 6, Fig 6.61, Page 230

 KEY Trip Level Pmax

a) Generator Reactive Power

-100 ■■
Reactive Power (VAr)
Thousands

-200 ■■

-300 ■■

-400 ■■
Qtrip = -37.8 MVAr
-500
0.000 500.966 1001.961 1502.923 2003.904 2504.884 3005.865 3506.846
Time (s)

b) Generator Real Power

26 ■■
Real Power (W)

24 ■■
Millions

22 ■■

20 ■■

0.000 500.%6 1001.961 1502.923 2003.904 2504.884 3005.865 3506.846

Time (s)

c) Generator Rate of Change of Power (RCP)

RCP (W/s)
Millions

-10 ■■

-20 ■■

-30 1.............. T..... ........ T-----------r-----------T----------- 1---------- T-.....

0.000 500.966 1001.961 1502.923 2003.904 2504.884 3005.865 3506.846
Time (s)

Figure 6.62
Pole Slipping Algorithm Measurands for Generator Steady State Operation During the
Field Trials on a 26.5 MVA Hydro-Generator.

M. J. Checksfield/PhD/University o f Bath Chapter 6, Fig 6.62, Page 231

 KEY Trip Level Pmax

a) Generator Reactive Power

2

1
i
0
i
Q h
<
>D 1

I •2

•3
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

b) Generator Real Power

2.5 "

2 ■■
1.5

NO TRIP

0.5

-0.5
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

c) Generator Rate of Change of Power (RCP)

X .____

-10 ■■
g
r ' -20 ■■
Qh
o
Oh
-30 ■■

-40 ■■

-50
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

Figure 6.63
Pole Slipping Algorithm Response to an ATP Simulated Stable Power Swing on a 750 kVA
Generator with Tdo' = 1.0 s, Slow AYR Model Included, No Governor Model - Test LT1 .

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.63, Page 232

 KEY Trip Level Pmax

a) Generator Reactive Power

-2 ■■

-3 ■■

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

b) Generator Real Power

3

l NO TRIP
1
Oh
o

■2
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

c) Generator Rate of Change of Power (RCP)

-io ■■
j>
T
Ph
' -20 ■■
V
PS
-30 ■■

-40 ■■

-50
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

Figure 6.64
Pole Slipping Algorithm Response to an ATP Simulated Stable Power Swing on a 750 kVA
Generator with Tdo' = 2.0 s, Slow AYR Model Included, No Governor Model - Test LT2.

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.64, Page 233

 KEY Trip Level Pmax

a) Generator Reactive Power

2

1
Reactive Power (VAr)

-4
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

b) Generator Real Power

4

3
Real Power (W)

2
NO TRIP

•2
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

c) Generator Rate of Change of Power (RCP)

-10 ■■
RCP (W/s)

-30 ■■

-40 ■■

-50
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

Figure 6.65
Pole Slipping Algorithm Response to an ATP Simulated Stable Power Swing on a 750 kVA
Generator with Tdo' = 3.0 s, Slow AYR Model Included, No Governor Model - Test LT3.

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.65, Page 234

 KEY Trip Level Pmax

a) Generator Reactive Power

2
1
Reactive Power (VAr)

0
1

•3

-4

■5
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

b) Generator Real Power

4

3
Real Power (W)

2 TRIP
l

■2
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

c) Generator Rate of Change of Power (RCP)

-20 ■■
RCP (W/s)

-40

-60 ■■

-80 ■■

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

Figure 6.66
Pole Slipping Algorithm Response to an ATP Simulated Pole Slip on a 750 kVA Generator
with Tdo' = 1.0 s, Slow AYR Model Included, No Governor Model - Test LTP1.

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.66, Page 235

 KEY Trip Level Pmax

a) Generator Reactive Power

Reactive Power (VAr)

-2 ■■
-3

-4 ■■

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

b) Generator Real Power

Real Power (W)

TRIP

-2 ■■

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

c) Generator Rate of Change of Power (RCP)

-20 ■■

-40 ■■
RCP (W/s)

-60 ■■

-80 ■■

-100 ■■

-120 ■■
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

Figure 6.67
Pole Slipping Algorithm Response to an ATP Simulated Pole Slip on a 750 kVA Generator
with a value o f Tdo1of 2.0 s, Slow AYR Model Included, No Governor Model - Test LTP2.

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.67, Page 236

 KEY Trip Level Pmax

a) Generator Reactive Power

« -2

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

b) Generator Real Power

4

2 TRIP

1
0
1
■2
■3

■4
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

c) Generator Rate of Change of Power (RCP)

o -50--

-100 ■■

-150 "

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

Figure 6.68
Pole Slipping Algorithm Response to an ATP Simulated Pole Slip on a 750 kVA Generator
with a value o f Tdo' of 3.0 s, Slow AYR Model Included, No Governor Model - Test LTP3.

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.68, Page 237

 KEY Trip Level Pmax

a) Generator Reactive Power

3

•5
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

b) Generator Real Power

3

2
NO TRIP /;
l

■2
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

c) Generator Rate of Change of Power (RCP)

-10 ■■

Oh
-20 ■■
u
04
-30 ■■

-40 ■■

-50
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

Figure 6.69
Pole Slipping Algorithm Response to an ATP Simulated Stable Power Swing on a 750 kVA
Generator with Tdo' = 1.0 s, Slow AYR Model and Fast Governor Model - Test FG1.

M. J. Checksfiel d/PhD/University of Bath Chapter 6, Fig 6.69, Page 238

 KEY Trip Level Pmax

a) Generator Reactive Power

2

1
Reactive Power (VAr)

0
1
•2
•3

-4

■5

-6
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

b) Generator Real Power

4

3 NO TRIP
2
Real Power (W)

1
0
1
■2

■3
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

c) Generator Rate of Change of Power (RCP)

-20
RCP (W/s)

■■

-40 ■■

-60 ■■

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)

Figure 6.70
Pole Slipping Algorithm Response to an ATP Simulated Stable Power Swing on a 750 kVA
Generator with Tdo' = 3.0 s, Slow AYR Model and Fast Governor Model - Test FG2.

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.70, Page 239

 KEY Trip Level Pmax

a) Generator Reactive Power

0.5 ■■
Reactive Power (VAr)

-2.5 ■■

0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40

Time (s)

b) Generator Real Power

2

1
Real Power (W)

•2

■3 NO TRIP
■4
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40
Time (s)

c) Generator Rate of Change of Power (RCP)

100

80 ■■

60 ■■
RCP (W/s)

40 ■■

20 ■■

-20 ■■

0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40

Time (s)

Figure 6.71
Modified Motor Pole Slipping Algorithm Response to an ATP Simulated Synchronous
Motor Stable Power Swing on a 750 kVA Hydro Machine in Pumping Mode - Test MOT1

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.71, Page 240

 KEY Trip Level Pmax

a) Generator Reactive Power

Q must still be less than Qtrip to trip for a motor pole slip
Reactive Power (VAr)

-2 ■■

-3 ■■

-4 ■■

0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40

Time (s)

b) Generator Real Power

Real Power (W)

-4 ■■
TRIP P must be greater than Pt to trip for a motor pole slip
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40
Time (s)

c) Generator Rate of Change of Power (RCP)

200
(AP/At) must be greater than (AP/At) . to trip for a motor pole slip

150 ■■
RCP (W/s)

50 ■■

0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40

Time (s)

Figure 6.72
Modified Motor Pole Slipping Algorithm Response to an ATP Simulated Synchronous
Motor Pole Slip on a 750 kVA Hydro Machine in Pumping Mode - Test MOT2

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.72, Page 241

 KEY Trip Level Pmax

a) Generator Reactive Power

0.5

g-O J

I
Oh
-
g> -1.5

18 -2
o4
-2.5

2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00

Time (s)

b) Generator Real Power

0.5

S ' 05*

%
-1.5

-2 ■■

-2.5 + +
2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00
Time (s)

c) Generator Rate of Change of Power (RCP)

100

80 ■■

60 ■■

S' 40"
£
Pk 20 "
V
&

-20 "

-40 ■■

2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00

Time (s)

Figure 6.73
Modified Pole Slipping Algorithm Response to an ATP Simulated Pole Slip on a 750 kVA Hydro
Motor, Pole Slip Caused by an Redcuction in Machine Terminal Voltage to 80 % - Test MOT3.

M. J. Checksfield/PhD/University of Bath Chapter 6, Fig 6.73, Page 242

 KEY Trip Level Pmax
a) ATP Machine Angle
^ 200

I)
o 100
3 o
oo -100
3
^ -200
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 20.00 22.00 24.00 26.00 28.00
Time(s)
b) Generator Reactive Power (PU)
- 0.2
S ' -0 3
&
Ifl/w —
%
*2
<u
o

-0.9 i-------1-------1-------1-------1 1-------I I —1-------1-------1-------1-------1-------1-------

0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 20.00 22.00 24.00 26.00 28.00
Time (s)
c) Generator Real Power (PU)
1
0.8 r
§! 06
| 0.4

& 0-2
b -
TRIPt |r v
IS 0
Pi

-0.4 i------ 1 i i------1 i » i------ i «— i— i— i— i—

0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 20.00 22.00 24.00 26.00 28.00
Time (s)
d) Generator Rate of Change of Power (RCP) (PU/s)
2

0 . ...................................................... ...................................... — ^ -----------------------------------------

PPh -2
^ -4
(J _A
Pi
-6

-8
0.00
+

2.00 4.00 6.00

r
8.00 10.00 12.00 14.00 16.00 18.00 20.00 22.00 24.00 26.00 28.00
Time (s)
*

Figure 6.74
Pole Slipping Algorithm Response to Additional Algorithm Test' SLW' - Slow Pole
Slipping of a 200 MVA Generator, Induced by a Weak Excitation, Governor and AVR
Modelled, Pole Slipping Rate = 0.05 % slip - Test SLW.

M. J. Checksfi el d/PhD/University of Bath Chapter 6, Fig 6.74, Page 243

 Chapter 7
SUMMARY OF RESULTS

This chapter presents a summary of the test results presented in Chapter six.

7.1 RESPONSE OF THE POWER BASED POLE SLIPPING ALGORITHM

The power based pole slipping algorithm operated for all of the pole slipping tests
performed. It successfully restrained for all of the stable power swing, fault, loss of prime
mover, synchronisation, load change and adjacent generator pole slip tests. It did not
nuisance trip during the field trials conducted on a hydro and a steam turbine generator.

7.1.1 Differentiation Between 'Transient* and 'Steady State' Pole Slips

The algorithm can successfully categorise pole slips as either ’transient' or steady state.
Using the magnitude of (AP/At) as well as the ratio Pt/Pmax, the algorithm successfully
categorises all transient pole slips. If however, the algorithm only used the ratio Pt/Pmax
for the assessment, it could fail to recognise some 'transient' pole slips because it is
possible that no peak in output power will occur if the fault which causes the pole slip is
cleared at a load angle between 180° and 360°.

7.1.2 Adaptive Tripping Times for Minimising Circuit Breaker Operating Duty
If there is concern over generator circuit breaker switching duty when a generator is
disconnected during pole slipping, the algorithm can be set for delayed circuit breaker
tripping. This will minimise the asynchronous switching duty placed on the circuit
breaker as illustrated by the POWSIM test results. Instead of tripping when the algorithm
detects that a pole slip is imminent, the algorithm can delay the "trip circuit breaker"
command until the reactive power criterion resets. In an ideal system, this occurs at a
load angle of 270°, and ideal point to commence circuit breaker arc interruption.

The delayed switching option can also be used adaptively. During steady state pole slips,
the generator loses synchronism slowly and therefore the algorithm can be set to trip the
instant it detects that a pole slip is about to occur. This will typically occur at load angles

M.J.Checksfield/PhD/University of Bath 1996 Chapter 7, Page 244

of 110°, where the switching duty on the circuit breaker is not too high. If the algorithm
diagnoses a transient pole slip, tripping can be delayed until the reactive power criterion
is reset, corresponding to a load angle of 270°.

7.1.3 Analysis of the Optimum Values for the ’lag' and ’Ptol' Settings
The lag setting used by the algorithm was chosen to be set to one second. The 5 kVA
laboratory generator test results show that a shorter setting may result in incorrect updates
in Pt. For all of the tests conducted, the one second setting resulted in correct updates.

The default value for Ptol of 5 % of the generator rating worked satisfactorily for all tests.
The most demanding test conditions occurred during the 625 kVA diesel generator field
tests. Significant oscillations occurred in the real power which could have resulted in the
algorithm failing to update Pt because the magnitude of the oscillations exceeded the 'Ptol'
setting. However, the algorithm updated Pt correctly for all of the diesel generator field
tests. The real output power of the 5 kVA, 26.5 MVA and 353 MVA generators showed
no significant oscillations. The diesel generator provided the most testing conditions.
Because the algorithm functioned satisfactorily on the diesel generator, it follows that it
can be expected to function correctly for all other types of generator.

7.1.4 Algorithm Response to Adjacent Generator Disturbances

The 5 kVA and 625 kVA generator tests showed that the algorithm will remain secure
during pole slipping and loss of excitation of an adjacent generator connected to the same
bus as the protected generator. During pole slipping, the adjacent generator's reactive
power demand increases. The healthy generator will attempt to supply some of the
increased reactive power demand and will therefore increase its reactive power output.
This increases algorithm security because the protected generators reactive power output
moves away from the trip setting, Qtrip.

7.1.5 Algorithm Response to Fault Tests

Fault tests were conducted on the 5 kVA laboratory and 200 MVA PPSS generators, and
three fault disturbance records were captured by the 353 MVA steam turbine generator
field trial relay. The algorithm remained secure for all of the faults it was subjected to.
It was noted that the reactive power criterion is not satisfied at the beginning of, or during
a fault. Faults present an inductive load to a generator, the generator's reactive power
output is therefore generally positive during a fault, and therefore does not satisfy the

M.J.Checksfield/PhD/University of Bath 1996 Chapter 7, Page 245

negative reactive power trip setting.

The most testing situation for algorithm security occurs during clearance of short circuit
faults which can cause the generator load angle to increase above 90° thus satisfying the
reactive power criterion. The fundamental basis of the algorithm ensures that it remains
secure. After fault clearance, if the generator is not going to lose synchronism, it will not
have reached the CSP, the real power criterion will therefore not be satisfied. This
condition has to occur because the generator requires deceleration energy to remain stable.
However, for some faults it was observed that the real power criterion may be satisfied
for a very short time at fault clearance due to the fast transients which occur. These occur
for under one power system cycle, and algorithm operation is therefore blocked by the one
and a quarter cycle time delay applied to the algorithm.

7.1.6 Algorithm Operational Limits

The setting curves presented in Chapter four were shown to be correct by the PPSS tests.
Test 'PSL3GA' (section 6.3.1) showed that the value of external reactance at which the
algorithm’s reactive power criterion fails can be read from the setting graphs. The graphs
therefore enable the applications engineer to see if the power based algorithm is suitable
for the system to be protected. For most generators, the value of external reactance
between the generator terminals and the infinite bus is low enough to enable the algorithm
to successfully operate.

The PPSS test ’PSL3NN’ (section 6.3.1) showed that the algorithm can detect pole slips
up to a rate of 8 Hz. This equates to a slip of 16 %. This test confirmed the theory of
section 4.4.2, which predicted that the maximum rate of pole slipping that the algorithm
could detect was 8.3 Hz.

Test 'SLW' described in section 5.6 demonstrated that the algorithm could also detect slow
pole slipping rates. Slow pole slips occur when the generator is operating at low power,
with low excitation. The generator pole slipped in this test at a rate of two to three cycles
per minute. A pole slip rate of three slip cycles per minute equates to a slip of 0.1 %.
This is the lowest slip rate specified in conventional impedance based pole slipping
schemes. The algorithm is able to detect such slow pole slip rates because the section of
the pole slip cycle where the algorithm will operate occurs quickly. The section of the
pole slip which occurs after a load angle of approximately 130° occurs quickly because

MJ.Checksfield/PhD/University of Bath 1996 Chapter 7, Page 246

even at low mechanical input powers, the combined effect of the prime mover input power
and the power which the generator draws from the utility causes a rapid increase in load
angle.

7.1.7 Algorithm Response to Stable Power Swing Tests

The PPSS tests provided the most testing stable power swing conditions. The algorithm
restrained for all of the stable power swings tested. The severest test conditions occurred
when the power swings were caused by a long duration fault at low initial excitation and
low power output with the AVR set to manual. Test 'PSW1GN' (section 6.3.2) is an
example of such an operating condition. These conditions are severe because the initial
low level of excitation is further decreased by the demagnetising effect of armature
reaction during the fault. With the AVR set to manual, no field forcing occurs during the
fault, which would normally offset the effects of armature reaction. As a result, the
generator output power characteristics are no longer dominated by field excitation terms.
The field excitation terms cause the sinusoidal power-load angle characteristics. Instead
the characteristics have a significant asynchronous induction generation component. The
sinusoidal characteristics which the power based algorithm and impedance based
protection schemes are based upon therefore do not occur. The resulting asynchronous
characteristics produce real and reactive power variations which are very testing for all
types of pole slipping protection.

The algorithm did not operate during any of these severe power swing tests. This was
attributed to its adaptive (AP/At)min trip setting which enhances algorithm security without
reducing sensitivity. During transient conditions such as stable power swings, a high
(AP/At)min setting is used because large fluctuations in (AP/At) occur. The setting is
increased by the transient saliency scaling factor (AP/At)^. During steady state conditions,
a sensitive value of (AP/At)min is used because only small fluctuations in (AP/At) occur.

7.1.8 The Effect of Using a Lower ’Slip* Setting

The PPSS tests showed that a reduction in the 'slip' setting from the default value of
-0.5 % to -0.25 % resulted in faster tripping times for steady state pole slips. The highest
gain in tripping time occurs for steady state pole slips which occur from low initial
operating powers. The largest practical gain in tripping response was 140 ms for test
'PSL1GAF' (section 6.3.1). In order for the change in the 'slip' setting to be effective, the
(AP/At)fact setting had to be reduced from -25 % to - 10 % of the generator rating.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 7, Page 247

The reduction in slip setting did not result in the algorithm nuisance tripping for any of
the tests conducted. The transient saliency setting, (AP/At)^ provided adequate algorithm
security during transient conditions. Care must be taken however, if the slip setting is
reduced when a low value of the setting (AP/At)^ is used. If the algorithm were to be
released commercially, the slip setting may no longer be a setting. It could be included
in the algorithm code. It would therefore be advisable to use a limit on the value of
(AP/At)^ of twice the generator rating so that the algorithm was always secure during
transient conditions.

7.1.9 Algorithm Performance for Detecting Loss of Field Conditions

The algorithm successfully detected all of the loss of field conditions where the generator
lost synchronism and operated asynchronously.

The algorithm operated for all of the PPSS loss of field tests. The 200 MVA PPSS
generator was of round rotor construction, it therefore had virtually no saliency and could
not generate reluctance torque to keep it in synchronism. It would not detect loss of field
from zero initial power however. There would be no prime mover power available to
accelerate the rotor and cause the generator to lose synchronism.

The algorithm only operated for the high initial operating power 625 kVA diesel generator
loss of field tests. It failed to detect the low and medium initial power loss of field
conditions because the saliency present in the diesel generator rotor structure provided
enough reluctance torque to prevent the generator from losing synchronism in the
timescale of the test.

Full loss of field cover could be achieved for all types of generator by complimenting the
power based pole slipping algorithm with an additional simple power based algorithm11601.

7.1.10 Algorithm Tripping During Synchronous Operation

As predicted in Chapter four, the power based algorithm nuisance tripped during the tests
where the generator was operated at load angles of greater than 90° with the aid of a fast
AVR. This is of little concern however, as generators are not operated in this mode in
practice due to insufficient transient stability margins.

The algorithm also tripped between pole slips for the 625 kVA diesel generator test 'A3'.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 7, Page 248

The real power oscillated following the first pole slip which occurred in this test, and
these oscillations resulted in the real and rate of change of power criterions being satisfied.
Normally this would not cause a problem because the reactive power criterion would not
be satisfied. However, in this case the generator was about to slip another pair of poles,
the reactive power criterion was therefore satisfied. This trip is not serious because it
occurred between two pole slips. It could be a concern if a 'trip after 'n' pole slips'
function is implemented in the algorithm.

This trip could have been prevented by increasing the reset time used in the 'Pmax'
function. After Pmax has been updated to a value above nominal, it is held at the new
value for one second and then returned to its nominal value. The reset in Pmax results
in (AP/At)^,, being omitted from the (AP/At)min calculation. This increases algorithm
sensitivity because the algorithm is in its 'steady state' mode. The trip occurred just after
a reset in Pmax, if Pmax had remained at its high value, the 'transient' rate of change of
power setting would not have been satisfied and the algorithm would not have tripped.
It is therefore recommended that the reset time is increased from one to two seconds.
This would ensure that any transient disturbances have decayed by the time the reset
occurs.

7.1.11 Algorithm Modification to Detect Motor Pole Slipping

The 'MOT' tests described in section 5.6 confirmed the theory of section 4.6; the algorithm
can be easily modified to detect synchronous motor pole slips. If the signs of the real
power and rate of change of power criterions are reversed, the algorithm successfully
detects synchronous motor pole slipping whilst restraining for synchronous motor stable
power swings. The algorithm code would only require minor modifications to enable it
to provide synchronous motor and generator pole slipping protection in one function. This
would be required for pumped storage machines, which are often operated in both modes
and are prone to pole slipping because of the low CCT that hydro machines have in the
pumping mode.

MJ.Checksfield/PhD/University of Bath 1996 Chapter 7, Page 249

7.2 RESPONSE OF THE IMPEDANCE BASED POLE SLIPPING RELAYS

Analysis of the impedance based relays showed that they may nuisance trip during stable
power swings. The single blinder scheme is the most secure of the schemes, followed by
the double blinder, then the mho scheme. The impedance based schemes were also shown
to fail to detect pole slips whose impedance locus does not follow the 'ideal' trajectory.

The single blinder scheme will operate correctly with the right settings. However, finding
the correct settings requires detailed computer based simulations.

The single blinder scheme requires a full poie slip cycle before it wili trip and disconnect
the generator. It therefore trips after the pole slip has happened.

7.2.1 The Mho Pole Slipping Protection Scheme

The mho scheme operated for all of the PPSS pole slip tests, but failed to operate for the
625 kVA diesel generator pole slips. For these tests, the impedance locus passed
underneath its operating characteristic.

Having detected the first few pole slips, it was shown to fail to detect some of the faster
PPSS pole slips because the impedance locus travelled through its characteristic too
quickly. This problem is made worse if the locus travels across the top or bottom of the
mho circle because the locus will remain inside the trip zone for a shorter time.

The mho scheme nuisance tripped for 8 of the 12 PPSS stable power swing tests. If the
reach of the mho scheme had been reduced to prevent these nuisance trips, or a longer
time delay applied to the scheme, it would have failed to detect some of the pole slips.

7.2.2 The Single Blinder Protection Scheme

This scheme failed to detect some pole slips because the impedance locus passed
underneath its supervisory mho characteristic. If the supervisory mho reach was increased,
the scheme may have detected some of the pole slips. However, increasing the size of
the mho circle increases the chances of nuisance tripping during recoverable power
swings.

For numerous tests, the single blinder scheme also failed to detect the first pole slip cycle

MJ.Checksfield/PhDAJniversity of Bath 1996 Chapter 7, Page 250

following faults because the impedance locus crossed the right hand blinder element
before the supervisory mho element had picked up. The problem could be remedied by
increasing the reach of the supervisory mho element, but care has to be taken since this
may compromise the security of the scheme. A more complex scheme described by
Goody[79] can also be used which overcomes this problem.

Some pole slips satisfied the initial stages of the schemes logic, but then failed to satisfy
the final stages. The impedance locus would travel horizontally at first, entering the mho
characteristic, then crossing the right hand blinder unit. However, as it reached the jX
axis, it would 'bend' downwards, leaving the supervisory mho characteristic before it
crossed the left hand blinder. The scheme would therefore fail to trip. This type of
problem can be overcome by using a quadrilateral type impedance characteristic.

The scheme also failed to detect some of the fast pole slips because the impedance locus
crossed the left and right blinders in under 40 ms. In test 'PSL3NN' (Section 6.3.6.1) the
single blinder failed to operate at a mean pole slipping rate of 6.5 Hz.

The scheme did restrain for all of the stable power swing tests, showing that it is the most
secure of the conventional impedance based schemes.

7.2.3 The Double Blinder Protection Scheme

This scheme suffered from the same problems as the single blinder scheme. It was also
shown to nuisance trip during one of the PPSS stable power swing simulations. After
fault clearance, the impedance locus travelled to the right of the right blinder unit, then
travelled to the left of the right inner blinder unit before moving back out towards the load
area of the RX plane.

This nuisance trip demonstrated that the double blinder scheme does not have the inherent
security of the single blinder scheme because the impedance locus does not have to enter
and exit its characteristic on opposite sides to cause a trip.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 7, Page 251

7.3 THE EFFECTIVENESS OF STANDARD PROTECTION SCHEMES AT
DETECTING POLE SLIPPING

The response of under-voltage, over-frequency and IDMT over-current relays during pole
slipping conditions was analyzed. The analysis showed that none of the relays could be
relied on to trip during pole slipping because of the fluctuations in most power system
quantities during pole slipping.

All three types of relay were reset at some point in the pole slip cycle, which meant that
their time delay criteria were never satisfied. The problem occurs because as the pole
slipping generator reaches the in-phase position, its current and voltage return to their
normal values, resetting the relays. The problem can be overcome if a delay on reset
function is included in the relay. This would enable the relay to 'integrate' over several
pole slip cycles until their trip criterion is satisfied, but may cause difficulties for other
protection duties.

The fluctuations in real and reactive power cause the generator terminal voltage phasor
to swing back and fourth. Frequency measuring devices track these swings in phase
angle, which cause the measured frequency to rise and fall with the period of pole
slipping. These relays are therefore also reset during every pole slip cycle and are
prevented from tripping.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 7, Page 252

 Chapter 8
CONCLUSIONS & FURTHER WORK

8.1 CONCLUSIONS

With the increasing number of synchronous embedded generators being installed to

operate in parallel with utility distribution systems, generator stability is being recognised
as an area of concern. Generator instability causes pole slipping which can damage the
machine and disturb the local power system. Stator winding damage, overheating,
excitation system insulation failure and shaft fatigue may occur if a pole slipping generator
is not disconnected from the network.

The likelihood of embedded generator pole slipping is much higher than that of large 'grid'
type generators because of lower inertias, shorter time constants and longer fault clearance
times. The critical clearance time for most embedded generators is between 200 and
300 ms whereas the fault clearance times of the IDMT type over-current protection used
on most distribution systems can be up to one second.

The protection requirements of embedded generators are best met by modem multi­
function microprocessor relays. Research has therefore been conducted into a new pole
slipping algorithm which uses the computing power available in microprocessor relays to
enhance the performance provided by conventional pole slipping schemes and reduce the
setting complexity.

The traditional technique for detecting pole slipping is to examine the variations in
apparent impedance at the terminals of the generator. The simplest protection scheme
uses a mho relay. It can trip before the generator load angles reaches 180° and also
provides a degree of back up protection for faults occurring in the generator and
transformer. However, it can nuisance trip during recoverable power swings and can fail
to detect fast pole slips or pole slips whose impedance locus does not take the 'ideal'
trajectory.

The single blinder scheme is the most secure of the impedance based pole slipping
protections. It requires a complete pole slip cycle to trip and therefore trips after pole
slipping has occurred. The main disadvantage with the single blinder scheme is finding

M.J.Checksfleld/PhD/University of Bath 1996 Chapter 8, Page 253

settings which work for all pole slipping conditions.

The double blinder scheme is less secure than the single blinder scheme because unlike
the single blinder scheme, it does not require the impedance locus to enter and exit its trip
characteristic on opposite sides. It can therefore nuisance trip during stable power swings.

The pole slipping impedance locus of embedded generators can be far from ideal. This
can make it difficult to set the impedance based schemes so that they operate for all pole
slipping conditions without them nuisance tripping during stable power swings or faults.
To obtain reliable settings, detailed computer based simulations of the embedded
generator, its control equipment and the local electrical system are required. These can
be expensive and time consuming.

The level of protection required for embedded generators is specified in Engineering

Recommendation 'G59'[14] and Engineering Technical Report 113I15]. Under-voltage, over-
frequency, loss of grid, over-current and other relays are included in this specification. It
has been suggested that these relays will disconnect a pole slipping generator*151. Tests
have shown however, that these relays do not generally operate during pole slipping
because they are reset once every slip cycle as the generator passes through the in phase
position.

The alternative to installing relays to disconnect the generator when pole slipping occurs
is to ensure that pole slipping will not occur. However, this requires costly changes to
the utility protection systems and generator which make this option unfeasible.

A new power based pole slipping algorithm has been developed which monitors the
operating condition of the generator and adapts its settings according to the generator's
mode of operation. The algorithm is based on the fact that once the generator operates
beyond its Critical Stability Point (CSP), it will become unstable and pole slipping will
therefore occur. The algorithm uses real and reactive power and rate of change of real
power as inputs. It trips if the generator exceeds its CSP and its load angle is increasing.

The algorithm uses sensitive settings to detect pole slips which occur due to a loss of
steady state stability, whilst it automatically selects more secure settings when transient
conditions such as recoverable power swings occur. It offers the immediate advantage of

M.J.Checksfield/PhD/University of Bath 1996 Chapter 8, Page 254

detecting when an embedded generator is committed to a pole slip, rather than has actually
pole slipped. It therefore offers the opportunity of disconnecting the machine from the
network before possible damage or major disturbance to the power system occurs.

The algorithm’s settings are determined from readily available generator data. It is
therefore simple to set compared with conventional impedance based relays and obviates
the need for power system simulation studies.

The new algorithm has been implemented in a commercially available microprocessor

relay and tested using a variety of test platforms such as computer based simulations,
laboratory power system models and industrial generators. The algorithm has been shown
to operate correctly using computer based simulations of a 588 MVA steam turbine
generator, a 200 MVA steam turbine generator and a 750 kVA diesel generator. It has
also operated correctly during tests on a 5 kVA laboratory generator, a 625 kVA industrial
diesel generator, a 26.5 MVA hydro generator, and a 353 MVA steam turbine generator.

Tests have shown that the algorithm trips before a generator pole slips whilst remaining
secure during stable power swings, power system short circuit faults, generator
synchronisation, and generator loss of prime mover. Tripping may occur if a generator
is operated at load angles above 90°, but this mode of operation is not used in practice
because the reserve in hand for transient stability is too low.

The algorithm has also been shown to operate for generator loss of field conditions if the
generator operates asynchronously. It will not trip for loss of field conditions where the
generator maintains synchronism due to reluctance torque. The algorithm can be
complimented by a time delayed PQ trip zone characteristic^601 to provide full loss of
field protection.

The tests show that the algorithm can differentiate between 'steady state’ and ’transient'
pole slips. This information can be used to delay tripping until after the pole slip, if the
circuit breaker switching duty is a concern.

Although this research has focused specifically on embedded generators, the algorithm will
function correctly on large grid generators.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 8, Page 255

It has been shown that the algorithm can be easily modified to detect synchronous motor
pole slips. This is important for pole slipping protection of pumped storage hydro
machines. These machines have short values of critical clearing times which increase the
likelihood of pole slipping.

8.2 SUGGESTIONS FOR FUTURE W ORK

The pole slipping tests have assumed that the protection algorithm is applied to only one
generator. However, embedded generation units can use multiple generators operating in
parallel. If the generators are similar, it is expected that one relay could be used to
provide pole slipping protection for the whole group of generators. Further simulation
studies are required to confirm this theory. This approach would only be used if one
overall protection scheme was used for the whole group.

Tests have shown that the algorithm can be modified to detect motor pole slipping
protection. However thorough testing of the motor pole slipping algorithm has not been
tested.

If the algorithm were to be released commercially, most of the future work required would
be development rather than research. At present the algorithm code is written using 32
bit variables for the intel 80186 processor. The code would need optimising by converting
to 16 bit variables. The codes efficiency could be further improved, since the calculation
of the rate of change of power trip criterion is performed every 5 ms. The variables, Pt
and Pmax which are used in the calculation may only change once a second. The code
could therefore be written so that (AP/At)min is only calculated when a change in Pt or
Pmax occurs.

A setting calculation program could also be written for a commercial version of the
algorithm so that the applications engineer need only enter the generator variables Xd', Xq
and the generator rating, and the relay would then calculate the settings itself.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 8, Page 256

 Chapter 9
REFERENCES.

[1] The Energy Act 1983', Her Majesty's Stationary Office, London.
[2] 'The Electricity Act 1989', Her Majesty's Stationary Office, London.
[3] 'Public Utility Regulatory Policies Act - PURPA 1978', US Federal Energy
Regulatory Commission.
[4] FIELDING G & BRADLEY J, 'Local Generation - The Devolution of Power',
IEE Review, March 1990, pp 117-120.
[5] SMITH W P & FISCHER R N, 'Cogeneration: Small Turbine-Generator Case
Study', IEEE Transactions on Industry Applications, Vol IA-22, No.l, Jan/Feb
1986, pp 121-125.
[6] HOGWOOD E E & RICE D E, 'The Electrical Aspects of Cogeneration System
Design', IEEE Transactions on Industry Applications, Vol IA-23, No.4,
July/August 1987, pp 712-722.
[7] DALEY J M, 'Design Considerations for Operating On-Site Generators in Parallel
with Utility Service', IEEE Transactions on Industry Applications, Vol IA-21,
No.l, Jan/Feb 1985, pp.69-80.
[8] JEWELL T & RIZY TM, 'Interconnection Problems Associated with Small
Dispersed Electric Generators and Inverter Devices', ISA Transactions, 1984, Vol
23, No.2, pp 21-26.
[9] POPE J W, 'Parallel Operation of Customer Generation', IEEE Transactions on
Industry Applications, Vol IA-19, No.l, Jan/Feb 1983, pp 32-36.
[10] PRINCE W R, 'Current Operating Problems Associated with Non-Utility
Generation', IEEE Transactions on Power Systems, Vol.4, No.4, October 1989,
pp 1534-1541.
[11] NOBILE P A, 'Power System Studies for Cogeneration: What's Really Needed ?',
IEEE Transactions on Industry Applications, Vol IA-23, No.5, Sep/Oct 1987,
pp 777-785.
[12] NICHOLS N, 'The Electrical Considerations in Cogeneration', IEEE Transactions
on Industry Applications, Vol.IA-21, No.4, May/June 1985, pp 754-761.
[13] DOUGHTY R L, GISE L, KALKSTEIN E W & WILLOUGHBY R D,
'Electrical Studies for an Industrial Gas Turbine Cogeneration Facility', IEEE

M.J.Checksfield/PhD/University of Bath 1996 Chapter 9, Page 257

 Transactions on Industry Applications, Vol.25, No.4, July/August 1989,
pp 750-765.
[14] The Electricity Council, 'Engineering Recommendation G59 - Recommendations
for the Connection of Private Generating Plant to the Electricity Boards'
Distribution System', London, June 1985.
[15] The Electricity Council, 'Engineering Technical Report ET113 - Notes for
Guidance for the Protection of Private Generating Sets up to 5 MW for Operation
in Parallel with Electricity Boards' Distribution Networks', London, 1989.
[16] American National Standard, 'IEEE Guide for Interfacing Dispersed Storage and
Generation Facilities with Electric Utility Systems', ANSI/IEEE Std 1001-1988,
February 1989, IEEE New York, USA.
[17] ROOK M J, GOFF L E, POTOCHNEY G J & POWELL L J, 'Application of
Protective Relays on a Large Industrial-Utility Tie with Industrial Cogeneration',
IEEE Transactions on Power Apparatus and Systems, Vol.PAS-100, No.6,
June 1981, pp 2804-2812.
[18] WAGNER C L, FEERO W E, GISH W B & JONES R H, 'Relay Performance in
DSG Islands', IEEE Transactions on Power Delivery, Vol.4, N o.l, Jan 1989,
pp 122-129.
[19] DUGAN R C & RIZY D T, 'Electric Distribution Protection Problems Associated
with the Interconnection of Small, Dispersed Generation Devices', IEEE
Transactions on Power Apparatus and Systems, Vol.PAS-103, NO.6, June 1984,
pp 1121-1127.
[20] FARDANESH B & RICHARDS E F, 'Distribution System Protection With
Decentralised Generation Introduced into the System', IEEE Transactions on
Industry Applications, Vol.IA-20, N o.l, Jan/Feb 1984, ppl22-130.
[21] Summary Report of an IEEE Working Group Report, 'Intertie Protection of
Consumer-Owned Sources of Generation, 3 MVA or Less', IEEE Transactions on
Power Delivery, Vol.5, No.2, April 1990, pp 924-929.
[22] BOWE T R, IQBAL S, DAPKUS D & RIZY D T, 'A Decision Analysis Model
to Determine the Appropriate Level of Protection for Small Power Producer/Utility
Interconnection', IEEE Transactions on Power Delivery, Vol.PWRD-1, No.3,
July 1986, pp 78-89.
[23] HARLOW J H, 'A Multi-function Protective Relay for the Cogeneration Industry1,
IEEE Computer Applications in Power, October 1990, pp 25-30.
[24] POWELL L J, 'An Industrial View of Utility Cogeneration Protection

M.J.Checksfield/PhD/University of Bath 1996 Chapter 9, Page 258

 Requirements’, IEEE Transactions on Industry Applications, Vol IA-24, N o.l,
Jan/Feb 1988, pp 75-81.
[25] FROMM F, FRANC Z, KULENDIK B & STEINER C, 'Generator Protection with
the Digital MODURES 216', Brown Boveri Review 1987, Vol.74, No. 12,
pp 693-700.
[26] RIFAAT R M, 'Protection Scheme Considerations for Common Generator Bus
Configurations used in Cogeneration Plants', IEEE Wescanex 93 -
Communications, Computers and Power in the Modem Environment: Conference
Proceedings 1993, Ch. 60, pp 1-6.
[27] ZIEGLER G, 'Developments in Generator Protection - Design and Application
Aspects of a New Numerical Relay Range', Fifth International Conference on
Developments in Power System Protection, IEE Conf Pub 368, 1993, pp 111-114.
[28] PETTIGREW R D, 'Generator Protection Using Multi-function Digital Relays',
International Conference on New Developments in Power System Protection and
Local Control, May 25 - 28, 1994, Beijing, China, pp 390 - 399.
[29] XUESHEN Z, RONGSHANG G, ZHAOXIN Z, ZHENHUA Z, WEIJIAN W,
JUNHONG L, & ZHENHUA L, 'Development of Microcomputer-Based Complete
Protection for Large Generator-Transformer Set', International Conference on New
Developments in Power System Protection & Local Control, May 25-28, 1994,
Beijing, China, pp 110 - 115.
[30] KOULISCHER M, MAUN J C, & HENNIGER K, 'A Multi-Functions Digital
Protection for Generators', CIGRE Symposium, Boummouth 1989, Paper 2A-05.
[31] IEEE Committee Report, 'Survey of experience with Generator Protection and
Prospects for Improvements using Digital Computers', IEEE Transactions on Power
Delivery, Vol.3 , No.4, Oct 1988, pp 1511-1521.
[32] Modem Power Station Practice: Operation and Efficiency. Volume 7, C.E.G.B,
Pergamon Press, Oxford, 1971.
[33] SEETHARAMAN C K, VERMA S P & EL-SERAFI A M, ’Operation of
Synchronous Generators in the Asynchronous Mode', IEEE Transactions on Power
Apparatus and Systems, Vol.PAS-93, pp 928-934.
[34] CHALMERS B J, 'Asynchronous Performance Characteristics of Turbo-
Generators', Proc. IEE Paper No. 3940 S, Vol. 109 A, Aug 1962, pp 300-308.
[35] Transient Phenomena in Electrical Power Systems. VENIKOV V A, Pergamon
Press, Oxford, 1964.
[36] Power System Stability Volume 2: Power Circuit Breakers and Protective Relays.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 9, Page 259

 KIMBARK E W, John Wiley & Sons, New York, 1964.
[37] Power System Stability and Control. KUNDUR P, McGraw-Hill Inc, USA 1994.
[38] SHACKSHAFT G & NEILSON R, ’Results of Stability Tests on an
Under-exited 120 MW Generator’, Proc.IEE, Vol. 119, No.2, Feb 1972, pp 175-187.
[39] Industry Survey Analysis, 'Distribution Line Protection Practices', IEEE
Transactions on Power Apparatus and Systems, Vol.PAS-102, No. 10, Oct 1983,
pp 3270-3287.
[40] FIELDING G, 'Specification of Control and Protection Equipment for Independent
Generation Systems', IEE Second Annual Conference on Embedded Generation,
May 1995.
[41] FIELDING G, EVANS G W, 'Industrial Feeder Protection', Tenth International
Conference on Electricity Distribution, Part 1, 5- Consumers Installations, 1989,
pp 460 -466.
[42] CLARK H K & FELTES J W, 'Industrial and Cogeneration Protection Problems
Requiring Simulation', IEEE Transactions on Industry Applications, Vol 25, No.4,
July/August 1989, pp 766-775.
[43] MACKENZIE W F (Chairman), 'Loss-of-Field Relay Operation During System
Disturbances - Working Group Report - June 1971', IEEE Trans, on Power
Applications and Systems, Vol. PAS-94, No.5, Sept 1975, pp 1464-1472.
[44] YALLA M, 'IEEE Tutorial on the Protection of Synchronous Generators: Loss of
Field Protection', IEEE Power System Relaying Committee, 1995, ref 95-TP-102.
[45] Protective Relays Application Guide. GEC Measurements, Third Edition - 1987,
pp 316-319.
[46] LANDOLL L E, 'Summary of the Guide for AC Generator Protection',
ANSI/IEEE C37.102-1987, IEEE Transactions on Power Delivery, Vol.4, No.2,
April 1989, pp 957-964.
[47] SIMPSON R H, 'Protective Relaying for Multi-Source Generator Buses', IEEE
Transactions on Industry Applications, Vol.26, No.2, March/April 1990,
pp 330-341.
[48] MASON T H, FAIRNEY W, ARNOLD J J & THELWELL M J, 'Asynchronous
Operation of Turbo-Generators', CIGRE Vol.l , 1972, Sec 11-02.
[49] COOPER C B, EASTON V, JOHN M N, PARTON K C, SHACKSHAFT G &
WHITELEGG T M, 'Problems Associated with Limited Pole Slipping of Turbo-
Generators Following System Faults', CIGRE 1966, Vol.3, No.306, pp 1-16.
[50] MASON T H, AYLETT P D & BIRCH F H, 'Turbo-Generator Performance

M.J.Checksfield/PhD/University of Bath 1996 Chapter 9, Page 260

 Under Exceptional Operating Conditions', Proc. IEE, Vol. 106, Jan 1959,
pp 357-373.
[51] Electrical Power Systems : GUILE A E and PATTERSON , Vol 2, pp 106-141,
Oliver & Boyd, 1969.
[52] ABOLINS A, LAMBRECHT D, JOYCE J S & ROSENRERG L T, 'Effect of
Clearing Short Circuits and Automatic Re-closing on Torsional Stress and Life
Expenditure of Turbine-Generator Shafts', IEEE Transactions on Power Apparatus
and Systems, Vol.PAS-97, N o.l, Jan/Feb 1976, pp 14-25.
[53] MASRUR M A, AYOUB A K & TIELKING J T, 'Studies on Asynchronous
Operation of Synchronous Machines and Related Shaft Torsional Stresses',
Proc.IEE, Part-C, Vol.138, Jan 1991, pp 47-56.
[54] MASON C R, 'Relay Operation During System Oscillations'. AIEE Transactions,
Vol. 56, July 1937, pp 823-832.
[55] BANCKER E H & HUNTER E M, 'Distance Relay Action During Oscillations',
AIEE Transactions, Vol. 53, July 1934, pp 1073-1080.
[56] MECHRAOUI A & THOMAS D W P, 'The Influence of Power Swing on
Transmission Line Distance Protection, Measurement of Power Flow and Voltage
Phasor Oscillation at the Relaying Point', UPEC 93, Stafford, 1993, pp 585-588.
[57] YALLA M V V S, 'A Digital Multi-function Protective Relay', IEEE Transactions
on Power Delivery, Vol.7, N o.l, Jan 1992, pp 193-200.
[58] BENMOUYAL G, BARCELOUX S & PELLETIER R, 'Field Experience with a
Digital Relay for Synchronous Generators', IEEE Transactions on Power Delivery,
Vol.7, No.4, Oct 1992, pp 1984-1992.
[59] STRANNE G, 'Numerical Generator Protection', IEE Discussion Meeting, London,
17 December 1992.
[60] STALEWSKI A, GOODY J L H & DOWNES J A, 'Pole Slipping Protection’,
Developments in Power System protection, IEE Conference Publication No. 185,
pp 38-45.
[61] Applied Protective Relaying. Westinghouse Electric Corporation, 1979. Silent
Sentinels Publication.
[62] ILAR F, 'Innovations in the Detection of Power swings in Electrical Networks',
Brown Boveri Revue, Part 2, 1981, pp 87-93.
[63] CHECKSFIELD M J, 'An Investigation into Pole Slipping Protection for Small and
Medium Sized Embedded Generation', Final Year Project Report No. 12/93,
University of Bath, 1993.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 9, Page 261

[64] CHECKSFIELD M J and REDFERN M A, 'An Investigation into Pole Slipping
Protection for Small and Medium Sized Embedded Generation', Proc. 28th UPEC,
Stafford 1993, Vol 2, pp 957-960.
[65] BERRY T, CHAN K, DUNN R W, and NG F, 'Real Time Power System Training
Simulator', Proc. 25th UPEC, University of Bath 1990, pp 671-674.
[66] REDFERN M A, BARRETT J, HEWINGS D & USTA O, 'A Laboratory Facility
for Research into Digital Protection Algorithms Used for the Protection of Small
and Medium Sized Synchronous Generators', Proc. 27th UPEC. Vol.l,
September 1992, pp 16-19.
[67] HEWINGS D B, 'An Investigation into Enhanced Passive Techniques for the
Protection of Stator Windings in Small and Medium Sized Generators', MSc
Thesis, University of Bath 1992.
[68] IMHOF J A, 'Out of Step Relaying for Generators - Working Group Report', IEEE
Transactions on Power Apparatus and Systems, Vol.PAS-96, No.5, Sep/Oct 1977,
pp 1556-1564.
[69] SMAHA D W, 'IEEE Tutorial on the Protection of Synchronous Generators: Out
of Step Protection of Generators', IEEE Power System Relaying Committee, 1995,
ref 95-TP-102.
[70] Power System Stability and Control: Prabha Kundur, EPRI Power System
Engineering Series, Mc-graw-Hill Inc, London 1993.
[71] WOODWORTH M, 'Electrical Control Aspects of Small Scale CHP', Presented at
a Lecture for the IEE Power Group, London, 26 October 1990.
[72] BERDY J, 'Loss of Excitation Protection for Modem Synchronous Generators’,
IEEE Transactions on Power Apparatus and Systems, Vol.PAS-94, No.5, Sep/Oct
1975, pp 1457-1463.
[73] MASON C R, 'A New Loss of Excitation Relay for Synchronous Generators',
AIEE Trans. Vol 68, pp 1240-1245, 1949.
[74] LGPG 111 Integrated Generator Protection Service Manual R5942. GEC
ALSTHOM T & D PROTECTION AND CONTROL LTD. Stafford. July 1995
[75] ARNDT C R & ROGERS M, 'A Study of Loss-of-Excitation Relaying and
Stability of a 595-MVA Generator on the Detroit Edison System', IEEE Trans, on
Power App. and Systems, Vol. PAS-94, No.5, Sept 1975, pp 1449-1456.
[76] MORRIS W C, 'One Slip Cycle Out of Step Relay', AIEE Trans. Vol 68,
pp 1246-1250, 1949.
[77] Applied Protective Relaying: Westinghouse Electric Corporation, 1979. Silent

MJ.Checksfi el d/PhD/University of Bath 1996 Chapter 9, Page 262

 Sentinels Publication. Out of Step Relaying Chapter, Chapter 19.
[79] GOODY J, 'Overcoming Problems Associated with Impedance Measurements in
Pole Slipping Protection for Dinorwig', IEE Proc. Vol 133, Part C, No. 1,
January 1986, pp 44-48.
[80] HARKER K, 'The North Wales Super-grid Special Protection Schemes', IEE
Electronics and Power, September 1984, pp 719-723.
[81] ILAR M, ZIDAR J, FIORENTIS M, 'Innovations in Generator Protection', Brown
Boveri Review, Vol 65, 1978, No 6, pp 379-387.
[82] HABIBULLAH B, 'Wivenhoe Pumped Storage Station - System Protection Against
Loss of Synchronism', Electrical and Electronics Journal, Australia, Vol 5, No 2,
June 1985 pp 120- 128.
[83] AIEE RELAY SUBCOMMITTEE, 'Interim Report on Application and Operation
of Out-of-Step Protection', AIEE Transactions, Vol. 62, Sept 1943, pp 567-573.
[84] AKIYAMA K, SHIRAISHI K, TAMAKI H, MATSUSHIMA T, SAWAI K,
ISHIBASHI A, 'Development of a New Out-of-Step Separation System to Balance
Supply and Demands After Power System Separation', International Conference on
New Developments in Power System Protection & Local Control, May 25 - 28,
1994, Beijing, China, pp 234 - 239.
[85] EYSSEM J and PETROIANU A, 'An Investigation of Characteristics to be Used
for Detecting Out of Step Conditions', ESKOM Conference 1995.
[86] CENTENO V, DE LA REE J, PHADKE A G, MICHEL G, MURPHY J,
BURNETT R, 'Adaptive Out of Step Relaying Using Phasor Measurement
Techniques', IEEE Computer Apps. In Power, Vol. 6, No. 4, Oct. 93, pp 12-17.
[87] GISH W, 'Small Induction Generator and Synchronous Generator Constants for
DSG Isolation Studies', IEEE Trans. Power Systems, Vol PWRD - 1, No. 2,
April 86, pp 231 - 239.
[88] TAYLOR C W, HANER J M, HILL L A, MITTELSTADT A & CRESAP R L,
'A New Out-of-Step Relay with Rate of Change of Apparent Resistance
Augmentation', IEEE Transactions on Power Apparatus and Systems, Vol.PAS-
102, No.3, March 1983, pp 631-639.
[89] SHIWEN S, 'Microcomputer Based Out-of-Step Protection for Large Generator',
IEE International Conference on Advances in Power System Control, Operation
and Management, November 1991, Hong Kong, pp 839-842.
[90] EYSSEN J VAN, 'An Investigation of Characteristics for Tripping Protection
Detecting Out of Step Conditions: An ESKOM Case Study', 30th UPEC, Vol 2,

M.J.Checksfield/PhD/University of Bath 1996 Chapter 9, Page 263

 1985, pp 415-418.
[91] WINKLER W, 'A Microprocessor Based Protective System for Generator-
Transformer Units', IEE Conf Pub 302. Fourth International Conference on
Developments in Power System Protection, pp 56-60.
[92] VILLIERS L N F De, 'Out of Step Operation in A Weakly Interconnected
Network', ESKOM Conference 1995.
[93] BAOHUI Z, XIAONING K & YAOZHONG G, 'Out-of -Step Relay Based on
Microprocessor', International Conference on New Developments in Power System
Protection & Local Control, May 25 - 28, 1994, Beijing, China, pp 320 - 324.
[94] ABDELAZIZ A Y et al, 'Fast Prediction of Out of Step Condition by Artificial
Neural Networks for Power System Transient Stability Assessment', 30th UPEC,
Greenwich 1995, Vol 2, pp 789-793.
[95] LIE T T, 'A Global Detection Relay for Synchronous Generator Transient and Loss
of excitation Stability Using Neural Networks', Third APSCOM, Hong Kong 1995,
pp 351-356.
[96] EL-ARABATY A M et al, 'Out of Step Detection Based on Pattern Recognition',
International Journal of Electric Power and Energy Systems, 1994, Vol 16(4),
pp 269-275.
[97] OHURA Y et al, 'A Predictive Out of Step Protection System Based on
Observation of the Phase Difference Between Substations', IEEE Trans. Power
Delivery, Vol 5, No 4, November 1990, pp 1695-1704.
[98] ROEMISH W R & WALL E T, 'A New Synchronous Generator Out-of-Step Relay
Scheme', Parts 1 and 2, IEEE Transactions on Power Apparatus and Systems,
Vol.PAS-104, No.3, March 1985, pp 563-582.
[99] BROWN R D & McCLYMONT K R, 'A Power Swing Relay for Predicting
Generation Instability', IEEE Winter Power Meeting, New York, N.Y., Jan/Feb
1965, pp 119-224.
[100] ROEMISH W R, CLEMANS C L, LLOYD L W & SCHLEIF F R, 'An
Acceleration Relay for Power Systems', IEEE Trans, PAS 1971, pp 1150- 1154.
[101] SCHLEIF F R, LLOYD L W, WORLD R W & GISH W B, 'A Swing Relay for
the East-West Intertie', IEEE Transactions on Power Apparatus and Systems,
Vol.PAS-88, No.6, June 1969, pp 821-824.
[102] THORP J S et al,'Some Applications of Phasor Measurements to Adaptive
Protection', IEE Proc. Vol 133, Part C, No 1, January 1986, pp 44-48.
[103] A Power Based Digital Algorithm for the Protection of Embedded Generators.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 9, Page 264

 Omer Usta, Ph D Thesis, University of Bath 1992.
[104] A Digital Loss of Grid Protection. James Barrett, Ph D Thesis, University of
Bath, 1994.
[105] CUDWORTH C J & SMITH J R, 'Steam Turbine Generator Shaft Torque
Transients : a Comparison of Simulated and Test Results', Proc.IEE, Vol. 137, Pt.C,
No.5, Sep 1990, pp 327-333.
[106] CDV 22,62 Voltage Controlled Over-Current Relays. Manual R 5121, GEC
ALSTHOM T & D PROTECTION AND CONTROL LTD. Stafford.
[107] Distribution Switch Gear. R W Blower, Collins, London 1986.
[108] DINELEY J L, 'Power System Stability Part 2 - Hardware, Generators and Their
Controls', IEE Power Engineering Journal, July 1991, pp 191-196.
[109] Power System Control and Stability. ANDERSON P M and FOUAD A A, Iowa
State University Press. Vol 1, 1977.
[110] Power System Protection Volume 3 - Synchronous Machines. KIMBARK E W,
J Wiley & Sons, Inc. London 1956.
[111] DEMELLO F P & CONCORDIA C, 'Concepts of Synchronous Machine Stability
as Affected by Excitation Control', IEEE Trans. Vol. PAS-88, No.4,
April 1969, pp 316.
[112] Electrical Power Systems: GUILE A E and PATTERSON, Volume 2, pp 304-309.
Oliver & Boyd, 1969.
[113] DINELEY J L and FENWICK P J, 'The Effects of Prime Mover and Excitation
Control on the Stability of Large Steam Turbine Generators, IEEE Trans. PAS 93,
1974, pp 1613-1623.
[114] BYERLY R T AND KIMBARK E W, 'Stability of Large Electric Power Systems',
IEEE Press 1974, New York, pp 459-467, pp 468-478, pp 479-488.
[115] Electric Power Systems. WEEDY B M, Third Edition, J Wiley & Sons,
Chichester 1987.
[116] ISHIGURO et al, 'Coordinated Stabilizing Control of Exciter, Turbine and
Breaking Resistor', IEEE Trans, on Power Systems, Vol PWRS-1, No 3, August
1986, pp 74-80.
[117] AL-BAIYAT S A & RAHIM A H M A, 'Dynamic Braking Resistor-Reactor
Switching Strategies Through A Novel Linear Transformation Technique', Electric
Machines & Power Systems, Vol. 21, No. 5, Sept 93, pp 543- 568.
[118] PEELO D F et al, 'Application of a 138 kV 200 MW Braking Resistor', IEE Power
Eng Journal, Aug 1994, Vol.8, No.4, pp 188-192.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 9, Page 265

[119] DINELEY J L, 'Power System Stability Part 3 - Improved Modelling and New
York', IEE Power Engineering Journal, January 1992, pp 43-46.
[120] BUSEMANN F & CASSON W, 'Results of Full Scale Stability Tests on the
British 132 kV Grid System', Proc. IEE, Paper No.2575S, Feb 1958, pp 347-362.
[121] NICKLE C A & PIERCE C A, 'Stability of Synchronous Machines - Effect of
Armature Circuit Resistance', AIEE Transactions, Vol. 49, 1930, pp 338-351.
[122] SESHANNA P, 'Characteristics of an Asynchronized Synchronous Generator',
Electric Machines and Electromagnetics 1982 Vol.7 No.l PP.9-25
[123] Synchronous Machines. Charles Concordia. J Wiley & Sons Inc, London 1951.
[124] NILSSON N E & MERCURIO J, 'Synchronous Generator Capability Curve
Testing and Evaluation', IEEE Trans, on Power Delivery, Vol. 9, No. 1, January
1994, pp 414 - 421.
[125] DINELEY J L, 'Power System Stability Part 1 - Fundamental Theory', IEE Power
Engineering Journal, January 1991, pp 29-35.
[126] DOHERTY R E & NICKLE C A, 'Synchronous Machines-III - Torque-Angle
Characteristics Under Transient Conditions', AIEE Transactions, Vol.46,
Jan 1927, pp 1-8.
[127] Transient Phenomena in Electrical Power Systems: Problems and Solutions.
VENIKOV V A, Pergamon Press, Oxford, 1964.
[128] The Unified Theory of Electrical Machines. JONES, Butterworth & Co Ltd,
London 1967.
[129] Design Operation and Testing of Synchronous Machines. JAIN G C, Asia
Publishing House, London 1966.
[130] Advanced Electrical Power and Machines. BROS AN & HAYDEN, Pitman &
Sons, London 1966.
[131] CRARY S B & WARING M L, 'Torque Angle Characteristics of Synchronous
Machines Following System Disturbances', AIEE Transactions, Vol. 51, 1932,
pp 764-773.
[132] REDFERN M A & CHECKSFIELD M J, 'A New Pole Slipping Protection for
Numerical Relaying', 11th CEPSI Conference Proceedings, paper 12-29, Kuala
Lumpur, Malaysia, October 1996, pp 12.29/1 - 12.29/10.
[133] WALKER J H, 'Operating Characteristics of Salient Pole Machines', IEE Proc.
Vol 100, Part 2, 1953, pp 13-24.
[134] SZWANDER W, 'Fundamental Electrical Characteristics of Synchronous Turbo-
Generators', Joumal.IEE, No.91, Part II, Oct 1943, pp 185-194.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 9, Page 266

[135] Modem Power Station Practice - Vol.4 - Generator and Electrical Plant, CEGB:
1963 - Permagon Press, pp 126-153, and pp 527-533.
[136] EBERLY T W and SCHAEFER R C, 'Minimum/Maximum Excitation Limiter
Performance Goals for Small Generation', IEEE Trans. Energy Conversion, Vol 10,
No 4, December 1995, pp 714-721.
[137] NILSSON N E & MERCURIO J, 'Synchronous Generator Capability Curve
Testing and Evaluation', IEEE Trans, on Power Delivery, Vol. 9 . No. 1, January
1994, pp 414-421.
[138] ALDEN R T H et al, 'Analysis of Damping & Synchronising Torques, Part 1 -
General Calculation Method, Part 2 - Effect of Operating Conditions and Machine
Parameters', IEEE Trans, Vol PAS-98, No 5, September 1979, pp 1696-1706.
[139] Power System Stability Volume 1: Elements of Stability Calculations.
KIMBARK E W, John Wiley & Sons, New York, 1964.
[140] ELKATEB M M & FIELDING G, 'Coordinating Protection and Control of
Dispersed Generation', Tenth International Conference on Electricity Distribution,
Part 1, 5- Consumers Installations, 1989, pp 131 -135.
[141] WALLACE A R and STAPLETON S A, 'Renewable Energy Generation Schemes
Embedded in Rural Grid Systems', 30th UPEC, Greenwich 1995, Vol 2,
pp 609-612.
[ 142] CONCORDIA C, 'Effect of Prime Mover Speed Control Characteristics on Electric
Power System Performance', IEEE Trans. Vol. PAS-88, No. 5, May 1969, p 752.
[143] CONCORDIA C & CARTER G K,'Negative Damping of Electrical Machinery',
AIEE Trans. Vol. 60, March 1941, p 116.
[144] CHECKSFIELD M J & REDFERN MA, Assessment of Embedded Generator
Stability when Connected to a HV Utility Distribution System', UPEC 95,
Greenwich, September 1995, Vol 1, pp 219-222.
[145] OMAHEN P & PREPELUH F, 'Analysis of the Bulb-Unit Hydro Power Plant
Stability and its Protection Scheme Arrangement', UPEC 96, Iraklio, Crete,
September 1996, pp 555-559.
[146] CRICHTON L N, 'A System Out-of-Step and it's Relay Requirements', AIEE
Transactions, Oct 1937, Vol.56, pp 1261-1267.
[147] GANTNER J & GAILLET B, 'Protection Against Out of Step Operation of Large
Synchronous Machines', Electra No. 50, pp 77-93.
[148] MATSUKI J, OKADA T & UENOSONO C ,'L oss of Synchronism Process of
a Synchronous Generator Described by its Internal Flux and Force Distributions',

M.J.Checksfield/PhD/University of Bath 1996 Chapter 9, Page 267

 IEEE Transactions on Energy Conversion, Vol.7, N o.l, March 1992, pp 177-181.
[149] UENOSONO C, MATSUKI J & OKADA T, ’Analysis of Step-Out Process of
a Three-Phase Synchronous Machine by Air Gap Fluxes', Electrical Engineering
in Japan, Vol. 100B, N o.l, Jan, 1980, pp 41-48.
[150] KHAN K R & CORY B J, 'Developments in Digital Generator Protection', 3rd
International Conference on Developments in Power System protection, 1985,
Vol.249, Ch.47, pp 223-226.
[151] REDFREN M A, USTA O & FIELDING G, 'Protection Against Loss of Utility
Grid Supply for a Dispersed Storage and Generation Unit.' IEEE Paper 92 SM
376-4 PWRD presented at the PES Summer Meeting, Seattle, 1992.
[152] REDFERN M A, HEWETT R J, CHI WAY A A A W, 'An Investigation into
Reverse Power Flow Protection for Small and Medium Sized Embedded
Generation.' 28 UPEC, Stafford, 1993, pp 109-112.
[153] BONWICK W J & HESSION P J, 'Fast Measurement of Real and Reactive Power
in Three Phase Circuits', Proc. IEE, Part.A, Vol.139, No.2, March 1992, pp 51-55.
[154] Power System Protection, Volume 4 - Protection and Signalling. IEE Series,
London, UK, 1995, ISBN 0 85296 838 8.
[155] Digital Filters. HAMMING R W, Second Edition, 1983, Prentice-Hall Inc,
Engelwood Cliffs, New Jersey.
[156] KWONG W S, CLAYTON M J and NEWBOLD A, 'A Microprocessor Based
Current Differential Relay for Use with Digital Communication Systems', GEC
Measurements Ltd, UK.
[157] Polarised Polyphase Current Differential Protection for Distribution Feeder Circuits
Using a Digital Voice Frequency Communications Channel. CHIWAYA A A W,
Ph D Thesis, University of Bath, 1994.
[158] Power Circuit Breaker Theory and Design. FLURSCHEIM C H, IEE Publications,
Peter Peregrinus Ltd, 1975.
[159] FISCHER J D, 'Alternatives in Synchronous Motor Pull Out Protection', IEE
Conference Record of Industrial and Commercial Power Systems, Technical
Conference 1987. pp 48-51.
[160] CHECKSFIELD M J & REDFERN M A, UPEC 96, Iraklio, Crete, September
1996, pp 9-12.
[161] CHAN K W et al, 'Real Time Electro-Mechanical Transient Simulator for On Line
Applications', International Conference on Digital Power System Simulators, Texas,
USA, April 1995, p 259.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 9, Page 268

[162] REDFERN M A, BARRETT J and HEWINGS D, 'A Laboratory Facility for
Research into Digital Protection Algorithms used for the Protection of Small and
Medium Sized Synchronous Generators', UPEC 92, Vol 1, Sept. 1992, pp 16-19.
[163] ElectroMagnetic Transients Program (EMTP) Volume 3, Workboook 3.
ALVARADO F L, DOMMEL H and BRANDWAJN V, Ref EL-4651, EPRI
Research Project 2149-6, June 1989.
[164] Computer Analysis of Power Systems. ARRILLAGA J & ARNOLD C P, John
Wiley & Sons, Chichester 1990.
[165] Electrical Machine Dynamics. GUPTA S & LYNN, Macmillon Press Ltd, 1980.
[166] EMTP Rule Book 1989, MEYER et al.
[167] EMTP Theory Book 1989, MEYER et al.
[168] ROSS G & HALL M C, 'Synchronous Machine and Torsional Dynamics
Simulation in the Computation of Electromagnetic Transients', IEEE Transactions
on Power Apparatus and Systems, Vol.PAS-97, No.4, July 1978, pp 1074-1086.
[169] PARK R H, 'Two Reaction Theory of Synchronous Machines - Part I', AIEE
Transactions, Vol. 48, No. 2, July 1929, pp 716-730.
[170] PARK R H, 'Two Reaction Theory of Synchronous Machines - Part II', AIEE
Transactions, Vol. 52, No. 2, June 1933, pp 352-355.
[171] ElectroMagnetic Transients Program (EMTP) Workbook 1. ALVARADO F L,
LASSETER R H and LONG W F, Ref EL-4651.EPRI Research Project 2149-6,
September 1986.
[172] HASSAN I D, WERONICK R, BUCCI R M & BUSCH W, 'Evaluating the
Transient Performance of Standby Diesel-Generator Units by Simulation', IEEE
Transactions on Energy Conversion, Vol.7, No.3, Sep 92, pp 470-477.
[173] TANG L and ZAVADIL R M, 'Simulation of the Operational and Control
Characteristics of Hybrid Wind/Diesel Energy Systems Using the EMTP', ASME
Transactions on Wind Energy, SED - Vol 14, 1993, pp 59-66.
[174] STAVRAKAKIS G S & KARINIOTAKIS G N, 'A general Simulation Algorithm
for the Accurate Assessment of Isolated Diesel-Wind Turbine System Interaction',
Part 1, A General Multi-Machine Power System Model. IEEE Paper
Reference 95 WM 092-7 EC.
[175] STAVRAKAKIS G S & KARINIOTAKIS G N, 'A general Simulation Algorithm
for the Accurate Assessment of Isolated Diesel-Wind Turbine System Interaction',
Part 2, Implementation of the Algorithm and Case Studies with Induction
Generators, IEEE Paper Reference 95 WM 093-5 EC.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 9, Page 269

[176] Micro-Machine Model Studies of a Ships Power Supply System, HILL B H,
Ph D Thesis, University of Bath 1984.
[177] A Real Time Simulation of a Complex Engineering System Incorporating a Diesel
Generator. EVANS P A, BSc report No 22/86, University of Bath, 1986.
[178] MALATESTAS P B et al, 'Modelling and Identification of Diesel-Wind Turbines
Systems for Wind Penetration assessment', IEEE Trans on Power Systems, Vol 8,
No 3, August 1993, pp 1091-1097.
[179] ROY S et al, 'An Adaptive Control Scheme for Speed Control of Diesel Driven
Power Plants', IEEE Trans. Energy Conversion, Vol 6, No 4, December 1991,
pp 605-611.
[180] The Analysis and Experimental Assessment of the Transient Behaviour of Marine
and Off Shore Electro-Mechanical Systems. STRONACH A F, Ph D Thesis,
University of Aberdeen, 1980.
[181] HUNG W W et al, 'Dynamic Simulation of Gas-Turbine Generating Unit', IEE
Proc. Part C, Vol 138, No 4, July 1991, pp 342-350.
[ 182] HEYDEMAN J & LULF E N, 'Microprocessor Based Under-Frequency Relaying',
IEE Conf Publication No 249, International Conference on Developments in Power
System Protection, pp 24-28.
[ 183] The Diesel Engine - It's Theory, Basic Design and Economics, ARMSTRONG and
HARTMAN, Chapter 19, New York, 1959, The Macmillan Company.
[184] ANDERSON P M & MIRHEYDAR M, 'Analysis of a Diesel Driven Generating
Unit and the Possibility for Voltage Flicker', IEEE PES Summer Meeting 1994,
Paper No. 94 SM 416-8 EC.
[185] CLARKE E, 'Impedances Seen By Relays During Power Swings With and Without
Faults', AIEE Transactions, Vol. 64, 1945, pp 372- 384.
[186] Performance & Control of Electrical Machines. O'KELLY D, McGraw-Hill,
London, 1991.
[187] St. PIERRE C R, 'Loss-of-Excitation Protection for Synchronous Generators on
Isolated Systems', IEEE Transactions on Industry Applications, Vol.IA-21, N o.l,
Jan/Feb 1985, pp 81-98.
[188] Example 2, Rotating Machines Advance Level Course Notes, 17th European EMTP
Meeting, November 6-7, 1989, Leuven.
[189] Circuits, Devices and Systems. SMITH R. John Wiley & Sons, Chichester, 1984.
[190] PHADKE A G, 'Synchronized Phasor Measurements in Power Systems', IEEE
Computer Applications in Power, April 1993, Vol. 6, No.2, pp 10-15.

M.J.Checksfield/PhD/University of Bath 1996 Chapter 9, Page 270

 Appendix A
DERIVATION OF THEORETICAL IMPEDANCE, POWER
VARIATIONS AND ALGORITHM SETTING EQUATIONS

A l.l CALCULATION OF THE IMPEDANCE LOCI PRODUCED DURING POLE

SLIPPING

The method used to calculate the theoretical impedance locus during pole slipping is well
established, it was first published by Clarket36,185l

If it is assumed that the impedance of all parts of the system have the same angle, then
the total impedance between the two voltage sources can be lumped together, as shown
in Figure A l. Note that the generator is represented as a voltage source behind transient
reactance for these studies. The value of generator impedance, Zg used is therefore Xd'
+ R = Zd\ The armature resistance, R is relatively small in large synchronous machines.
Zd* is normally expressed in generators as a pure reactance, Xd'. Zd' has been used in this
analysis for completeness. The source impedance, Zs comprises of the generator
transformer impedance, inter-connection impedance, and bus source impedance. Referring
to Figure A l, the total combined impedance, Ztot is given by;

Ztot = Zg +Zs A0 )
The value of m is given by;

m = Zg / (Zg +Zs) = Z g/Z lot A(2)

The generator impedance, Zg is given by;

Zg = Z d ' =mZtot A(3)

The voltage and current at the relaying point, i.e the generator terminals is given by;

relay voltage , Er = (1 -m )E g Z 5 +mEs A(4)

relay current , Ir = {Eg Z 5 - E s ) / Ztot A(5)

The impedance 'seen' by the relay at the generator terminals, Zr, is therefore given by;

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 271

 Zr = * . ( l- /w ) £ g Z 8 + mEs A(6)
Ztot
Ir Eg Z 8 - Es

There are three power system conditions which need analysing using equation A(6).
These are when the generator internal voltage, Eg is less than, equal to, or greater than
the utility source voltage, Es.

Condition 1 Eg = Es = 1,0 pu
With this condition, A(6) can be written;

Zr (1 - /« ) Z 8 + m - m (Z 8 - 1 ) + Z 8 1 A(7)
= -m
Ztot Z 8 -1 Z 8 -1 1 - Z -8

Zr 1 +Z8 1 +Z8 A(8)

--m + ---------------------------- = - m + --------------------------
Ztot (1 - Z - 8 ) (1 + Z 8 ) 1 +Z8 -Z -S -1

Zr 1 +Z8 1 + cos 8 + jsin 8

= - m + ---------------- = - m + ------------------- --------- A(9)
Ztot ZS - Z - S 2j sin 8

Zr (1
— —m
1 -j
f 1 + cos 8 " (1
— -m
} . 1
Ztot U J ^ 2 sin 8 , 2

This equation represents a straight vertical line, since the real part is constant, whilst the
imaginary part varies as a function of 8, (recall cot = 1/tan). If A(10) is multiplied through
by Ztot, to give the relay ’seen' impedance, the lengths are multiplied by the magnitude
of Ztot, and the line is rotated anti-clockwise through the impedance angle, a. In an 'ideal'
reactive transmission system, the system impedance angle, a , is equal to 90°. Table Al
contains calculated values for the components of A(10) for a unity per unit system.

If the two generators are in phase with each other, the current is zero and the seen
impedance is infinite. If they are in complete anti-phase, the voltage becomes zero at the
middle of the impedance Ztot, and the current magnitude becomes high, giving the
appearance of a three phase short circuit at the mid-point.

For the purpose o f setting impedance relays, pole slipping loci for different ratios of Eg/Es
are required. The maximum is dependent upon the ceiling limit of the AVR. Consider the
case where the system is purely reactive and the value of Ztot is j l pu, and m is 0.75.
Solving A(6) for values of Eg/Es of 2 pu and 0.5 pu will give two curves which will

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 272

generally provide a bounds for the theoretical pole slipping impedance locus;

Table Al - Solution of A(10) for Eg/Es = 1.0, Ztot = 1Z90° pu, m = 0.75;

8 cot (8/2) Real component of Imaginary component of

Ztot * A(10) Ztot * A(10)

0 oo oo -0.25
30 3.732 1.866 -0.25
60 1.732 0.866 -0.25
90 1.000 0.5 -0.25
120 0.577 0.289 -0.25
150 0.268 0.134 -0.25
180 0 0 -0.25
210 -0.268 -0.134 -0.25

Condition 2 Eg = 2.0 Es = 1.0 pu

Applying the above figures to A(6), then;

Zr = 0.25 * 2.0 ZS + 0.75 = 0.5 Z8 + 0.75 A, n .

~Ztot 2.0 ZS -1 .0 2.0Z8 -1 .0

The maximum value of Zr/Ztot occurs at 5 = 0°, and is therefore;

maximum Zr/Ztot = (0.5 + 0.75) / (2.0 - 1.0) = 1.25

The minimum value of Zr/Ztot occurs at 8 = 180°, and is therefore;

minimum Zr/Ztot = (-0.5 + 0.75) / (-2.0 - 1.0) = 0.25/-3.0 = -0.08333

When the ratio of Es to Eg is not equal to unity, the pole slipping impedance locus is no
longer a straight line, it is a circle. In this case the impedance does not take on an infinite
value when 8 = 180°. A proof of this is given in reference1361.

Since the impedance locus is a circle, the centre of the circle must lie midway between
the maximum and minimum points, i.e between 1.25 and -0.08333. The centre of the

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 273

circle is therefore at;

centre of locus = -0.08333 + (1.25 - - 0.08333)/2 = 0.58333

The centre of the locus of Zr on the impedance plane is therefore at;

0.58333 * Ztot = 0.58333 * j l pu = j0.58333 pu.

The radius of Zr must be half the difference between the two points. The radius of the
impedance locus is therefore;

radius of locus = | Ztot | *(1.25 - - 0.08333)/2

= lpu * 0.6666 = 0.6666pu.

Condition 3 Eg = 0.5 Es = 1.0 pu

Using A(6) and the above figures, then;

Zr = 0.25 * 0.5 Z5 + 0.75 = 0.125 Z 5 + 0.75 w 1?x

~Ztot 0.5 Z5 -1 .0 0.5Z6 -1 .0

The maximum value of Zr/Ztot occurs at 8 = 0°, and is therefore;

maximum Zr/Ztot = (0.125 + 0.75) / (0.5 - 1.0) = -1.75

The minimum value of Zr/Ztot occurs at 5 = 180°, and is therefore;

minimum Zr/Ztot = (-0.125 + 0.75) / (-0.5 - 1.0) = -0.41666

Since the impedance locus is a circle, the centre of the circle must lie midway between
the maximum and minimum points, i.e between -1.75 and -0.41667. The centre of the
circle is therefore at -1.08333. The centre of the locus of Zr on the impedance diagram
is therefore at -1.08333 * Ztot = -1.08333 * j l pu = -j 1.08333 pu. The radius of Zr must
equal half the difference between the two points, i.e the radius is equal to;

1 pu * (1.75 - 0.41666)/2 = 0.6667 pu

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 274

Al.1.1 The Affect of Resistance on Impedance Loci
Embedded generators can be connected to the main utility network through cables which
have a higher resistive component than overhead lines.

The effect of a resistive component in the impedance, Ztot, is to rotate the loci derived
for a purely reactive system clockwise. Consider the same system of section A l.l, but
with a value of Ztot of 1Z45° pu, i.e, 1/V2 + j 1/V2 pu. Compared to the original
impedance of 1Z90° pu, there is a 45° difference. The impedance loci for this condition
will therefore be those shown in Figure A2, rotated clockwise around the origin by 45°.
For convenience, Figure A2 has been repeated in Figure A3c, Figure A3d shows the
impedance loci for a value of Ztot of 1Z45° pu. This shows that the impedance loci for
this condition are rotated versions of those in A3c.

A l.1.2 The Affect of Variation in the Generator to Source Impedance Ratio

Figure A3 also shows the effects of changing the ratios of Zg to Zs. Impedance loci for
values of m of 1.0, 0.75 & 0.5 have been given. When m is equal to 1.0, the impedance
Ztot is determined entirely by the generator impedance, Zg. This scenario occurs when
a generator is operating on an 'ideal' infinite bus. When m is equal to 0.5, the generator
and source impedances are equal. This situation occurs if two identical machines operate
in parallel with each other. A value of m of 0.5 is unlikely to occur in a practical
embedded generation environment because the rating of the generator is generally small
when compared to the rating of the utility bus.

A2.1 REAL AND REACTIVE POWER TRANSFER CHARACTERISTICS

Using the diagram in Figure A l, the real and reactive power characteristics can be derived
from;

Real Power, P =Real part o f E rlr*

F J A(13)
Reactive Power, Q =Imaginary part o f E r lr *

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 275

From A(5) the complex conjugate of Ir is given by;

_ (Eg Z5 - E s ) Eg Z 8 Es
Ir =
Ztot Z a Ztot Z a Ztot Z a
:.Ir = I ^ - Z S - a - -E L -Z -a A(14)
Ztot Ztot
.\Ir* = _ ^ L _ Z a - § - Z a
Ztot Ztot

where a is the impedance angle, i.e a = sin ' 1 (X/Ztot). The total apparent power of the
circuit is given by Erlr*, using A(4) & A(14);

Erlr* = ((l-/w )£gZ 5 +mEs) * f ES Z a - 8 _- Es Z a

Ztot Ztot A(15)

.-.Erlr' ■ 0 - '" ) g g 2z a - ( l l ^ ) M l z (8 + a ) + mEgEs Z ( a - S ) - H ^ l z a

Ztot Ztot Ztot Ztot

Converting from polar to rectangular format, A(15) becomes;

E rlr' = cos(«) ♦ sin(a)

Ztot Ztot
m Es2 t x .m E s2 . , .
-------- cos(a) - j sin(a)
Ztot Ztot AO 6}
- cos(5 + a ) - j ^ - m)RsEs sin(g + a )
Ztot Ztot
+ mEgEs c o s(a - 8 ) + j mEgEs s in (a - 8 )
Ztot Ztot

Using A(13) and A(16), the real power, P characteristic is given by;

P =1 1 cos (a) - a * , cos (a)

Ztot Ztot

_ (1 -m )E g E s cos(8 + ex) + mEgEf_ cos( a - 8 )

Ztof Ztof
A(17)
p cos(a) _ (l-m )£g£s cos(S + ct)
Z/o/ Zfo/

+ mEgEs_ cos(8 + a )
Zfo/

Using A(13) and A(16), the reactive power, Q characteristic is given by;

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 276

 ~ (1 -m )E g 2 . t . mEs2 . , .
Q = sin(a) -------- sin(a)
Ztot Ztot

_ (1 - m)EgEs s.n(5 + a ) + mEgEs_ sin (a _ S)

Ztot Ztot
A(18)
:.Q = ^ : m-}E-g 2- r mEs2 Sin(a)
Ztot

- l ' : m) EgEs sin(S + a) - mEgEs Sin(5 - a)

Ztot Ztot
A(17) & A(18) have been plotted for various values of Eg, Es, m & a in Figure A4.
These graphs show how variations in m, and the ratio Eg/Es effect the real power load
angle characteristics. This analysis does not allow for generator saliency, a more detailed
analysis of the real and reactive power characteristics for synchronous machines is
therefore provided below.

A3.1 SYNCHRONOUS GENERATOR REAL AND REACTIVE POW ER

CHARACTERISTICS WHEN OPERATING ON AN INFINITE BUS

Figure A5 contains the voltage vector diagrams for a synchronous generator operating in
the steady state. To derive real and reactive power load angle equations from these vector
diagrams, the following assumptions are required[123,128,130’1861;
a) a balanced supply to sinusoidally distributed stator windings.
b) saturation is neglected
c) stator slots cause no appreciable variation in rotorinductance with change
in rotor angle
d) the generator is operating on an infinite bus.
Figure A5a shows a simplified machine equivalent circuit, showing the conventions used
for this derivation. The generator convention has been adopted, the internal machine
voltage, E therefore leads the terminal voltage, V. Since the generator is operating on an
infinite bus, its terminal voltage is equal to the infinite bus voltage, V. The machine or
rotor angle, 8 is positive for a generator and the output power will also be positive.
Figure A5b shows a method for finding 5 and E, given the generator terminal voltage,
current, and power factor angle, <|>, together with the machine constants, Xd, Xq, and R,
the armature resistance. The values V, I and <|>, are drawn, then the vectors IR, IXq, IXd
are added. A line is drawn which lies on points 'o' and 'c*, then an additional line is drawn

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 277

from 'd' so that it is perpendicular to the *oc' line, meeting it at 'e\ Point 'e' is the tip of
the internal machine voltage vector, E.

Figure A5c shows the voltage vector diagram broken down into its quadrature and phase
components. Resolving along quadrature and phase axes gives;

Fcos(8 ) =E - R l q - X d l d A(19)

Ksin(S) =Xqlq - R i d A(20)

The real and reactive power characteristics at the generator terminals are given by;

S =P +jQ - V I * A(21)

where S is the apparent power. Solving A(19) & A(20) for Id and Iq gives;

= EXq - VXqcos(&) - KKsin(S)

A(22)
R 2 +XqXd

. _ KAysin(S) +ER - V R cos(8 )

A(23)
R 2 +XqXd

In complex notation the infinite bus voltage (and terminal voltage) is expressed as;

V = Vcos(5) - F/sin(5) A(24)

The generator current is expressed as;

I =Iq - j l d /* =Iq * jld A(25)

The apparent power, S is therefore given by;

S =[Tcos(5) -yFsin(8 )] *
R *X qX d A(26)

[ VXd sin(8 ) - VR cos(8 ) +ER - jV X q cos(8 ) - j V R sin(8 ) +jE X q ]

The real power, P is given by the real part of A(26);

1 Xd -Xq ' A(27)

P = sin(28) - R +EVZ s in ( 5 +0)
R 2 +XdXq .

where Z = V (R2 + Xq2 ) and 0 = sin 1 (R/Z). The reactive power, Q is given by;

l Xd -Xq A(28)
EVZ cos(8 +0) + V cos(2S) -
R 2 +XdXq

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 278

A3.1.1 Transient Power Characteristics
When a synchronous generator is perturbed by a power system disturbance such as a fault,
load change, or line switching operation, it can be said to pass through three different
modes of operation. These are the sub transient, transient, and steady state modes. These
modes of operation are a result of the high inductance to resistance ratio in the field and
damper windings. This has the effect of keeping the field and damper winding flux
linkage constant over a period determined by the time constant of the winding[123]. If an
abrupt change occurs in the stator current, the change causes an mmf which tries to
establish a flux through the field pole core. If such a flux occurred, it would tend to link
the field winding, and since the field winding wants to maintain its flux at a constant zero
level, an opposing mmf is produced by induced field current. This tends to maintain the
flux linkage of the field at zero. The only flux which can therefore be established
immediately is that which does not link the field winding. This flux passes through the
low permeance leakage paths, which are mostly in air. The resultant flux per ampere
value is low, and the reactance which results is the direct axis transient inductance, Ld'.
This value will be much smaller than Ld, the direct axis synchronous reactance due to the
flux being forced to travel through air rather than iron. The same process also occurs with
the damper windings. However, the time constant of the damper windings is much
smaller than the field winding time constant. The sub-transient damper winding effects
therefore last for a fraction of the field winding transient time.

Figure 6a shows a voltage vector diagram for the machine in steady state. The diagram
shows the quadrature and phase components of terminal voltage, V and terminal current, I.
It also shows some additional voltages. This diagram is very similar to Figure A5b. Both
diagrams can be used to find the position of the quadrature and direct axes relative to
V and I. Eq does not have any simple physical meaning, but is useful because its tip lies
on the quadrature axis and therefore identifies the axis position. When drawing IXq and
Eq, it is also possible to draw IXd and the corresponding quantity Ed. Once the location
of Ed is known, E can be found by drawing a line from Ed which is perpendicular to the
quadrature axis. E can then be found at the intersection of the two lines. Figures A5c
and A6a show that the value, E can be found from;

E = Vq +IdXd +IqR A(29)

If the field flux linkage is assumed constant, a point on the vector diagram which is
proportional to field flux linkage can be identified11231. Figure A6b shows this point,

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 279

which is denoted as Eq\ It can be seen that Eq' is given by;

Eq' = Vq +I d X d 1 +IqR A(30)

Equations A(29) and A(30) are identical in form, the difference between them is that Xd
has been substituted with Xd1, and E has been substituted with Eq'. Figure A6b also
shows that Eq' can be found from;

Eq1 = E - (Xd - X d 1) Id A(31)

The above substitutions do not just apply to equations A(29) and A(30). They can also be
applied to equations A(27) and A(28) because these equations were derived from the same
vector diagram[123]. The transient power load angle equations are therefore given by;

r( X d ' - X q ' A(32)

P = E q'V Z sin(5 +0) +V sin(28) - R
R 2 +Xd'Xq \ ^ >

fX d 1 -Xq (Xd* +Xq' A(33)

Q = E q 'V Z cos(5 +0) +F cos(25) -
R 2 +Xd’Xq { “ ,

Note that an equivalent value of Xq' does not exist since there is no field winding in the
quadrature axis. A value of Xq' does exist for solid round rotor machines, but results
from flux linking with the iron paths in the solid rotor structure. In such cases the duration
of the quadrature axis transient period and the decrease in impedance from Xq to Xq' is
much less pronounced than in the direct axis11101.

As well as a transient period following a sudden change in stator currents, a sub-transient

period also exists. This sub-transient period is shorter and occurs before the transient
period. In defining Xd' it is assumed that no rotor circuits exist except the main field
winding. This is not the case as salient pole machines will have damper windings fitted
to both axes[110]. If round rotor machines do not have dampers fitted, their steel rotor core
acts as a path for eddy currents to flow. When a sudden change in stator currents occurs,
induced currents occur in the damper circuits as well as the main field winding. These
induced currents serve to oppose the stator mmf, and are able to maintain the flux linkage
of the rotor damper circuit constant at zero for a short time. Because the rotor circuits are
closer to the air gap than the field winding circuits, the flux is forced into an even lower
permeance path than during the field winding transient. The resulting stator flux
linkage/amp is lower than the transient field winding case, and this value is defined as the
sub-transient reactance. This may occur in both the quadrature and direct axes giving rise

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 280

to Lq" and Ld".

The time that the generator direct and quadrature axes reactances remain at their sub-
transient values is small due to the low X/R ratios of the damper circuits. The generator
reactance will remain at its transient value for much longer due to the higher value of X/R
which occurs in the field winding.

A3.1.2 Calculation of Synchronous Generator Steady State and Transient Real

and Reactive Power Curves

The following section explains how to derive the quantities necessary for calculating the
steady state and transient real and reactive power machine angle characteristics.

A3.1.2.1 No Load Calculations

For no load excitation, the generator is said to be 'floating', and the value of E will
therefore be equal to V, the terminal voltage. For this reason the following names are
often used for E; 'open circuit voltage', 'no-load excitation', 'excitation e.m.f, 'internal
machine voltage', and 'voltage behind synchronous reactance' [186]. Note that at this
condition, the angle 5 will be zero. The angle 8 is known as the 'machine angle', 'rotor
angle', 'rotor load angle', or 'pole angle'. If the generator is not operating against an
infinite bus, then for the case of a simple tie-line with series resistance, r and reactance,
X, the tie line impedance can be added to the stator impedance[123]. The term 'load angle'
is used more in the context of the angle between voltages at either end of a transmission
line. It is important to know which angle is being referred to when reading texts on the
subject.

The values used to calculate the no load characteristics are as follows ;

E= 1
V = 1
Eq’ = 1
E and Eq' are also unity at no load since the current I is zero, see A(29) and A(31). The
reader is reminded that the generator terminal voltage is equal to the infinite bus
voltage, V in this case because the generator is operating on an infinite bus.

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 281

A3.1.2.2 Full Load Calculations
The following equations can be used to find E and Eq' for a generator operating at full
load excitation11231. For a generator operating at current magnitude 111, with a power factor,
pf, then complex value of the current I is given by;

i - m A (34)

The quantity Eq must then be calculated. From Figure A6a, this is given by;

(jEq) = V +(R +jXq) I A(35)

The direct axis current, Id must then be found from (Figure A5c);

Id = |/| sin(S +<()) A(36)

From Figure A5c, the angles 8 and <|), are given by;

5 = tan
{ItnjEqY and (j) = tan 4 rIm{I) A(37)
\Re{Eq)) {Re(I))

From Figure A6a, the value of E can then be found using;

E = Eq + (Xd - X q ) I d A(38)

From Equation A(31), the value of Eq' can then be found. Once E and Eq' are known,
the power load angle characteristics can be plotted using Equations A(27,28,32,33).

A3.1.3 Calculation of Generator Characteristics for a Diesel Generator

Using the above procedure, the power characteristics for the 625 kVA diesel generator
described in Appendix C were calculated. The parameters required for the calculation are
given below;
Xd = 2.25 pu, Xd' = 0.22 pu, Xq = 1.1 pu, pf = 0.8, R = 0.02 pu, V = 1.0 pu.

No Load Values.
The no load values for both generators are V = E = Eq' = 1 .0 pu.

Full Load Values (all values in pu)

The current, I is given using A(34);
I = 0.8 - j0.6
The quantity, Eq is given using A(35);

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 282

 jEq = 1 + (0.8 - jO.6) ( 0.02 + j l . l ) = 1.676 + j0.868 = 1.887
The angles 5 and <|> are given using A(37) ;
8 = tan’ 1 ( 0.868/1.676) = 27.34° <J>= tan' 1 (0.6/0.8) = 36.87°
Using (A3 6) the direct axis current is therefore ;
Id = 1 * sin (27.34 + 36.87) = 0.9
The value of E is calculated using A(38);
E = 1.887 + (2.25 - 1.1) (0.9) = 2.922
The value of Eq' can then be found from A(31);
Eq’ = 2.922 - (2.25-0.22)(0.9) = 1.095.

Applying these figures to equations A(27,28,32,33) gives the following no load and full
load, steady state and transient power load angle equations;

Pnl(ss) = 0.444sin(5 +1.04°) + 0.2323sin(25) -0.0080

Qnl(ss) = 0 . 444 c o s (5 +1.04°) + 0.2323cos(25) -0.6767

Pnl(tr) = 4.539sin(8 +1.04°) - 1.815sin(25) -0.0825

Qnl(tr) = 4 . 539c o s (8 +1.04°) - 1.815cos(25) -2.723

PJJ(ss) = 1.299sin(8 +1.04°) + 0.2323sin(25) -0.0080

Qfl(ss) = 1.299cos(5 +1.04°) + 0.2323cos(25) -0.6767

PJJ{tr) = 4.970sin(8 +1.04°) - 1.815sin(25) -0.0825

QJliir) = 4.970cos(5 +1.04°) - 1.815cos(25) -2.723

Where 'nl' refers to no load, 'fl* refers to full load, (ss) refers to steady state, and (tr) refers
to transient. Figures 3.1 and 3.2 contain plots of these equations. Chapter three discusses
the nature of these curves.

A4.1 ANALYSIS OF GENERATOR REAL AND REACTIVE POWER

CHARACTERISTICS WHEN OPERATING AGAINST A SIGNIFICANT
SOURCE IMPEDANCE, SALIENCY EFFECTS INCLUDED

For reasons of clarity, the analysis of generator real and reactive power characteristics has
so far dealt with either operation against a 'finite' bus, with no saliency, or vice versa.
This section will provide a full analysis which takes into account the effects of saliency
and operation on a 'finite' bus. Such an analysis could not be found in any other texts.
This is not surprising however, as the equations derived are complex and are not easy to

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 283

use.

Figure A7a shows the schematic representation of a generator operating against a source
impedance. Figure A7b shows the corresponding phasor diagram for this scenario, and
Figure A7c shows the voltage phasor diagram which relates the generator internal
voltage, E, the generator terminal voltage, Vg, and the infinite bus voltage, V. Note that
two angles have been used for the machine load angle. The angle between generator
internal voltage and terminal voltage, known as the machine angle is denoted as §m. The
angle between generator terminal voltage and infinite bus voltage, known as the system
load angle, is denoted as 5. Using the same form of analysis as in section A3.1,
Figure A7b shows that the direct and quadrature axis currents are given by;

_ E(Xq +X) - V(Xq ^ c o s ( S ) - V(R +r)sin(8 ) A(40)

(R +r) 2 +(Xq +X)(Xd +X)

_ V(Xd ^ s i n ( S ) * E(R +r) -V ( R + r )cos(S) A(41)

(R +r)2 + (Xq +X)(Xd +X)

In complex notation the generator terminal voltage is expressed as;

F = F c o s(S m) - 7F sin (8„) A(42)

The current expressed in complex notation is identical to equation A(25). Resolving along
the direct and quadrature axis of Figure A7c gives;

F c o s ( 5 j =E - I d X d - I q R A(43)

F s i n ( 5 j =IqXq - IdR A(44)

Substituting A(43) and A(44) into A(42), and multiplying by the complex conjugate, as
defined by equations A(25), A(40), and A(41) yields the following equations for real and
reactive power;
Q =EId - I d 2X d - I q 2Xq
* A(45)
P =Elq - IdlqXd - I q 2R +IdIqXq - I d 2R

substituting equations A(40) and A(41) into A(45) gives;

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 284

 q - £ E(Xq +X) - V(Xq -kY)cos(5) - V(R +r)sin(8 )
(Xd +X)(Xq +X) + (R +r)2

_ (E(Xq +X) - V(Xq +X)cos(8 ) - V(R +r)sin(5))

Xd A(46)
[(Xd *X)(Xq *X) + (R +r)2]2

_ (V(Xd +JQsin(S) - V(R +r)cos(S) + (R +r)E)2 y

[(Xd +X)(Xq +X) +(R + r)2]2

V(Xd +X)sin(8 ) - V(R +/•)cos(6 ) + (R +r)E)

P =£
(Xd +X)(Xq +X) +(R +rf

(E(Xq +X) - V(Xq +X)cos(5) - V(R +r)sin(8 ) )

[(Xd *X)(Xq +X) +(R +r)2]2

[(V(Xd +i')sin(8 ) - V(R +r)cos(8 ) + (R *r)E)Xd]

(V(Xd tA)sin(S) - V ( R + r ) cos(8 ) +(R + r ) E ) \

[(Xd +X) (Xq +X) + (R +r)2]2

(E(Xq +X) - V(Xq +A')cos(8 ) - V(R +r)sin(8 ) )

[(Xd +X) (Xq +X) + (R +r)2]2

[(V(Xd +X)sin(8 ) -V ( R + r ) cos(8 ) + (R +r)E)Xq\

_ (E(Xq +X) - V(Xq +^)cos(8) - V(R +r)sin(S))

R
(Xd+X)(Xq +X) +(R +r)2

Since the equations and analysis are in the same form as that of section A3.1, the transient
equations can be obtained by substituting Xd’ for Xd, and Eq' for E.

A5.1 THE EFFECT OF MACHINE LOSSES UPON THE POLE SLIPPING

ALGORITHM AND ASSESSMENT OF THE CORRECT VALUE FOR Pfact

The effect of machine losses must be considered because the fundamental basis of the
algorithm relies on power input and output differences. The accelerating power on the
generator rotor, Pa for any operating condition can be expressed in the form;

MJ.Checksfield/PhD/University of Bath 1996 Appendix A, Page 285

 Pa = Pm - P - P.loss A(48)

where Pm is the mechanical input power, P is the electrical output power, and Ploss
represents the losses inside the generator. The algorithm estimates the input power from
the electrical output power when Pa is zero, (i.e. the generator is in a steady state). If Pfact
is set to unity to simplify the argument, the trip setting, Pt is given by;

Pt = P = Pm - Phss A(49)

If the losses remained constant for both steady state and transient operation, they would
have no effect on the algorithm's operation, because they would be taken into account in
the trip setting Pt. However, if the losses increased during transient disturbances, due to
increased synchronous and asynchronous current components, the algorithm may be
adversely effected. This therefore requires investigation.

Let Padd_ioss be the additional component of losses during transient operation, due to higher
current levels. The real power criterion is satisfied when P < Pm, i.e. for the situation
where Pa goes from negative to positive. The effect of the additional losses on P„ can be
seen below;

Pa = Pm - P loss
1 - P add-loss - P A(50)

The extra component Padd.]oss can be seen to make the accelerating power more negative.
As expected, the additional losses therefore improve stability. This stands to reason,
additional losses during a transient provide the generator with an additional load, and
therefore decrease the accelerating area as defined by the equal area criterion.

The algorithm however does not take this additional loss component into account. The real
power criterion in the algorithm will therefore be satisfied earlier than it actually should
be. This shows that in theory, if the losses increased sufficiently during a critical power
swing, the algorithm could nuisance trip because the real power criterion was satisfied
before it actually should have been.

It is stressed that this is an absolute worst case consideration. With the algorithm having
three criteria, and a time dependent criterion, the likelihood of nuisance tripping is
decreased considerably. Algorithm security is threatened most during stable power swings
where the generator rotor swings up to the CSP, and there is an appreciable increase in

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 286

losses. This scenario could cause a threat to algorithm security if 'Pfact' is not chosen
correctly.

Jain[129] discusses synchronous machine losses in detail. There are many types of losses,
such as iron losses due to main and parasitic fluxes, hysteresis losses, eddy-current losses,
I2r armature losses, friction and windage losses. The loss which increases most during
transient operation is the armature I2r loss, due to the increased armature currents which
flow.

The highest value of embedded generator armature resistance can be expected to be

around 0.03 p.u. as found in table Cl of appendix C. The highest rise in armature current
expected as the generator approaches the CSP is approximately 2 pu. The highest value
of additional losses created in the generator during a transient disturbance are therefore
approximately equal to 24 % of generator rating. This is a very high figure, since most
generators typically have a value of armature resistance far below 0.03 per unit.

The value of Pfact chosen was 0.9, since this allows for 10 % additional losses during the
transient. This value for Pfact of 0.9 has worked correctly although it is much lower than
the 24 % figure. The algorithm has functioned correctly on machines with a value of
stator resistance of 0.032 pu. The development algorithm implemented in the commercial
relay hardware had several settings which, if the algorithm were to be released
commercially, would not be available. These settings were chosen at the development
stage so that the algorithm could be 'fine tuned' if necessary. One such setting was 'Pfact'.
It was never necessary to change this from the default setting of 0.9 however.

A5.2 THE VALUE OF RATE OF CHANGE OF POWER AT THE CSP

In order to choose the correct value of margin for error for the rate of change of power
trip level, (AP/At)min, an analysis of the factors which determine the rate of change of
power at the CSP is required.

The type of pole slip will have a great affect on the magnitude of AP/At. A pole slip
which results from a severe fault, will cause the machine power output to reach a high
value before the actual pole slip. At the CSP, the real power will be decreasing from this

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 287

maximum value. This type of pole slip will therefore produce high values of AP/At, and
this must be considered when choosing the setting (AP/At)min.

If the pole slip is a result of insufficient excitation it will be a steady state pole slip, and
no peak will occur in the output power before the pole slip. The value of Pmax will
therefore be at its nominal value of 1.4 * Pt.

The foregoing discussion shows that there are two ways in which a generator can lose
synchronism. These different ways will produce different variations in (AP/At), an
adaptive technique which alters the algorithm setting (AP/At)min according to which type
of pole slip is going to occur will therefore provide a faster acting, more secure algorithm.

A5.2.1 Mathematical Analysis of (AP/At)mlnat the CSP

The real power output of the generator can be expressed as a sum of three different
expressions. These expressions are related to,
a) the sinusoidal variation in output power with load angle, Ps
b) the added effect transient saliency has on this, Ptrs
c) the effect of induction generator action, i.e. asynchronous effects. , Pasyn

The synchronous parts given by (a) and (b) are described by equations A(27) and A(32),
whilst the asynchronous part is given by equation 3(7) of Chapter 3. Equation 3(7) is too
complicated to be used in a protection relaying environment because it requires too many
generator parameters, as well as information on the state of operation of the generator.
The asynchronous parts will therefore be neglected from the analysis.

When the generator is in a steady state, the algorithm assumes a sinusoidal power load
angle relationship when calculating the trip setting (AP/At)min. The magnitude of this
sinusoid is determined by the algorithm measurement, Pmax. Figure A7 shows such a
curve, along with the algorithm variables Pt and Pmax. This derivation of the value of
(AP/At) at the CSP neglects any saliency effects. These effects will be included later. The
sinusoidal power load angle relationship is given by;

P =Pmax sin (5) A(51)

The load angle at which P = Pt for load angles greater than 90° (unstable operation) is
given by :

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 288

 5 =180 - sin"1( ^ ) A(52)
c Pmax
The derivative of real power with respect to time, can be expressed as :

dP_ ^dP_ d_5 A(53)

dt db dt
The derivative of load angle with respect to time is the slip, s and is defined as;

— = -s A(54)
dt

i.e. negative values of slip exist for speeds above synchronous speed. If A(51) is
differentiated with respect to load angle then:

— = Pmax cos (5) A(55)

db

The derivative of real power at the algorithm CSP, where P = Pt is given using equations
A(52,53,54,55);
AD
dP_
= -s Pmax cos 8= (AP/A/) . A(56)
dt @CSP

The value of (AP/At)min will therefore change with generator loading. The value of
(AP/At)min becomes more negative as the generator loading increases. This stands to
reason, since as the loading increases the values of (AP/At) occurring at the CSP will
increase because the nominal value of Pmax increases with generator loading. With Pmax
its nominal value of 1.4 * Pt, (AP/At)min for steady state operation is given using A(52,56);

(AP/At) I = 0.98 *5 *Pt A(57)

min I {steady state)

Equation A(56) is used by the algorithm to calculate the rate of change of power trip
setting during non-transient conditions.

In terms of algorithm operation, generator transient conditions are defined as occurring

when the ratio Pt/Pmax falls below 0.6. This shows that the generator output power has
risen significantly above the nominal value of Pmax. This will only occur when a
transient disturbance on the generator has caused the generator output power-load angle
curve to take on transient values. Chapter three showed that during transient conditions,
the machine output power is much greater than nominal, and transient saliency
components occur in the power load angle curve. These effects must be considered if the
algorithm is to remain secure during stable power swings.

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 289

A5.2.2 Including Transient Saliency Effects in the (AP/At)min Calculation
When the generator is operating in its transient mode, the effects of transient saliency can
be very pronounced. Figures 3.1a and 3.2a in Chapter 3 show the steady state and
transient characteristics for both round rotor and salient pole machines. The peak in the
transient power load angle curve occurs around 120°, for both types of machines.
Transient saliency therefore increases the value of (AP/At) at the CSP. If resistance is
neglected, then A(32) becomes;

Eq 'V A(58)
P = sin(8 ) sin(28)
Xd {Xq X d ')

The derivative of A(58) with respect to 5 is given by;

dP _ Eq 'V A(59)
cos (8 ) V 2 cos (28)
db Xd' * If Xq
- Xd

Using A(53), A(54) and A(59) gives an expression for the rate of change of power,
neglecting asynchronous effects, which is given by;

dP Eq 'V A(60)
= -s cos(S) V 2 cos (28)
~dt Xd'

The Pmax measuring function within the algorithm measures the peak produced by the
two terms contained in A(60). There is no simple way of telling how much each term has
contributed to that peak however.

During transient conditions, algorithm security is the greatest concern since recoverable
power swings produce massive variations in all of the algorithm measurands, and the
machine characteristics become far from ideal. With this in mind a conservative approach
has been taken in including the effects of transient saliency into the calculation of
(AP/At)min which is as follows.

If the voltage, V is assumed to be 1 p.u. and it is assumed that the CSP occurs at a load
angle which makes the magnitude of cos(28) equal to unity, then an estimate of the
contribution from the transient saliency term to the (AP/At) signal at the CSP can be made.
Figures 3.1a and 3.2a shows that even at full load the intersection of the unity real power
line and the transient power load angle characteristic is such that the CSP will occur at
around 160°. This makes the cos(28) term approximately equal to unity. The above

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 290

assumption is therefore valid. The above assumptions make the approximate contribution
of the transient saliency term to (AP/At) at the CSP equal to;

(AP/At) tran ( 1 1 *Sgen A(61)

[Xq X d ')

This term is added to the existing value of (AP/At)min which was calculated by assuming
a sinusoidal power load angle characteristic. This additional transient saliency term,
(AP/At)^ is therefore only added to (AP/At)min when the ratio Pt/Pmax falls below 0.6 as
defined by equation 4(18) in Chapter 4.

A5.2.3 Errors Produced by the Assumptions Used to Derive (AP/At)min

An error occurs because steady state saliency is neglected when assuming a sinusoidal
power load angle relationship. This is only a consideration in salient pole machines.
Neglecting this effect results in a value of (AP/At)min being estimated which is more
negative than it should be. This tends to make the algorithm more secure. The degree of
steady state saliency produced in a salient pole generator is small compared to the
transient saliency term, the effect of neglecting it is therefore small and does not affect
algorithm operation.

The other main error in deriving (AP/At)min is neglecting asynchronous effects. These are
too complicated to be taken into consideration. A qualitative analysis of their effects on
the value of (AP/At) at the CSP will therefore be given. The generator’s instantaneous
asynchronous power output is given by;
P =

_ F 2 X d -X d ' sTd' 1
1 + \/( 1 +(sTd/)2).sin ^28o - ta n -1 -1st
2 X d X d ' 1 + (sTd')2 sTd
A(62)
_ v 2 X d '-X d " sTd"
1 - yj{ 1 +(sTd u)2). sin ^28o - tan-1 ^— L_.j-2.sfjj
2 X d ' X d " 1 +(sTd")2

Xq-X q" sT q "

2 X q X q " 1 +(sTq")2 )

If the mean components of A(62) (non-sinusoidal components) are differentiated with

respect to time, it can be seen that the sign of the dPM/dt term is given in the form;
. ds n ds A(63)
= -A — - B — - c *
dt dt dt dt

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 291

where the constants A, B, C are all positive. Figure A9 shows the generator rotor angle,
slip, s, and rate of change of slip, ds/dt for a stable and an unstable power swing.
Figure A9c shows that ds/dt will be positive for a stable power swing as the load angle
swings up to and back away from the CSP. The contribution of the asynchronous power
to (AP/At) at the CSP will therefore be negative, making (AP/At) more negative around
the CSP area during stable power swings.

Inspection of the unstable, pole slipping ds/dt curve shows that it changes sign as the
generator approaches and passes the CSP. The curve shown is for the case where stability
is only just lost. If a more severe fault caused instability, then the ds/dt curve may change
sign earlier because the generator would accelerate into the pole slip in a more severe
manner. The effect of the asynchronous power in this case is therefore to increase the
value of (AP/At) at the CSP, i.e the asynchronous power makes (AP/At) less negative at
the CSP.

It can therefore be concluded that neglecting steady state saliency and asynchronous power
generation effects in the calculation of (AP/At)min, results in a value of (AP/At)min which
is less negative than it should be. This suggests that the rate of change of power trip
setting derived by the algorithm would tend to be too sensitive, and may result in
algorithm nuisance tripping during recoverable power swings. However, sections A5.2.1
and A5.2.2 both state that the methods used in deriving (AP/At)min result in a value of
(AP/At)min which is more negative than it should be. This is especially true for the transient
saliency calculation. These methods therefore help to cancel out the inaccuracies
produced by neglecting asynchronous power effects and steady state saliency in the
calculation of (AP/At)min. Tests show that the algorithm is stable during recoverable power
swings, yet able to detect pole slipping. The settings are therefore satisfactory.

A5.2.4 Units Used in Equations

The units of the expressions found in this chapter require clarification. Slip, s is in
fractional units, 8 is in radians, and the units of time are in radians. In order to convert
from this time domain into the conventional time domain of seconds, the following
relationship is used; A(64)
[ rad ] o [ sec ]

The values of slip used in this work should therefore be entered in terms of s * co0,
rather than purely s, since slip is defined as the rate of change of load angle. C0o is defined

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 292

as the angular power system frequency, and is equal to 2 n f0. All voltages and machine
reactances used in equations are in per unit, on the generator base, whilst the generator
time constants also require conversion using equation A(64) tl27] .

A5.3 DERIVATION OF LIMITS OF OPERATION OF ALGORITHM AS THE

SYSTEM INFEED IS VARIED

Four operational limit graphs were constructed to show the value of source impedance
where the reactive power trip criterion would fail to work properly. One graph was
constructed to shows the effect of varying the reactance to resistance ratios in the tie line,
the other to show the effect of varying the generator internal voltage, E. The quantities
referred to are shown diagrammatically in Figure A7a. The following assumptions were
made in order to derive the graphs. The results are also given in tabular from in Tables
A2 and A3. Chapter 4 details the methods used when using these results to apply the
power based pole slipping algorithm to a generator.

a) The direct axis reactance which should be usedto determine if the algorithm will
operate satisfactorily, is the direct axis transient reactance, Xd'. This is the lowest
operational reactance likely to occur during pole slipping. Assuming the lowest operational
reactance produces a worst case scenario. Using Xd' assumes that the generator is
operating at 50 % slip1451. A slip of 0.33 % results in the generator reactance equalling 2
Xd'.

b) The value of inter-tie reactance, X, quoted is the highest theoretical value for
which the algorithm will operate. The value is derived such that 60° of the theoretical
reactive power load angle curve falls below the reactive power trip criterion, Qtrip, where
Qtrip = -1/Xq * Sgen.

c) The values of voltage behind generator reactance used were 1.25, 1.5 and 1.75 p.u.
Although it is referred to as E, technically, it should be Eq', the voltage behind transient
reactance, as defined in section A3.1.1. A value of 1.25 should be chosen if the AVR
fitted to the generator is not a high performance fast AVR[79l Normally a value of 1.5 is
the highest value expected. However, if a very fast high performance AVR is used on a
generator with a short field time constant, a value of 1.75 may occur due to field forcing

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 293

action. Section 3.1.3 contains methods for calculating Eq1.

d) The per unit resistance of the generator was assumed to be 0.03 p.u. A parametric
study on the effect of varying the stator resistance showed that a 1000 % increase in this
figure produced a 1.6 % variation in the number of degrees of the reactive power curve
which were below Qtrip.

e) For the first graph, curves have been calculated for two different amounts of inter­
tie resistance. The first amount was calculated for an inter-tie resistance of 10 % of the
inter-tie reactance. For the second situation, the inter-tie resistance was set equal to the
reactance. The effect of varying the inter-tie R/X ratio was found to be small.

f) The synchronous machine assumptions usually made apply for this analysis, i.e,
a sinusoidal, balanced voltage supply, and sinusoidally distributed windings.

g) The value of V, the infinite bus voltage was set to 1 pu in all cases.

Table A2 - Limits of Operation of Algorithm for 50 % slip case (Xg = Xd' = O.IXq) -
Effect of a Varying Tie Line Resistance .

criteria \ Xq (pu) 0.5 1.5 2.5

X(pu) for E = 1.25 , r = 0.1X, 0.0320 0.09295 0.1545

X(pu) for E = 1.25, r = X, 0.03545 0.09845 0.16215

X(pu) for E = 1.5, r = 0.1X, 0.02734 0.07942 0.13195

X(pu) for E = 1.5, r = X, 0.02968 0.083 0.13695

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 294

Table A3 - Limits of Operation of Algorithm for 50 % slip case (Xg = Xd', R = 0.1X)-
Effect of a Varying Generator Internal Voltage, E, and Xd'.

criteria \ Xq (pu) 0.5 1.5 2.5

X(pu) for E = 1.25 , Xd' = O.IXq 0.03201 0.09295 0.15445

X(pu) for E = 1.25, Xd' = 0.2Xq 0.05590 0.16510 0.27470

X(pu) for E = 1.25, Xd1= 0.3Xq 0.07423 0.2207 0.3674

X(pu) for E = 1.25, Xd' = 0.6Xq 0.10525 0.3152 0.5255

X(pu) for E = 1.25, Xd' = 0.9Xq 0.1123 0.3374 0.5628

X(pu) for E = 1.5 , Xd' = O.IXq 0.02734 0.0794 0.13195

X(pu) for E = 1.5, Xd' = 0.2Xq 0.04890 0.1444 0.2402

X(pu) for E = 1.5, Xd' = 0.3Xq 0.06633 0.1971 0.3281

X(pu) for E = 1.5, Xd' = 0.6Xq 0.0994 0.2975 0.496

X(pu) for E = 1.5, Xd’ = 0.9Xq 0.1117 0.3353 0.5593

X(pu) for E = 1.75 , Xd' = O.IXq 0.02382 0.0692 0.1150

X(pu) for E = 1.75, Xd' = 0.2Xq 0.04330 0.1279 0.2128

X(pu) for E = 1.75, Xd’ = 0.3Xq 0.05970 0.1773 0.2952

X(pu) for E = 1.75, Xd' = 0.6Xq 0.09335 0.2793 0.4654

X(pu) for E = 1.75, Xd' = 0.9Xq 0.1092 0.3275 0.5465

M.J.Checksfield/PhD/University of Bath 1996 Appendix A, Page 295

 Ztot

\
mZtot (l-m)Ztot

Relaying
Point E s Z i)
Voltage

Figure A1
Two Machine Representation of Power System
Used in Analysing Swing Loci During Pole Slipping

Eg/Es = 2.0 locus

0.5
Reactance, jX (pu)

Eg/Es = 1.0 locus 210180 150 120 90 5 = 60

-0.5

Eg/Es = 0.5 locus

1 -0.5 0 0.5 1
Resistance, R (pu)

Figure A2
Impedance Plane Pole Slipping Loci for System Shown in Figure Al, m = 0.75, Z = j 1 pu.

M. J. Checksfi el d/PhD/University of Bath Appendix A, Fig A1/A2, Page 296

 a) Impedance Plot for m =1.0, a = 90 b) Impedance Plot for m =1.0, a = 45

Eg/Esj= 2.0
lodus
Eg/Es = j 2.0
0.5 0.5 - locu$......

I
"8
Eg/Es = 1.0
-0.5 -0.5 -
locus
Eg/Es HO.5
locuSS Eg/Es =|0.5 Eg/Esj= 1.0
locusj V lodus
1 -0.5 0 0.5 I 1 -0.5 0 0.5 1
Resistance, R (pu) Resistance, R (pu)

c) Impedance Plot for m =0.75, a = 90 d) Impedance Plot for m =0.75, a = 45

Eg/Es = 2.0 \
Eg/Es = 2
locus
0.5 -

j locus

-0.5 - -0.5

E g/E s = 1.0
|g/Es = 0.5 X locus
i locus
1 -0.5 0 0.5 1 1 -0.5 0 0.5 1
Resistance, R (pu) Resistance, R (pu)

e) Impedance Plot for m =0.5, a= 90 f) Impedance Plot for m = 0.5, a = 45

Eg/Es = Eg/Es = 2|.0
locus locus !

0.5 - 0.5

Eg/Es = 1.0
locus

ijg/Es = 1.0
-0.5 - -0.5 \;lo c u s......

Eg/Es = 0.5 g/Es = 0.5

! locus locus j
1 -0.5 0 0.5 1 1 -0.5 0 0.5 1
Resistance, R (pu) Resistance, R (pu)
Figure A3
Impedance Plane Pole Slipping Loci for System Shown in Figure Al.
M. J. Checksfield/PhD/University of Bath Appendix A, Fig A3, Page 297
 a) Real Power Eg = Es = 1.0 b) Reactive Power Eg = Es = 1.0

o c = 9 0 4 = 0 .5,0.75 jl.O
2

1
Real Power (pu)

p
Cu
a> 0
£
o
CU
oc=45 73
<u 1
04

•2
-2 -

a= 4 5 ;m = 1 .0

■3
0 90 180 270 360 0 90 180 270 360
8 (degrees) 8 (degrees)

c) Real Power Eg = 0.5, Es = 1.0 d) Reactive Power Eg = 0.5, Es = 1.0

2
oe=90,m=0.5,0.75,1.0 oc=45,m=0.5

a=45,ijn=0.5 p
Real Power (pu)

a.
<u 0
o£
cu
73
<D 1
04

oc=90,m=0.75
■2
-2 -

a = 4 5 ,m : o p 9 0 ,m = 1 .0
■3
0 90 180 270 360 0 90 180 270 360
8 (degrees) 8 (degrees)

e) Real Power Eg = 2.0, Es = 1.0 f) Reactive Power Eg = 2.0, Es = 1.0

2 2
dc=90,m=0.$,0 .7 5 ,1 .0
oc=45,m=0.75
1 1
3
Real Power (pu)

3
u-
o ill Io 0
Cu
oc=45,m=d.
o4

■2
■2
oc=45,m =(

0 90 180 270 360 0 90 180 270 360

8 (degrees) 8 (degrees)

Figure A4 - Real and Reactive Power Load Angle Relationships for Figure A1.
M. J. Checksfi el d/PhD/University o f Bath Appendix A, Fig A4, Page 298
 j R Xd

—diUIlr-T

Figure A5a
Synhronous Machine Equivalent Circuit

IX(

LXq

NOTE : IXd = cd
o

IR

Figure A5b
Voltage Vector Consruction for Finding E and 5 given V, I, <(), Xd, Xq, and R

Id Xd

NOTE:
points a,b,g, e are
common to figures
IqR A5b and A5c

Figure A5c
Steady State Voltage Vector Diagram for Salient Pole Synchronous Generator

M. J. Checksfield/PhD/University of Bath Appendix A, Fig A5, Page 299

 IXd

IXq

Vd IR

Figure A6a
Steady State Voltage Vector Diagram Showing Eq, Ed and E

IXd

JXq

IXd

IR

Figure A6b
Voltage Vector Diagram Showing Eq', Eq, and E

M. J. Checksfield/PhD/University of Bath Appendix A, Fig A6, Page 300

 infinite bus
generator line, transformer
generator impedance and source impedance
Xd, Xq, R

m
V ( 8 - 5 m)
VZO

Figure A7a
Schematic Diagram Showing Generator Operating
Against an Infinite Bus Through an Inter-tie Impedance

Against An Infinite Bus Through an Inter-tie Impedance

Figure A7c
Voltage Vector Diagram Showing Direct and Quadrature
Axis Voltages for the Inter-tie Impedance.

M. J. Checksfi el d/PhD/University o f Bath Appendix A, Fig A7, Page 301

 Pmax Assumed Sinusoidal
Power Power Load Angle Curve

Critical
Stability
Algorithm Point
Real Power (CSP)
TripSetting,
Pt

5c 180 Machine
Load
Figure A8 Angle
Diagram Showing Assumed Sinusoidal Power Load Angle Variation and
the Pole Slipping Algorithm Variables, Pt and Pmax

Unstable Case Stable Case

a) Generator Rotor Angle
CSP

t <L>
’3b
9

b) Generator Slip

t
c) Generator Rate of Change of Slip

t
Time(s) — ►
Figure A9
Rotor Angle, Slip, and Rate of Change of Slip for Stable and Unstable Power Swings.

M. J. Checksfield/PhD/University of Bath Appendix A, Fig A8/A9, Page 302

 Appendix B
SETTING OF CONVENTIONAL IMPEDANCE RELAYS

The procedures used to set conventional impedance based pole slipping schemes are
contained in this chapter. Settings for the laboratory power system model 5 kVA
generator, 625 kVA industrial diesel generator, and PPSS 200 MVA steam turbine
generator are all included. Settings for the loss of field, mho, single blinder and double
blinder schemes have been derived.

B l.l LOSS OF FIELD SCHEME

The setting method used was based on recommendations taken from

references [42,43,44,45,46,68,72,73,75,187]. The basic loss of field scheme used for
embedded generators is the loss of field characteristic 1 shown in Figure 2.2. The
recommended settings are;
mho offset = -Xd'/2
mho diameter = Xd
time delay of 0.5 to 0.6 seconds.
The offset referred to here is the distance from the origin of the R-X plane and the top of
the loss o f field characteristic.

Bl.1.1 Loss of Field Settings for PPSS Test Generator

The 200 MVA PPSS generator specified in Appendix C will have the following loss of
field settings;
offset = -Xd'/2 = -j0.23832pu/2 = -jO. 120 pu
circle diameter = Xd = 1.7 pu , centre = -jO. 12 - j 1.7/2 = -j0.97 pu
circle centre = -0.120 pu - 1.7 pu / 2 = -0.97 pu
time delay = 0.5 seconds.
Because the value of Xd is significantly larger than 1 pu, a time delay of 0.5 seconds is
used in order to prevent nuisance tripping for stable swings that cause the offset mho
element to pick up[44].

M.J.Checksfield/PhD/University of Bath 1996 Appendix B, Page 303

Bl.1.2 Loss of Field Settings for 5 kVA Laboratory Generator
The data for the 5 kVA laboratory generator is given in Appendix C. The loss of field
settings are as follows;
offset = -Xd'/2 = -j 1.71 pu/2 = -j0.855 pu
circle diameter = Xd = 2.32 pu, /. centre = -j0.855 - j2.32/2 = -j2.015pu
time delay = 0.5 seconds.

Bl.1.3 Loss of Field Settings for 625 kVA Diesel Generator

The available data for this generator is given in Appendix C. The loss of field settings are
as follows;
offset = -Xd’/2 = -j0.22 pu/2 = -jO.llOpu
circle diameter = Xd = 2.25 pu, centre = -jO.110 - J2.25/2= -j 1.235pu
time delay = 0.5 seconds.

B1.2 MHO ELEMENT POLE SLIPPING SCHEME

The Mho element scheme is the simplest of the impedance based pole slipping projections.
A trip signal is issued the instant the locus enters the circle, or after a time delay if the
reach of the mho circle is large. There are various recommendations for the settings used.
The IEEE ( PR 3 & PR 64 section 9) recommend that the reach should be set short of the
HV terminals of the generator transformer down to an impedance value which is equal to
the direct axis transient reactance. If a larger mho characteristic than this is used, a time
delay should be employed to ensure that nuisance tripping does not occur for faults or
swings which appear beyond the high side terminals of the transformer. The IEEE
recommend that the angle at which a swing is non-recoverable is 120°.

Bl.2.1 Mho Element Settings for PPSS Test Generator

The pole slipping impedance loci for the 200 MVA PPSS generator were calculated using
the procedure described in Appendix A. These impedance loci are shown in Figure Bl.
Initially the mho circle was set so that it covered the generator transient reactance and up
to 80 % of the transformer. This meant that the swing locus entered the circle at a load
angle of 110°. This was unacceptable because the protection scheme was not secure
enough. The mho circle was therefore adjusted so that the Eg/Es = 1 locus entered the
mho characteristic at an load angle just greater than 120°. This protected all of the

M.J.Checksfield/PhD/University of Bath 1996 Appendix B, Page 304

generator, but none of the transformer. A time delay of 20 ms was also used, in order to
simulate delays inherent to an 'instantaneous relay' and to enhance relay security.

Figure B2 shows the mho characteristic (solid circle) with the impedance swing loci
(dotted lines) also shown, along with the system and generator impedances (extra thick
solid lines). This figure also contains part of the loss of field characteristic and shows
where the protection zones of the two schemes overlap. The intersection of the mho circle
and the Eg/Es = 1 .0 impedance locus shows that the locus enters the circle at a load angle
just greater than 120°. The diameter of the mho circle was set to 0.240 pu, its centre was
located at 0, - j 0.120 pu.

Bl.2.2 Mho Element Settings for the 5 kVA Laboratory Generator

The pole slipping impedance loci for the 5 kVA laboratory generator are shown in
Figure B5. If the mho circle was set to cover the whole of the transient reactance, the
Eg/Es = 1 .0 locus entered the circle at 92°. This was unacceptable, as the scheme would
be likely to nuisance trip for recoverable swings. The mho circle was set so that the
impedance locus entered it at 105°. This should provide cover for almost all impedance
loci down to a ratio of Eg/Es of 0.19. The chosen characteristic is shown in Figure B6,
as well as the loss of field characteristic. The two characteristics overlap providing an
enhanced protection zone. It is important to not however, that the loss of field scheme
has a 0.5 second time delay, it may therefore fail to detect fast pole slips whose
impedance loci cross the area below the mho circle. In theory such impedance loci should
occur rarely, since the effective generator voltage has to fall below 0.19 pu for this to
happen. Since the 120° load angle criterion was not satisfied, a time delay of 40 ms was
employed with the mho scheme to enhance security towards recoverable swings.

Assuming a uniform pole slipping rate, and using the Eg/Es =1. 0 impedance locus, the
locus remains in the mho characteristic for 2 * (180°-105°) = 150°. The highest rate of
pole slipping that the scheme can detect is therefore;
1507(360° * 40 ms) = 10.4 Hz
The diameter of the mho circle equalled 1.402 pu, its centre was located at 0, - j0.709 pu.

Bl.2.3 Mho Element Settings for the 625 kVA Industrial Diesel Generator
The pole slipping impedance loci for the 625 kVA diesel generator were calculated using
the procedure described in Appendix A. These loci are shown in Figure B9. If the mho

M.J.Checksfield/PhD/University of Bath 1996 Appendix B, Page 305

circle was set to cover the whole of the transient reactance, then the Eg/Es = 1 . 0
impedance locus would enter the circle at 95°. This was unacceptable because the scheme
may nuisance trip for recoverable swings. The mho circle was set so that the impedance
locus entered at a load angle of 105°. This provided cover for swings down to a ratio of
Eg/Es = 0.19. Figure BIO shows the mho circle characteristic, the loss of field
characteristic, and the impedance loci lines. A time delay of 40 ms was employed to
enhance scheme security towards recoverable power swings because the 120° criterion was
not satisfied, . Assuming a uniform pole slipping rate, and using the Eg/Es = 1.0 locus,
the locus remains inside the characteristic for 2 * (180°-105°) = 150°. The highest rate
o f pole slipping that the scheme can detect is therefore;
1507(360° * 40 ms) = 10.4 Hz
The diameter of the mho circle equalled 0.1818 pu. The centre of the mho circle was
located at -0.0076, -j0.0909 pu

B1.3 SINGLE BLINDER POLE SLIPPING PROTECTION SCHEME

With the single blinder scheme, tripping occurs after the pole slipping impedance locus
has entered the mho circle, and crossed both of the blinder units in a specified time. The
trip signal is issued after the locus has crossed the second blinder. Reference [68]
recommends that the blinder reaches are set so that tripping occurs at a load angle of 90°.
The angle of the blinder units should be set so that they are parallel with the overall
equivalent system impedance line. The supervisory mho unit should be set so that
impedance loci which pass though the generator transformer are detected. The minimum
travel time between blinder units required for the scheme to trip is usually set to 40 ms[78].

Bl.3.1 Single Blinder Settings for the PPSS Test Generator

The location of the right and left blinder units was chosen so that the Eg/Es = 1 swing
impedance locus crossed them at 100° and 260° respectively. This was necessary in order
to reduce the reach of the supervisory mho circle. This increases the security of the
scheme towards recoverable swings which may pass underneath the transient reactance on
the RX plane. An example of such as swing is shown in Figure 2.4. With the blinder
characteristics set, the supervisory mho circle size was then selected so that it intersected
the blinder units at the HV terminals of the generator transformer. This enables the
scheme to trip for pole slips whose system centre appears in the generator transformer.

M.J.Checksfield/PhD/University of Bath 1996 Appendix B, Page 306

Pole Slips up to an Eg/Es ratio of 1.5 can be detected. The setting characteristics are
shown in Figure B3, the transformer HV terminals are shown by the dash-dash line. The
angle of the blinder units was set to the overall system impedance angle (86.4°). The left
blinder crosses the R axis at a value of -0.152 pu, the right hand blinder crosses the R
axis at a value of -0.159 pu. The supervisory mho element circle was centred at 0, -j0.120
pu with a diameter of 0.415 pu. The recommended impedance locus traverse time of 40
ms refers to blinder settings at 90° and 270°, this infers a mean pole slipping rate o f ;
180/(360° * 40 ms) = 12.5 Hz.
Using this mean rate of pole slipping on the blinder settings of 100° and 260°, gives a
setting time of;
160/(360° 12.5 Hz) = 35.6 ms

Bl.3.2 Single Blinder Settings for the 5 kVA Laboratory Generator

The left and right blinders were set so that the Eg/Es = 1.0 swing locus crossed them at
270° and 90° respectively. The supervisory mho circle was then set so that swings
appearing inside the generator were detected. The angle of the blinder units was set to
the system impedance angle of 88.4°. The left blinder crosses the R axis at a value of
-0.872 pu, the right blinder crosses the R axis at a value of 0.890 pu. The supervisory
mho circle diameter was set to 2.43 pu, its centre located at 0, - j0.840 pu. Figure B7
shows the single blinder characteristic. Theoretically, the scheme should detect pole slips
for an Eg/Es ratio of 1.5 to 0.67. The recommended traverse time of 40 ms was used as
the minimum time the impedance locus could remain between the blinders. This
corresponds to a pole slipping rate of 12.5 Hz (section Bl.3.1).

Bl.3.3 Single Blinder Settings for the 625 kVA Diesel Generator
The left and right blinders were set so that the Eg/Es = 1. 0 swing locus crossed them at
270° and 90° respectively. The supervisory mho circle was then set so that swings
appearing inside the generator were detected. The angle of the blinder units was set to
the system impedance angle of 83°. The left blinder crosses the R axis at a value of
-0.116 pu, the right blinder crosses the R axis at a value of 0.119 pu. The supervisory
mho circle diameter was set to 0.306 pu, its centre was located at -0.009, - j0.100 pu.
Figure B ll shows the single blinder characteristic along with the pole slipping loci and
the supervisory mho circle. The recommended time of 40 ms was used as the minimum
traverse time between blinders.

M.J.Checksfield/PhD/University of Bath 1996 Appendix B, Page 307

B1.4 DOUBLE BLINDER SCHEME

The inner blinder elements of this scheme should be set so that only pole slipping
conditions will cause them to pick up[68]. A load angle of 130° was chosen for the inner
element settings, this was 10° above the IEEE recommended value. This should therefore
improve scheme security. The double blinder scheme produces a trip signal when the
supervisory mho element resets. Before a trip signal will be issued, the impedance locus
must first enter the mho circle, then cross the outer and inner blinder elements in a preset
time, then stay in between the inner blinder elements for a preset time, and then cross the
inner and outer blinders in a preset time. Unlike the single blinder scheme, a trip will
result if the locus enters and leaves from the same side on the impedance plane.
The reach of the supervisory mho element should be chosen to cause the least stress on
the generator circuit breaker. A load angle of 90° is therefore recommended as the
optimum tripping point. The outer blinder element reach should be chosen so that the
scheme operates for swings whose system centre appears inside the generator or its
transformer.

Bl.4.1 Double Blinder Settings for the PPSS Test Generator

The inner blinder elements were set so that the Eg/Es = 1 impedance locus crossed them
at 230° and 130° respectively. The right inner blinder crosses the R axis at 0.090 pu, and
the left inner blinder crosses at -0.083 pu.

The supervisory mho element was initially set so that tripping occurred at 90°. The
location of the outer blinders was then chosen so that the scheme would operate for the
Eg/Es =1. 5 pole slipping locus. The outer blinders were also placed so that the scheme
would operate for swings whose system centre occurred inside the transformer. The
location of the outer blinders was therefore fixed by the intersection of the supervisory
mho element, the transformer terminals and the Eg/Es =1. 5 impedance locus.

Power systems analysis showed that with these settings, the scheme failed to operate for
fast pole slips because the supervisory mho element dropped out before the impedance
locus had crossed the left outer blinder unit. This problem could be remedied in two ways.
The reach of the outer blinder elements could be reduced, or the reach of the supervisory
mho relay could be increased. Decreasing the reach of the outer blinder elements would
have produced too small a time setting ( less than 5 ms) between the outer and inner

M.J.Checksfield/PhD/University of Bath 1996 Appendix B, Page 308

blinders, consequently the reach of the mho element was increased.

Further analysis showed that decreasing the reach of the outer and inner blinders elements
and keeping the supervisory mho circle at a 90° load angle setting would have been more
beneficial. Figure B4 shows the impedance plane representation of the double blinder
scheme for the PPSS test system.

The adjustments to the scheme's settings highlight the difficulty in obtaining satisfactory
settings for such a scheme. The settings used were as follows; the outer blinder elements
cross the R axis at 0.156 pu and -0.150 pu. The supervisory mho centre was located at
0, - j 0.091 pu, its diameter equalled 0.410 pu. These figures were determined
graphically. The mho element setting will make the scheme trip at a load angle of 85°.

No recommendations could be found on the timer settings required for the double blinder
scheme. For the scheme to operate, the locus must take at least the preset time to travel
from the outer blinders to the inner blinders. In addition to this time constraint, the locus
must also stay in between the inner blinders for longer than a preset time, and must also
take longer than a preset time to travel back from inner to outer blinder. The times used
were based on the figures recommended for the single blinder scheme.

In the single blinder scheme, the mean rate of pole slipping is 12.5 Hz (Section Bl.2.3).
The double blinder scheme setting times were based on this rate. The inner elements are
located at 130 and 230°, whilst the outer elements are located at 101 and 259°. The
angular difference between the inner and outer elements is therefore 29°. The minimum
traverse time for the locus entering and leaving the characteristic is therefore;
297360° * 1/12.5 Hz = 6.4 ms
This time applies to the blinders on both sides of the RX. The time setting for in between
the inner blinders is dictated by an angular separation of 100°, the minimum time setting
is therefore;
1007360° * 1/12.5 Hz = 22.2 ms.

Bl.4.2 Double Blinder Settings for the 5 kVA Laboratory G enerator

The left and right inner blinder elements were set so that the Eg/Es =1.0 impedance locus
crossed them at load angles of 230° and 130°. The right inner blinder crosses the R axis
at 0.419 pu, and the left inner blinder crosses at -0.410 pu.

M.J.Checksfield/PhD/University of Bath 1996 Appendix B, Page 309

The supervisory mho element was set so that tripping occurred when the Eg/Es = 1 . 0
locus crossed it at 90° or 270°. The diameter was therefore equal to 1.761 pu, its centre
was located at 0, - jO.855 pu. The location of the outer blinders was then chosen so that
the scheme detected the Eg/Es = 1.5 impedance locus was. This resulted in the
Eg/Es = 1 . 0 locus crossing the outer blinder elements at 101° and 259°. The right outer
blinder crosses the R axis at 0.769 pu, the left outer blinder crosses at -0.765 pu.
Figure B8 shows the double blinder scheme for the 5 kVA laboratory generator.

The 12.5 Hz pole slipping rate was used as the basis for the double blinder schemes timer
settings. The inner elements are located at 130 and 230 °, whilst the outer elements are
located at 101 and 259°. The angular difference between the inner and outer elements is
therefore 29°. the minimum traverse time for the locus entering and leaving the
characteristic is therefore;
29°/360° * 1/12.5 Hz = 6.4 ms
This time applies for both sides. The time that the impedance locus must remain in
between the inner blinders is dictated by an angular separation of 100°, the minimum time
is therefore;
1007360° * 1/12.5 Hz = 22.2 ms.

Bl.4.3 Double Blinder Settings for the 625 kVA Diesel Generator
The inner blinder elements were set so that the Eg/Es =1 . 0 locus crossed them at 230°
and 130° respectively. The right inner blinder therefore crosses the R axis at 0.058 pu,
and the left inner blinder crosses at -0.058 pu.

The supervisory mho element was set so that tripping occurred when the Eg/Es = 1.0
impedance locus crossed it at 90° or 270°. The mho element diameter was therefore
0.232 pu, its centre was located at -0.008, - j0.098. The location of the outer blinders was
then chosen so that the scheme would operate for the Eg/Es =1.5 locus. This resulted in
the Eg/Es =1. 0 impedance locus crossing the outer blinder units at 106° and 254°. The
right outer blinder crosses the R axis at 0.095 pu, the left outer blinder crosses at
-0.092 pu. Figure B12 shows the impedance plane representation of the double blinder
scheme for the 625 kVA diesel generator.

The 12.5 Hz pole slip rate criterion was used as the basis for providing the timer settings
for the scheme. The inner elements are located at 130 and 230 °, whilst the outer elements

M.J.Checksfield/PhD/University of Bath 1996 Appendix B, Page 310

are located at 106 and 254°. The angular difference between the inner and outer elements
is therefore 24°. The impedance locus must therefore remain between the inner and outer
blinders for a time of least;
24°/360° * 1/12.5 Hz = 5.3 ms
This time applies for both sides of the scheme. The time that the locus must remain
between the inner blinders is dictated by an angular separation of 100°, the minimum time
is therefore;
100°/360° * 1/12.5 Hz = 22.2 ms.

M.J.Checksfield/PhD/University of Bath 1996 Appendix B, Page 311

 0.3
KEY Eg/Es = 2.8
system or
generator 0.2
impedance
Eg/Es = 1.5
pole slip
locus 3
3 Transformer
90 load angle X
-X— value 8 Eg/Es = 1.0 150 120 90

I -0-1

-0.2 Eg/Es = 0.67

Eg/Es = 0.19
-0.3
-0.3 - 0.2 - 0.1 0 0.1 0.2 0.3
Resistance, R (pu)

Figure B1
Pole Slipping Loci for ATP 200 MVA PPSS Test System.

KEY
system or
generator
impedance
pole slip
locus
3 - 0.1
90 load angle
-X- value
protection - 0.2
jM h o / Loss of Field
relay Characteristic Characteristic
characteristic
-0.3

-0.4
- 0.2 - 0.1 0 0.1 0.2
Resistance, R (pu)

Figure B2
Single Mho Pole Slipping Protection Scheme for PPSS Test System

M. J. Checksfield/PhD/University of Bath Appendix B, Fig B 1,B2, Page 312

 KEY blinder blinder
system or 0.2
generator
impedance 0.1
pole slip
locus
0
protection
relay
characteristic - 0.1

- 0 .2 - - s i p f i f c
visory
mho

-0.3 -0.2 -0.1 0 0.1 0.2 0.3

Resistance, R (pu)

Figure B3
Single Blinder Pole Slipping Protection Scheme for 200 MVA PPSS Test System

0.3
KEY left outer left inner
blinder blinder
system or right inner right outer.
0.2
• generator blinder blinder
impedance
pole slip
locus 3
protection 3
■ relay
characteristic
£ super- I
visoreyl
8 -o.i mho

- 0.2

-0.3
-0.3 -0.2 -0.1 0 0.1 0.2 0.3
Resistance, R (pu)

Figure B4
Double Blinder Pole Slipping Protection Scheme for 200 MVA PPSS Test System

M. J. Checksfi el d/PhD/University of Bath Appendix B, Fig B3,B4, Page 313

 0.5
KEY Eg/Es = 2.8
system or
generator
0
impedance
Eg/Es = 1.5
pole slip
locus 0.5
3
90 load angle X 210 180 150 1120
-X— value 8c
Eg/Es = 1.0
§ 1
$
Eg/Es = 0.67
1.5

Eg/Es = 0.19
•2
1 -0.5 0 0.5 1
Resistance, R (pu)

Figure B5
Pole Slipping Loci for the Laboratory 5 kVA Generator.

KEY / Mho
0
Characteristic
system or
s is '
generator
300 ■150*120 90
impedance
pole slip -1
locus
3
a
'w'
90 load angle
X
-X -- value
■2
protection
relay
characteristic

3
Loss of Field
Characteristic
1 0 1
Resistance, R (pu)
Figure B6
Single Mho Pole Slipping Protection Scheme &
Loss of Field Scheme for the Laboratory 5 kVA Generator.

M. J. Checksfield/PhD/University of Bath Appendix B, Fig B5,B6, Page 314

 KEY “\ N Mho
\ /
§f ••• Characteristic.
system or • i _ I
generator •
\ \ ••
impedance V /
■■

90
pole slip
locus

load angle
-X-— value
3
3
300 /
7s
^
V
— __
;150‘120*90| \ 60
---------X-; r ^ r " X*—7 H - j X X ------

H /
X
£
N / \

•
protection
\
L 1• \ /\
M %
—— relay •• Xx. * \ ^*1 \ •%
$ ••
characteristic -2 1 ••
N%
%
Left | Right \
Blinder! Blinder
-3
-2 0 1
Resistance, R (pu)

Figure B7
Single Blinder Pole Slipping Protection Scheme for the Laboratory 5 kVA Generator.

0.5
KEY left oiuter rigfit outer
system or blinder blinder
■■■» generator
impedance
pole slip -0.5 -
locus 150 120 90
3
90 load angle 3
X
-X*— value
protection
i
Bo
“ relay 8 Mho
characteristic
Characteristic
-2 -

left inndr right inner

blinder Hlinder

1.5 1 -0.5 0 0.5 1 1.5

Resistance, R (pu)

Figure B8
Double Blinder Pole Slipping Protection Scheme for the Laboratory 5 kVA Generator.

M. J. Checksfield/PhD/University of Bath Appendix B, Fig B7,B8, Page 315

 o.i
Eg/Es = 2.8
KEY
0.05
system or Eg/Es = 1.5
m^ m generator
impedance 0
pole slip
locus -0.05
X
90 load angle 2 1 0 lO Q
, 150...J120
0.1
-X-— value -

protection I
-0.15
relay
characteristic
- 0.2
Eg/Es = 0.67
Eg/Es = 0.19
-0.25
0.1 -0.05 0 0.05 0.1 0.15 0.2 0.25
Resistance, R (pu)

Figure B9
Pole Slipping Loci for the Industrial 625 kVA Diesel Generator.

KEY
system or
generator
impedance - 0.1
pole slip
locus .'• f A
protection Mho Loss of Field
relay Characteristic Characteristic
characteristic

-0.4

- 0.2 - 0.1 0 0.1 0.2

Resistance, R (pu)

Figure B10
Single Mho Pole Slipping Protection Scheme & Loss of Field Scheme
for the Industrial 625 kVA Diesel Generator.

M. J. Checksfi el d/PhD/University of Bath Appendix B, Fig B9,B 10, Page 316

 KEY
system or
generator
£L° I 240 210’'" '
impedance
-o.i
pole slip
locus
90 load angle
--X-— value - 0.2

protection
relay
characteristic Mho
-0.3
Characteristic
Left Blinder Right Blinder
- 0.2 - 0.1 0 0.1 0.2
Resistance, R (pu)

Figure B ll
Single Blinder Pole Slipping Protection Scheme for the Industrial 625 kVA Diesel Generator.

0.1
I Mho Characteristic \
KEY
system or
• generator
impedance
pole slip
locus
protection x -o.i
“ relay
characteristic i
&
- 0.2 -

IV iV I /
inner b lin d e rs /^
outer blinders
-0.3
- 0.2 - 0.1 0 0.1 0.2

Resistance, R (pu)

Figure B12
Double Blinder Pole Slipping Protection Scheme for the Industrial 625 kVA Diesel Generator.

M. J. Checksfield/PhD/University of Bath Appendix B, Fig B 11,B 12, Page 317

 Appendix C
DATA FOR GENERATORS AND SYSTEMS USED
FOR ALGORITHM TESTS

This appendix contains data on the PPSS test system, the 5 kVA generator laboratory
power system model, the diesel generator field test system, and typical embedded
generator data.

C l.l PPSS TEST SYSTEM

Figure Cl shows the power system model used for testing the power based pole slipping
algorithm which had been installed in a commercially available generator protection
relay1741. The model was simulated using the ATP[166], and was based on a system used in
an EPRI stability evaluation example[,63]. Several modifications were made to this system
so that all of the simulations required could be performed. The equivalent source
impedance and interconnection impedances were reduced so the generator could operate
at load angles in excess of 90 degrees1501. The generator AVR was also adjusted for some
of the tests. Chapter 5 contains details on the changes made to the AVR.

C l.1.1 Governor Model

The governor model was also changed from the original hydro governor unit to a tandem-
compound single reheat model1164,1881. This model was faster than the original hydro unit,
but does not constitute a ’fast' governor, such as the type found in a system with fast
valving. The governor model was modified so that it would work for a 2 pole machine
rather than a 4 pole, this involved changing the first multiplier unit which converts the
velocity into per unit values. The block diagram for the governor is shown in Figure C2.

Cl.1.2 AVR Model

The AVR model is shown in Figure C3. This is an IEEE type 1 model with no saturation.
The limiter originally had infinite limits, these limits were changed to +/- 2 so that the
maximum field voltage that the AVR could produce was equal to three times the full load
value[1091. The AVR parameters had to be changed for the simulations where the generator
was operated at machine angles of greater than 90 degrees. This was necessary because
the generator damping was too low for the generator to remain operating in a stable

M.J.Checksfi el d/PhD /University of Bath 1996 Appendix C, Page 318

manner.

C l.1 .3 G enerator M odel

The generator used in the simulation had the following parameters.
Rating = 200 MVA
line to line terminal volts (rms) = 13.8 kV
Frequency = 50 Hz (3000 rpm, 2 pole )
Field current for rated volts on air gap line = 935.016 Amps
Field current for rated volts on no load curve = 1000 Amps
Field current for 1.2 pu volts on no load curve = 1440 Amps
R - armature resistance = 0.001096 pu
XI - Leakage reactance = 0.15 pu
Xd - direct axis synchronous reactance = 1.70 pu
Xq - quadrature axis synchronous reactance = 1.64 pu
Xd' - direct axis transient reactance = 0.238324 pu
Xq1 - quadrature axis transient reactance = 1.64 pu (Xq)
Xd" - direct axis sub transient reactance = 0.18469 pu
Xq" - quadrature axis sub transient reactance = 0.185151 pu
Tdo' - o/c d-axis transient time constant = 6.194876 s
Tqo' - o/c q-axis transient time constant = 0. (No Q axis field winding)
Tdo" - o/c d-axis sub transient time constant = 0.028716 s
Tqo" - o/c q-axis sub transient time constant = 0.074960 s
Xo - zero sequence reactance = 1.50 pu
HICO - inertia constant = .181128 million pound-feet2

The inertia constant expressed in terms of H is given by[166];

W R2 (million l b f ) = 2 W * SbasejMVA) *23.7304 C(l)

Wbase2 (rads/s)

The value of H for the generator is therefore;

H = 0.181128 * (100 iz)2 / 2 * 200 * 23.7304 = 1.88 seconds.

Some other useful formulae for converting between different units of inertia are given
below11831; ^ _ JJ [ft
[7T n]2 £^2)
1800G

M.J. Checksfiel d/PhD /University of Bath 1996 Appendix C, Page 319

Where J is the moment of Inertia in Kgm2 and n is the shaft speed in rpm, G is the
generator rating in VA.

J {units o f K gm 2) = (units o f Ibmft2) C(3)

Another relationship between H and J is;

(freCl \2
„ 19 74 7 ^ C(4)
t i —---------------------

Where freq is the system frequency in Hz, N is the number of pole pairs. T is sometimes
referred to as GD2. Generally J is specified in metric units and GD2 in imperial, though
physically they are a measure of the same quantity.

C l.1.4 Generator Infinite Bus Details

The generator was connected to an infinite bus via a 90 mile distributed parameter
transmission line and a thevenin impedance. The thevenin impedance was modelled as
a mutually coupled RL type (ATP type 51,52,53), this provides positive and zero phase
sequence modelling so that unbalanced faults can be modelled correctly. The distributed
parameter transmission line (ATP type -1,-2,-3) provides accurate simulation at frequencies
other than the fundamental power system frequency. The generator transformer was
modelled as a lossless transformer with no magnetising impedance, the leakage reactance
values used were based on an assumed 3 % reactance divided between the windings. The
impedance between the 'type 14' ideal voltage source and the generator is comprised of
the following;
Thevenin impedance : 0.0075 £2, 5 mH (= j 2*n *50 * 5 * 10‘3 £2)
i.e 0.0075 + jl.5 7 £2
Transmission Line : 0.0243 Q/mile * 90 miles = 2.187 £2
0.92238 mH/mile * 90 miles = 83.142 mH
1.e 2.187 + j26.12 £2
Transformer : 11.69mH * 2 = 23.38 mH
i.e j7.35
TOTAL. 2.195 + i35.04 £2 = 35.11 Z86.4° £2

The bus rating at the terminals of the generator is therefore;

(230 * 103 )2 / 35.11 = 1506.7 MVA,
which in terms of multiples of the generator rating is 7.5. These values can be converted

M.J.Checksfield/PhD/University of Bath 1996 Appendix C, Page 320

into per unit values on the generator base using;

Z( pu ) = Z(fl) * base VA c(5)

base line voltage 2

The generator base impedance is given by its rating, 200 MVA. The base line voltage
used for the above impedances is 230 kV. The per unit impedance is therefore;
Z (pu) = 35.11 * 200 * 106 / (23 0 * 103)2 = 0.132 74 pu
Some pole slipping schemes require the zone of protection to cover the generator and its
transformer. The per unit impedance of the transformer is therefore required, this is given
by;
X (pu) = 7.35 * 200 * 106 / (230 * 103)2 = j 0.02779 pu
Referring to Figure A l, the parameters for the PPSS test system are;
Ztot = 0.00833 + j 0.134278 + 0.001096 pu + j 0.238324 pu
Ztot = 0.009426 + j0.372602 = 0.37272Z88.6 pu
m = 0.238324 / 0.37272 = 0.63941.

Cl.1.5 Load & Fault Information

The load located at bus 13 on Figure Cl is approximately 235.4 MW, 12.75 MVAr, when
a voltage of 132 kV exists across it. The short circuit faults described for the PPSS test
system were placed at BUS 12 of Figure Cl. The generator therefore supplied the fault
through its step up transformer, and 3.75 miles of mutually coupled line (ATP type 1,2,3).
The short circuit faults were achieved by connecting the relevant phases to ground, they
were cleared by disconnecting the load and the 3.75 mile line.

C1.2 LABORATORY POW ER SYSTEM MODEL

Figure C4 shows a schematic diagram of the laboratory power system model. More
information on this test system can be found in references [66,67,104]. The laboratory
power system model has some inaccuracies. One inaccuracy is that the compound wound
DC machine used as the generator prime mover does not behave like a regulated turbine.
A regulated turbine will have a droop characteristic so that the turbine output will fall
from full load to no load if the speed rises 4 % above synchronous speed[109]. This will
typically occur with a time constant of 1 second. Below this 1 second region, the turbines
torque will drop with an increase in speed, the turbine torque coefficient is approximately

M.J.Checksfield/PhD/University of Bath 1996 Appendix C, Page 321

equal to the loading of the machine in pu[109]. A compound DC machine on the other hand,
will have a reasonably flat torque speed characteristic around synchronous speed[189]. It
will therefore draw more power from its supply if there is an increase in speed, unlike a
turbine.

The data available for the 5 kVA laboratory generator1671 is as follows;

Rating 5 kVA
line to line terminal volts (rms) 200 V
Frequency 50 Hz ( 3000 rpm, 2 pole )
Field current for rated volts on air gap line 3.75 Amps
Field current for rated volts on no load curve 4 Amps
Field current for 1.2 pu volts on no load curve 6.35 Amps
R - armature resistance 0.03 pu
XI - Leakage reactance *** 0.24 pu *
Xd - direct axis synchronous reactance 2.32 pu
Xq - quadrature axis synchronous reactance 2.23 pu
Xd1 - direct axis transient reactance 1.71 pu
Xq' - quadrature axis transient reactance 1.71 pu (Xq)
Ld- direct axis damper inductance 0 pu
L q - quadrature axis damper inductance 0 pu
LF - field winding inductance 7.72 pu
Tdo' - o/c d-axis transient time constant = Lf/Rf 0.808 s
Tqo' - o/c q-axis transient time constant 0. (No Q axis field winding)
Tdo" - o/c d-axis sub transient time constant 0 s (no damper)
Tqo" - o/c q-axis sub transient time constant 0 s (no damper)
Xo - zero sequence reactance 0.24 pu
HICO - inertia constant, H = 0.35 s
From the tests conducted, it was apparent that the machine had no damper windings. The
inductance base for the machine is 0.025 H, the resistance base is 7.96 ohms.

* Leakage Reactance - This parameter can not be measured1110], it is the reactance resulting
from the difference between the total flux produced by stator current acting alone, and the
space fundamental of the flux in the air gap. It forms part of Xd and Xq, in addition to
all other positive and negative sequence reactances. The value for XI is typically between

M.J.Checksfield/PhD/University of Bath 1996 Appendix C, Page 322

5 and 10 % of Xd for generators with good coupling between stator and field. Generators
with good coupling will have a large difference between Xd and Xd', this is typically an
order of magnitude.

Other Laboratory Power System Model Data

Prime Mover —> 8 Horse Power Compound DC motor (6 kW)
Bus Details —> The laboratory is supplied by 0.5 km of cable from a 250 kVA
source, i.e source impedance approximately equals jO. 16 at 200 V,
the approximate cable impedance is equal to 0.15 + j0.15ft
Total source impedance —> 0.15 + j0.31 ft
Converting into PU on generator base = ((0.15+j0.31 ft)* 5 kVA)/2002
= 0.0188+j0.0388 pu

Referring to Figure A l, the parameters for the laboratory 5 kVA generator are;
Ztot = 0.0188 + j 0.0388 + 0.03 pu + j 1.71 pu
Ztot = 0.0488 + j 1.7488 = 1.7495Z88.40 pu
m = 1.710 / 1.7495 = 0.9774.

C1.3 625 KVA INDUSTRIAL DIESEL GENERATOR

Data for the basic parameters of the generator was obtained from the manufacturer. The
generator was manufactured over 40 years ago, the manufacturers therefore do not hold
much of the data required for a full accurate simulation. The missing data has been
estimated from inspection of the generator, or typical values have been used

Rating 625 kVA

Line to line terminal volts (rms) 415 V
Frequency 50 Hz (1500 rpm, 4 pole)
R - armature resistance 0.02 pu (estimated)
exciter current for rated volts on no load 0.625 Amps (13.5 Volts)
exciter current for rated volts at full load 1.75 Amps (48.0 Volts)
XI - Leakage reactance 0.11 pu (estimated 5 % of Xd)
Xd - direct axis synchronous reactance 2.25 pu
Xq - quadrature axis synchronous reactance 1.1 pu (estimated)

M.J.Checksfield/PhD/University of Bath 1996 Appendix C, Page 323

Xd' - direct axis transient reactance = 0.22 pu
Xq1 - quadrature axis transient reactance = 1.1 pu (no Q axis damper)
Xd" - direct axis sub transient reactance = 0.15 pu
Xq" - quadrature axis sub transient reactance = 0.20 pu (estimated)
Tdo' - o/c d-axis transient time constant = 3.0 s (estimated)
Tqo' - o/c q-axis transient time constant = 0. (No Q axis field winding)
Tdo" - o/c d-axis sub transient time constant = 0.024 s (estimated)
Tqo" - o/c q-axis sub transient time constant = 0.016 s (estimated)
Xo - zero sequence reactance = 0.05 pu (estimated)
Inertia, WR2 of generator rotor (not diesel engine) = 853.43 pound-feet2

Using equation C l, this value of WR2 is equivalent to a value of H of 0.7 kWs/kVA.

Prime Mover —> Dorman Diesel Engine.

Generator Bus Details —>ratio of bus rating to generator rating = 48:1 i.e 30 MVA
BSP bus details 150 MVA i.e source impedance = (11 kV)2/150 MVA = j0.8067 £2
Interconnection impedance * 1.8 H- j 1.8 £2
Total source impedance =1. 8 + j3.607
Converting into PU on generator base = ((1.8 + j3.607 £2)*625 kV A )/ll kV2
= 0.0093 + j 0.0186 pu

Referring to Figure A l, the parameters for the 625 kVA generator are;
Ztot = 0.0093 + j 0.0186 + 0.02 pu + j 0.22 pu
Ztot = 0.0293 + j0 .2 3 86 = 0.240Z83.00 pu
m = 0.2209 / 0.240 = 0.9204.

C1.4 DATA FOR DIFFERENT SIZED SYNCHRONOUS MACHINES

Tables Cl and C2 below contain parameters for a range of synchronous machine sizes.
The tables provide information which can be used to compare small and large generators,
whilst also providing information which can be used when simulating embedded
generators.

M.J.Checksfield/PhD/University of Bath 1996 Appendix C, Page 324

 Table Cl - Generator Data for Different Sized Synchronous Machines.

rating 5 65 69 156 200 300 781 750

[67] [173] [87] [87] [175] [175] [87]
(kVA)

Xd (pu) 2.32 1.52 2.02 6.16 3.45 2.6 2.43 2.63

Xd’ (pu) 1.71 0.289 0.171 0.347 0.24 0.22 0.254 0.21

Xd" (pu) 1.71 0.230 0.087 0.291 0.165 0.15 0.207 0.15

Xq (pu) 2.23 1.52 1.06 2.49 2.07 1.56 1.12 1.57

Xq” (pu) 2.23 0.230 0.163 0.503 0.248 0.225 0.351 0.18

XI (pu) 0.12 0.15 - - - - - -

Tdo' (se c ) 0.808 3.36 0.950 1.87 1.725 1.182 1.9 2.35

Tdo" (se c ) 0 0.06 0.078 0.013 0.015 0.015 0.024 0.025

Td' (sec) 0.596 0.639 0.08 0.105 0.12 0.10 0.199 0.188

Tqo" (se c) 0 0.06 0.045 0.020 - - 0.016 -

R (pu) 0.03 0.008 0.011 0.034 0.095 0.067 0.017 0.032

H (sec) 0.35 0.82 0.329 0.205 1.5 1.8 0.5 0.22*

p.f. 0.85 0.8 0.8 0.8 0.8 0.8 0.8

* The inertia constant, H is for the generator only, it does not include the inertia of the
prime mover system.

M.J.Checksfield/PhD/University of Bath 1996 Appendix C, Page 325

 Table C2 - Generator Data for Different Sized Synchronous Machines.

rating 1.044 4.51 9 17.5 25 35 75 384

[87] [109] [109] [109] [109] [109] [109]
(MVA)

Xd (pu) 2.38 2.95 0.911 1.07 1.25 1.0 1.05 1.798

Xd' (pu) 0.264 0.25 0.408. 0.66 0.232 0.260 0.185 0.324

XdM(pu) 0.201 0.17 0.329 0.330 0.120 0.235 0.130 0.260

Xq (pu) 1.1 1.35 0.580 0.66 1.220 0.620 0.980 1.778

Xq" (pu) 0.376 0.31 - - 0.120 0.264 0.130 0.255

XI (pu) - 0.15 - 0.310 0.134 0.170 0.070 0.193

Tdo' (se c ) 2.47 5.5 4.2 5.4 4.750 7.1 6.1 5.21

Tdo” ( s e c ) 0.018 0.055 - - 0.059 - 0.038 0.042

Td' (se c ) 0.27 0.467 1.88 3.33 0.88 1.846 1.07 0.940

Tqo" (se c ) 0.009 0.27 - - 0.210 - 0.099 0.042

R (Pu) 0.013 0.01 - 0.003 .0014 0.004 0.003 .0014

H (se c ) 0.535 1.05 2.61 6.7 5.02 7.25 6.18 2.62

p.f. 0.8 - 0.9 0.8 0.8 0.9 0.8 0.85

* The inertia constant, H is for the generator only, it does not include the inertia of the
prime mover system.

M.J.Checksfield/PhD/University of Bath 1996 Appendix C, Page 326

 Thevenin 90 Mile
equivalent Distributed Generator
impedance Parameter line. Transformer
BUS7 BUS1 GEN3

Type 14 Fault Type 59

Ideal Clearing Synchronous
Voltage Breaker Machine
Transformer
Source
modelled as
inductance
BUS 13 IUUUCUU1WS BUS 12 BKR1
■dHHb-
LOAD Mutually Coupled

i
Fault
Switch RLC Model of 3.75
mile 230 kV line.

F ig u re d
ATP Power System Model Used for Producing 200 MVA Generator PPSS Test Files.

Unity Unity

VEL TACOM DIFROME KDIFROM PS1 PS2 PS21

0.4

0.0031831
-0.4

PGV3 PGV2 PGV1 PGV

1 +0.5 s 1 + 0.25 s 0.5 s

^o"
PM1 PMT PM
0.483 0.243 0.274
PM2
PM3

Figure C2
Steam Turbine Governor Model Used in
ATP Power System Simulations for 200 MVA Generator PPSS Test Files.

M. J. Checksfield/PhD/University of Bath Appendix C, FigC l/C 2, Page 327

 Unity
Input
Filter Regulator Limiter
Exciter
Vdc 0.52359878 400
1 + 0.03 s 1 + 0.02 s V2 1 + 0.015 s
V3

0.03 s DVf
V3
Stabilising Unity
Feedback

|GEN3GA| + |GEN3GB| + |GEN3GC| * SQRT (3/2)

Vdc =
Vline Vf

Figure C3
AVR Model Used in ATP Power System Simulations for 200 MVA Generator PPSS Test Files.

Reserve 2 , , Reserve 1

Main 2 CB5 Main 1

Bus
CB2 Section

CB4 CB6
variable
CB1X XCB3 srX - X CB7 XCB8
resistive
local load
& Fault variable resistive
Switch grid load

X Bench CB

Protection DC DC
Relay Motor Motor Incoming
Incoming Generator 2 Generator 1 Feeder 1
Feeder 2
Figure C4
Schematic Diagram of Laboratory Model Power System
Used to Test Pole Slipping Algorithm in MPR.

M. J. Checksfield/PhD/University of Bath Appendix C, FigC3/C4, Page 328

 Appendix D

PUBLISHED WORK

PAPER ONE

"AN INVESTIGATION INTO POLE SLIPPING PROTECTION FOR

SMALL AND MEDIUM SIZED EMBEDDED GENERATION"

M J Checksfield and M A Redfem

Presented at
28th Universities Power Engineering Conference (UPEC),
September 1993, Stafford, UK, pp 957 - 960.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 329

 AN INVESTIGATION INTO POLE SLIPPING PROTECTION
FOR SMALL AND MEDIUM SIZED EM BEDDED GENERATION.

M.J.Checksfield and M A Redfem

University of Bath, UK.

The probability o f pole slipping o f small and medium sized synchronism following an autoreclosure on a double circuit line5. The
embedded generation is being viewed with concern. Research has smaller of the power/load angle curves (points 0 to 3) represents the
therefore been directed at investigating new techniques for detecting machine output when only one of the two feeder circuits is in
pole slipping and the formulation of protection algorithms which can operation. For synchronism to be lost, the acceleration energy
be included into an integrated protection scheme for these generators. resuiting from a disturbance, represented by the area 1-2-3-3’, must
This paper describes a method for detecting pole slipping using be larger than the retardation energy, represented by the area 3’-4-5,
power measurements taken at the generator's terminals. This is shown and hence there is a net acceleration o f the rotor. If the rotor is still
to correctly detect pole slipping using both computer simulation and advancing when the load angle reaches point 5, then synchronism
a laboratory model power system while remaining stable for will be lost, and the rotor will continue to accelerate until one pair
excessive swing conditions and during power system faults. of poles has slipped and the generator has travelled through the
motoring region back into a generating condition. If the accelerating
1.1. INTRODUCTION. forces are not absorbed, a further slip may occur. Figure 1 shows the
slip increasing as synchronism is lost, as well as the asynchronous
Small and medium sized embedded generation sets are increasingly power output o f the machine increasing due to induction generator
being connected to the utility power supply system. Transient action. If the governor reduces the input power to the generator, Pm,
stability problems, including those resulting from pole slipping, are it is possible for stable asynchronous operation to occur or even
being recognized as a cause for concern. The high clearance times synchronous operation to be restored.
associated with IDMT relays used to protect the system and the
generator against short circuit faults together with the generator’s low
inertia and high resistance accentuate the probability of pole slipping.
POW ER MECHANICAL
Pole slipping is the process by which an imbalance between the POWER INPUT.
mechanical input power and the electrical output power of a
synchronous machine causes it to slip with respect to its primary
magnetic flux which is synchronised to the electrical power system.
The forces this produces within the machine can damage it in several
ways. The high currents experienced can loosen stator windings,
since these are generally only braced for three phase fault levels. If
prolonged pole slipping is permitted, injurious heating of the stator ASYNCHRONOUS
POWER OUTPUT.
and rotor will result The high shaft stresses experienced can also
endanger the shaft, if it does not have a sufficiently high safety
margin to withstand the overload1 and pulsating torques. From the
power system viewpoint if the generator is connected to a low
capacity utility infeed, the resultant voltage fluctuations produced can LOAD
cause considerable trouble to other consumers, and introduce risks of ANGLE
protection mal-operation. For these reasons it is therefore
recommended that for a pole slipping condition, the generator is
disconnected from the utility with no intentional delay and at least
within the first slip cycle2. PERIOD OF
SYNCHRONOUS SYNCHRONOUS
Investigations have therefore been undertaken into a new pole
OSCILLATIONS POWER OUTPUT
slipping protection algorithm which is suitable for a small or medium
sized synchronous embedded generator and which can be included
in an integrated protection scheme for these units. This project is an FIGURE 1. Equal Area Diagram for Generator Loss of
extension of other work into aspects of this protection package3,4. Synchronism.

2.1. POLE SLIPPING. The above scenario illustrates that if a generator has sufficient load
removed for a significant period of time, then pole slipping will
Pole slipping o f a synchronous machine occurs when there is occur, since the excess input power from the prime mover accelerates
insufficient electromagnetic torque to hold the rotor in synchronism the rotor. The most severe disturbance for a generator is the close-up
with the stator magnetic flux compared to the mechanical input three phase fault, for which all of the generator's load is lost and
torque. This can arise due to an external disturbance on the system, therefore all of the prime mover power is used to accelerate the rotor.
a faulty excitation system, or an excess of prime mover input power. For less severe short circuit faults, a greater fault duration is required
to incur pole slipping.
The Equal Area Criterion is often used for stability assessment, and
demonstrates the conditions where synchronism is lost and pole Pole slipping can also occur if the generator is under excited, since
slipping occurs. Figure 1 shows the power/load angle relationship this reduces the maximum possible synchronous electrical power
and the rate of change of load angle (slip, s) for a generator losing which can be taken out of the generator. If this maximum is less than

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 330

 the prime mover input power, synchronism will not be maintained. electric power output o f the machine, P, is equal to the mechanical
shaft input power, Pm.
If excitation is lost completely then the machine runs asynchronously
rather than pole slipping6. In this condition the output power It is possible however, for a negative value o f (AP/At) to occur
pulsations will be significantly reduced to the amplitudes dictated by during a stable swing. This would be the condition when the load
the machine's electric and magnetic characteristics. angle is swinging back from point 4 towards point 1 as shown in
figure 1. To differentiate between negative values o f (AP/At) at the
3.1. CONVENTIONAL METHODS FOR DETECTING POLE two points, reactive power is used. The steady state reactive
SLIPPING. power/load angle relationship for a round rotor machine is given by;

The most widely used technique for detecting pole slipping is to

monitor the path o f the apparent impedance o f the machine and its ~ E*V. V2
rate o f change with distance type relays. As a generator falls out of Q = (cos 5) - —
step with the power system, the impedance falls to a minimum and
then increases as the generator completes one slip cycle.

The basic scheme2 uses two load blinder characteristics set either This shows that for load angles between 90s and 270°, the reactive
side o f and parallel to the system impedance. When pole slipping power, Q, will always be less than the load angle independent term.
occurs, the apparent impedance travels across the impedance plane, This value, Q^p, is used to identify the stable negative derivative
(R+jX), and a trip condition exists if the impedance crosses the two terms from the unstable derivative terms. The above equation is
load blinder units in succession. If just one load blinder unit is altered when transient conditions are taken into account, however this
tripped, as occurs for an internal or a close-up external fault, no pole effect gives no detrimental effect in terms o f algorithm stability.
slipping trip will occur. Refinements of this scheme use a MHO
characteristic to prevent mal-operation due to disturbances else where In practice, a margin for error is allowed for in the rate expression
or lenticular characteristics in place o f the load blinders. (AP/At). A minimum value (A P/A t)^ is designated, based on a
minimum value o f slip and an assumed sinusoidal power/load angle
More sophisticated schemes7 have been developed to predict the relationship. It can be seen that once the power output is less than
probability o f pole slipping and to provide inputs into the generator's the input power, then for load angles greater than the peak o f the
and prime mover's control system. power/load angle curve, the trip criteria are defined by;

4.1. THE NEW APPROACH TO POLE SLIPPING

PROTECTION.

Pole slipping is a power based phenomena and hence algorithms

have been examined which are based on the Equal Area Criterion. In AND
the design o f the protection algorithm, the relay should trip when it
observes the conditions which will produce a pole slip or correspond
to an actual pole slip, but remain stable during fault conditions and
tr ip
other system abnormalities. By being able to recognise the conditions
were a pole slip is inevitable, the algorithm offers the prospect of
tripping before the actual pole slip and hence disconnecting the AND
machine from the system before it can be damaged.

The instantaneous values o f real and reactive power8 used for the trip
criterion were calculated using;- (— ),„„T £ (— ) s ( — ) .
At At At
P = va* i a vb*h vC
* lrC where the value P( is given by the expected output o f the generator.

Embedded generators are generally operated at or close to their full

rated load since they are used primarily to provide base load or peak
lopping. Adaptive relaying would have to be considered if the
Q =— (va*(ic-ib>+vb*(ia-iJ+Vc*(h-iaV generator operates with a range o f output levels.
\/3
Faults cause considerable technical difficulties since they cause
dramatic changes in the relay measurands. These transitions produce
where va, ia, vb, ib, vc, and ic represent the sampled values of the negative values o f (AP/At) which are a potential source o f instability
terminal voltage and current. The sign convention used defines that to the algorithm. Fortunately, the faults generally introduce sinusoidal
for an under-excited generator, the reactive power is negative. terms of twice the power system frequency into the power
measurements. Since pole slipping is a relatively slow process in
The Equal Area diagram, figure 1, shows that for a stable swing, the comparison to faults, then an imposed minimum tripping time o f one
load angle cannot exceed the critical swing point, shown as point 5. power system cycle will inhibit most fault tripping. In addition, the
For the generator to recover from the disturbance, the rate of change rapid transitions experienced with faults produce negative values o f
o f load angle (slip) and hence the rate o f change of electrical power (AP/At) far in excess o f that possible due to pole slipping. This
output, (AP/At), must be zero at point 5. For an unstable generator, provides an additional mechanism to inhibit tripping during power
the load angle will continue to advance after point 5, and therefore system faults, whenever (AP/At) is less than (AP/At)^.
(AP/At) will be negative at point 5. The measurement of (AP/At)
therefore provides the first criterion for the pole slipping protection. The relaying measurands were filtered to remove excessive
The critical swing point can be identified as the point at which the harmonics and unbalance. A half cycle moving average filter was

MJ.Checksfield/PhD/University of Bath 1996 Appendix D, Page 331

 used for the real and reactive powers, followed by a lull cycle than 90°. Once past the maximum value, the real power falls rapidly
moving average for the derived rate of change o f power signal. as the now unstable machine rotor accelerates into pole slipping.

5.1. SIMULATION STUDIES. Three pole slip cycles are shown in the diagram, o f which the second
and third cycles have a considerable amount o f distortion on them.
The performance o f the protection algorithm was demonstrated using This was caused by the measured currents exceeding the maximum
a computer based simulator9 and a laboratory power system model3. permitted for the microcomputer system's input circuits and by non-
linearities in the power system model. As with the computer
The computer simulator modelled the behaviour o f the machines, simulation, the algorithm tripped just before the machine actually
control systems and transmission network o f an interconnected power slipped and before the machine would be damaged. Similar results
system. To investigate the performance o f the pole slipping were obtained for the insufficient excitation pole slip.
protection algorithms, a simplified system was modelled which
consisted o f a grid network represented by a generator o f 58.8 GVA Of all fault types investigated, two phase to earth and three phase to
connected by two parallel transmission lines to the protected earth faults presented the greatest threat to algorithm security. For
generator o f 588 M V A with an associated generator transformer. This these faults, both o f the fault blocking mechanisms, the one cycle
model was similar to the laboratory model, since the ratio of utility minimum operating time and the maximum negative value o f (AP/At)
to machine ratings was 100:1, although ideally a much smaller were employed. The maximum negative value o f (AP/At), i.e.
generator should have been used. (AP/At)m.„, was calculated to be 75 kW/s for the 5 kVA machine.

Pole slipping was triggered by applying a 3 phase fault on the The results for a two phase to earth fault are shown in figure 5. The
generator transformer busbar with a fault duration just greater than dangerous condition arises following removal o f the fault which
that which would produce the critical conditions for stability. This occurred at about 1.0 second in the record. At this time, the
technique was also used to trigger a power swing but for this the derivative, (AP/At), was not negative for the one cycle required for
fault duration was chosen to be just less than the critical clearance tripping, and was also less than (AP/AtL..v indicating a fault
time required for pole slipping. condition. Tripping was therefore blocked for this fault condition.

The laboratory model3 embedded generation system consisted o f two 7.1. CONCLUSIONS.
SkVA generators connected to a double bus system, 'local' load and
a 200V laboratory three-phase 'utility' supply. The 5kVA generators With the growing number o f small and medium sized embedded
were two pole, round rotor machines, driven by dc motors. generators, pole slipping protection is becoming more important
Machine winding damage, system problems, and shaft failure can all
Pole slipping o f the laboratory machine was induced by two result if adequate protection is not provided. A power based
methods. First by increasing the power into the dc machine, thus algorithm, suitable for inclusion into an integrated protection scheme
applying excess input torque, and second by reducing the generator’s has therefore been developed.
excitation to a level insufficient to maintain synchronism. Ideally
pole slipping should have been induced by a long duration fault, but Initial tests have shown that this algorithm can successfully detect
due to laboratory supply limitations this could not be done. The pole slipping, whilst maintaining stability against faults and power
algorithm's security against faults was tested by applying a variety of swings. This approach has the advantage that it can detect when the
faults to the local load busbar. machine has passed the point o f no return and can therefore trip the
breaker before the machine actually slips.
6.1. TEST RESULTS.
8.1. REFERENCES.
The response to pole slipping on the computer simulation is shown
in figure 2. This shows pole slipping produced by a fault whose 1. MASRUR M A, AYOUB A K and TIELKING J T. 'Studies on
duration was 1 mS greater than the critical fault clearing time. The Asynchronous Operation o f Synchronous Machines and Related Shaft
AVR and governor were both set to manual for this test The Torsional Stresses.' IEE Proc. Part C, Vol 138, Jan 1991, pp 47-56.
algorithm detected the pole slip condition at a load angle o f 167°, i.e. 2. IEEE Power System Relaying Committee. 'Out o f Step Relaying
just before the machine actually slipped. Although the actual tripping for Generators Working Group Report,' IEEE Trans PAS, Vol 96,
time can be seen, the speed o f pole slipping is dependent upon the No.5, Sept 1977, pp 1556-1564.
degree o f unbalance and will inherently vary from case to case. 3. REDFERN M A, BARRETT J, HEWINGS D, 'A Laboratory
Expressing the tripping time in terms o f load angle provides a more Facility For Research into Digital Protection Algorithms used for the
suitable measure for comparison. Protection of Small and Medium Sized Synchronous Generators.'
Proc. 27th UPEC, Vol 1, Sept 1992, ppl6-19.
The simulated power swing which produced the largest possible 4. REDFERN MA, USTA O and FIELDING G, 'Protection against
stable swing o f 152° is shown in figure 3. For the period from 0.32 Loss of Utility Grid Supply for a Dispersed Storage and Generation
to 0.92 seconds, the reactive power is greater than Qtnp> since the Unit' IEEE Paper 92 SM 376-4 PWRD presented to the PES
load angle is less than 90°. However, during this time the real power Summer Meeting, Seattle, 1992.
is greater than Pt, which inhibits tripping. This indicates that the rotor 5. VENIKOV V A. 'Transient Phenomena in Electrical Power
is still retarding and the generator could remain stable. For the period Systems', Permagon Press, 1964.
when both P and (AP/At) satisfy their trip criterion, Q is greater than 6. CEGB. 'Modem Power Station Practice', Vol 4, 'Generator &
Qtrip restraining operation. Electrical Plant', Permagon Press, 1963.
7. SHIWEN S. 'Microcomputer Based Out-of-Step Protection for
The response to a pole slip o f the laboratory model is shown in Large Generator1, IEE APSCOM Hong Kong, Nov 1991, IEE Proc.
figure 4. This slip was invoked using excess input shaft torque. The No 348, Nov 1991, pp 839-842.
plots show the real output power increasing and the reactive power 8. BONWICK W J and HESSION P J. 'Fast Measurement of Real
decreasing as the rotor angle begins to advance. Note that the and Reactive Power in Three Phase Circuits.' IEE Proc A, Vol 139,
maximum output power occurs just after the reactive power has No 2, March 1992, pp 51-55.
fallen below Q^p, as expected. Transient machine theory predicts that 9. DUNN R, BERRY T, CHAN K, NG F. 'Real Time Power System
the peak power output would be produced at a load angle just greater Training Simulator.' Proc.25th UPEC, 1990, pp 671-674.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 332

 Real pow« (MW) Real R o w (W) a) Real Power Output
) Real Power Output upoo
12J>00
10poo
•poo ^ —
r / /..V a
op00
4 POO
2 POO pt
V
V•W .........
n
-2 poo
•J OA 04 CM IP IJ 14 u
Ttm e(s)

Reactive power (MVAr) Reactive Power (VAr) b) Reactive Power Output

b) Reactive Power Output
4poo
2poo

Tune(i)
c) RxU of Change o f R ed Power
(Expanded Tune Scale) c) Rate o f Change o f Real Power
D erivative (M W /s) Derivative (W/s)
2.000 .........^ P e a k - 60,000 300POO
j 200POO
loopoo
-2.&V,
— — ---------------- -^ 7
-4.000 •toopoo
4 000
•1 #4
f T“ p -200 POO
-300 POO H
-400 POO
0-2 OA OA OA 14 1J IA Id

T im e(f)
d) Machine LoadAngle
Load tog k (degrees) Derivative (W/s) d) R ite o f Change o f Real Power
lorxx (expanded time scale)

------- ^
■ 1 ^
^Trip
02 04 •« «« i
TVnc(s)
Tune(s)
FIGURE 2 - Algorithm Quantities During Pole Figure 4 - Algorithm quantities during pole
Slipping (Simulated). slipping o f the 5 kVA machine.

Real power (MW) a) Real Power Outpi* a) Real Power Output

No Trip
No Tnp

Reactive power (MVAr) Reactive Power (VAr)

b) Reactive Power Output b) Reactive Power Output

Q tnp No Trip
No Tnp

c) Rate of Change of
Real Power
Derivwivc (MW/s) (Expanded Tune Scale) c) Rate of Change o f Real Power
Oenvwive (W/s)
2.000
___ ________________ ^
-------------------- (Af> j
UN min

4.000 (AP/At)™. No Trip

No Trip

d) Rate o f Change o f Real Power

(Expanded Time Scale)
Load angle (degrees) Dcnv«jve (W/s) peak = 150.000
d) Machine Load Angle

(AP/A t)
No Trip

T m e (s) No Tnp

FIGURE 3 - Algorithm Stability Check Against a Power Swing FIGURE 5 - Algorithm Stabilty Check for a
(Simulated) Two Phase to Earth Fault.

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 333

 PAPER TWO

"A NEW POLE SLIPPING PROTECTION ALGORITHM FOR

DISPERSED STORAGE AND GENERATION USING THE
EQUAL AREA CRITERION”

M A Redfem and M J Checksfield

Presented at :-
IEEE Power Engineering Society (PES) Summer Meeting,
July 1994, San Francisco,
Paper No. 94 SM 425-9 PWRD.

Published in :-
IEEE Transactions on Power Delievery,
Volume 10, Number 1, January 1995, pp 194 -202.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 334

 A NEW POLE SLIPPING PROTECTION ALGORITHM
FOR DISPERSED STORAGE AND GENERATION
USING THE EQUAL AREA CRITERION.

M. A. Redfem and M. J. Checksfield

School of Electronic and Electrical Engineering

University of Bath
Bath, BA2 7AY, UK.

A b str a c t - T h e p ro b a b ility o f p ole slip p in g o f D isp ersed S torage Since the protection functions are generally independent, there
a n d G en era tio n (D S G ) sy n ch ro n o u s m ach in es is b ein g view ed w ith is little difficulty in using a common hardware system, thus
c o n cern , sin c e th is ca n d a m age th e m a ch in e as w ell as p rom ote providing the economy required to meet modem requirements.
p o w er sy stem in sta b ility . T h is is p a rticu la rly th e ca se for sm all In addition to the required protection functions, disturbance
an d m ed iu m siz ed m a ch in es b ecau se th eir low in ertia s and high
and event recording, continuous self monitoring, ease of
resista n ces m a k e them in h eren tly m ore u n stab le. R esearch h as
testing and SCADA interface are all possible features of this
th e refo re b een d irected in to n ew tech n iq u es for d etectin g pole
slip p in g an d th e fo rm u la tio n o f p rotection algorith m s w h ich can
approach.
be in clu d ed in to an in te g ra ted m icro co m p u ter b ased p rotection
sc h e m e fo r th ese g en era to rs. T h is p ap er d escrib es a m eth od for The algorithm described in this paper takes advantage of the
d e te c tin g p o le slip p in g u sin g p ow er m ea su rem en ts taken at the processing capabilities of modem micro processor relaying
g en era to r's term in a ls. T h is tech n iq u e h a s b een sh ow n to correctly platforms and has been designed to share sub-functions with
d etect p o le slip p in g u sin g b oth co m p u ter sim u la tio n and a other protection functions required for the protection of DSG
la b o ra to ry m o d el p o w er sy stem . T h e a lgorith m rem ain s sta b le for synchronous generation! 12,13,14]. The algorithm uses the Equal
reco v e ra b le sw in g co n d itio n s and p ow er system faults. Area Criterion to assess the stability of the machine and
determine when it is committed to a pole slip. This project is
Keywords : - Power system protection, Dispersed storage and an extension of other work into aspects of this protection
generation units, Pole slipping protection, Equal area criterion, package[15,l6].
Machine stability, Digital protection.
Pole slipping is the process by which an imbalance between
the mechanical input power and the electrical output power of
I. INTRODUCTION. a synchronous machine causes its rotor to accelerate, and
hence the rotor magnetic flux to slip with respect to the stator
Pole slipping o f any size of synchronous machine will always flux which is synchronised to the electrical power system. The
cause problems. The current trend of connecting an increasing forces this produces within the machine can damage it in
number of small and medium sized synchronous generators to several ways. If the machine slips fast enough so that it
operate in parallel with the utility power supply puts such operates past the peak of it's power/slip curve, the high
machines in a potentially damaging situation. currents experienced can loosen stator windings. These are
generally only braced for three phase fault levels. Also, if
Generators on distribution systems using the widely adopted prolonged pole slipping is permitted, injurious heating will
IDMT overcurrent relays can face fault clearance times of 0.5 result, the stator end teeth suffering the maximum temperature
to 1 second or more[l,2,3] which can cause generator rise[4J. Induced slip frequency currents will flow in the rotor
instability. Future generator protection will be provided by body, damping circuits and excitation windings. Their
microprocessor based integrated relaying schemes. magnitude will be high since they must compensate the high
stator mmf [5]. The rotor is therefore also at risk from
overheating, pole slipping tests performed for one minute on
a 60 MW turbo-generator[6] resulted in blueing of the rotor
end rings, indicating that operation had come close to
damaging the machine. If the generator uses an ac/dc
excitation scheme, the induced voltage that results when the
excitation rectifiers block the induced reverse rotor current can
stress insulation and cause eventual breakdown if the exciter
safety margins are not suitably specified[4].

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 335

The severe pulsating torques produced by pole slipping can
torsionally excite sections o f the shaft, exposing them to
oscillatory stress[7], Studies have found[8] that maximum
MECHANICAL
torques and torsional stresses occur during the initial period o f P O W ER
POWER INPUT.
torsional oscillation after each torsional impulse, and that shaft
damage generally occurs over this period. Additionally, it was
found that the cumulative effects o f smaller oscillations might
in the long run lead to significant shaft damage also. If the
shaft material is not sufficiently overdesigned, then its fatigue
life can be used up after relatively few pole slipping events.
ASYNCHRONOUS
POWIS OUTPUT. .
From the power system viewpoint, if the generator is
connected to a low capacity utility infeed, the resultant voltage
fluctuations produced can cause considerable trouble to other 360
►
consumers, such as induction motor stalls and pole slipping o f LOAD
synchronous motors. Risks o f protection mal-operation also ANGLE

result with overcurrent, undervoltage and distance relays being

particularly at risk[9]. ►
PERKS) OF
For these reasons it is therefore recommended that for a pole SYNCHRONOUS
SYNCHRONOUS
slipping condition, the generator is disconnected from the OSCILLATIONS pow er o irn v r
utility with no intentional delay [10] and at least within the first
slip cyclefll].

Fig. 1. Equal area diagram for generator loss of synchronism

II. POLE SLIPPING

Pole slipping o f a synchronous machine occurs when there is For synchronism to be lost, the acceleration energy resulting
insufficient electromagnetic torque to hold the rotor in from the loss o f one line, represented by the area 1-2-3-3',
synchronism with the stator magnetic flux. The loss o f must be larger than the retardation energy, represented by the
synchronism point can be defined as the point where the area 3'-4-5, causing net acceleration o f the rotor above
generator's real power output becomes zero. At this point, synchronous speed. If the rotor is still above synchronous
almost no flux links the stator and field windings, showing speed when the load angle reaches point 5, then synchronism
that the magnetic coupling between the two is at its weakest will be lost, and the rotor w ill continue to accelerate until one
state[17], pair o f poles has slipped. The generator has travelled through
the motoring region during this loss o f synchronism before it
Pole slipping can arise due to an external disturbance on the returns to a generating condition. If the accelerating forces are
system such as an insufficiently cleared fault, a faulty not absorbed, a further slip may occur. Fig. 1 shows the slip
excitation system, or an excess o f prime mover input power increasing as synchronism is lost, as w ell as the asynchronous
caused by a sudden and large loss o f load. power output o f the machine increasing due to induction
generator action. If the governor reduces the input power to
The Equal Area Criterion is often used for stability assessment, the generator, Pm, it is possible for stable asynchronous
and demonstrates the conditions where synchronism is lost and operation to occur or synchronous operation to be restored.
pole slipping occurs. Fig.l shows the power/load angle
relationship and the rate o f change o f load angle (slip, s) for The above scenario illustrates that if a generator has sufficient
a generator losing synchronism following a loss o f power load removed for a significant period o f time, then pole
transfer capability due to a disturbance on a double circuit slipping will occur, since the excess input power from the
line[18]. For this scenario, the generator is connected to a prime mover accelerates the rotor. The most severe
double circuit line and the disturbance is caused by switching disturbance for a generator is the close-up three phase fault,
one o f the lines out o f service temporarily. for which all o f the generator's load is lost and therefore all o f
the prime mover power is used to accelerate the rotor. For
The smaller o f the power/load angle curves, the broken line, less severe short circuit faults, a greater fault duration is
(points 0, 2, 3) represents the machine output when only one required to incur pole slipping.
o f the two feeder circuits is in operation. The larger o f the
power/load angle curves, the solid curve, represents the Pole slipping can also occur if the generator is under excited,
transfer capacity when both lines are in service. since this reduces the maximum possible synchronous

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 336

electrical power which can be taken out o f the generator. If This scheme has the disadvantage that it can mal-operate for
this maximum is less than the prime mover input power, recoverable swings. Generally the relay is set to a sensitivity
synchronism can not be maintained. If excitation is lost o f 120° load angle, since return to stable operation is usually
com pletely then the machine runs asynchronously rather than unattainable once this angle has been exceeded. The relay
pole slipping[6], this is a loss o f field condition. In this does provide fast back up protection for multiphase faults
condition the output power pulsations w ill be significantly close to the generator.
reduced to the amplitudes dictated by the machine's electric
and magnetic characteristics. A more secure distance type schem e[ll] is shown in Fig. 3
This uses two load blinder characteristics set either side o f
III. CONVENTIONAL METHODS FOR DETECTING and parallel to the system impedance, and a supervisory mho
POLE SLIPPING. relay. When pole slipping occurs, the apparent impedance
travels across the impedance plane, and a trip condition exists
The most widely used technique for detecting pole slipping is if the impedance enters the mho supervisory relay and then
to monitor the path o f the apparent impedance o f the machine crosses the two load blinder units in succession. If just one
and its rate o f change with distance type relays. Fig.2 shows load blinder unit is tripped, as occurs for an internal or a
the impedance characteristic for machine 'A' pole slipping due close-up external fault, the relay w ill not trip. The locus must
to a loss o f load. enter from one side and leave from the other for tripping to
occur.
The impedance is not limited to just one trajectory, but is free
to drift across the R/X plane according to variations in Tripping is sometimes set to occur follow ing reset o f the mho
generator impedances and generator internal voltages. The element, enabling control o f the angle at which tripping occurs
point where the locus crosses the system impedance line AB for more favourable circuit breaker arc interruption. This is
corresponds to a load angle o f 180°, i.e the point where loss achieved by adjusting the diameter o f the mho element,
o f synchronism occurs. however, careful study o f the mho setting must be made to
prevent tripping for recoverable swings that cross both blinder
The simplest scheme uses mho distance relay characteristics. elements.
The relay looks into the HV terminals o f the generator step up
transformer, a trip signal being given if the impedance enters Double blinder schemes with supervisory mho elements are
the circle. also available, which allow the impedance to enter and leave
on the same side. This makes for faster tripping than the
single element scheme, but can mal-operate for recoverable
swings unless very careful selection o f the relay settings is
made.
Z. z.W'—
-AA/V infinite
bos
if generator A
forward pole slips,
then impedance
locus travels in Circle oentre lies
direction shown by on the extension
arrows of line AB

Load A ngie - d
Syatcm Impedance - EF
Trmnrfonner Im podnnr, - DE
O en ea to t Impedance - AB

BUnder

Fig. 2. Pole slipping impedance loci. Fig. 3. Single blinder pole slipping protection.

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 337

More sophisticated schemes have been developed using lens requires the circuit breaker to have adequate arc interruption
characteristics^] rather than circular mho or ohm capabilities.
characteristics. This type have similar operating characteristics
to the single blinder scheme. The instantaneous values o f real and reactive power[21] used
for the trip criterion were calculated using;-
Microcomputer based schemes have been developed which
sectionalise the R/X plane into multiple zones, making the P = + vb*ib + vc+ic (1)
scheme inherently more stable. Fig. 4 shows one such
example described by Shiwen[20]. This scheme issues a trip
signal after the locus has travelled through zone 0 to zone 4,
staying in each zone for a time greater than that o f the Q = - j= ( v * + V ( * a -*'c) + V O W o ) ) (2 )
impedance characteristic associated with a short circuit fault, v/3
but less than the minimum time it would stay in any zone for
the out o f step condition. Again tripping does not occur until where v , , is , vb , ib , vc , and ic represent the sampled values
pole slipping has actually occurred, but the scheme does o f the terminal voltage and current. The sign convention used
provide preventative control signals for fast valving or defines that for an under-excited generator, the reactive power
dynamic breaking if the locus travels through the first two is negative.
zones in the correct manner.
The Equal Area Diagram, Fig. 1, illustrates that for a stable
Restrict* tripping to swing, the machine operating point cannot exceed point 5. If
swing* within g q s a l o r
point 5 can be identified in terms o f machine terminal
quantities, then machine stability can be determined. Point 5
corresponds to the electrical output o f the machine, P,
equalling the mechanical input from the prime mover, Pm.
Instability occurs if the machine operates past point 5, and this
can therefore be identified in terms o f real power as the
Z4 Z3 condition when
Zone 4 Zone 2 Zone 1 Zone 0

Far speed up pole P < Pm (3 )

d ip t, Z 1 is crow d
first

For riow down

pole dips, Z4 1* The above criterion can also be satisfied for machine operating
points between points 0 and 1 shown on Fig. 1. To
differentiate between the two, reactive power measurement is
used. The steady state reactive power/load angle relationship
Fig. 4. Impedance microcomputer based pole slipping protection. for a round rotor machine is given by

0 , W (cos#) - I f (4)
IV. THE NEW APPROACH TO POLE SLIPPING
PROTECTION.
This shows that for load angles between 90° and 270°, the
Pole slipping is a power based phenomena and hence power reactive power w ill always be less than the load angle (8)
algorithms have been examined which are based on the Equal independent term. Since point 5 o f Fig. 1 occurs for load
Area Criterion. In the design o f the protection algorithm, the angles greater than 90°, then if
relay should trip when it observes the conditions which will
produce a pole slip or correspond to an actual pole slip, but
remain stable during fault conditions and other system
( - <w <5>
abnormalities. By being able to recognise the conditions where
a pole slip is inevitable, the algorithm offers the prospect o f the machine must be operating at point 5 and not point 1, and
tripping before the actual pole slip and hence disconnecting the this forms the second criterion. The transient reactive power
machine from the system before it can be damaged. This characteristics vary in magnitude from the above steady state

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 338

values. This does not corrupt the above method since the an additional mechanism to inhibit tripping during power
value o f reactive power corresponding to a 90° load angle is system faults, whenever (AP/At) is less than (AP/At)m„, where
identical to the steady state value. The transient reactive (AP/At)mtx is the maximum negative value expected from pole
pow er characteristics above and below this load angle are such slipping, based on a maximum slip value and a sinusoidal
that in steady state terms it appears as if the load angle is power load angle relationship.
correspondingly higher or lower.
The relaying measurands are filtered to remove harmonics and
Finally, it must be ascertained whether or not the load angle unbalance present on the power system. A half cycle moving
is still advancing. To determine this, rate o f change o f real average filter is used for the real and reactive power
pow er is used. It can be seen from the synchronous power measurements, followed by a full cycle moving average for the
output characteristic o f Fig. 1, that if the load angle is still derived rate o f change o f power signal.
increasing when point 5 is reached, the machine output power
w ill be decreasing, i.e. the rate o f change o f real power will be
negative. This, therefore is the third criterion for determining V. SIMULATION STUDIES.
machine stability and hence predicting pole slipping.
The performance of the protection algorithm was investigated
In practice a margin for error is allowed for in this rate using a computer based simulator[24] and a laboratory power
expression (AP/At), to ensure the algorithm does not mal- system model[i5]. The computer based simulator was
operate. A minimum negative value, (AP/At)min is designated, particularly useful in that the results it produces are of an
based on a minimum value o f slip and an assumed sinusoidal 'ideal' nature and demonstrate the fundamental characteristics
power/load angle relationship. The conditions which of pole slipping. The laboratory power system model
determine that a generator is about to pole slip are therefore provides a test bed for identifying problems with the algorithm
not immediately evident from theory.

A. Computer Based Simulation Studies.

Q* <6>
The power system simulation software was used to investigate
( A P / A / ) * ( A P /A O ,^ the algorithm's performance with pole slipping and power
swings. It lends itself especially to power swing analysis since
Dispersed storage and generation units are generally operated the worst possible swing can be simulated. The computer
at, or close to their full rated load[22] to provide the greatest simulator models the behaviour o f the machines, control
return on investment. A fixed value o f Pm, designated as the systems and transmission network o f an interconnected power
trip setting Pt , can therefore be adopted for the algorithm real system. It uses a partitioned approach to solve the sets of
power setting which corresponds to the expected generator equations, which describe the behaviour o f the machines,
output. control systems and transmission network. The differential
equations for each machine group are solved independently to
If the synchronous machine is expected to operate over its full calculate the current injection into the transmission network.
working range, an adaptive relaying technique would be used The algebraic network equations are then solved and the whole
which utilises the mechanical power input. A technique calculation repeated until the required accuracy is obtained.
similar to that used in [23], which uses an acceleration The machine model used is based on Park's transformation, the
disturbance detector could easily be implemented. damper winding and eddy current effects being represented by
a short circuited winding on both the direct and quadrature
Faults cause considerable technical difficulties since they cause axis.
dramatic changes in the relay measurands. These transitions
produce negative values of (AP/At) which are a potential The classical single machine and infinite bus system was
source of instability to the algorithm. Fortunately, the faults modelled, which consisted o f a grid network represented by a
generally introduce sinusoidal terms of twice the power system generator o f 58.8 GVA connected by two parallel transmission
frequency into the pow er measurements. Since pole slipping is lines to the protected generator of 588 MVA with an
a relatively slow process in comparison to faults, then an associated generator transformer. This model is similar to the
imposed minimum tripping time o f one power system cycle laboratory power system model used, in that the ratio o f utility
will inhibit fault tripping. In addition, the rapid transitions to machine ratings is 100:1. This computer model has
experienced with faults produce negative values of (AP/At) far characteristics much more akin to those o f a grid system, since
in excess o f that possible due to pole slipping. This provides the machine is large, making the assumptions used in deriving

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 339

the Equal Area Criterion correct. For smaller machines, symmetrical function:
effects such as stator resistance have to be taken into account.
These effects do not however affect the functioning o f the
algorithm since it is the principle o f the Equal Area Criterion P =P sin 6 (8)
on which the algorithm is based and not absolute values.

Pole slipping was triggered by applying a 3 phase fault on the Where Pmax is the maximum output o f the machine at full load
generator transformer busbar with a fault duration just greater excitation, then the load angle at which P = Pm for load
than the critical clearance time (CCT). This is the maximum angles greater than 90° (unstable operation) is given by :
fault duration allowable before instability occurs. This
technique was also used to trigger a power swing but for this 6 = 180 - sin'1 (—=-) (9)
the fault duration was chosen to be just less than the critical 9 P
clearance time.

B. Laboratory Power System Model. The derivative o f real power w ith respect to time, can be
expressed as :
The laboratory model DSG system[l5] consisted of two 5kVA
generators connected to a double bus system, 'local' load and dP = dP db (10)
a 200V laboratory three-phase 'utility' supply. The 5kVA dt db dt
generators were two pole, round rotor machines, driven by dc
motors. Although the transformer supplying the laboratory is The derivative o f load angle w ith respect to time is the slip, s
rated at 500 kVA, giving a utility to machine rating of 100:1, and is defined as :
the cable connecting the two has a relatively large impedance,
making the system less 'stiff. This makes the laboratory db
= -s (11)
simulation more realistic for DSG machines, since in practice dt
these may be located at the end o f long distribution feeders.
i.e. negative values o f slip exist for speeds above synchronous
Pole slipping o f the laboratory machine was induced by two speed. If (8) is differentiated w ith respect to load angle then:
methods. In the first, a three phase fault was placed on the
DSG utility intertie o f sufficient duration to cause instability. dP (12)
=P cos b
The second method involved reducing the generator's db
excitation slowly until synchronism was lost.

The algorithm's security against faults was tested by applying Hence the derivative o f real pow er at the point where P = Pm
phase to earth, phase to phase, twin phase to earth and three can be expressed as :
phase faults to the local load busbar. Full three phase fault
current was not permitted and the fault current magnitude was
7 = - s f . COS J , = ( A / y A / ) «nlf (13)
therefore reduced using a current limiting resistor. at

VI. ALGORITHM SETTINGS A value for the margin for error can therefore be obtained by
deciding on a minimum value o f slip permitted at the point P
The value for Pt is set by the expected steady state machine = Pm. A value of 0.5 % was chosen as providing an adequate
output power. safety margin whilst still rem aining lower than values expected
from pole slipping generators. The above procedure can also
The value for Q ^. is given by : be applied to find the value required for fault blocking
(AP/At)m„. Table I shows the algorithm settings for both of
V2 the machines tested.
(7)
Q° » ‘ - T d pM-

The value o f (AP/At)min is derived as follows, if it is assumed

that the power output o f the generator can be expressed as the

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 340

 the load angle plot w hich is at its maximum at this point).
A. Test M achine Settings. The section BC then represents travel from point 5 to point 1
on Fig. 1, i.e. the rotor is backswinging towards a stable
TABLE I operating point. As it does so, the reactive power rises above
ALGORITHM SETTINGS. Q^p, restraining algorithm operation and ensuring no mal-
operation. The relay therefore does not trip.
Machine 5 kVA 588 MVA
Rating B Laboratory Studies.

Pmax 6.5 kW 720 MW Fig. 7 shows the 5 kVA laboratory machine pole slipping
following a close up three phase fault. Inspection o f the three
Xd 2.32 p.u. 2.77 p.u. phase fault section o f the algorithm measurands confirms that
the algorithm remains stable for such a disturbance. The
Qtrip -2160 VAr -260 MVAr
critical point occurs at 1.4 seconds on the record, when the
Pt 4 kW 500 MW real and reactive criteria are satisfied but the fault is blocked
(AP/Dt)miii -2000 W/s -260 MW/s by the extremely large negative value o f AP/At.

(DP /Dt)m»x -75 kW/s -9500 MW/s After the fault has cleared, it can be seen that the reactive
power is less than Q ^ , i.e. the load angle has increased above
90° during the fault period. A fter this, real and reactive power
gradually decrease as the generator slowly goes out o f step.
Tripping occurs when AP/At falls below the safety margin. It
VII. TEST RESULTS. is interesting to note, that during the fault period, a significant
amount of real power came from the generator. This was due
A. Simulation Studies. to the high fault resistance present which was required to limit
the fault current. Even with only 1 kW o f load removed, the
The response to pole slipping on the computer simulation is machine still lost synchronism. The algorithm successfully
shown in Fig. 5. The algorithm detected the pole slip condition operated once the machine was committed to a pole slip.
at a load angle o f 167°, i.e. just before the machine actually
slipped. It can be seen that the reactive power criterion was Of all fault types investigated, the two phase to earth and three
satisfied first, followed by the rate criterion, this indicates that phase to earth faults presented the greatest threat to algorithm
the load angle was in excess o f 90° and the rotor was security. For these faults, both of the fault blocking
travelling above synchronous speed. Tripping finally occurs mechanisms, the one cycle minimum operating time and the
once the real power falls below Pt, indicating that no further maximum negative value o f (AP/At) were employed.
decelerating energy was available and that instability was
unavoidable. Although the actual tripping time can be seen, The results for a two phase to earth fault are shown in Fig. 8.
the speed o f pole slipping is dependent upon the degree of The dangerous condition arises following removal o f the fault
imbalance and will inherently vary from case to case. which occurred ju st after 1.0 second in the record. At this
Expressing the tripping time in terms o f load angle provides time, the derivative, (AP/At), was not negative for the one
a more suitable measure for comparison. cycle required for tripping, and was also more negative than
(AP/At)m„ indicating a fault condition. Tripping was therefore
The simulated power swing which produced the largest blocked for this fault condition.
possible stable swing o f 152° is shown in Fig. 6. For the time
period between 0.3 and 0.8 seconds the reactive power is less VIII. CONCLUSIONS.
than Q ^ , indicating that the load angle is greater than 90°, as
the load angle plot confirms. However, during this time, real With the growing number o f small and medium sized DSG
power is greater than P„ restraining operation o f the algorithm. generators, the probability o f pole slipping is growing and
This condition indicates that decelerating energy is still therefore pole slipping protection is becoming more important.
available. The section marked AB represents the portion 1-4-5 Machine winding damage, system problems, and shaft failure
on the power/load angle curve o f Fig. 1. Since this is a stable can all result if adequate protection is not provided. A power
case, but close to the critical condition, point five is practically based algorithm, suitable for inclusion into an integrated
reached. Inspection o f the derivative curve shows that a zero protection scheme has therefore been developed using the
value o f AP/At exists at this point, showing that the rotor is Equal Area Criterion.
once again at synchronous speed (this can be ascertained from

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 341

 Load angle (degrees) a ) M a c h in e L o a d A n g le L o ad a n g le (d e g r e e s ) a ) M a c h i n e L o a d A n g le

| T rip

0.2 0.4 0.6 08 I 1.2 1.4 1.6 1.8

T im e ( s ) T im e (s )
R e a l p o w e r (M W ) R e a l p o w e r (M W ) b ) R e a l P o w e r O u tp u t
b ) R e a l P o w e r O u tp u t
IJOO
/ ^ . ___^ B ____ No
-------------- ---- T rip
Pt

1.000
0.2 0.4 0.6 0.8 I 1.2 1.4 1.6 1.8
T im e ( s ) T im e (s )

R e a c t i v e p o w e r (M V A r) R e a c tiv e p o w e r (M V A r)
c) R e a c tiv e P o w e r O u tp u t c ) R e a c tiv e P o w e r O u tp u t

-wo Q trip

T im e ( s ) T im e ( s )

d ) R a te o f C h a n g e o f R e a l P o w e r
d) R a te o f C h a n g e o f R eal P o w er
D e r i v a ti v e ( M W /s ) D e riv a tiv e (M W /s)
0 4.000

/ \ / I

0 ( d P / a t ) m jn \

0
^ T rip
0
.
0.6 08
T im e ( s ) T im e ( s )

FIGURE 5 - Algorithm Quantities FIGURE 6 - Algorithm Stability Check Against

During Pole Slipping (Simulated). a Power Swing (Simulated).

P o w e r (W ) a ) R e a l P o w e r O u tp u t
F a u lt o n F a u lt o ff L oss of 7.000
S y n c h ro n is m a) Real Fewer Output 6J000
P o in t \
5.000 No
4.000 T rip
3j000
2.000
1.000
T rip
0.6 0.8 I 12 1.4 1.6 1.1
T im e ( s )

R e a c tiv e P o w e r (V A r) b ) R e a c tiv e P o w e r O u tp u t
6£00
b) Reactive Power Output
Raactivs Power (VAr) 4,000
4.000 r 2JXX)
2.000
L /x
-2.0QCL -2.000
No
-4,000
-e.ooo
Qtrip -4,000 Q trip t T rip
-6,000
- 10,000
0 0.4 0.8 1.2 1.6 2 24
TripT 2.8 3.2 3.6
-6tfX>
02 04 0.6 0.8 1
T im e ( s )
1.2 1.4 1.6 IA

n m e(« )

c) Rate of Change of Raal Power c ) R a te o f C h a n g e o f R e a l P o w e r

D e r iv a tiv e (W /s )
Derivative (W/s)
200.000 200.000

100,000

I Ir - - - - - - - iL _ ^
r
0 ____________ .............................................
-100.000 (AP/At).
-100.000
(A P /A O jjja x I
0 0.4 0.8
-*1.2
f T rip
No
T rip
1.6 2 2.4 2.8 3.2 3.6
Time («)
T im e ( s )

D e riv a tiv e (W /s ) d ) R a te o f C h a n g e o f R e a l P o w e r
d) Rate of Change of Real Power (Expanded Time Scale)
Derivative (W/s) 5.000 (E x p a n d e d T im e S c a le )

-5.000
-5.000 (A P /A t) No
-10.000 T rip
,
1 000

3.09 3.49 102 104 I 06 I 08 II 112 1.14 I 16 I 18 1.2 1.22 1.24

Time (s) T im e ( s )

FIGURE 7 - Laboratory Machine Pole Slipping Due to FIGURE 8 - Algorithm Stabilty Check for a
Three Phase Fault. Two Phase to Earth Fault.

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 342

Tests have shown that this algorithm can successfully detect [12] J. H. Harlow, "A Multifunction Protective Relay for the Cogeneration
Industry," IEEE Computer Applications in Power, Oct 1990, pp 25-30.
pole slipping, whilst maintaining stability against faults and
[13] F. Fromm, Z. Franc, B. Kulendik and C. Steiner, "Generator Protection
power swings. This approach has the advantage that it can with the Digital MODURES 216," Brown Boveri Review 1987, Vol.74,No.l2,
detect when the machine has passed the point of no return and pp 693-700.
can therefore trip the breaker before the machine actually slips, [14] M. V. V. S. Yalla, "A Digital Multifunction Protective Relay," IEEE
preventing the pending damage. Transactions on Power Delivery, Vol.7, N o.l, Jan 1992, pp 193-200.
[15] M. A. Redfem, J. Barrett, D. Hewings, "A Laboratory Facility For
Research into Digital Protection Algorithms used for the Protection o f Small
and Medium Sized Synchronous Generators." Proc. 27* UPEC, Vol 1, Sept
IX. ACKNOWLEDGEMENTS. 1992, pp 16-19.
[16] M. A. Redfem, O. Usta and G. Fielding, "Protection against Loss of
The authors are pleased to acknowledge the help and Utility Grid Supply for a Dispersed Storage and Generation Unit" IEEE
Transactions on Power Delivery, Vol. 8, No. 3, July 1993, pp 948-954.
encouragement provided by the University of Bath, GEC
[17] J. Matsuki, T.Okada and C. Uenosono, "Loss o f Synchronism Process
ALSTHOM Protection and Control and the Science and o f a Synchronous Generator Described by its Internal Flux and Force
Engineering Research Council. Distributions," IEEE Transactions on Energy Conversion, Vol.7, N o.l, March
1992, pp 177-181.
[18] V. A. Venikov, Transient Phenomena in Electrical Power Systems,
Pergamon Press, 1964.
X. REFERENCES.
[19] A. Stalewski, J. L. H. Goody and J. A. Downes, "Pole Slipping
Protection," Developments in Power System protection, IEE Conference
[1] N. Nichols, "The Electrical Considerations in Cogeneration," IEEE
Publication, No. 185, pp 38-45.
Transactions on Industry Applications, Vol.LA-21, No.4, May/June 1985, pp
[20] S. Shiwen, "Microcomputer Based Out-of-Step Protection for Large
754-761.
Generator," IEE APSCOM Hong Kong, Nov 1991, IEE Conf. Proc. No 348,
[2] Harrison K. Clark and James W. Feltes, "Industrial and Cogeneration
Nov 1991, pp 839-842.
Protection Problems Requiring Simulation," IEEE Transactions on Industry
[21] J. W. Bonwick and P. J. Hession, "Fast Measurement o f Real and
Applications, Vol IA-25, No.4, July/August 1989, pp 766-775.
Reactive Power in Three Phase Circuits," IEE Proc A, Vol 139, No 2, March
[3] G. Fielding and G. W. Evans, "Industrial Feeder Protection," 10*
1992, pp 51-55.
International Conference on Electricity Distribution, Part 1, Subject Area 5 -
[22] R M. Rifaat, "Protection Scheme Considerations for Common Generator
Consumers Installations, IEE Conference Proceeding No. IEE 305, 1989, pp
Bus Configurations used in Cogeneration Plants," IEEE Wescanex 93 •
460-466.
Communications, Computers and Power in the Modem Environment :
[4] T.H. Mason, W.Faimey, J. J.Amold and M. J. Thelwell, "Asynchronous
Conference Proceedings 1993, Ch. 60, pp 1-6.
Operation o f Turbo-Generators," CIGRE Vol.l , 1972, Sec 11-02.
[23] William R Roemish and Edward T. Wall, "A New Synchronous
[5] C. B. Cooper, V. Easton, M. N. John, K. C. Parton, G. Shackshaft and T.
Generator Out-of-Step Relay Scheme," IEEE Transactions on Power Apparatus
M. Whitelegg, "Problems Associated with Limited Pole Slipping of Turbo-
and Systems, Vol.PAS-104, No.3, March 1985, pp 563-582.
Generators Following System Faults," CIGRE 1966, Vol.3, No.306, pp 1-16.
[24] R Dunn, T. Berry, K. Chan and F. Ng, "Real Time Power System
[6] T. H. Mason, P. D. Aylett and F. H. Birch, "Turbo -Generator
Training Simulator," Proc.25* UPEC, 1990, pp 671-674.
Performance Under Exceptional Operating Conditions," IEE Conf. Proc.
Vol. 106, Jan 1959, pp 357-373.
[7] A. Abolins, D. Lambrecht, J. S. Joyce and L. T. Rosenberg, "Effect of
Miles A Redfem (M*79) received his BSc degree from Nottingham University
Clearing Short Circuits and Automatic Reclosing on Torsional Stress and Life
and PhD degree from Cambridge University in 1970 and 1976 respectively.
Expenditure o f Turbine-Generator Shafts," IEEE Transactions on Power
Ill 1970, hi joined British Railways Research, and in 1975, moved to GEC
Apparatus and Systems, Vol.PAS-97, N o.l, Jan/Feb 1976, pp 14-25.
measurements where he held various posts including Head o f Research and
[8] M. A. Masrur, A. K. Ayoub, J. T. Tielking, "Studies on Asynchronous
Long Term Development and Overseas Sales Manager. In 1986, he joined
Operation o f Synchronous Machines and Related Shaft Torsional Stresses,"
Bath University with interests in Power Systems Protection and Management
IEE Conf. Proc, Part-C, Vol.138, Jan 1991, pp 47-56.
He is currently a member o f IEE Professional Group P ll , Power System
[9] F. liar, "Innovations in the Detection o f Power swings in Electrical
Measurement, Protection and Control, and is a corresponding member o f the
Networks," Brown Boveri Revue, Part 2, 1981, pp 87-93.
IEEE Line Protection Sub-Committee.
[10] Summary Report o f an IEEE Working Group Report, "Intertie Protection
o f Consumer-Owned Sources o f Generation, 3 MVA or Less," IEEE
Transactions on Power Delivery, Vol.5, No.2, April 1990, pp 924-929.
Matthew Checksfield received his B Eng degree from Bath University in 1993.
[11] J. A. Imhof (Chairman) et at, "Out o f Step Relaying for Generators -
He is currently studying for his PhD degree at Bath University and is
Working Group Report," IEEE Transactions on Power Apparatus and Systems,
sponsored by GEC ALSTHOM Protection and Control.
Vol.PAS-96, No.5, Sep/Oct 1977, pp 1556-1564.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 343

 PAPER THREE

"A CONDITION MONITORING POLE SLIPPING PROTECTION

ALGORITHM BASED ON THE EQUAL AREA CRITERION”

M J Checksfield and M A Redfem

Presented at
29th Universities Power Engineering Conference (UPEC),
September 1994, Galway, Ireland, pp 747 - 750.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 344

 A CONDITION MONITORING POLE SLIPPING PROTECTION
ALGORITHM BASED ON THE EQUAL AREA CRITERION.

M. J. Checksfield and M. A. Redfem.

U niversity o f B ath, UK.

A B ST R A C T The study indicated that the maximum torques and torsional stresses
This paper introduces a new condition monitoring power based occur during the first period o f oscillation caused by pole slipping,
technique for detecting pole slipping o f synchronous generators. The and it is this period where the shaft damage usually occurs. It is
particular advantage o f this approach is that it reduces the complexity therefore advantageous to disconnect the generator from the system
o f installation. As with techniques previously reported[i], the before a pole slipping event occurs in the hope that the torsional
algorithm provides a trip signal 'on the way in' to a pole slip and impulse will be prevented. It was also indicated that the cumulative
hence before the pole slip actually occurs. In doing so, som e o f the effects o f smaller disturbances taken together, due to less severe
undesirable effects produced by pole slipping can be avoided. A transients, might in the long run lead to significant shaft damage.
range o f various operating conditions are used to demonstrate the
algorithm's ability to restrain against faults and power swings and to The effect o f pole slipping on the rest o f the system can result in
operate when a pole slip is imminent plant outages and even a cascading loss o f synchronism o f machines
nearby. The voltage fluctuations produced can cause motor starter
1 IN TR O D U C T IO N . contactors to open, causing unnecessary loss o f plant I f a serious
Pole slipping relays have existed for over SO years, and nearly all voltage depression is experienced, induction motors could stall,
have been impedance based, relying on the variation in apparent depressing system voltage further. Synchronous motors m ay also
impedance at the terminals o f the protected generator. Pole slipping pole slip as a result o f the voltage depression, suffering all o f the ill
occurs when an imbalance between generator input and output effects o f generators. The lowest voltages will occur at the location
powers causes the rotor to accelerate and lose synchronism from the where the contribution from the pole slipping generator is equal to
stator magnetic flux. the contribution from the rest o f the system , referred to as the system
centre. If the system centre occurs inside the generator, the resultant
Impedance relays generally work by detecting when the apparent voltage fluctuations will not be severe. However, if the utility infeed
impedance o f the machine crosses the equivalent system impedance is o f low capacity compared to the generator rating, the voltage
line in a specified time limit(2 ]. Choosing the correct settings makes fluctuations will be centred in the system , causing disturbances to
the installation o f these types o f relays difficult^]. Research is other users.
therefore being conducted into a new type o f pole slipping algorithm
which does not require such careful analysis before installation. Power system protection also suffers from the effects o f pole
slipping, since the voltage and current variations at the electrical
2 R EA SO N S FO R USIN G POLE SLIPPING PROTECTION. centre are similar to those caused by a three phase short circuit The
Pole slipping protection is necessary to minimize the disturbance pole relays most affected are undervoltage, overcurrent and impedance
slipping can cause to other consumers connected to the power system type relays. If adequate blocking is not applied to these relays then
and to prevent damage to the generator. The extent o f the damage unnecessary system outages will occur.
is dependent on the rating o f the bus that the generator is connected
to, the speed o f the pole slip, and the overcapacity rating o f the 3 TIIE EQ UAL AREA C RITER IO N.
machine. When the utility infeed capacity is high in comparison to If the power/load angle characteristic o f a generator operating on an
the generator rating, the currents experienced during pole slipping infinite bus is known, then the Equal Area Criterion is often used to
can be greater than three phase fault levels. Since machine windings determine which conditions cause transient instability[6]. Consider
are only braced for three phase fault levels, pole slipping could the case where a generator is initially operating with a prime mover
therefore loosen and deteriorate the windings^]. input power o f Pm, and the prefauh electrical load curve as shown
in Figure 1, giving an initial load angle o f 51. If a three phase fault
The induced slip frequency currents that flow in the rotor body, occurs near to the generator, then the generator terminal voltage will
damping circuits and excitation windings can result in serious collapse and the generator will be less able to export power. The
overheating if prolonged operation is permitted. electrical output from the generator is now characterised by the 'fault
duration' power/load angle curve. This reduction in output power
Pole slipping can also cause damage to excitation systems due to the results in an imbalance between generator input and output powers,
large induced currents that flow in the field winding, which try to which accelerates the rotor causing an increase in load angle. If
reverse in direction every pole slip cycle. Rectifier excitation subsequently, at a load angle o f 52, the fault is removed by
systems will not permit the rotor current to reverse, thus producing disconnecting the faulty line, therefore restoring the voltage, the
a rapid flux change which produces a large reverse voltage across the electrical output is characterised by the post fault load curve.
rectifiers. This can cause reverse breakdown of the semiconductor
devices if they are not sufficiently over rated[4]. The post fault load/angle curve differs from the pre-fault, due to the
system transfer impedance increasing on removal o f the faulty line.
Generator shafts are also at risk due to the pulsating torques The generator output power w ill now be higher than the input power,
produced torsionally exciting them and exposing them to oscillatory causing a decelerating force on the rotor. It can be shown that for
stress. Particularly at risk are turbine generator shafts since the a given load angle swing, the area between the shaft power input
bladed rotating parts o f the turbine interact with the couplings and line, Pm, and the electrical output power o f the machine, is
reduced diameter shaft extensions to form a torsionally resonant proportional to the total energy involved in that swing. Area 'A' in
system. One study[s], found that the generator-exciter shaft section Figure 1, therefore represents the accelerating energy into the rotor,
experiences the highest stresses. It was found that 100 % o f the and area 'B' the decelerating energy. Even though a retarding force
fatigue life o f this shaft section was used up after one pole slip for is now acting on the rotor, the load angle still continues to advance
a shaft over design rating o f two. since the rotor is travelling above synchronous speed. If area A were

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 345

equal to area B, the rotor would once again be a synchronous speed
on reaching load angle 53. The load angle would then begin to Q s Qtrip A nd
decrease as shown, due to the retarding force acting on the rotor still.
(2)
P * Pm ( = Pt )

Generator Where Pt is the algorithm setting for Pm. Since in theory it is

Power C r itic a l possible to just reach the CSP and still m aintain stability, a third
O u tp u t S ta b ility criterion, the rate of change o f real power is used to detect that the
Point load angle is still increasing. From the power/load angle
characteristics in Figure 1, it can be seen that at the CSP, if the load
angle is still increasing, then the real pow er output will be
decreasing, i.e the rate o f change o f pow er (AP/At) will be negative.
Fault- This forms the final criterion:
Duration (AP/At) s (A P/A f)^ (3)
Where (AP/At)B11 is a margin for error, allowed for in this expression
Load to ensure secure reliable relay operation. D erivation o f this m argin
Angle for error will be discussed later.

Extra constraints are also necessary to elim inate m al-operation due

Time to faults. Faults introduce oscillations in pow er o f double system
of frequency, which will produce a change in sign in (AP/At) every half
Load Synchronism cycle. Since pole slipping is a slow process com pared with faults,
Synchronism
Angle Synchronism Lost, Machine
just if the above three criteria (equations 2 and 3) m ust be satisfied for
Guaranteed will Pole Slip
Swing Maintained/ at least one cycle, faults will be blocked due to the change in sign
occurring every half cycle.
Figure 1. Equal Area Diagram Showing Load Angle
5. CONDITION M ONITORING TECHNIQ UES.
doweveP, if ar</a 'A' d e re ecfttal to ^ re a 'A + C '% en tHl retaliation" This algorithm is primarily intended for inclusion in a m icroprocessor
energy produced by the machine when swinging from 52 to 54 based integrated generator protection package[7j. The only
would only ju st be sufficient to counteract the accelerating energy information available to the relay is the three voltage and three
absorbed by the rotor in swinging from 51 to 52. In this instance, current signals that are produced by the relay’s V T s and C T s. Since
stability would have ju st been maintained. The load angle would no prime mover input power signal is available, a m ethod is required
swing up to a value 54, this is known as the critical stability point [6] which estimates the prime m over input power, Pm from these voltage
(CSP). Above this, synchronism would have been lost, i.e. if area 'A' and current signals, in order to provide an algorithm setting, P t This
were larger than area 'B + C' . is achieved using the following algorithm:

if \p - p n ^ \ * P« ; then Pt = P (4)
This point forms the basis for the new power based algorithm. Once
the CSP is exceeded, the rotor will again start to accelerate and the where lag is a predefined time, and n is an integer varied from 0 to
m achine will slip one pair o f poles. If the excess kinetic energy is n, Pto) is the expected oscillation in output pow er due to irregular
not rem oved from the rotor by the next time the CSP is reached, prime mover torque, chosen as nom inally 5 % o f machine rating. P
another pole slip will occur. is the present value o f real power, this evaluation is performed every
lag seconds. The value o f lag and n are chosen to provide a
4 TH E POW ER BASED TECHNIQUE. compromise between reliable evaluation o f Pm, and expeditious
A pole slipping algorithm m ust be capable o f detecting when a pole update times. A value o f lag o f 1 second was chosen in conjunction
slip is about to occur, or actually is occurring. It must remain stable with a value for n o f 3, resulting in 3 seconds o f steady state
for all other pow er system conditions. The two situations which operation within P,ol being required before Pt is updated.
m ost com m only cause difficulties are external faults and power
swings. Adaptive (AP/At ) .„ Setting.
As well as a continuously changing real pow er setting, an adaptive
Figure 1 dem onstrates that once the machine has passed the CSP, a (AP/At)mjo is used in order that algorithm perform ance be optimised
pole slip will occur. The power based algorithm functions by for both steady state and transient loss o f synchronism pole slips. A
detecting if the load angle is still increasing once the CSP has been value for (AP/At)^ is derived by assum ing that the pow er output o f
reached. In order to determine when the m achine has reached the the generator can be expressed as the sym m etrical function;
CSP, both real and reactive pow er are used. The reactive power
characteristics o f a generator are such that at a load angle o f 90°, the P = P n*s 8111 6 (5)
m achine's reactive power dem and is always given by :
Where Pm„ is the maximum m achine output ju st before pole slipping.
The load angle at the CSP is given by (using Pm = Pt):

bp = 180 - sin'1 ( - ^ - ) (6)

W here V is the terminal voltage, Xq is the quadrature axis PmMX
synchronous reactance. If the generator reactive power falls below
this value, then it can be concluded that the generator load angle is The derivative of real power can be expressed as:
greater than 90°. This therefore provides a trip criteria in that if the
dP = dP db (7)
reactive pow er falls below the level corresponding to the load angle
dt db dt
being 90° (Qtrip) then the m achine is operating in an area where
pole slipping is likely. Also at the CSP, the generator real output The derivative of load angle with respect to tim e is the slip, s and is
pow er equals the prime m over input power, Pm. Operation at the expressed as:
CSP can therefore be identified when: ** = -5 (8 )
dt

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 346

i.e. negative values o f slip exist for speeds above synchronous speed. producing a very fast and violent pole slip.
If equation 5 is differentiated with respect to load angle then ;
7. TEST RESULTS.
dP , Figure 2 shows the algorithm results for the laboratory pole slipping
(9)
db ““ test The machine was initially operating at it's minimum value o f
Hence the derivative o f real power at the CSP can be expressed as; excitation, indicated by the initial negative value o f reactive power.
The field was weakened by inserting the parallel resistor at a time o f
s P ^ ccsO S O -sh r1 ) - (A P /A r )^ (10) 0.8 seconds, which caused a glitch in the real power signal. The
reactive power progressively decreased as the load angle increased
due to the rotor's field not being strong enough to hold it in
A value for the margin for error can therefore be obtained by synchronism.
deciding on a minimum value o f slip permitted at the CSP. This
2a) Generator Reactive Power.
value will not ultimately be the minimum value o f slip at which a Reactive Power (VAr)
trip signal w ill be issued due to the errors introduced in assuming a
constant sinusoidal power/load angle characteristic. In reality, errors
will be introduced by non-adherence o f the machine to the constant
-4,000
flux linkage theorem, asynchronous induction generator effects, and
transient saliency. Only transient saliency can be accounted for,
however to do this requires more generator data, which is often not
available to the com missioning engineer. The errors will cause a 1inie(s)
decrease in the observed negative value o f (AP/At), so the value o f 2b) Generator Real Power.
slip chosen can be selected accordingly. Equation 10 is used by the Real Power (w)
algorithm to continuously adjust the value o f (AP/At)nia according to
what the variables Pt and PBn dictate.

Determination of PM .
TRIP Signal
If the generator pole slips due to a steady state loss o f synchronism, - 1,000
no peak w ill occur in the real power before the event, as shown in
um e(s)
Figure 2b. If the pole slip is due to a transient loss o f synchronism
2c) Generator Rate of
(due to a fault, or line switching etc) then a peak will occur before
Change o f Power.
the pole slip, as shown in Figure 4b. This will cause larger values o f Rate of Change of Power (W/s)
(AP/At). The value o f P__ used in deriving (AP/At)mia is nominally
set to 1.25*Pt, in order that the cosine expression in (10) does not
fall below -0.6, giving a minimum value o f (AP/At)-h o f 0.6*s*P--t.
-4,000 (AP/At)min
An absolute minimum value o f P must also be chosen to ensure
(AP/At).* = 0 never occurs. If generator output power does rise 0.5 1.0 1.5 2.0 15 3.0 3.5 4.0 4.5
above 1.2* Pt for more than half a power system cycle, then the value Hiiie(s)
for P . . used in deriving (AP/At^,* is updated to this value for a time Figure 2. Generator Pole Slipping Due to Weak Field.
o f 1 second, which ensures that the correct value o f PM is used if a
pole slip does occur. The half cycle constraint ensures that P .„ is An update in Pt occurs just before the real power starts to drop away
not incorrectly updated for short spikes in power produced by suddenly, this was due to the power staying within the tolerance, Plol
disturbances such as faults. o f equation 4. Once the real power started to drop suddenly with
further load angle increase, only the rate o f change o f power criterion
6. SIMULATION STUDIES. was blocking algorithm operation. This was finally satisfied at a
The protection algorithm was tested using a computer based dynamic time o f 3.3 seconds, which was the first time (AP/At) was less than
simulation package and a laboratory embedded generation model. (AP/At).* for one cycle. The algorithm therefore produced a trip
signal a significant time before the pole slip had occurred, with the
The laboratory embedded power system was comprised o f two 5 real power output o f the machine still high at a value o f 2 kW. The
kVA generators driven by 8 hp dc machines, connected to a 'local' trip signal produced by the algorithm was intermittent until 3.9
load and a 200 V laboratory 'utility' supply. Pole slipping was seconds, this was due to a noisy (AP/At) signal. This distortion was
induced in the protected machine by suddenly inserting a resistor in due to noise being induced in the data acquisition system, due to
parallel with the field supply, so that the field was weakened inadequate levels o f shielding. The action o f differentiating increases
sufficiently to cause pole slipping within the time frame o f the these noise levels to an observable magnitude and provided an
micro-computer based relay being used for data recording. The full opportunity to test the algorithm in extreme noise conditions, where
range o f fault types were also placed on the local load busbar to test it successfully functioned.
the algorithms stability against such disturbances.
Other tests using a variety o f fault types did not cause any mal-
The computer based simulation package was used to model power operations. The two phase to earth and three phase faults provided
swings and pole slips. A simplified one machine (588 MVA) and the most testing conditions for the algorithm.
infinite bus (58.8 GVA) power system model was used, which
possessed 'ideal' power system characteristics. For the power swing, Using results from the simulator, Figure 3 shows that the algorithm
the generator power was first reduced from rated power (500 MW) remained stable for the simulated three phase fault induced power
to half rated power. A three phase fault was then placed on the swing, which caused the load angle to rise to 160°. Figure 3a shows
generator transformer to trigger a power swing that came close to the that the reactive power increased from 0 to 250 MVAr over the fault
CSP. This technique was used to test the adaptive Pt algorithm. period (5.5 to 5.8 seconds), and then jumped to -500 MVAr when
the fault was removed. This satisfied the first criterion for the
The Pole slip started with a change in output power from 250 MW algorithm, since Q < Qtrip, indicating a load angle o f greater than
to 500 MW , and was triggered also by a three phase fault, the 90°. Figure 3b shows thait the active power criterion was not
duration o f which pushed the generator far beyond the CSP, therefore satisfied over this period, indicating that decelerating power was still

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 347

available and thus blocking algorithm operation. Inspection o f the 8. CONCLUSIONS.
adaptive prime m over setting, Pt in Figure 3b shows the three second The new power based algorithm successfully operated for pole
update time taken in adjusting from the old value o f 300 MW to the slipping conditions, providing a trip signal before the machine
new value o f 2 5 0 MW. After this update at 3 seconds on the record actually slipped. It successfully restrained for power swings and
no changes occurred in Pt, indicating that the algorithm used external faults. The use o f condition monitoring methods simplifies
(equation 4 ) functioned correctly. the settings o f the algorithm. This only requires the generator rating
and quadrature axis synchronous reactance. Monitoring the prime
3a) Generator Reactive Power. mover input power and maximum output power also provides a
Reactive Power (MVAr) means o f adjusting the algorithm settings depending upon the type
250 o f pole slip that could be potentially occurring and thus optimising
0 h AT the algorithm's performance.
-250
-500 Qtrip ly 4a) Generator Reactive Power.
0 1.0 20 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 Reactive Power (MVAr)
Time (s) 2501- h
0
3b) Generator Real Power. -250
Qtrip
Real Power (MW) _ -500
9001 Pnjax
600 1.0 2.0 3.0 4.0 5.0 6.0
300 Tlme(s)
0 Real Power 4b) Generator Real Power,
-300 (MW) ftn ax -----
0 1.0 20 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0
Time (s) 500
250
0 PC
3c) Generator Rate of Change of Power. -250
Rate of Change of Power (MW/s) -500
50,000 Peak*" 1 I -750
2000
1.0 20 3.0 4.0 5.0 6.0
Time (s)
-2 00 0 m 1 1 / ^
-4,000 (AP/At)* -49,000 Peak IM y -12,000 Peak 4c) Generator Rate o f Change o f Power.
0 1.0 20 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 Rate o f Change of Power (MW/s)
Time(s) Peak 50,000
3d) Generator Load Angle.
Load Angle (Degrees)
(AP/At)* Peak -49,000
1601 /\
NOTRIP.
20 3.0 4.0 5.i
Tkne(») Peak -9,000
0 1.0 20 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0
Time(s) 4d) Generator Load Angle.
Load Angle (Degrees)
Figure 3. Generator Power Swing Due to Three Phase Fault 150
75
0
Figure 4 show s the algorithm measurands for the simulated pole slip
using the computer model, which caused the algorithm to correctly
-75
-150
1.0 20 3.0
TRIP <a 8 = 170

4.0 5.0
L
6.0
operate. Observation o f Pt in Figure 4b shows that it was updated Tim e(s)
from 250 MW to 500 MW and then no further change occurred. Figure 4. Generator Pole Slipping Due to a Three Phase Fault
Also at this time, P increased to satisfy the PmB[ = 1.2*Pt
requirement Before this time it can be seen that Pnn tracked the 9. REFERENCES.
output power o f the machine, since this was higher than 1.2*250 [1] REDFERN M A and CHECKSFIELD M J, 'An Investigation
MW, i.e. the old Pt setting. This resulted in an increase in inherent into Pole Slipping Protection for Small and Medium Sized Embedded
stability o f the algorithm, due to the trip setting for the rate o f Generation', Proc. 28* UPEC, Stafford, Sept 1993, pp 957- 960.
change o f power criterion (AP/At)mk, becoming progressively more
negative as Pt/Pnai decreased. [2] IMHOF J. A. (Chairman) et al, "Out o f Step Relaying for
Generators - Working Group Report," IEEE Trans, on Power App.
At a time o f 5.7 seconds the reactive power, Q falls below Qtrip. and Systems, Vol.PAS-96, N o.5, Sep/Oct 1977, pp 1556-1564.
Shortly after this the real power, P falls below the Pt setting, and the
[3] CLARK H K and FELTES J W, 'Industrial and Cogeneration
rate o f change o f power signal (AP/At) falls below (AP/At)mia,
Protection Problems Requiring Simulation', IEEE Transactions on
resulting in the algorithm producing a trip signal. The algorithm
Industry Applications, V ol 25, N o.4, July/August 1989, pp 766-775.
produced a trip signal at a load angle o f 170°. Inspection o f Pmax
also shows the temporary 1 second update to the value dictated by [4] GUILE A E and PATERSON W. 1977. Electrical Power
the peak in real power just before the pole slip. Systems Volume 2: Pergamon Press.

This action can also be seen in Figure 3c, where P can be seen to [5] MASRUR M A, AYO UB A K and TIELKING J T, 'Studies on
be returning to the nominal value o f 1.2*Pt, shortly before being Asynchronous Operation o f Synchronous Machines and Related Shaft
updated by the next peak. Inspection o f the (AP/At)-in signal o f Torsional Stresses', Proc.IEE, Part-C, Vol. 138,Jan '91, pp47-56.
Figure 4c shows how it decreases when Pm„ is increased due to a
[6] CEGB: Modem Power Station Practice - Vol.4 - Generator and
peak. This de-sensitises the (AP/At) part o f the algorithm for large
Electrical Plant, 1963 - Permagon Press, pp 126-153.
sw ings in power, thus improving algorithm security. However, if a
peak does not occur, as in the case o f steady state loss o f [7] ZIEGLER G, 'Developments in Generator Protection - Design
synchronism, (AP/At).;. is at it's lowest value for a given value o f and Application Aspects o f a N ew Numerical Relay Range', IEE
Pt, providing a more sensitive algorithm when required. Conference Publication N o. 368, 1993, pp 111-114.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 348

 PAPER FOUR

"A POWER BASED CONDITION MONITORING ALGORITHM FOR

DETECTING DISPERSED GENERATOR LOSS OF SYNCHRONISM”

M J Checksfield and M A Redfem

Presented at
International Power Engineering Conference (IPEC) 1995,
Singapore, February 1995, Volume 1, pp 328 - 333.

MJ.Checksfleld/PhD/University of Bath 1996 Appendix D, Page 349

 A POWER BASED CONDITION MONITORING ALGORITHM FOR
DETECTING DISPERSED GENERATOR LOSS OF SYNCHRONISM.

M. JLChecksfield and M. A. Redfem.

University of Bath, UK.

Abstract For these reasons, research has been carried out into a
new LOS algorithm which can be included in a modem
This paper introduces a new loss o f synchronism (LOS) microprocessor based integrated protection scheme for
relay which uses condition monitoring and adaptive dispersed generators. The new approach reduces the
techniques to reduce the complexity o f installation. As complexity of installation and provides a trip signal
previously reported [1] this approach uses the Equal Area before the LOS actually occurs.
Criterion as its basis to provide a trip signal before the
LOS occurs. By doing this, it reduces the likelihood of
machine damage and voltage disturbances due to 2 LOSS OF SYNCHRONISM PROTECTION
generator LOS. The algorithm's derivation is given and
its performance tested using a variety o f system LOS protection is required to reduce the damage caused
operating scenarios from power swings to faults. to the generator by the LOS or pole slipping condition,
and to ensure that the quality of supply o f consumers
Keywords connected electrically close to the effected generator
Loss o f synchronism, pole slipping, dispersed generator remains acceptable. The degree of these ill-effects
protection, microprocessor based relay. depends upon the relative rating of the bus to which the
generator is connected, the speed w ith which the
generator loses synchronism, and the overcapacity rating
1 INTRODUCTION of the generator.

The majority o f traditional LOS relays use the variation When the dispersed generator is connected to an infinite
in 'apparent' impedance at the generator's terminals to utility bus, then the currents experienced during LOS can
detect a LOS condition[2]. If the impedance crosses two be in excess o f those experienced during three phase
measurement characteristics on an impedance plane faults. Generator windings are normally designed to
within a specified time limit, a trip signal is issued. withstand three phase fault levels, and the higher current
levels experienced therefore loosen and abrade the
However, there are two problems with this approach. windings [3].
The relays function by observing the impedance variation
produced by the complete LOS condition, and therefore High utility infeeds also increase the risk o f generator
only provide a trip signal once the LOS has occurred. shaft damage during LOS. Turbine generators are
This means that an unstable generator may already have especially vulnerable since the pulsating torques
undergone damage and would cause voltage disturbances experienced can torsionally excite resonant sections of
before it was disconnected from the system. the turbine shaft. One study [4] found that the generator-
exciter section was most at risk after the fir st torsional
Choosing the correct boundaries and time limits on the oscillation produced by a LOS. It can therefore be
impedance plane can also be a difficult process, concluded that if shaft damage due to LOS is to be
especially when dispersed generators are being minimised, the generator must be disconnected from the
considered since they can produce unpredictable power system before the LOS occurs, in the hope that
impedance variations [3]. If the wrong relay settings are the torsional impulse will be avoided.
chosen, the LOS condition may go undetected, or worse
still, the relay may mal-operate due to a power swing or Synchronous machines are not designed to operate
power system fault. asynchronously, the resulting induced currents
experienced during a LOS can cause stator and rotor

M.J. Checksfiel d/PhD /University of Bath 1996 Appendix D, Page 350

 overheating. Induced field winding currents can stress the Generator Pre-Fault
generator exciter insulation and semiconductor devices Power
due to the excessive voltages they produce. The Outputi 1 Critical
semiconductor devices do not allow reverse field current Stability
flow, therefore when the field current tries to reverse Point
during the LOS, it is blocked and a high voltage results. Pm (CSP)

The effect o f LOS on other consumers connected close Post Fault

to the effected generator is most pronounced if the utility
infeed is o f a relatively low capacity. The 'system , Fault-
centre' is an expression often used to describe the point Duration
in the power system where the contribution from the
generator equals that from the rest o f the system. At this 0 81 82 90 83 180 Load
point the voltage fluctuations experienced w ill be at their A ngle
most severe, the voltage may even fall to zero. For the
Figure 2 - The Equal Area Criterion.
normal generator/infinite bus situation the system centre
appears inside the generator. However, if the utility
infeed is low, then the system centre may appear in the Consider the case where a transient disturbance results
load network. from a three phase fault on one side o f a double circuit
line as shown in Figure 1. The fault is removed by
The voltage fluctuations at the system centre can cause disconnecting the faulty circuit, leaving the healthy side
plant outages, and in extreme cases, cause a cascading o f the line to transfer power at a reduced capacity. The
LOS o f nearby machines [5], including both generators Equal Area Representation o f this scenario is shown in
and motors. If the voltage depressions last for too long Figure 2.
a period, then induction motors may stall depressing
system voltage further. The Pre-fault curve corresponds to the initial operating
condition, where both lines are transferring power, in this
If the system centre occurs in the zone o f a distance condition the dispersed generator's power transfer
relay, LOS may cause it to mal-operate since the system capability is at its maximum. The fault duration curve
centre takes on the appearance o f a fault, i.e a low represents the generator's power/load angle characteristic
voltage and high current. The LOS condition can also during the fault period. The three phase fault severely
cause other protection relays to mal-operate, such as reduces system voltage so that power transfer is
undervoltage and overcurrent relays. If adequate impaired. The post fault curve depicts the power transfer
blocking is not applied to the effected relays, then capability once the faulty line has been disconnected, the
unnecessary outages could occur. power transfer capability is reduced due to the increase
in the interconnecting impedance.

3 GENERATOR STABILITY ASSESSMENT The generator's prime mover input power is given by the
line Pm. The interception o f the prefault output curve
and prime mover input curve gives the initial load angle
The Equal Area Criterion is one o f the simplest methods
for assessing generator stability for a given transient 81. When the fault occurs the generator's output
disturbance. If the power/load angle curves for all states characteristic falls down to the fault duration curve. The
o f power system operation are known, then generator resulting discrepancy between input and output powers
stability and hence the possibility o f LOS can be produces a net accelerating torque on the generator rotor.
ascertained for a given disturbance. The disturbances As the rotor accelerates the load angle increases, until the
most likely to cause LOS are severe faults, large load fault is removed, which in this example corresponds to
switching operations, and excitation system faults. a load angle o f 82.

Dispersed Double Circuit Utility At the instant the fault is removed, the generator's output
Generator Interconnecting Line 'Infinite' characteristic jumps up above the line Pm to the post
Bus. fault curve. This results in a net retarding force on the
r-X— L ........— I— X~~i generator rotor, but because the rotor is now travelling
g y -@ > above synchronous speed, the initial retarding force is
D—X- <S> used to remove kinetic energy from the rotor, in order
to return the rotor to synchronous speed. If this is
Fault
1 achieved before the rotor reaches load angle 83, then the
retarding force will slow the rotor to below synchronous
Figure 1 - Dispersed Generator Utility Connection.
speed, thus causing the load angle to decrease back

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 351

towards a stable operating value. If however, the rotor where Pt is the algorithm's trip setting for the prime
does not reach synchronous speed by load angle 83, then mover input power, Pm .
it will again start to accelerate, due to the output power
characteristic o f the generator falling below the input The third criterion uses the rate o f change of power
power line Pm, causing an accelerating torque once (AP/At) to deduce that the load angle is still increasing in
more. This is the unstable situation which results in LOS. an unstable manner at the CSP. If the machine had
maintained stability after just reaching the CSP, the load
The Equal Area Criterion states that the accelerating angle would be decreasing.
energy on the rotor due to a disturbance can be
represented by area 'A', whilst the decelerating energy is From Figure 2 it can be seen that at and beyond the
represented by area 'B'. For synchronism to be lost, the CSP, the real power output o f the machine decreases
net accelerating energy for a given disturbance must with load angle increase. Therefore, if the rate o f change
exceed the net decelerating energy, i.e a LOS will occur o f power, (AP/At) is negative at the CSP it can be
if area 'A' is greater than area 'B'. It can be seen from concluded that the load angle is still increasing and that
Figure 2 that the limiting point for stability occurs when the generator is unstable and a LOS will occur. This
the generators output power falls below its input power, forms the third and final criterion for detecting LOS :
this point is referred to as the Critical Stability Point
(CSP). If the generator operates past this point, then (AP/At) < (AP/At)mln (3)
stability cannot be maintained, and a LOS will occur.
where (AP/At)min is a margin for error allowed for in this
expression to ensure algorithm mal-operation due to
4 THE NEW LOS ALGORITHM measurement errors or noise does not occur.

The algorithm predicts LOS by detecting if the load

angle is still increasing once the CSP has been exceeded. 4.1 Fault Blocking
In designing a LOS algorithm, it must be able to predict
when a generator is committed to a LOS. It must also With the criteria set out in Equations (2 & 3) the
remain stable for all other power system conditions, algorithm successfully predicts LOS. However it
including faults and power swings. mal-operates for power system faults. To overcome this
fault blocking constraints are required. Unbalanced
Figure 2 shows that once the CSP has been exceeded, the faults introduce oscillations in power o f twice power
load angle is greater than 90°. The reactive power system frequency, which produce a corresponding change
output characteristics of a generator are such that at a in sign o f (AP/At) every half cycle. A LOS condition is
load angle o f 90°, the reactive power demand of the slow compared to faults and so if the criteria of
machine is always given by : Equations (2 & 3) have to be satisfied for one power
system cycle continuously, imbalanced faults will be
blocked due to the change in sign of (AP/At), whilst the
algorithm is only delayed from tripping by one cycle for
where V is the machine terminal voltage, and X q is the a LOS condition.
quadrature axis synchronous reactance. This value is
denoted as the reactive trip criteria, Qtrip. If the In addition to this, faults also produce values o f (AP/At),
generator's reactive power demand falls below this which are far in excess o f those witnessed due to LOS.
negative value, then it's load angle must be greater than If a maximum negative magnitude is chosen such that a
90°, indicating that it is operating in a potentially LOS condition will never produce it, but faults will, then
unstable area. this can be used as another form o f fault blocking. If
(AP/At) falls below this limit, referred to as (AP/At)mac
Generator real power is used in conjunction with the then algorithm operation can be blocked.
reactive criterion to determine whether or not the
generator is operating at, or beyond the CSP. From
Figure 2 it can be seen that when the CSP has been 5 CONDITION MONITORING TECHNIQUES
exceeded, the generators real power output, P is less
than the prime mover input power level, Pm indicating
In a dispersed generator microprocessor based relaying
that no more decelerating power is available. Operation
environment, the only information normally available to
at, or beyond the CSP is therefore recognised when :
the relay is the 3 Voltage and 3 Current signals provided
by the relay's transducers. Dispersed generators may be
operated over a range o f output powers and so a method
P < Pm ( = P t) of monitoring the prime mover input power, Pm is

MJ.Checksfield/PhD/University of Bath 1996 Appendix D, Page 352

necessary in order to determine the algorithm trip setting, The value of P used by Equation (5) is nominally set
Pt. The algorithm has been designed with a to 1.25*Pt, so that Equation (5) evaluates to minimum
microprocessor relaying platform in mind, a limited setting level of 0.6*s*Pmax ( = 0.75*s*Pt). If the
amount o f processing power is therefore available to generator's output rises above 1.25*Pt for greater than a
achieve this goal. The input power, Pm is effectively half a power system cycle, then the value of Pmat is
found by monitoring the power output o f the generator, updated to the new maximum for a duration of one
if this remains constant for a sufficient amount o f time, second. The one second limit is used so that the new
then the input value must approximately equal this value, value of Pnm is only used for the period where a LOS
the only discrepancy being produced by the losses in the associated with that value is likely to occur. The half
generator. The following algorithm performs this cycle constraint is necessary so that incorrect values of
function : Pma d° not result due to spikes on the power waveform.

if \ P * P », ; it'enP * = P (4)
6.0 SIMULATION STUDIES
where lag is a predefined time, and n is an integer varied
from 0 to n. PM is the expected oscillation in generator
A laboratory dispersed generation model and a computer
output power, due to irregular input torque, chosen as
based dynamic simulation package were used to test the
nominally 5% o f machine rating and P is the machine
algorithm's performance at predicting LOS and
output power. This evaluation is performed every lag
restraining for faults and power swings.
seconds. The value o f n and lag are chosen so that
update times are satisfactorily short and accuracy is
The laboratory dispersed generation model consisted of
sufficiently high.
two 5 kVA synchronous generators driven by 8 horse
power dc motors. These were connected to a 'local' load
and a 200 V laboratory U tility' supply. This was useful
5.1 Adaptive Trip Settings for testing the algorithm in a real life situation, since
harmonics and heavily alternating loads nearby made the
5.1.1 (A P / A t ) Setting. Assuming a sinusoidal supply far from ideal. To test the condition monitoring
power load angle curve, then for a given value of slip at section o f the algorithm the generator was ran initially at
a given power input level, (AP/At)min can be calculated low power, when data recording started, the input power
using : was then increased. A short while after, the LOS was
forced by quickly inserting a resistor in parallel with the
^ = -■* P^cosClSO - sin v( - p - ) ) = (AP/A/)miI (5)
generator field winding which weakened the field
®CSP mtx
sufficiently to cause a LOS.
It should be stressed that the value o f slip chosen will
not be the exact value at which the relay produces a trip The laboratory model was also used to test the algorithm
output, due to the assumptions used in deriving Equation during power system fault conditions. All of the fault
(5). Inaccuracies due to transient saliency and induction types possible were placed temporarily on the local load
generator action will cause (AP/At) to be more negative busbar by switching in a 'fault' resistance.
than predicted by Equation (5).
The computer based dynamic simulation package was
Equation (5) is used by the algorithm to continuously used to test the algorithm's performance against power
update the trip setting (AP/At) mtn depending on what the swings and LOS. The model used was the simple one
variables Pmac and Pt dictate. machine (588 MVA), infinite bus model (58.8 GVA). A
power swing was caused by placing a fault on the
5.1.2 Determination of Pmm . The maximum generator transformer terminals of sufficient duration to
power output o f the generator is monitored so that the cause the generator rotor to swing up to the CSP and
algorithm can adjust its setting according to whether a back down to a stable operating area.
transient or steady state LOS will occur. For a transient
LOS, the generator's output during the First swing will be To test the algorithm for detecting LOS, the simulated
much greater than its nominal output. This results in a generator was initially set to ran at maximum output
greater magnitude in the signal (AP/At), the algorithm power. The input power was then reduced to half rating
therefore adjusts its setting (AP/At)min accordingly. For in order to test the condition monitoring section of the
a steady state LOS, no peak will occur in the power algorithm. A three phase fault o f slightly longer duration
output before the LOS, the algorithm trip setting should than the one used to trigger the power swing was then
therefore be at its most sensitive level to detect the LOS used to cause a LOS.
in the shortest time possible.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 353

 7 T E ST RESULTS angle had increased to a value above 90°. Inspection of
the real power plot shows that after insertion o f the
Figure 3 shows the algorithm variations for the parallel resistor, the generators field was weakened
laboratory system weak field LOS test. Inspection of sufficiently so that the generator could not maintain the
plots a and b show that the generator was initially level o f output required, the real power criterion was
operating overexcited at very low power. At a time of therefore satisfied on inserting the resistor.
Is on the record the power input to the generator was
increased so that the machine operated at approximately The final criterion (AP/At) , can be seen to fall below its
2.5 kW. Observation o f the (AP/At)min curve in plot c trip setting (AP/At)mtn at a time o f 4.3s. This meant that
shows that because Pt was not updated during this all three criteria were satisfied for the one cycle required
period, the trip setting (AP/At)min gradually decreased, and a trip signal was therefore produced. The trip signal
resulting in a de-sensitising o f the algorithm. This is can be found on the real power plot, it can be seen that
beneficial since the likelihood o f LOS is higher during the trip was issued a significant time before the LOS
adjustment to a higher output level. occurred. The usual point where LOS is said to have
occurred is when the machine goes from generator to
a) Generator Reactive Power
motor action, i.e when the power output o f the machine
goes negative. Unfortunately, in this case the whole
LOS could not be recorded, due to limitations in the data
> -1500 acquisition system.
-3000 Qtrip
Figure 4 shows the algorithm's response to a two phase
en -4500
to earth fault. O f all the fault types tested, this and the
-6000
2.0 3.0 4.0 5.0 three phase faults provided the toughest conditions for
Time (s) testing algorithm stability. The reactive power criterion
b) Generator Real Pow er was satisfied for a small period on removal o f the fault,
at 1.4 s on the record. During this time the real power
'S' 3000 criterion is satisfied for a very short amount o f time,
ts
jS
> 2000 before the real power output o f the generator jumps
TRIP —
above the trip setting Pt. It is this short interval where
algorithm mal-operation is most likely. However,
inspection o f the (AP/At) plot shows that during this
00 1.0 2.0 3.0 4.0 5.0 period, the (AP/At) signal jumps to a negative value off
Time (s) o f the scale for a very short time, before escalating to a
c) Generator Rate o f Change very high positive value for a short time. These
o f Pow er oscillations took less than one power system cycle, and
^ 3000 so the algorithm was continually restrained.
| 1500
The algorithm also successfully restrained for all o f the
^% °
-1500
(AP/At)mm
other fault types tested.
-3000
0 1.0 2.0 3.0 4.0 Figure 5 shows the dynamic simulator results. Initially
Time (s) the generator was operated at full power, it was then
reduced to half rated operation. The condition
Figure 3 - 5 kVA Generator Loss o f Synchronism (LOS).
monitoring part o f the algorithm updated at a time o f
3.75s on the record. Inspection o f the (AP/At)mln curve
At a time o f 3 s on the record, the condition monitoring shows that until the update it stayed at its high power
part o f the algorithm updates the value o f Pt to the new level, resulting in the algorithm being de-sensitised for a
operating level. This caused a drop in magnitude o f small duration. This is unavoidable, but is o f no great
(AP/At)min , but it did not drop back to the initial value, concern since in the event o f a LOS it would just result
due to Pt being at a higher level. This is the desired in the algorithm producing a trip signal delayed by a
effect since at higher input powers, more severe LOS can small time. At 4.75 s on the record the fault was placed
occur and more negative values o f (AP/At) are produced. on transformer bus, and removed again at 5.2s.

Shortly after the update, the resistor was connected in On removal o f the fault the generator's reactive power
parallel with the field, causing the LOS to occur. was less than Qtrip, satisfying the reactive criterion
Inspection o f the reactive power plot shows that this indicating that the load angle was above 90°. At the
criterion was satisfied at 3.75 s, when the reactive power same time, the real power output is much greater than
fell below the trip setting, Qtrip indicating that the load the nominal value before the fault.

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 354

 a) Generator Reactive Power
1500 _ 500

I ° 250

^ -1500 Qtrip
I -250
O
^ -3000

0 0.5 1.0
j
1.5 2.0 2.5
^ -500
0
Qtrip

1.0 2.0 3.0 4.0 5.0 6.0

Time (s) Time (s)

b) Generator Real Power b) Generator Real Power

5000 1000

4000 500

b 3000 T R IP —^
£ 2000 N O T R IP o -500

-1000
0 0.5 1.0 1.5 2.0 2.5 3.0 2.0 3.0 4.0 5.0 6.0
Time (s) Time (s)
c) Generator Rate o f Change c) Generator Rate o f Change
o f Power o f Power
^ 2000
C/5
(A P /A t)
I °
3-2000 (AP/At)min 5-3000

5-3 -4000
0 0.5 1.0 1.5 2.0 2.5 3.0
Time (s) Time (s)

Figure 4 - 5 kVA Generator 2 Phase to Earth Fault Test. Figure 5 - 588 MVA Generator Loss o f Synchronism
Due to a Three Phase Fault.

This invoked the adaptive Pmax part o f the algorithm,

which tracked the real power output up to its maximum 9 REFERENCES
value, and then held this value for 1 second before
returning back to 1.25*Pt.
[1] M. A. Redfem and M. J. Checksfield. A N ew Pole
This action resulted in (AP/At) mtn falling to a lower level, Slipping Protection Algorithm for Dispersed
thus desensitising the algorithm to the pending transient Storage and Generation Using the Equal Area
LOS. The (AP/At) signal still fell below the trip setting Criterion. Presented at the IEEE PES Summer
(AP/Al)min for one cycle, and during this time the other M eeting, San Francisco, 94 SM 425-9 PWRD, July
two criteria were satisfied. A trip signal therefore 1994.
resulted.
[2] J. A. Imhof et al. Out o f Step Relaying for
The algorithm successfully restrained for the power Generators - Working Group Report. IEEE Trans.
swing test, which produced a severe swing in load angle P ow er Apparatus and System s. Vol PAS-96, N o.5,
up to a value o f 160°. Sep/Oct 1977, pp 1556 - 1564.

[3] H. K. Clark and J. W. Feltes. Industrial and

8 CONCLUSIONS Cogeneration Protection Problems Requiring
Simulation. IEEE Trans. Industry Applications.
The new algorithm successfully predicted LOS before it Vol IA-25, N o.4, July/August 1989, pp766-775.
actually occurred. In doing so the new algorithm offers
the advantage over traditional relays o f disconnecting the [4] M. A. Masrur et al. Studies on Asynchronous
generator before the LOS, minimising the potential Operation o f Synchronous Machines and Related
damage. It successfully restrained for all o f the fault Shaft Torsional Stresses. IEE Conf. Proc. Part C,
types and for power swings. The algorithm's adaptive Vol. 138, Jan 1991, pp 47-56.
nature helps improve algorithm performance for steady
state and transient LOS by monitoring the input and [5] L. J. Powell. An Industrial View o f Utility
maximum powers. In addition to this, it simplifies the Cogeneration Protection Requirements. IEEE
installation procedure. Since only the q-axis synchronous Transactions on Industry Applications, Vol IA-24,
reactance and the machine rating are required to set the N o .l, Jan/Feb 1988, pp 75-81.
LOS protection function.

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 355

 PAPER FIVE

"ASSESSMENT OF EMBEDDED GENERATOR STABILITY WHEN

CONNECTED TO A HV UTILITY DISTRIBUTION SYSTEM"

M J Checksfield and M A Redfem

Presented at
30th Universities Power Engineering Conference (UPEC),
September 1995, Greenwich, UK, pp 219 - 222.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 356

 ASSESSMENT OF EMBEDDED GENERATOR STABILITY WHEN
CONNECTED TO AH V UTILITY DISTRIBUTION SYSTEM.

M J Checksfield & M A Redfem.

University of Bath, UK.

ABSTRACT. 2.1 The Generator and It's Control Equipment

Concern is mounting about the likelihood of instability of embedded The two main generator parameters which have the most affect on
generators which are being installed to operate in parallel with utility stability are the inertia of the generator, and the direct axis transient
distribution systems. This is due to their low inertias, high resistances reactance. The most useful measure of generator inertia is the inertia
and low time constants. A stability assessment of a 625 kVA constant, H in units of kW-s/kVA. The main advantage of using H
embedded generator connected to an 11 kV distribution system is compared with other measures of rotational inertia is that its value is
used to illustrate the factors affecting stability. fairly constant for a wide range of generator sizes and types, and it
therefore makes comparisons between machines much easier.
1. INTRODUCTION.
The installation of small and medium sized synchronous generators 2.1.1 Inertia Constant, H & Transient Reactance Xd'.
to operate in parallel with a Regional Electricity Companies (REC) From a stability viewpoint the higher the generator's value of H, the
distribution system has become commonplace. These embedded more stable it will be. For a given fault clearing time, the machine
generators pose t e c h n i c a l problems which normally do not have to load angle at which the fault is cleared decreases with an increase in
be considered with the large grid generators connected to inertia. The machine requires more acceleration energy to make its
transmission circuits. One concern is the transient stability margins rotor advance and therefore appears less responsive and more able to
of these machines. Transient stability is the ability of the generator withstand a mismatch between generator input and output powers.
to produce forces which act to restore the generator back to a state
of equilibrium following a transient disturbance. Transient The direct axis transient reactance, denoted Xd' is also a vital
disturbances can be produced by load changes, switching operations measure of transient stability because it determines the amplitude of
and electrical faults. the generator's power output - load angle characteristic. For the
simplest case of a round rotor machine with resistance neglected, the
A transient disturbance, such as a fault will reduce a generator's transient P 5 curve is given by;
ability to provide output power. The input power from the prime
mover will not follow this change however. This m i s m a t c h between P = sin(6) (1)
Xd'
input and output results in an accelerating torque on the generator's
rotor which causes the load angle of the generator to increase. If the where E is the voltage behind transient reactance, V is the infinite
load angle increases beyond a certain angle, called the critical bus voltage which the generator is connected to, and 5 is the rotor
clearing angle, 5c, then the generator will not be able to provide the angle. As can be seen from equation (1), the lower the transient
necessary decelerating forces to return the generator back to a stable reactance, the more output power the generator can produce. This
operating point. A further increase in load angle will cause the serves to provide a greater decelerating force on the generator rotor
generator to lose stability and 'slip poles' until either it's rotor returns following a transient disturbance.
to synchronous speed and re-synchronises to the REC's power
system, or the generator is disconnected by protective relays. 2.1.2 Open Circuit Time Constant, Tdo' & Resistances.
To a lesser extent than H and Xd', other synchronous machine
The time taken for the load angle to swing from its initial value to parameters have an effect on stability. One of these is the direct axis
the critical clearing angle (5c) for a 3 phase short circuit fault, is open circuit time constant Tdo'. Following a transient disturbance,
referred to as the Critical Clearing Time (CCT). Faults must therefore the reactance of the generator goes from the normal steady state
be cleared within this time in order to maintain stability. In most synchronous value, Xd to the transient value, Xd'. This can be
cases of transient stability assessment, if the machine does not lose explained using the constant flux linkage theorem. Following an
stability during the first swing in load angle, it will maintain stability increase in stator current, and therefore armature reaction, a decrease
for that disturbance. in the flux flowing from the stator to the rotor occurs, the field
current therefore changes so that it can maintain constant flux linkage
For most embedded generators, the point of connection to the REC's in the field circuit. This increased field current decays eventually,
distribution system can be thought of as a connection to an infinite the time it takes to decay is given by the time constant Tdo', where;
bus, that is a bus whose frequency and voltage remain un-affected by
the embedded generator's operation. This simplifies the discussion Tdo' = *£ (2)
of stability greatly since the one machine - infinite bus is the most Rf
basic system from a stability analysis viewpoint. If the REC's bus Lflf is the field winding self inductance and Rf is the field
is not infinite, then the system can be represented as a two machine resistance[ 1]. The field circuit parameters are dictated by the values
system, which is also relatively simple to analyze. given by the combination of fielding winding, iheostat and exciter
armature. Tdo' determines the time taken for the reactance of the
2. FACTORS AFFECTING GENERATOR STABILITY. machine to change from the transient value Xd' back to the
The factors which affect synchronous generator stability can be split synchronous value Xd. With this change comes a corresponding
into two different categories, those which are determined by the decrease in the output power capabilities of the generator and hence
generator and its control equipment, and those which are dependent its potential to produce decelerating forces in response to a
upon the configuration and operation of the system to which the disturbance. It follows that the higher the resistance compared with
generator is connected. the inductance of the field, the shorter the time constant, and the less
time the generator has for producing high output powers. The

M.J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 357

 average value of the time constant is around 6 seconds, and when the protection associated with distribution feeders, the clearance time
this is compared with the mechanical period of interest for stability can be in excess of one second if graded inverse definite minimum
of 1 second the condition of constant flux linkage is valid. However, time overcurrent relays are employed[7,8]. With this amount of time
if the generator is exposed to a severe fault, which is cleared in a the load angle increases so much so that the likelihood of the
time of around a few hundred milliseconds[l] the constant flux generator maintaining stability is low.
theorem does not hold due to the decrease in flux which the fault
causes. Embedded generators tend to have higji per unit resistances, 2.2.2 Transfer Reactance Between Generator and Infinite Bus.
which results in them having smaller time constants. The value of connecting impedance between the generator and the
infinite bus has a crucial affect on generator stability. A study
2.1.3 Damping Power. conducted on a 4.5 MVA generator [5] revealed that the CCT was
Another factor which can influence stability is the damping power reduced by 40 ms when the ratio of the bus at the generator
produced by the generator. This is dependent mainly upon the type terminals to the generator rating was reduced from 100:1 to 5:1.
of damper winding in the machine, and the magnitude of armature This can be explained using equation (1). Any reactance between the
resistance. As the amount of resistance increases, the damping terminals of the generator and the infinite bus can be added to the
torque produced by the machine decreases, the synchronous torque reactance of the generator, including that of a step up transformer.
also decreases, but not by as much[2]. As well as enhancing stability Therefore it can be seen from equation (1) that an increase in
by acting like an induction machine and providing positive damping, connection impedance results in a decrease of the maximum output
the damper windings also serve to improve stability by providing power of the generator, and therefore the amount of decelerating
negative sequence damping and enhancing positive sequence torque it can produce. As a general rule[5], the critical clearing time
damping during unbalanced faults, in addition to producing some DC is related to inertia and reactance in the following manner;
braking[l].
H (3)
C CT a
2.1.4 Automatic Voltage Regulator (AVR) (.X d' + Xt) (Xd + Xt)
As mentioned in section 2.1.2, the effect of armature reaction during
a fault tends to decrease flux linkage, especially if the reactive power Where Xt is the transfer reactance. The situation may be made worse
output of the generator is large[3]. If the generator has an AVR in following the clearance of a fault since one of a set of parallel
operation the action of the AVR would tend to force the excitation interconnections would have to be disconnected in order to clear the
system to boost the exciters mmf and therefore maintain flux levels. fault, resulting in a higher transfer impedance.
Anderson[3], states that the dominant factor with AVR's in
maintaining stability is the ability to offset the affects of armature 2.2.3 Effect of Resistance on Stability.
reaction during faults. Even though AVR's can significantly improve As well as the reactance of the generator and interconnection altering
the flux levels during faults, their effect cm the magnitude of the first the stability limits, the resistance of the generator and interconnection
swing in load angle is generally only a reduction of a few degrees. can also adversely effect it. Normally, with large synchronous
The situations where they are most beneficial is where the faults machines, the per unit values of resistance can be neglected.
have long clearance times. However, with smaller embedded machines the value of resistance
both in the generator and the interconnection becomes more
One possible drawback to using high performance fast AVR's is their appreciable. As with reactance the resistance of the interconnection
effect on the damping of the generator. A fast AVR may reduce the can be added to that of the armature circuit of the generator, hi
magnitude of the first swing, but in doing so reduces the damping section 2.1.3 it was stated that resistance reduces the damping
(section 2.1.3) for subsequent swings. If the reduction is severe capabilities of the generator, lessening the stability margin. The
enougjh the generator may maintain stability for the first swing and greatest concern with high resistance values in an embedded
then lose synchronism in subsequent swings. This is known as generation environment is the affect on dynamic stability. It is
dynamic instability, and additional stabilising feedback signals need possible that a high value of resistance found in a long length of
to be incorporated into the AVR to avoid it[4]. cable, combined with an already lowly damped machine, low
performance AVR and poorly damped governor could result in a
2.1.5 Generator Governors. negative value of damping[9,10,l 1]. This could potentially cause the
Depending upon the type of prime mover/govemor set used, the machine to oscillate at its natural frequency, with the amplitudes of
governor may offer a significant improvement in the CCT. If the the oscillations increasing until the generator dynamically loses
governing system has a small overall time lag, then the CCT will be stability. The only situation where increased resistance in beneficial
increased by its effect. Studies conducted in reference [5] show that to stability is with a close up three phase fault, where the power
the governor can increase the critical clearance time by dissipated in the resistance reduces the accelerating power going into
approximately 40 ms. If, however, the governor and prime mover the rotor.
system has a long time lag, it's time constants will render it
practically in-active during the first swing of a transient disturbance, 2.2.4 Affect of Fault Type on Stability.
and it will therefore have little effect on the CCTf6]. As the impedance of a fault decreases, so does the amount of power
transfer possible from the generator, causing a lower stability limit
2.2 The System Connected to the Generator. for a given fault duration. In order of fault severity, the most severe
The two main factors which can affect the transient stability margins fault is the 3 phase fault followed by 2 phase to earth, two phase,
of a generator are the effective transfer impedance between generator and single phase to earth. The difference in severity of faults
and infinite bus and the protection clearance times for faults. becomes smaller as the fault duration is decreased, however, with a
fault of 200 ms there is an appreciable difference!6],
2.2.1 Protection Clearance Times.
Protection clearance time is the most important factor in maintaining 2.2.5 Affect of Grounding on Stability.
stability, since the longer the fault exists on the system, the longer The grounding of a system affects it's zero sequence impedance, Zo
the power transfer capabilities of the generator are disrupted. The which in turn affects the impedance of unbalanced faults and hence
protection clearance times normally associated with high power the level of power mismatch caused by the fault. As Zo increases
transfer transmission circuits are usually in the region of several with reference to the fault location, the severity of the fault
power system cycles, this means that only a small amount of decreases. Taking it to the extreme of an infinite value of Zo, it can
accelerating energy is produced as a consequence of the fault. With easily be seen that this would transform a two phase to earth fault

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 358

into a two phase fault. It is therefore beneficial from a stability CCT of 280 ms is very alarming since to obtain satisfactory grading,
viewpoint to ground through impedances rather than using solid the REC's generally set their overcurrent relays to clear faults in a
grounding techniques. Grounding resistors close to generators are time of 0.5 to 1 second [7,8]. With clearance times far in excess of
more beneficial than reactors since they provide an additional the CCT, generator instability is very likely.
breaking effect to the generator during unbalanced faults. Embedded
generators are generally resistively earthed in order to keep ground ~ 300
fault currents low and to reduce third harmonics, this therefore | 250
improves stability with respect to unbalanced faults. Unstable Case
Q 200
© Stable Case
2.2.6 Affect of Local Load on Stability. ■a 150
The type of load must also be considered since loads such as
induction motors can continue to depress system voltage after a fault
^ 100f0.0 0.5 2.0
has been cleared[12]. If an induction motor is braked sufficiently Time (s)
during the fault due to the voltage dip, then the machine's reactive
load will increase as the machine attempts to regain speed. The Figure 2: Machine Angle for Stable & Unstable Simulations
voltage reduction produced will reduce the generator's power transfer
capabilities, and hence the amount of decelerating torque produced Also of interest, in terms of quality of supply are the fluctuations
will be lower. experienced in the load and generator terminal voltages following
loss of stability. These can be seen in Figures 3 & 4. It is evident
2.3 Comparison of Stability Parameters of Typical Embedded from Figure 3 that the mean load voltage for the unstable case is
and Grid Synchronous Generators. lower than for the stable case. The severity of the depression is of
Table 1 illustrates how the constants which affect the stability of a the order of 2 %.
generator vary for embedded generators[13] and conventional 'grid'
size machines[3]. 1.01.
------ Unstable Case
Stable Case
Analysis of the inertia constants show that for embedded generators,
the value for H is generally 10 times less than it is for grid machines. 1
This highlights the responsiveness of embedded machines to
disturbances, an embedded machine will swing much further than a
conventional grid machine for a given disturbance. 0.0 0.25 0.5 0.75 1.0
Time (s)
Examination of armature resistance (ra) values reveals that this also
tends to be 10 times larger for embedded machines, resulting in Figure 3: Load Voltage for Stable and Unstable Simulations.
embedded machines theoretically having less damping. Damping
however is dependent cm numerous factors so this does not This scenario has a high 'infinite' bus to generator ratio of 200:1. The
necessarily mean the damping will be proportionally 10 times less. infinite bus therefore maintains the load voltage within acceptable
limits. If the ratio of infinite bus to machine ratings dropped, or the
Inspection of the reactances and time constants shows that although impedance between the 'infinite' bus and load increased, the
the transient values are practically the same whatever the size of the magnitude of the voltage fluctuations would increase. This can be
generator, the open circuit time constants for embedded generators explained if the generator and infinite bus are each thought of as
are over 2.5 times shorter. This means the flux levels will not be contributing to the load voltage. Their respective contributions to the
maintained for long in the embedded machines following a voltage depaid upon their apparait rating from the load's viewpoint,
disturbance, unless a high response ratio, high performance AVR is the higher the rating the more they contribute. When the generator
fitted [3], which will keep the flux high by field forcing action. pole slips, it's contribution to the voltage is 180° out of phase, and
therefore its effect is to decrease the voltage. The higher its rating in
3.0 EMBEDDED GENERATOR STABILITY ASSESSMENT. proportion to the infinite bus, the more it will decrease the load
The industrial embedded generation system, shown in Figure 1, was voltage.
modelled using the Electromagnetic Transients Program (EMTP).
The data used in the simulation is in Appendix A. The generator In this simulation, the load was static, and did not have any
only uses an AVR when operating on its own, when it is paralleled sensitivity to voltage variations If the load had beoi an induction
with the REC system, the AVR is taken out of service. Due to motor load for example, the decreased voltage may have caused it to
insufficient data, the governor model was not simulated. stall, or it's starter contactors to open.

Embedded REC It can be seen from Figure 4 that the generator terminal voltage
Generator Infinite' Bus
Cable Cable suffers more severe voltage depressions, and if a local load was
625 kVA 125 MVA.
415/11,000 1 km 2km connected, the voltage depressions would have severely disrupted it.
GCHr-c As well as induction motor loads suffering from voltage disturbances,
Load thought must be bome over to the effect on undervoltage protection
Fault
2 MW relays which may operate if the voltage is depressed for long enough

. ___ Unstable Case

Figure 1: Embedded Generator/Utility Connection. s 1 Stable Case
A*, 0.8
]
|) 0.6
A three phase fault was placed at the HV terminals of the step up 2
transformer and the fault duration gradually increased until the ^ 04
generator went unstable. For a fault clearance time of 270 ms the 0.2
generator remained stable, for a clearance time of 280 ms, it was 0.0 0.25 0.5 0.75 1.0
unstable and pole slipping resulted. The machine load angle curves Time (s)
for these simulations can be found in Figure 2. The value for the Figure 4: Generator Voltage for Stable and Unstable Simulations

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 359

 3.1 Effects of Embedded Generation System Configuration on 5.0 REFERENCES.
Stability. [1] Kimbark, E W : 1956. "Power System Stability -
Some embedded configurations have multiple embedded generators Synchronous Machines" Vol 3, Chapman & Hall Ltd, London.
located within one site. If the generators are connected to the same [2] Gupta, S & Lynn J : 1980. "Electrical Machine
bus, they can usually be grouped together, there behaviour appearing Dynamics" The Macmillan Press Ltd, London.
as though they were one larger generator. From a stability viewpoint [3] Anderson, P M & Fouad, AA : 1977. "Power System
adding extra generation to a site will reduce the ratio of infinite bus Control and Stability" The Iowa State University Press, Iowa, USA.
to machine ratios, if the ratio becomes too low, then the generators [4] Demello F P & Concordia C : "Concepts of
will not receive enough synchronising power following disturbances Synchronous Machine Stability as Affected by Excitation Control"
and will be more prone to instability. On the other hand, the IEEE Trans. Vol. PAS-88, No.4, April 1969, p 316.
inertias of the two machines combine to make a larger inertia, which [5] Electricity Council: "Notes of Guidance for the Protection
makes the 'group' of generators less twitchy'. of Private Generating Sets for Operation in Parallel with Electricity
Board Distribution Networks" Engineering Technical Report 113,
The effect of step up transformers also affects stability, since the London, 1989.
impedance of the transformer can be added to that of the generator's [6] Kimbark, E W : 1964. "Power System Stability -
stator circuit. If the transformer has a high impedance then the Elements of Stability Calculations" Vol 1, 5th edition, John Wiley
interchange of synchronising power between generator and grid will & Sons, London.
be reduced. The transformer may improve stability for close up [7] Clark H K & Feltes J W : "Industrial and Co-generation
terminal faults, since the transformer will partially load the generator. Protection Problems Requiring Simulation" IEEE Trans, on Industry
Applications, Vol 25, No.4, July 1989, pp766 - 775.
3.2 Availability of Data and Use of Computerised Techniques [8] Fielding G & Evans G W : "Industrial Feeder Protection"
for Stability Analysis. Tenth international Conference on Electricity Distribution, Parti, 5 -
If it is possible to obtain manufacturers data on the generator, Consumers Installations, 1989, pp 460 - 466.
transformer & connecting lines, then a comprehensive stability [9] Concordia C : "Effect of Prime Mover Speed Control
evaluation can be performed using one of the many transient stability Characteristics on Electric Power System Perfomance" IEEE Trans.
evaluation packages available. However, in many cases, it may be Vol. PAS-88, No. 5, May 1969, p 752.
impossible to obtain the full set of data. In these instances, an [10] Nickle C A & Pierce C A : "Stability of Synchronous
estimate of the CCT can be found using just the inertia, synchronous Machines - Effect of Armature Circuit Resistance"
and transient reactances and bus capacity. If a computer based ATHE Transactions, Vol. 49, 1930, pp 338-351.
technique is used, then the other parameters required by the package [11] Concordia C & Carter G K : "Negative Damping of
will have to be estimated. If a computer package is not available, Electrical Machinery" ATF.F. Trans. VoL 60, March 1941, p 116.
then a fair estimate of the CCT can be found using the Equal Area [12] Arcidiacono V "Studies on Damping of
Criterion and step by step calculations, as described in reference [6]. Electromechanical Oscillations in Multimachine Systems with
Longitudinal Structure" IEEE Trans. Vol. PAS-95, No. 2, March
4.0 CONCLUSIONS. 1976, p 450.
Embedded generators connected to distribution systems using [13] Gish W : "Small Induction Generator and Synchronous
overcurrent protection can be prone to instability. The long fault Generator Constants for DSG Isolation Studies" IEEE Trans. Power
clearance times associated with overcurrent protection, coupled with Systems, Vol PWRD - 1, No. 2, April 86, pp 231 - 239
low generator inertias, short time constants, high resistances and high
reactance values in the generator and interconnection make them APPENDIX A DATA USED FOR TRANSIENT STABILITY
much less stable than traditional 'grid' generators. EVALUATION.

Prolonged disturbance to the surrounding power system may result All per unit values given on generator base (625 kVA)
following a fault, due to the effects of a synchronous generator
operating in an unstable manner. These problems can only be Generator : S = 625 kVA, H = 2.84 s, Xd = 2.25 pu, Xd' = 0.22 pu,
remedied by either reducing fault clearance times so that the Xd" = 0.15 pu, Xq — 2.15 pu, Xq' = 2.15 pu (No q axis damper
generator remains stable, or installing protective relays to disconnect modelled), Xq" = 0.225 pu, Tdo' = 1.9 s, Tqo' = 0 s, Tdo" = 0.024
the generator when it is operating unstably. s, Tqo" = 0.018 s, Xo = 0.051 pu, Ra = 0.017 pu, XI = 0.1 pu

The Critical Clearance Time (CCT) for the typical case studied was Transformer Parameters : S = 625 kVA, X = 0.1, ratio 415 V/l 1 kV.
270 ms, when this is compared with the 0.5 to 1 second fault REC parameters : S —125 MVA, 11 kV.
clearance times typically produced by overcurrent relays, then it can Load Parameters : P = 2 MW, Q —0.2 MVAr
be concluded that there is a high likelihood that embedded generators Interconnection : R = .00655 pu, X = .00689 pu (for 1km)
will go unstable.
Table 1 - Typical Stability Analysis Data for Embedded and Grid Type Machines

Type Embedded Embedded Embedded Embedded Grid Grid Grid Grid

Rating 156 kVA 781 kVA 1044 kVA 4510 kVA 100 MVA 233 MVA 414 MVA 552 MVA

ra (pu) 0.011 0.034 0.017 0.013 0.0035 0.0016 0.0019 0.0047

H(seconds 0.329 0.205 0.500 0.535 4.985 4.122 3.704 5.453

Tdo' (pu) 0.950 1.87 1.90 2.47 5.90 5.14 5.43 3.65

Xd (pu) 2.02 6.16 2.43 2.38 1.18 1.57 1.77 1.78

Xd' (pu) 0.171 0.347 0.254 0.264 0.220 0.324 0.274 0.258

M.J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 360

 PAPER SIX

"THE EFFECTIVENESS OF STANDARD SCHEMES FOR THE

PROTECTION OF DISPERSED GENERATORS FROM
ABNORMAL OPERATING CONDITIONS"

M A Redfem and M J Checksfield

Presented at
3rd International Conference on Advances in Power System Control,
Operation and Management (APSCOM)
November 1995, Hong Kong, pp 506 - 511.

M.J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 361

 THE EFFECTIVENESS OF STANDARD SCHEMES FOR THE
PROTECTION OF DISPERSED GENERATORS FROM
ABNORMAL OPERATING CONDITIONS.

M A Redfem & M J Checksfield.

University of Bath, UK.

ABSTRACT. in a portion o f the utility's load being left connected to

the embedded generator alone. In this condition the
Questions are increasingly being asked about the isolated part o f the system may continue to operate
response o f standard protection schemes used for independently o f the utility, forming its own power
protecting embedded synchronous generators during island.
abnormal operating conditions. Loss o f grid and pole
slipping are two such abnormal operating conditions, Two problems can result following the loss o f grid.
both o f which can cause potentially damaging shaft The embedded generator may not be able to maintain
torques and disturbances to the power supply system. the voltage and frequency within acceptable limits,
This paper discusses a standard protection scheme and thus disrupting consumer loads which are still
highlights the operating conditions where standard connected to the power island. Additionally, the
schemes may not operate as expected. It then presents embedded generator is at great risk from an out o f
new protection techniques w hich have been developed synchronism re-closure, reconnecting it to the utility
and compares these to the standard schemes. supply.

Depending upon the degree that the two systems are

1.0 INTRODUCTION. out o f phase at the time o f re-closure, the resulting
shaft torques may cause mechanical shaft failure. It is
W hen a small embedded synchronous generator is therefore vital that the loss o f grid condition be
connected to operate in parallel with a utility detected before the utility circuit breakers are
distribution supply, the protective relay schemes re-closed. Since automatic re-closure schemes can
normally used for protecting large synchronous operate in as little as 0.5 seconds, the loss o f grid
generators are generally too expensive. The size and condition must be detected and the generator
hence cost o f an embedded generator dictates that a disconnected within 0.5 seconds.
low cost protection scheme is required.
1.2 Pole Slipping.
In the United Kingdom, the m i n i m u m protection
requirem ents for connecting a synchronous generator Pole slipping occurs when there is an imbalance
to operate in parallel w ith the distribution system are between the generator's input and output powers. This
detailed in 'Engineering recommendation G59' [1] and imbalance causes the generator rotor to accelerate, and
'Engineering Technical Report ET 113' [2]. These to try to operate at a speed which is above that o f the
docum ents stipulate that the generator be disconnected electrical supply to which it is connected.
from the utility supply whenever an operating
condition exists which will produce an unacceptable The imbalance can occur due to delays in clearing
deviation in the voltage or frequency at the point of short circuit faults, faulty excitation systems, faulty
supply. governing systems, or sudden losses o f load.

Loss o f grid and pole slipping can both cause The power surges associated with pole slipping can
unacceptable deviations in the supply voltage and have a detrimental affect on the quality o f supply of
frequency. the system to which the generator is connected. They
can also damage the generator. If the utility system
1.1 Loss o f Chid. does not have infinite bus characteristics, then as the
generator pole slips, the utility voltage will fluctuate.
Loss o f Grid protection is required to protect against If voltage sensitive loads such as induction motors are
an accidental isolation o f the utility supply and the connected nearby, the voltage depressions may cause
embedded generator systems. This isolation can result them to stall, depressing system voltage further.

M.J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 362

 Pole slipping can also cause rotor and stator over
heating due to the eddy currents which are produced
when the machine is operating asynchronously. The V rm s.k = —
v t is
Z 2.( .k -
.
r )
0)
currents and shaft torques experienced during pole r-0
slipping can exceed those produced by three phase where N is the number of samples per power system
fault levels. Since generators are generally only cycle, and Z is the voltage sample value.
designed to w ithstand three phase fault levels, the extra
forces produced during pole slipping may loosen the 2.2 Under/ Over Frequency Relays.
stator windings and cause permanent damage.
This protection operates when the measured frequency
exceeds the preset limits for the specified time. In this
2.0 A STANDARD PROTECTION SCHEME. study, the frequency was measured using a zero
crossing technique and linear interpolation to determine
There are several suggested protection schemes for the length of each h alf cycle.
embedded generation. The scheme used for this study
is based on G 59 and is shown in Figure 1. The The sampled values o f voltage used for the frequency
technical details are listed in Table 1. The time delays measurement were first pre-filtered w ith a two cycle
have been calculated w ith the assumption that it takes cosine filter. This removes dc offsets and any half
100 ms for the circuit breaker to operate. The multiples o f the fundamental frequency present in the
performance o f the relays shown was analysed using raw sampled data. A sampling rate o f 1000 H z was
protection algorithm versions of the functions in a used. Following the calculation o f the frequency, an
digital com puting environment. h alf cycle moving average filter was applied to remove
any high frequency components w hich may have
corrupted the rate o f change o f frequency
Local
<S> 25 C h ec k Sync
measurement.
27 U n d e r V oltage
2.3 Voltage Controlled Overcurrent Protection.
59 O v e r V oltage
46N N .P.S. O v ercu rren t.
51V H 46N ) 51V V oltage C o n tro lle d Voltage controlled overcurrent protection is necessary
O v ercu rren t. for synchronous embedded generators, due to their
8 1 U )r(8 1 0 810 O v e r F req u e n c y decaying fault current characteristics. This type of
81U U n d e r F req u e n c y protection has two characteristics, one long IDMT
RF RF R a te o f C h an g e o f characteristic for overloads or internal faults where the
F req u e n c y
terminal voltage is maintained, the other a faster IDMT
characteristic for faults which severely depress the
F ig 1 - S ta n d a r d E m b e d d e d G e n e r a to r P r o te c tio n S c h e m e terminal voltage. The standard inverse IDMT
characteristic was simulated using the RM S phase
Since loss o f grid and pole slipping are both balanced current as the actuating quantity. A reset characteristic
phenomena, the negative phase sequence protection was also modelled such that the integrand
function was not analysed. exponentially reset over a time o f five power system
cycles.
Table 1 - Protection Settings.
The voltage control characteristic was set such that if
PROTECTION SETTING DELAY the rms voltage dropped below 30 % o f its nominal
UNDER VOLTAGE - 10 % 0.4 seconds value, then the plug scale multiplier (PSM ) setting was
OVER VOLTAGE + 10% 0.4 seconds multiplied by a factor of 0.4.
UNDER FREQUENCY -4 % 0.4 seconds
OVER FREQUENCY + 1% 0.4 seconds 2.4 Rate o f Change of Frequency Protection.
VOLTAGE CONTROLLED PSM= 2 * In TMS = 0.1
OVERCURRENT
RATE OF CHANGE OF 0.3 Hz/s 0.25 seconds This protection operates when the rate o f change of
FREQUENCY frequency exceeds the preset limits for the preset time.
The rate o f change o f frequency was calculated using;

2.1 U ndercover V oltage Protection. (AF/At) = ( Fn - Fn ., ) * 1 /dt (2)

This protection operates when the relaying voltage where Fn is the present sample value o f frequency, and
exceeds the preset limits for the specified time. The Fn_, is the previous sample value o f frequency, dt is the
rms phase voltage which was used as the input to the time between samples. After the rate o f change signal
relay was derived from the instantaneous voltage had been calculated, a full cycle moving average filter
samples using; was applied to the signal to remove short fluctuations.

M.J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 363

3.0 THE NEW PROTECTION ALGORITHMS. (ATP). The test model contained a 625 kVA
synchronous generator connected to a utility
New types o f generator protection are increasingly distribution system through parallel 11 kV cables, as
using m icrocom puter based relays and integrated shown in Figure 2.
protection functions. Instead o f having one individual
relay for each function o f the protection scheme, one
Loss of E m bedded
microprocessor relay is used w hich contains all of the REC Grid B reakers G enerator
necessary protection functions required by the scheme. (150 MVA) (625 kVA)
There are num erous advantages to this approach, such
2 Km c a b le s 1 Km 11 kV c a b le s
as disturbance and event recording, SCADA interface,
self monitoring and alternative setting groups.
© - M---------------- *
Islanded Local
Load
New algorithm 's have been developed that take
advantage o f the processing capability provided by
these m icroprocessor relays, and offer the advantage of
improved protection perform ance over conventional Fig. 2 Embedded Generation System Used for Tests.
techniques.
The generator's inertia constant, H was 2.84 s, its open
Research has been conducted into two such algorithms circuit time constant , Tdo' was 1.9 s, its synchronous
for detecting loss o f grid and pole slipping [3,4]. The reactance, Xd was 2.25 p.u., whilst it's transient
new algorithm s developed, both use real time power reactance was 0.15 p.u. The impedance per km o f the
measurem ents and the rate o f change of power in their interconnecting lines was 0.0066 + j0.0069 p.u., and
algorithms. the transformer's reactance was jO.l p.u. All per unit
values have the generator rating as their base.
Using pow er m easurem ents offers advantages over
impedance m easurem ents since precise synchronisation When the generator is operating in parallel with the
o f the voltage and current samples and hence phasors utility system, its AVR is disabled. The AVR was not
is not required. Power m easurem ent is also more modelled in these simulations. The governor present
convenient than frequency measurem ent since it is on the generator was o f a slow acting type and hence
much less prone to input w aveform distortion and can would not have affected the simulations in the period
tolerate low sampling rates. o f interest.

In sharing the same m easuring quantities the 4.1 Simulation o f Loss o f Grid.
algorithm's also reduce the processing burden placed
on the microprocessor. Two different loss o f grid situations were simulated in
order to evaluate the standard protection scheme and
3.1 Power Based Loss o f Grid Protection. the new power based loss o f grid algorithm's
performance. For both tests, the generator was initially
This technique [3] detects the loss of grid by operating at 470 kW.
monitoring the behaviour o f the power system to load
disturbances. It detects a loss o f grid by differentiating For the first test, following the loss of grid, the load
between the responses produced w hen the embedded connected to the embedded generator was 940 kW.
generator is connected to the utility supply, and when The generator therefore experienced a 100 % increase
the em bedded generator operates in isolation. in loading when the loss o f grid breakers were opened.

3.2 Power Based Pole Slipping Protection. For the second test, following the loss o f grid, the load
connected to the embedded generator was 420 kW.
This technique [4] uses real power, it's rate o f change, The generator therefore experienced a 10 % decrease
and reactive pow er to predict when a generator is in loading following the loss o f grid. For both tests,
about to, or actually is pole slipping. It works by the loss o f grids occurred after 0.1 seconds from the
monitoring the operating point o f the generator to see start o f the simulation.
if the point o f no return has been passed. Once the
point o f no return has been passed, a pole slip is 4.2 Pole Slipping Simulations.
inevitable.
Two different tests were used to evaluate the standard
scheme and the power based pole slipping algorithm.
4.0 SIMULATION STUDIES. The generator was initially operated at full power.

The perform ance o f the protection schemes was For the first test, a three phase fault o f 255 ms
assessed using the Alternative Transients Program duration was applied at the term inals of the generator

MJ. Checksfiel d/PhD /University of Bath 1996 Appendix D, Page 364

 s te p u p t r a n s f o r m e r a t a t i m e o f 0 .1 s. T h i s c a u s e d th e U nder and o v e r v o lta g e re la y s w ill g e n e r a lly o n ly
g e n e r a t o r r o t o r to a c c e l e r a t e a n d t h e g e n e r a t o r to p o l e d e te c t a lo s s o f g rid i f th e re a l o r re a c tiv e pow er
s lip . T h e g e n e r a t o r w a s p e r m i t t e d to s l i p 6 p a i r s o f d e m a n d s re s u ltin g fro m t h e l o s s o f g r id a r e s e v e r e
p o l e s b e f o r e th e s i m u l a t i o n w a s s to p p e d . e n o u g h to c a u s e a n a p p r e c i a b l e e x c u r s i o n in v o l t a g e
T h e tim e d e la y o f 0 .4 s e c o n d s is u s e d s o t h a t th e
F o r t h e s e c o n d te s t , a t h r e e p h a s e f a u l t o f 2 3 5 m s p r o te c tio n c a n rid e th ro u g h d is tu r b a n c e s w h ic h c a u s e
d u ra tio n w a s a g a in a p p lie d a t th e t e r m i n a l s o f t h e s h o rt te rm v o lta g e flu c tu a tio n s .
g e n e r a to r s te p u p tra n s fo rm e r. T h is w a s n o t s u ffic ie n t
to c a u s e t h e g e n e r a t o r to p o l e s l i p , h o w e v e r it d id W i t h t h i s d r a m a t i c c h a n g e in l o a d f o l l o w i n g t h e l o s s o f
cause a la rg e s w in g in pow er o u tp u t fro m th e g rid , b o th th e u n d e rv o lta g e and ra te o f change of
g e n e ra to r. T h i s p o w e r s w i n g t e s t e d th e p r o t e c t i o n 's f r e q u e n c y r e l a y s tr i p p e d . A lth o u g h th e ra te o f c h a n g e
a b ility to re m a in s ta b le d u r in g la rg e p o w e r s y s te m o f f r e q u e n c y r e l a y is d e s i g n e d to o p e r a t e f o r a l o s s o f
d is tu rb a n c e s . g r i d , t h e u n d e r v o l t a g e r e l a y 's o p e r a t i o n s h o w s t h a t th e
q u a lity o f s u p p ly w a s c o m p ro m is e d .

5 .0 S IM U L A T IO N R E S U L T S . O b s e r v a tio n o f th e u n d e r f r e q u e n c y p lo t s h o w s w h y
t h i s p r o t e c t i o n d id n o t o p e r a t e . T h i s w a s d u e to th e
5 .1 L o s s o f G rid S im u la tio n s . g e n e r a t o r t e r m i n a l v o l t a g e d r o p p i n g to a v a l u e w h i c h
m a d e t h e e l e c t r i c a l lo a d e q u a l to t h e i n i t i a l l o a d o n th e
100 % L o a d In c re a se T e s t T h e r e s u lts fro m th is te s t g e n e ra to r o f 4 7 0 kW . O n c e th is c o n d itio n o c c u rre d
a r e p r e s e n t e d i n F i g u r e 3. F o r th is te st, th e p o w e r t h e g e n e r a t o r s t o p p e d s l o w i n g d o w n s i n c e its i n p u t a n d
b ased LOG trip p e d 33 ms a fte r th e lo s s o f g rid o u tp u t p o w e rs w e re m a tc h e d .
o c c u rre d . T h e ra te o f c h a n g e o f f r e q u e n c y trip p e d
0 . 3 4 4 s a f t e r t h e lo s s o f g r i d , a n d th e u n d e r v o l t a g e T h e v o l t a g e c o n t r o l l e d o v e r c u r r e n t a l g o r i t h m d id n o t
p r o t e c t i o n t r i p p e d a f t e r 0 .6 4 1 s. o p e ra te . A v e ry h ig h o v e r lo a d le v e l w o u ld b e re q u ire d
to m a k e th is p ro te c tio n o p e r a te , a t s u c h a h ig h le v e l o f
T h e in c re a s e d lo a d in g , c o u p le d w ith th e s lo w in g d o w n o v e rlo a d , th e u n d e r v o lta g e a n d fr e q u e n c y a lg o rith m s
o f t h e g e n e r a t o r c a u s e d t h e v o l t a g e to f a l l b e l o w t h e w o u ld d e fin ite ly o p e ra te .
u n d e r v o l t a g e t r i p le v e l.
1 0 % L o a d D e c r e a s e T e s t. T h e r e s u l t s f r o m t h i s te s t
c a n b e s e e n in F i g u r e 4. T h e fir s t p ro te c tio n fu n c tio n
TRIP BOUNDARY 3a) 'A' Phase Voltage RM S value
to t r i p w a s th e p o w e r b a s e d L O G , w h i c h o p e r a t e d 55
300
m s a f t e r th e e v e n t . A f te r th is , th e ra te o f c h a n g e o f
250
fr e q u e n c y tr ip p e d 0 .3 6 9 s a f te r th e e v e n t. T h e o v e r
200
f r e q u e n c y t r i p p e d la s t , 0 .8 8 3 s a f t e r t h e l o s s o f g r id .
4 150
100 TRIP
T h e s ta n d a rd s c h e m e s u c c e s s fu lly d e te c te d th e lo s s o f
3b) Frequency
g rid c o n d itio n w ith in th e tim e s p e c ifie d b y E n g in e e rin g
re c o m m e n d a tio n G 59. T h e p o w e r b a se d L O G o ffe re d
im p r o v e d p e r f o r m a n c e s i n c e it d e t e c t e d t h e l o s s o f g r id
49
c o n d itio n 0 .3 1 4 s b e fo re th e ra te of change of
f r e q u e n c y fu n c tio n .

3c) Rate o f Change of Frequency

V e ry c a r e f u l s e ttin g o f th e ra te o f c h a n g e o f fre q u e n c y
r e l a y is r e q u i r e d , a s t o o l o w a s e t t i n g w i l l r e s u l t in
m a l - o p e r a t i o n d u e to e v e r y d a y s y s te m d is tu rb a n c e s ,
w h ile to o h ig h a s e ttin g w ill re s u lt in sm a ll lo a d
c h a n g e lo s s o f g rid s g o in g u n d e te c te d .
TRIP
The change in lo a d in g o f th e g e n e ra to r w as not
3d) LOG Algorithm Output
su ffic ie n t to cause th e over v o lta g e a lg o rith m to
o p e ra te .

5 .2 P o le S l i p S i m u l a t i o n s .
3 -4
P o l e S li p T e s t T h e r e s u l t s f r o m t h e p o l e s l i p p i n g te s t
Time (s)
c a n b e s e e n in F i g u r e 5. T h e f i r s t p o l e s lip o c c u r r e d
b e t w e e n 0 .7 5 a n d 1 .2 5 s. T h e f i r s t p r o t e c t i o n f u n c t i o n

F ig u re 3 R e s u l t s to 1 0 0 % L o a d I n c r e a s e L o s s to t r i p w a s t h e r a t e o f c h a n g e o f f r e q u e n c y , a t a tim e

o f G r id S i m u l a t i o n . o f 0 . 3 5 7 s.

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 365

 TRIP BOUNDARY 5a) A Phase Voltage RMS value — TRIP BOUNDARY
4a) Frequency
250

100 TRIP

TRIP

4b) Rate of Change of Frequency

2
1
5c) Rate o f Change of Frequency

100

TRIP

•2
-100
4c) LOG Moving Average Value
| TRIP
0.5
5(1) Three Phase Real Power
1.5E+6 -0.5
1E+6
£ 5E+5

TRIP
TRIP
-2.5
000 0.50
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s) Time (s)

F ig u re 4 R e s u l t s to 1 0 % L o a d D e c r e a s e L o s s F ig u re 5 R e s u l t s to P o l e S l i p p i n g S i m u l a t i o n .
o f G rid S im u la tio n

I n s p e c tio n o f th e p lo t s h o w s th a t th is re la y re s p o n d e d T h e v o l t a g e p h a s e o f th e g e n e r a t o r a s it p o l e s l i p s
to t h e f a u l t , a n d o p e r a t e d b e f o r e th e f i r s t p o l e s lip . s w in g s th ro u g h 3 6 0 °, a n d th is a f f e c ts th e lo c a tio n o f
th e z e ro c ro s s in g s o f th e v o lta g e w a v e f o rm . T he net
T h e u n d e r v o lta g e p r o te c tio n tr ip p e d n e x t, a t a tim e o f re s u lt w a s t h a t th e lo c a tio n o f th e zero c ro s s in g s
0 . 5 0 2 s. T h i s a g a i n w a s n o t d u e to th e p o l e s l i p p i n g , m oved fo rw a rd s and b a c k w a rd s as th e p o le s lip
b u t th e s y s te m v o lta g e b e in g d e p re s s e d a fte r th e fa u lt o c c u rre d , re s u ltin g in th e o b serv ed freq u e n c y
w a s c le a r e d . flu c tu a tin g up and dow n. T h is flu c tu a tio n in th e
fre q u e n c y re p e a te d ly re set th e over freq u en cy
W hen a g e n e ra to r p o le s lip s , it's re a c tiv e pow er a l g o r i t h m a n d p r e v e n t e d it f r o m o p e r a t i n g .
dem and o s c illa te s w ild ly . W hen th e g e n e ra to r is
d e m a n d in g la rg e a m o u n ts o f re a c tiv e p o w e r, a v o lta g e If a freq u e n c y m e a su re m e n t te c h n iq u e had been
d e p r e s s io n re s u lts , w h ic h m a y c a u s e th e u n d e rv o lta g e e m p l o y e d w h i c h o n ly m e a s u r e d t h e f r e q u e n c y o f t h e
r e l a y to o p e r a t e . T h e u n d e r v o l t a g e a l g o r i t h m d i d n o t g e n e r a to r, th e n th e o v e r fr e q u e n c y a lg o rith m w o u ld
r e s p o n d to t h e p o l e s l i p p i n g in t h i s c a s e h o w e v e r , th e h a v e d e t e c t e d t h e p o l e s lip .
ti m e d e la y e n a b le d th e re la y to rid e th ro u g h th e
v o l t a g e d ip s . A lth o u g h th e c u r r e n t e x c e e d e d th re e p h a s e fa u lt le v e ls
d u rin g th e p o le s lip p in g , th e v o lta g e c o n tro lle d
T h e n e w p o w e r b a s e d a lg o rith m s u c c e s s fu lly trip p e d a t o v e r c u r r e n t a lg o rith m d id n o t o p e ra te . T h is w a s d u e
a t i m e o f 0 . 6 4 0 s, b e f o r e t h e p o l e s lip o c c u r r e d . The to t h e r e l a y b e i n g r e s e t b e t w e e n t h e c u r r e n t m a x i m u m s .
g e n e ra to r w as c o m m itte d to a p o le s lip , and as I f a lo n g e r r e s e t tim e h a d b e e n u s e d , a n d th e p o le s lip s
d e s i g n e d , t h e a l g o r i t h m p r e d i c t e d th e p o l e s lip b e f o r e had been fa s te r, th e n th e a lg o rith m m ay have
it o c c u r r e d . in te g ra te d th e c u rre n t o v e r se v e ra l s lip c y c le s and
tr i p p e d .
E x a m in a tio n o f th e freq u en cy p lo t s h o w s t h a t th e
f r e q u e n c y m e a s u r e m e n t ju m p s a t th e fa u lt o n a n d fa u lt T h e m o s t im p o r ta n t ta s k o f th e o v e r c u r r e n t re la y w ith
o f f p o in ts . T h i s w a s a r e s u l t o f th e zero c ro s s in g r e f e r e n c e to p o l e s l i p p i n g , is t h a t o f d i s c o n n e c t i n g th e
m e t h o d u s e d to m e a s u r e f r e q u e n c y , h o w e v e r , it d id n o t g e n e r a t o r fro m th e u tility s u p p ly w h e n a fa u lt o c c u r s
c a u s e th e r e l a y to t r i p b e c a u s e th e u n d e r a n d o v e r w h o se d u ra tio n is lo n g enough to c a u s in g p o le
f r e q u e n c y p r o t e c t i o n 's t i m e s e t t i n g e n a b l e d it to r id e s l i p p i n g . F o r t h e s e te s t s , th e f a u l t w a s c l e a r e d b e f o r e
th ro u g h th e g litc h e s th e o v e rc u rre n t re la y c o u ld o p e ra te .

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 366

 .... TRIP BOUNDARY 6a) A Phase Voltage RMS value T h e s ta n d a rd s c h e m e c a n h o w e v e r, m a l-o p e ra te d u rin g
300 s t a b l e s w i n g s in p o w e r f o l l o w i n g p o w e r s y s t e m f a u lts .
250 If th e g e n e r a to r s h o u ld b e d is c o n n e c te d fo r p o w e r
> 200 s y s te m fa u lts th e n th is is not a p r o b le m . I f th e
\7 ~ ~
S* 150
g e n e ra to r s h o u ld re m a in c o n n e c te d , th e n th e
Vol

u n d e r v o l t a g e r e l a y s e t t i n g s u s e d w o u l d h a v e to b e r e ­
o/* oo

c o n s id e re d .
V

| | TRIP
o

6b) Frequency O f m o r e c o n c e r n , is t h e r a te o f c h a n g e o f f r e q u e n c y
51.5
r e l a y , w h i c h m a y a ls o o p e r a t e d u r i n g s t a b l e s w i n g s in
p o w e r. T h is is due to th e o s c illa tio n s in th e
1 51 t
NO TRIP g e n e r a t o r 's v o l t a g e p h a s e a n g l e c a u s i n g t h e o b s e r v e d
/
g 50-5 / fre q u e n c y to o s c i l l a t e . It m a y b e p o s s ib le f o r th e
sr /
£ 50 — ■— — g e n e ra to r to s w in g fo r re a so n s o th e r th a n fa u lts ,
c o n c e r n m u s t t h e r e f o r e b e g i v e n to t h e o p e r a t i o n o f
49.5 6c) Rate of Change of Frequency s u c h r e la y s .
8
|6 f \ T h e tim e d e la y s a s s o c ia te d w ith th e s ta n d a rd s c h e m e
X 4
h a v e b e e n f o u n d to p r e v e n t s e v e r a l r e l a y s d e t e c t i n g
& 2 ___ _
p o le s lip p in g c o n d itio n s , w h ile n o t p r e v e n tin g re la y
-2 / TRIP | o p e r a t i o n d u r i n g p o w e r s w i n g s p r o d u c e d b y f a u lts . T h e
0.00 0.50 1.00 1.50 2.00 2.50 n e w p o w e r b a s e d p o le s lip p in g a lg o r ith m s u c c e s s fu lly
Time (s) d e te c te d th e p o le s lip p in g c o n d itio n w h ile r e m a in in g
s t a b l e d u r i n g t h e p o w e r s w in g .
F ig u re 6 R e s u l t s to P o w e r S w i n g S i m u l a t i o n .

P o w e r S w in g T e s t. The re s u lts fro m th e p o w e r 7 .0 REFEREN CES.

sw in g te s t c a n be seen in F ig u r e 6 N o n e o f th e
fu n c tio n s s h o u ld h a v e tr ip p e d fo r th is s im u la tio n s in c e [1 ] T h e E le c tric ity C o u n c il. " R e c o m m e n d a tio n s
th e g e n e r a to r w o u ld have re tu rn e d back to n o r m a l f o r t h e C o n n e c t i o n o f P r i v a t e G e n e r a t i n g P l a n t to th e
o p e ra tio n . E l e c t r i c i t y B o a r d 's D i s t r i b u t i o n S y s t e m s - E n g i n e e r i n g
R e c o m m e n d a t i o n G 5 9 ." J u n e 1 9 8 5 .
The u n d e rv o lta g e d id trip , th is w as due to th e
g e n e ra to rs re a c tiv e pow er dem and d e p re s s in g th e [2 ] T h e E le c tric ity C o u n c il. " N o te s o f G u id a n c e
s y s te m v o lta g e l o n g e n o u g h f o r it to o p e r a t e . fo r th e P ro te c tio n of P r iv a te G e n e ra tin g S e ts fo r
O p e ra tio n in P a ra lle l w ith E le c tric ity B o ard
T h e ra te o f c h a n g e o f f r e q u e n c y f u n c tio n a ls o o p e ra te d D is trib u tio n N e tw o rk s " E n g in e e r in g T e c h n ic a l R e p o rt
w h e n t h e g e n e r a t o r s w u n g f o l l o w i n g t h e f a u lt. The 113, L o n d o n 1989.
o s c i l l a t i o n s in r e a l a n d re a c tiv e p o w e r e x p e rie n c e d
d u rin g th e pow er sw in g caused th e freq u e n c y to [3 ] R ed fern M A, U sta O, & F ie ld in g G,
flu c tu a te a n d h e n c e th e ra te o f c h a n g e o f fre q u e n c y " P r o te c tio n A g a in s t L o s s o f U tility G rid S u p p ly F o r a
r e l a y to o p e r a t e . D is p e r s e d S to r a g e A n d G e n e r a tio n U n it" , P a p e r R e f 9 2
SM 3 7 6 -4 PW RD, P re s e n te d at th e IE E E Pow er
T h e o t h e r f u n c t i o n s , i n c l u d i n g th e p o w e r b a s e d p o le E n g in e e r in g S o c ie ty 1 9 9 2 S u m m e r M e e tin g , S e a ttle ,
s lip p in g f u n c tio n re m a in e d s ta b le . J u ly 12 - 16.

[4 ] R ed fern M A & C h e c k s f ie ld M J, "A N e w

6 .0 C O N C L U S IO N S . P o le S lip p in g P ro te c tio n A lg o rith m fo r D is p e rs e d
S to r a g e and G e n e r a tio n U sin g th e Equal A re a
T h e s ta n d a rd s c h e m e s u s e d fo r p ro te c tin g sy n c h ro n o u s C rite rio n ." , IE E E T r a n s a c tio n s o n P o w e r D e liv e ry ,
e m b e d d e d g e n e r a t o r s d e t e c t t h e m a j o r i t y o f l o s s o f g r id V o l. 1 0 , N o . l , p p 1 9 4 - 2 0 2 , Ja n u a ry 1995.
c o n d i t i o n s s a t i s f a c t o r i l y . T h e n e w p o w e r b a s e d lo s s o f
g r id a lg o rith m o ffers th e p ro sp e c t of enhanced
p e rfo r m a n c e o v e r th e s ta n d a r d s c h e m e , w h ils t b e in g
s im p le enough to be in c lu d e d in an in -e x p e n s iv e
in t e g r a t e d p r o t e c t i o n r e l a y .

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 367

 PAPER SEVEN

"A STUDY INTO A NEW SOLUTION FOR THE PROBLEMS

EXPERIENCED WITH POLE SLIPPING PROTECTION"

M A Redfem and M J Checksfield

To be presented at
IEEE Power Engineering Society (PES) Winter Meeting,
N ew York, USA , February 1997.

Accepted for Publication in the

IEEE Transactions on Power Delievery

Reference PE-486-PW RD-0-11-1996.

M.J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 368

 A STUDY INTO A NEW SOLUTION FOR THE PROBLEMS
EXPERIENCED WITH POLE SLIPPING PROTECTION.
M. A. Redfem and M. J. Checksfield

School of Electronic and Electrical Engineering

University of Bath, Bath, BA2 7AY, UK.

Abstract - Pole slipping o f synchronous generators operating in The new approach to pole slipping protection uses power
parallel with a utility supply system is a complex phenomenon. measurements to determine the status of the generator, whereas
This paper provides an overview o f a new solution for pole conventional schemes monitor the system's impedance. An
slipping protection, together with a review o f conventional immediate difficulty with the impedance techniques is that
techniques. Immediate advantages o f the new approach are that
they are difficult to set [1,3,4]. This problem is avoided with
it simplifies relay settings and avoids the need for extensive
the new relay. Most techniques require detailed power system
simulation studies. Results from simulation studies and field tests
are used to demonstrate the performance o f these systems. The
simulations to determine a reliable setting and performing a
new technique is shown to offer a significant improvement in detailed computer simulation is time consuming and therefore
performance over conventional schemes. expensive. Unfortunately most techniques for calculating the
impedance locus for pole slipping and stable swing conditions
Keywords: Power system protection, Pole slipping protection, do not take into account effects such as generator damping,
Out-of-step protection, Digital protection, Numeric protection. non-constant flux linkage, and control system operation[5].
This can therefore be a constraint on the choice of pole
I. IN T R O D U C T IO N slipping protection.

Pole slipping is a concern for all synchronous generation II. P O L E S L IP P IN G

designed to operate in parallel with utility systems. Recent
trends in power systems have meant that synchronous Pole slipping can be caused by an external disturbance on the
generators are being operated close to their stability limits[l]. system such as slow fault clearance, line switching operations
This together with the lower generator inertias, higher machine which increase the transfer impedance, a faulty excitation
reactances and the long fault clearance times especially system, or an excess of prime mover input power. The
encountered on distribution systems, increases the likelihood phenomenon is likely to occur when the generator is under
of synchronous generator pole slipping. excited, since this reduces the maximum electrical power
which can be taken out of the generator. Unfortunately,
Pole slipping causes dramatic fluctuations of the generator's generators are occasionally operated in an under-excited
currents and as a result can cause fluctuations in power system manner when the system is lightly loaded to prevent high
voltages. The mechanical forces and heating effects produced system voltages.
during pole slipping can damage the machine and associated
transformers. Pole slipping protection is required to initiate From a machine view point, pole slipping occurs when the
disconnection of the generator from its supply as soon as mechanical torque produced by the prime mover is greater
possible after it loses synchronism with the utility supply to than the electromagnetic torque used to produce the power
which it is connected. output. If this condition persists for too long the rotor is
unable to stay in synchronism with the stator magnetic flux.
The loss of synchronism point can be defined as the point
where the generator's real power output becomes zero. Once
this point has been reached the generator will briefly accelerate
into a motoring condition, during which time it will absorb
power from both the prime mover and the system. It will then
return to a generating condition. If it cannot then produce the
synchronising forces necessary to pull itself back into a
synchronous state of operation, it will repeat the cycle again
and again.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 369

 P o le s lip p in g c a n d a m a g e th e g e n e r a t o r in s e v e r a l w a y s . I f it u p to th e C r itic a l S ta b ility P o in t, C SP . I f th e f a u lt c le a r a n c e
is c o n n e c te d to a h ig h c a p a c ity in f e e d , th e c u r r e n ts e x p e r ie n c e d tim e is s lig h tly lo n g e r, th e n a r e a 'A ' w ill b e g r e a te r th a n a re a
c a n e x c e e d th r e e p h a s e f a u l t le v e ls . M a c h in e w in d in g s a re 'B ', a n d th e lo a d a n g le w ill s w in g p a s t th e C S P a n d in s ta b ility
g e n e r a lly d e s ig n e d to w ith s ta n d th r e e p h a s e f a u lt le v e ls . T h e s e a n d h e n c e p o le s lip p in g w ill o c c u r . T h e a lg o r ith m p r e d ic ts th a t
h ig h c u r r e n ts c a u s e d a m a g e b o th t h r o u g h m e c h a n ic a l f o r c e s o n p o le s lip p in g w ill o c c u r b y d e te c tin g i f th e lo a d a n g le is s till
th e c o n d u c to rs and o v e r h e a t in g . The g e n e r a to r 's u n it in c r e a s in g once th e C SP has been ex ceeded [2]. I f th e
tr a n s f o r m e r w in d in g s a re a ls o e x p o s e d to h ig h m e c h a n ic a l g e n e r a to r o p e r a te s p a s t th e C S P , s ta b ility c a n n o t b e m a in ta in e d
s tr e s s e s a n d o v e r h e a t in g a n d c a n th e r e f o r e b e d a m a g e d a s w e ll. a n d a p o le s lip is in e v ita b le .
Pre-Fault
P o le s lip p in g a ls o p r o d u c e s s e v e r e p u ls a t io n s in s h a f t to r q u e , Power Power
Curve
a n d i f th e s p e e d o f th e p o le s lip is s u c h th a t th e p u ls a t io n s Critical
e x c ite r e s o n a n c e in th e s h a f t, s h a f t f a ilu r e m a y re s u lt. T h e Stability
m a g n itu d e o f th e s h a f t to r q u e p u ls a t io n b u ild s u p i f th e Point
g e n e r a t o r is a llo w e d to c o n tin u e to p o le s lip , a n d th e g e n e r a t o r Mover (CSP)
s h o u ld t h e r e f o r e b e tr i p p e d im m e d ia te ly w h e n a p o le s lip
Post Fault
o c c u r s [1,3], Power
Power
Curve
T h e h ig h c u r r e n ts a s s o c ia t e d w i t h p o le s lip p in g c a n c a u s e
Fault-
v o lta g e f l u c tu a tio n s to th e lo c a l s u p p ly w h ic h in tu r n c a n c a u s e D uradon
m o to r s ta r te r c o n ta c to r s to o p e n a n d in d u c tio n m o to r s to s ta ll, Power
fu r th e r d e p r e s s in g th e s y s te m v o lta g e . In s e v e r e c a s e s , p o le Curve
s lip p in g o f a g e n e ra to r c a n le a d to a c a s c a d in g lo s s o f
180 L o a j
s y n c h r o n is m o f m a c h in e s n e a r b y .
Angle
Fig. 1 Equal Area Diagram Showing Critical Stability Point ( CSP)
III. N E W P O L E S L IP P IN G P R O T E C T IO N
for Clearance of a Fault on a Double Circuit Transmission Line

A n e w p o le s lip p in g a lg o r ith m h a s b e e n d e v e lo p e d [ 2 ] to ta k e In d e s ig n i n g a p o le s lip p in g a lg o r ith m , it m u s t b e a b le to

a d v a n ta g e o f th e c a p a b i l i t i e s p r o v id e d b y m o d e m n u m e r ic p r e d ic t w h e n a g e n e r a t o r is c o m m itte d to a p o le s lip , o r d e te c t
m ic r o p r o c e s s o r re la y s . U s i n g r e a l tim e p o w e r m e a s u r e m e n ts w h e n a p o le s lip is a c tu a lly o c c u r r in g . It m u s t a lso r e m a in
a n d th e ra te o f c h a n g e o f p o w e r , th e a lg o r ith m trip s w h e n th e s ta b le f o r a ll o th e r p o w e r s y s te m c o n d itio n s , in c lu d in g s h o rt
g e n e r a t o r is c o m m itte d to a p o le s lip , r a t h e r th a n w h e n a p o le c ir c u it f a u lts a n d r e c o v e r a b l e p o w e r s w in g s . T h e a lg o rith m
s lip h a s a c t u a lly o c c u r r e d . T h e n e w a p p r o a c h c a n th e r e f o r e u s e s r e a c t iv e p o w e r , r e a l p o w e r , a n d r a te o f c h a n g e o f re a l
d is c o n n e c t th e m a c h in e fr o m th e p o w e r s y s te m b e f o r e it c a n p o w e r to d e te c t o p e r a tio n p a s t th e C SP . R e a c tiv e p o w e r is
b e d a m a g e d , o r s e v e r e v o lta g e d is tu r b a n c e s a re c a u s e d . u s e d to te ll i f th e g e n e r a t o r is o p e r a tin g a t lo a d a n g le s o f
g r e a te r th a n 9 0 °. The tr ip c r ite r ia a re :-
The a lg o r ith m u s e s th e g e n e r a t o r p a r a m e te r s , X q - th e
q u a d r a tu r e a x is s y n c h r o n o u s r e a c t a n c e , X d - th e d ir e c t a x is T R IP W H E N :-
tr a n s ie n t r e a c t a n c e , a lo n g w ith th e g e n e r a t o r r a tin g , S g e n , a n d A ll m u s t
Q < Q trip
g e n e ra to r o p e r a tin g p o in t to d e r iv e th e trip le v e ls . By b e s a tis f ie d
and
d y n a m ic a lly a d ju s t in g th e tr ip s e tt in g s a c c o r d i n g to th e c u r r e n t c o n tin u o u s ly
P < Pt (I)
o p e r a tin g p o in t, th e a lg o r ith m c a n u s e s e n s itiv e trip le v e ls to and f o r 1.5 pow er
q u ic k ly d e te c t p o le s lip s w h i c h o c c u r d u e to s te a d y s ta te o r (AP I A t) < (A P I A t) s y s te m c y c le s .
d y n a m ic in s ta b ility . D u r in g t r a n s i e n t d is tu r b a n c e s , la r g e r trip
le v e ls a re a u to m a tic a ll y u s e d , e n s u r i n g th a t th e r e a re n o fa ls e W h e r e Q is th e th r e e p h a s e r e a c t iv e p o w e r , P is th e th re e
tr ip s d u r in g s ta b l e p o w e r s w in g s . p h a s e re a l p o w e r , a n d (A P \A t) is th e ra te o f c h a n g e o f re a l
p o w e r. T h e r e a c t iv e p o w e r trip le v e l, Q trip is c a l c u la te d fro m
F ig . 1 s h o w s th e E q u a l A r e a D ia g r a m c o m m o n ly u s e d to X q , a n d is e n te r e d a s a n e x te r n a l s e ttin g ;
a s s e s s s ta b i lity f o r a g iv e n d i s t u r b a n c e ^ ] . T h e d is tu r b a n c e in
th is c a s e is th e c le a r a n c e o f a s h o r t c i r c u i t f a u lt o n o n e o f tw o Q trip = - *Sgen (2)
Xq
tr a n s m is s io n lin e s w h ic h c o n n e c t th e g e n e r a t o r to th e m a in
u tility s u p p ly . T h e f a u lt is c l e a r e d a t lo a d a n g le 8 2 . The T h e re a l p o w e r tr ip le v e l, P t is d e r iv e d b y th e a lg o r ith m , an d
l im it in g c a s e f o r s ta b ility is w h e n a re a 'A ', th e a c c e le r a tin g is a u to m a tic a lly a d ju s te d a c c o r d i n g to th e g e n e r a t o r o p e r a tin g
e n e r g y in to th e r o to r is e q u a l to a r e a 'B ', th e d e c e le r a tin g p o in t. T h e ra te o f c h a n g e o f p o w e r trip s e ttin g , (A P /A t)min is
e n e r g y . W h e n th is c o n d itio n o c c u r s , th e lo a d a n g le w ill s w in g a lso c o n tin u o u s ly a d ju s te d b y th e a lg o r ith m a c c o r d i n g to th e

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 370

generator operating point. Since Pm ax is a measure of the maximum output power that
the generator has produced in the last second, it enables the
The 1.5 power system cycle time constraint is introduced to algorithm to adjust its setting according to whether a 'transient'
ensure that the algorithm remains stable during short circuit or 'steady state' pole slip will occur. A 'steady state' pole slip
faults. The majority of faults will not satisfy all of the trip occurs due to control system malfunction, or loss of steady
criteria, but the few that do only cause the criteria to be state stability, whereas a 'transient' pole slip occurs after the
satisfied for less than one power system cycle. generator has been severely disturbed, after a fault for
example. If the generator has been transiently disturbed, its
A - C o n d itio n M o n ito r in g T e c h n iq u e s output power during the first swing of its load angle, be it
stable or unstable, will be much greater than its nominal
The algorithm is designed for a commercially available output. This results in a greater magnitude in (AP/At) at the
microprocessor based relaying environment, and hence the CSP, the algorithm therefore needs to adjust its setting
only information available to the relay is 2 phase voltages and (A P /A t)^ accordingly. The algorithm adjusts this setting
3 current signals provided by the relay's transducers. Since using;
generators may be operated over a wide range of output I f ... < 0.6;
Pmax
powers, a method of estimating the prime mover input power (5)
is required so that the real power trip setting, Pt can be then (AP/At) . =(AP/At). +(AP/AtY
dynamically derived. The input power, Pm is estimated by
observing the power output of the generator, and if this where (A P /A t)^ is an external relay setting based on the
remains within a pre-defined tolerance band for a sufficient degree of transient saliency in the generator. Transient
amount of time, then the input power is approximately equal saliency arises due to differences in the values o f X d ’ and X q 1.
to the output power, the only discrepancy being produced by Normally X q ' is equal to Xq. (A P /A t)^ is given by;
the losses in the generator. Pt is derived from;
(APIA* - i - L - L )’ * s P (6)
</ I*. --V*o I and < P“ i ; Xq Xd'
(3)
For a steady state pole slip, no peak will occur in the power
then Pt = Pn *Pfact
output before the pole slip, the algorithm trip setting should
where Pn is the present real power sample value, Pn_^ is the therefore be at its most sensitive level to detect the pole slip
sample value of 'del samples ago, Pn _ (2*dd) *s sample since low values of (AP/At) will result.
value o f '2 *del samples ago, Ptol is the tolerance band, chosen
as nominally 5% of machine rating, and Pfact is a coefficient During steady state conditions the value of Pmax used by
to take into account the losses and inaccuracies inherent in equations (4) and (5) is nominally 1.4 * Pt. The ratio Pt/Pmax
measuring machine input power from output power. therefore is 0.71, and Pt/Pmax > 0.6. (A P /A t)^ is therefore
not included in the calculation of (A P /A t)^ during steady state
The protection algorithm is also designed to be conditions. If the generator's power output rises above the
accommodated in the finite processing resources of a multi­ existing value of Pm ax for greater than one power system
function microprocessor relay and is executed 4 times per cycle, then the value of Pmax is updated to the new maximum
power system cycle. The value of 'del typically chosen is 200 for a duration of one second. The one second limit is used so
which equates to 1 second, Pfact is nominally set to 0.9. that the new value of Pm ax is only used for the period where
a pole slip or power swing associated with that value is likely
B - A d a p tiv e T rip S e ttin g (A P /A t)mjn to occur. The one cycle constraint is necessary so that
incorrect values of Pm ax do not result from spurious spikes on
The adaptive trip setting, (AP/At)min is the theoretical value the power waveform. This mechanism also de-sensitises the
of (AP/At) at the CSP. The algorithm calculates this value algorithm during the period where generator output power is
using an assumed sinusoidal power load angle curve[2], which being increased.
gives;
IV . C O N V E N T IO N A L P O L E S L IP P IN G P R O T E C T IO N
(AP/Af)min - - s *P m ax ♦cos (180 -sin _/(—^ —) ) (4)
Pmax
A - G e n e r a to r P o le S lip p in g Im p e d a n c e C h a ra c te ristic s.
Note that the value of slip, s used in (4) will not be the exact
value at which the relay produces a trip output, due to Conventional methods of detecting pole slipping use
damping power and other effects not taken into account in the impedance relays to monitor the variations in apparent
derivation of (4). This equation is used to continuously update impedance at the generator or its unit transformer terminals
the trip setting (A P /A t)^ depending on the variables Pmax during the pole slip. The theoretical variations in impedance
and Pt. during pole slipping were studied by Edith Clarke[7] and are

M.J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 371

simplistically shown in Fig. 2. These loci are derived using the B - MHO Relay Scheme.
assumption that the ratio of generator to source voltages,
Eg/Es remains constant. The analysis also assumes that This is the simplest type of impedance based pole slipping
generator saliency is neglected, generator damper effects are protection and uses a single MHO characteristic[l,3]. The main
neglected, transient fault impedance effects have decayed, advantages are that tripping occurs before 8= 180°, and hence
shunt loads and shunt capacitance effects are ignored, effects before the most damaging current conditions. It also provides
of automatic voltage regulators and governors are neglected, back-up protection for faults occurring in the transformer and
and the source voltages behind their equivalent impedances are generator. Careful selection of the MHO circle's reach is
sinusoidal and at fundamental frequency [1,3,5,7]. required to avoid nuisance tripping during recoverable and
hence stable swings.

generator & line infinite The relay is normally set to cover the generator transient
source
generator transformer bus reactance, X d ' plus the reactance of the unit transformer. It is
. impedance impedance impedance
recommended that the angle of swing where the impedance
locus enters the characteristic is at least 120° [3].

Eg/5 E s /0 C - Single Blinder Scheme.

Eg/Es > 1 This is the most secure of the impedance based techniques
Zs
for detecting pole slipping and uses a pair of blinder
characteristics as well as a supervisory MHO circle[l,2,3]. The
MHO supervisory element restricts operation to swings
► appearing inside the generator and its transformer. For the
Eg/Es = 1 scheme to operate, the impedance locus must enter the MHO
circle and its trajectory must take longer than a specified time
to cross from one blinder unit to the other. This time
constraint is typically set to two power system cycles.
Eg/Es < 1
The scheme has the advantage over the MHO scheme that
the diameter of the supervisory MHO element can be increased
pole slipping loci in order to provide better sensitivity to pole slipping conditions
G
without introducing nuisance tripping for stable swings.

D - Double Blinder Scheme.

Fig. 2 Classical Pole Slipping Impedance Loci.[7]
The double blinder scheme uses two pairs of blinder
When a pole slip occurs and the two generator voltages are characteristics together with a supervisory MHO element[l,3].
the same, the load angle 8 increases causing the impedance to For the scheme to operate, the MHO element must first pick
travel from right to left across the R-X plane. When the load up, and then the locus must then stay between the inner and
angle reaches 180°, the loss of synchronism point has been outer blinders for greater than a preset time. It must then stay
reached and the impedance will lie on the system impedance between the inner blinder elements for greater than a second
line GS. This is known as the system centre, and at this point time setting. Finally when the locus leaves the inner
the impedance presented to the generator appears to be a three characteristics, it must stay between the inner and outer blinder
phase fault. As the pole slip cycle is completed and 8 elements for greater than a preset time. When all of these
approaches an in-phase value of 360°, the impedance locus stages have been satisfied, a trip signal is issued when the
will travel to the left of the system impedance line before supervisory MHO resets.
returning to the load area of the R-X plane.
The main difference between the single and double blinder
Fig. 2 also shows impedance loci for the cases where the schemes is that with the double blinder scheme, the locus can
ratio Eg/Es < 1 and Eg/Es > 1. The effect is to cause the enter and leave the characteristic from any direction. With the
impedance locus to take on a circular arc form. These curves single blinder scheme the locus must enter from one side and
show that as the generator internal voltage is increased the leave from the other for a trip to occur. The inner elements of
system centre is pushed away from the generator. AVR field the double blinder scheme must therefore be set so that they
forcing action has this effect. will only respond to non recoverable swings.

M.J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 372

 V. SIMULATION STUDIES the machine. Although contrived, this scenario is not unknown
and would result from a generator control system malfunction
The performance of the new and conventional protection or inexperienced operators.
schemes was investigated using computer simulations of a 200
M VA generator using the Electro-Magnetic Transients Program VI. DISCUSSION OF SIMULATION STUDIES
(EMTP) [6] and field trial tests using an industrial 625 kVA
diesel generator. A . C o m p u te r B a s e d S im u la tio n S tu d y R e su lts.

The three phase impedance used to demonstrate the An extensive series of simulation studies was conducted
impedance based relays' responses to the tests was derived including both pole slipping tests and recoverable or stable
from three phase real and reactive power, and voltage signals swings of the system. For these tests, the power based
with a moving average filter to reproduce the averaging algorithm operated for all of the pole slipping tests and
techniques used in an impedance relay. successfully restrained for all stable swing tests. The MHO
pole slipping scheme operated for all of the pole slipping
90 mile generator
simulations, but it also operated for 30 % of the stable swing
thevenin equivalent distributed transformer
source (33.7 GVA) parameter line. (3.33 % p.u) tests. The single blinder scheme operated for 75 % of the pole
slipping tests, and successfully restrained for all of the stable
swing tests. The double blinder scheme also operated for 75 %
transformer fault clearing of the pole slipping tests, but also operated for 15 % of the
(10% p.u) breaker 200 MVA
stable swing tests.
synchronous
machine
load
235 MW X
i 3.75 mile A .l Power Based Pole Slipping Protection Scheme
switch j 230 kV line.
13 MVAr
Fig. 4 shows the algorithm's response to a classical pole slip,
for which it trips after 1.52 s. The pole slip was induced by a
Fig 3 EMTP Power System Model. two phase fault which was cleared at a time of 1.495 s. The
algorithm detected the pole slip 25 ms after fault clearance.
A . C o m p u te r B a s e d S im u la tio n S tu d ie s. Note that the units are in relay secondary quantities, the
secondary rating of the generator was 190 VA.
The EMTP used the power system model shown in Fig. 3.
This model was based on the EPRI transient stability study T R IP B O U N D A R Y G enerator R eactive Pow er
model[8]. Stable swings were generally initiated by using a 300
fault whose duration was just short of the critical clearing time | 200
of the generator. This produces the most severe stable swing 100
possible. Pole slips were triggered using a range of faults
whose duration was longer than the critical clearing time. -100
-200
B. 6 2 5 kV A D ie s e l G e n e r a to r F ie ld T ria l Test. -300 Generator Real Power
400
In order to demonstrate the algorithm's operation in anger,
200
limited tests were conducted using a diesel generator. The
625 kVA industrial generator was connected to the utility
network via a 415/11,000 volt unit transformer and 2 km of
w -200 1521 ms
11 kV cable. The utility infeed capacity at the cable
Rate of Change of Real Power
termination was 150 MVA and the local load connected to the
generator was approximately 500 kW.
£I
^ -5000
Performing practical pole slipping tests on a generator was
particularly difficult because of the damage pole slipping could
i graph limited to
go -10000
cause. The operator would not allow the use of short circuit cVw improve clarity
faults on the power system to induce pole slipping. A steady -15000
0.000 500 996 1495 2000 2500
state pole slip was therefore induced by increasing the diesel Time (ms)
engine's output torque with the generator under-excited for one Fig. 4 Power Based Pole Slipping Algorithm Response
test as shown later. This produced a 2.5 MW power surge in to 200 MVA Generator Pole Slip.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 373

 The algorithm's ability to remain stable during a stable swing leaves temporarily at 1.11 s. It then re-enters the circle and
is shown in Fig. 5. Following the disturbance, parts of the trip remains inside the characteristic for a total time of 140 ms.
criteria could be satisfied, but not all of them together and The scheme tripped at 1.16 s.
hence the algorithm did not trip. At the 1.0 s point, the rate
of change of power satisfies its trip level, as does the reactive If the scheme's time delay had been increased to 140 ms, the
power, but the algorithm does not trip because the real power nuisance tripping would have been avoided, but the scheme
is above its trip level. At 1.4 s, the rate of change of power would fail to detect fast pole slips, whose locus may travel
satisfies its trip level once more, as does the real power, but through the relay characteristic in less than 140 ms.
here, the reactive power is above its trip level. By the time 0.2
the reactive trip criteria is once again satisfied, the rate of
change of power trip level is not satisfied and the algorithm 0.57
(Fault Ch
therefore remains stable. &line

TRIP
■TRIP BOUNDARY Generator Reactive Power (p.u.) 1.07,
0
&
£
8 1.16 1.1 I
& 0.5 - 0.1
Xd
S 1.28
-0.5 et
135
- 0.2
Generator Real Power (p.u)
2 mho
ciaracteristi
-0.3
Pmax - 0.2 - 0
0.1 0.1 0.2 0.3
Resistance, R (p.u.)
NO TRIP Fig 6 MHO Scheme Characteristic for
-0.5 200 MVA Generator Stable Swing Simulation.
Rate of Change of Real Power (p.u.)
A.3 Single Blinder Scheme

The settings for the single blinder scheme were chosen so

that the theoretical swing locus crossed the right and left
gM o blinders at 100 and 260° respectively. The MHO was set so
et -15 graph limited to
improve clarity that the scheme would operate for pole slips appearing on the
-20 generator side of the high voltage transformer terminals. The
-25 minimum permitted travel time between left and right blinders
0.00 100 Time (s) 200 3.00
for this scheme to operate was set to 40 ms, two cycles at
Fig. 5 Power Based Pole Slipping Algorithm Response 50 Hz.
to 200 MVA Generator Stable Swing.
A. 2 MHO Scheme Fig. 7 shows the single blinder scheme characteristic and the
impedance locus for a typical pole slipping condition. The pole
The MHO characteristic reach was set such that the slip was caused by a 3 phase fault, with the generator initially
theoretical swing locus entered at 120°. It therefore included at 50 % load, and both the AVR and governor were modelled.
all of the generator's impedance, but none of the transformer's.
A time delay of 20 ms was used to increase the stability of the As a result of the fault, the impedance locus travels between
scheme. the two blinder units and moves down the impedance plane
towards the origin, this can be seen at 0.27 s on Fig. 7. When
Fig. 6 shows the MHO characteristic and the impedance the fault is cleared, the impedance enters the supervisory MHO
trajectory for a stable swing test which resulted in a nuisance characteristic as the generator is part way through its first pole
trip. This swing was set up using a 2 phase to earth fault, with slip cycle, this corresponds to 0.94 s in Fig 7. The first stage
the generator initially at 50 % load, its governor was included, of the single blinder logic is not satisfied, since the impedance
but the AVR was modelled as being on manual control. locus must first enter the MHO circle before crossing the
Immediately after the fault was cleared, at 1.07 s, the blinder unit. If the fault had been cleared when the system
impedance trajectory jumps into the MHO circle, and then angle was at a lower value, then the impedance would have

M.J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 374

jumped to the right of the right hand blinder and the scheme single blinder scheme. The double blinder scheme requires a
would have operated successfully. After fault clearance, the complete pole slipping cycle to operate, and could therefore
impedance locus then moves to the left of the R-X plane, and fail to operate for the first pole slip following a close up fault.
leaves the MHO circle just before 1.12 s. The single blinder
scheme therefore did not trip for the first pole slip. Fig. 8 shows part of the double blinder characteristic and an
impedance locus for a stable swing produced by a 3 phase
For the series of tests, the scheme failed to trip for 25 fault. The generator was initially operating at 100 % load, and
percent of the pole slipping conditions. In all of these tests, the the AVR was included in the model, but not the governor.
generator was initially operating at low power, and as a result
only slipped one pair of poles. The scheme therefore did not The inner elements of a double blinder scheme should be set
detect these conditions as it requires a complete pole slipping so that only non-recoverable swings should cross them [1], and
cycle to operate. This increases the potential for damage to the they were therefore set to 130°. To reduce the stress on the
generator, since for pole slips following faults, the scheme breaker, tripping is delayed until the supervisory MHO element
may require two complete pole slipping cycles before tripping. resets. The MHO characteristic was therefore set to a value of
This may be seen as desirable in some instances, however it 85°. The outer blinder elements were then set so that the
is more flexible to have a facility built into the relay which swings with a ratio of Eg/Es of up to 1.5 would be detected.
delays tripping until the second pole slip cycle if required. The timing constraints were calculated assuming the maximum
mean rate of pole slipping to be 8 Hz.
0.4
lelt
0.2
blinier right right
inner outer
0.27 >linder blinder
0.2

1.28 supsrvisoiy
0 / mho

1.62 1.49 0.94 0.52

- 0.2
1.59
.57
0.85
supervisory 0.70
mho
-0.4
-0.4 - 0.2 0 0.2 0.4 - 0.1
Resistance, R (p.u.) 0 0.1 0.2
Resistance, R (p.u.)
Fig 7 Single Blinder Scheme Characteristic
for 200 MVA Generator Pole Slip. Fig. 8 Double Blinder Scheme Characteristic for 200 MVA
Generator Stable Swing Simulation.
The impedance trajectory for the second pole slip cycle
enters the diagram at 1.28 s as shown in Fig. 7, and then As shown in Fig. 8, the fault is cleared at 0.52 s. The
enters the MHO circle and crosses the right hand blinder unit, impedance trajectory then travels from inside the two inner
sealing in the first stage of the single blinder logic. The elements to the outside of the right outer blinder element. At
impedance locus then travels across from right to left and 0.57 s, the supervisory MHO is picked up, but the two right
crosses the left blinder unit, causing the scheme to trip at a blinder units are not. The locus then heads back to the left and
time of 1.59 s. The scheme therefore successfully trips for the crosses both of the blinders with a time delay of approximately
second pole slip. 0.13 s. This satisfies the first stage of the double blinder logic.
The locus then stays in between the inner blinder elements for
A.4 Double Blinder Scheme long enough to seal the next stage of the scheme's logic. It
then crosses back over the two right hand blinder units in
The double blinder scheme also failed to trip for 25 percent 0.12 s, satisfying the final stage of the scheme logic, and a trip
of the pole slipping simulations for similar reasons to the signal is produced at a time of 0.85 s.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 375

B. 625 KVA D iesel G en erator P ole Slip Test. The conventional impedance based schemes were set so that
they would operate for swings appearing on the generator side
The performance of the generator and the new power based of the transformer high voltage terminals for a range of values
algorithm was monitored by the disturbance recorder function of Eg/Es from 0.66 to 1.5. The value of X d' for the generator
included in the microprocessor relay package together with a was 0.22 p.u., the transformer reactance was 0.03 p.u. The
second package. During the test the operator permitted the MHO scheme was set so that the swing locus for a value of
generator to complete two entire pole slipping cycles before Eg/Es of 1 entered the characteristic at an angle of 120°. If the
reducing the diesel engine torque to allow the generator to characteristic had been set to reach from the transformer high
pull back into synchronism. voltage terminals all the way into the generator, the theoretical
locus would have entered at a value of 102°, making the
In the most dramatic of the tests, the new power based pole scheme liable to operate for recoverable power swings.
slipping algorithm successfully detected both of the pole slips,
however, analysis showed that the impedance based schemes The MHO scheme's failure to trip further highlights the
all failed to trip. difficulty in using it. Covering the whole of the generator Xd'
"TRIP BOUNDARY Generator Reactive Power can lead to too large a MHO characteristic, and this makes the
scheme prone to nuisance tripping. Adding a time delay
increases the stability of the scheme to recoverable swings, but
-500,000 as the time delay is increased, the scheme may fail to detect
fast pole slips whose locus travels through the characteristic
^ - 1, 000,000
quickly. If the circle size is decreased in order to increase
stability, the scheme can fail to detect pole slips whose
-1,500,000
impedance does not take the classical trajectory.
Generator Real Power

2 ,000,000
The single blinder scheme was set using the recommended
Pmax blinder settings of 90° and 210°. The supervisory MHO circle
1,000,000
was set so that theoretical swings appearing in the transformer
and generator could be detected. The scheme characteristics
I
a, and the impedance locus of the pole slipping generator are
TRIP, TRIP shown in Fig. 10. It must be noted that the time records are
- 1,000,000 not the same as used in Fig. 9 since the disturbance recorders
Generator Rate of Change of Real Power were not synchronised.
20 ,000,000
The first pole slip occurred between 0.2 and 0.3 s, and the
second occurred between 0.5 and 0.6 s. The first pole slip was
ft -20,000,000 not detected because the impedance trajectory did not enter the
-to,000,000 supervisory MHO characteristic. However, even if the
supervisory MHO had been made larger, the scheme would
50 550 1050 1552 still have failed to trip since the trajectory did not cross the
Time (ms)
left hand blinder. For the second pole slip, the swing
Fig. 9 Power Based Pole Slipping Algorithm Response
impedance travelled across the impedance plane from right to
to 625 kVA Diesel Generator Pole Slip.
left as predicted by theory, but again it did not enter the
Fig. 9 shows the power based pole slipping algorithm's supervisory MHO characteristic. This was due to the speed at
response to this test. The real power plot shows where the which the generator pole slipped. If the generator pole slips
algorithm tripped, which for the first pole slip occurred just at a slow speed, then the impedance of the generator is much
before 1000 ms on the record. Allowing for a circuit breaker larger. At a slip of 0.33 %, the effective impedance of the
operating time of 100 ms, this would have isolated the generator becomes 2*Xd' [9],
generator before the pole slip. The real power plot also reveals
the potentially damaging effects of pole slipping, since the The double blinder scheme suffered similar problems to the
peak in power output after the first pole slip reaches 2.7 MW, other impedance based techniques and failed to trip.
five times rated power. By tripping before the pole slip
occurred, the new algorithm would have avoided this A loss of field protection included in the relay did however
dangerous surge in the power output. Due to the finite trip. Its MHO characteristic was set to a value of X d + Xd'/2,
disturbance recording time, the transition from normal to which for this generator was -j 2.26 pu. This relay had a 0.5 s
underexcited generator operation was not captured. delay.

M.J.Checksfleld/PhD/University o f Bath 1996 Appendix D, Page 376

 VIII. ACKNOWLEDGEMENTS
N O T R IP left right
blinder Hinder
The authors are pleased to acknowledge the help and
0.44 encouragement provided by the University of Bath, GEC
Xd
ALSTHOM Protection and Control and the Engineering and
Physical Science Research Council.
1.23
o -0.5 IX. REFERENCES
0.62 0.29
[1] D.W. Smaha, "Out-of-Step Relay Protection o f Generators,"
IEEE Tutorial on the Protection o f Synchronous
Generators, Section 8, 95-TP-102, Power System Relaying
Committee 1995.
0.55
[2] M.A. Redfem & M.J. Checksfield, "A New Pole Slipping
Protection Algorithm for Dispersed Storage and Generation
Using the Equal Area Criterion," IEEE Trans. Power
-1.5 Delivery, Vol.10, N o.l, Jan 1995, pp 194 - 202.
[3] J.A. Imhof (Chairman) et al, "Out of Step Relaying for
Generators - Working Group Report," IEEE Trans. Power
1 -0.5 0 0.5 Apparatus and Systems, Vol. PAS-96, No.5, Sept 1977, pp
Resistance, R (p.u.) 1556 - 1564.
Fig 10 Single Blinder Protection Scheme Characteristic [4] H.K. Clark and J.W. Feltes, "Industrial and Co-generation
for 625 kVA Generator Pole Slip Test. Protection Problems Requiring Simulation," IEEE Trans.
Industry Applications, Vol. IA 25, No.4, July 89, pp
V II. C O N C L U S IO N S
766-775.
[5] E.W. Kimbark, Power System Stability - Power Circuit
Breakers and Protective Relays, Vol - II, New York: Wiley
The growth in microprocessor relaying has enabled new
& Sons, 1964, Chapter X.
adaptive pole slipping algorithms to be developed which [6] Alternative Transients Program (ATP) Rule Book, July 1987
automatically adjust their settings according to the generator Edition, Leuven EMTP Centre (LEC), Belgium.
operating point. The power based scheme discussed in this [7] Clarke E, "Impedances Seen by Relays During Power
paper successfully detected all of the pole slipping conditions Swings With and Without Faults" A1EE Trans. Vol 64,
and restrained for all of the stable swings. In addition a major 1945, pp 372 - 384.
advantage is the simplicity of setting the relay. [8] Alvarado F L, "Electromagnetic Transients Program
(EMTP), Volume 3: Workbook m", EPRI EL-4651,
Conventional impedance based pole slipping protection Research Project 2149-6, Chapters 7 & 8, June 1989.
requires detailed simulations of the system to be protected in [9] Stalewski A, "Pole Slipping Protection", Developments in
Power System Protection Conf, IEE Conf. Pub. No. 185,
order to derive reliable settings. Unfortunately, pole slipping
pp 3845.
impedance loci predicted from theory are not always reflected
in reality and complicate the choice of reliable settings. Miles A Redfem (M'79) received his BSc degree from Nottingham
University and PhD degree from Cambridge University in 1970 and
The MHO scheme was shown to mal-operate for stable 1976 respectively. In 1970, he joined British Railways Research, and
swings if the characteristic was set too large, and failed to in 1975, moved to GEC measurements where he held various posts
detect some pole slips if time delays were used in order to including Head o f Research and Long Term Development and
prevent nuisance tripping during stable swings. Overseas Sales Manager. In 1986, he joined Bath University with
interests in Power Systems Protection and Management. He is
The single blinder scheme is the most secure of the chairman o f the organising committee for the 1997 IEE conference
conventional schemes, but the price for this extra security "Developments in Power System Protection", and is a member o f IEE
Professional Group P ll , Power System Measurement, Protection and
against relay mal-operation is that a complete pole slip cycle
Control. He is a corresponding member o f the IEEE Line Protection
is required for the scheme to operate.
Sub-Committee.

The double blinder scheme is less secure than the single Matt Checksfield received his B Eng degree from the University of
blinder scheme, and may fail to detect the first pole slip Bath in 1993. He is currently studying for his PhD degree at the
following a fault. Like the single blinder scheme, it requires a University o f Bath and is sponsored by GEC ALSTHOM Protection
complete pole slip cycle to operate properly. and Control. His research interests are embedded generation,
protection, power system stability, and renewable energy systems.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 377

 PAPER EIGHT

"PROTECTION AGAINST POLE SLIPPING AND LOSS OF

EXCITATION FOR EMBEDDED GENERATORS"

M J Checksfield and M A Redfem

Presented at
31st Universities Power Engineering Conference (UPEC),
September 1996, Crete, pp 9 - 12.

M.J. Checksfiel d/PhD /University o f Bath 1996 Appendix D, Page 378

 PROTECTION AGAINST POLE SLIPPING AND
LOSS OF EXCITATION FOR EMBEDDED GENERATORS.

M J Checksfield & M A Redfem.

University o f Bath, UK.

ABSTRACT. failure o f the excitation power source, a faulty Automatic Voltage

Regulator (AVR), and personnel errors. Pole slipping is generally due
With the increasing amount o f embedded generators being installed, to a system problem, whereas loss o f excitation is a machine
concern has arisen regarding pole slipping and loss o f excitation. problem. Although both conditions require immediate tripping, it
Both conditions can cause machine damage and system problems. may be possible to re-synchronise and re-load the generator directly
Protection which disconnects the machine as quickly as possible for after pole slipping. It is therefore useful to know which phenomenon
either conditions is therefore required. This paper presents new caused the generator disconnection.
algorithms for detecting both loss o f excitation and pole slipping
which are simple to set and detect both conditions quickly and
securely. Results from industrial generator field trials are used to 2.C ASYNCHRONOUS OPERATING CHARACTERISTICS.
verity the new algorithms.
The synchronous real power, Ps and reactive power, Qs
characteristics for a disturbed synchronous generator operating
1.0 INTRODUCTION. against an infinite bus[6] are given in (1) and (2);

Ps
Eq' V
sin(S) 1 1 sin(25) (1)
With the growing number o f embedded generation sets being
Xd' Xq Xd'
installed to operate in parallel with utility distribution systems,
synchronous generator pole slipping and loss o f excitation are
Qs
EqV cos(8) - ± . J_]X1 , |_L_ _L]vi cos(2S) (2)
increasingly being recognised as areas o f concern.
Xd' Xq Xd'j 2 (Xq Xd'J 2
The likelihood o f pole slipping for embedded generators is high[l], Where Eq' is the voltage behind transient reactance, Xq is the Q axis
due to the low values o f generator inertia, short transient time synchronous reactance, Xd' is the D axis transient reactance, V is the
constants, high reactances, and long fault clearance times associated generator terminal voltage, and 5 is the generator load angle. The
with distribution systems. Stability studies have shown [1] that the first term containing Eq' in both equations is due to the excitation
critical clearance time for embedded generators is generally below supply on the field winding. The (25) terms are due to the effects of
300 ms. Since the fault clearance times associated with the rotor saliency. The middle term o f (2) is a result o f the
overcurrent type protection schemes are normally in the 400 to 1000 magnetisation energy that the machine requires to produce an air gap
ms range, generator instability and hence pole slipping is highly flux. The dominant terms during stable synchronous operation are
probable following clearance o f a fault which does not cause the the first terms. During asynchronous operation additional components
generator itself to be disconnected [2]. o f power are produced due to induction generator action. The mean
values [7] of these terms are given by (3) and (4);
Pole slipping can cause utility voltage disruptions which can create
problems for other consumers close to the affected generator [3]. Xd-Xd' Td' Xd'-Xd" Td"
Pas - - s —
Induction motor stalls, motor tripping, synchronous motor pole 2 Xd.Xd' Id (sTd')J Xd'Xd" 1+ (sTd")2
slipping and nuisance tripping o f other protection relays can also be ^ Xq-Xq" Tq"
caused. Pole slipping may damage the machine due to rotor and + Xq.Xq" 1 (sTq")2 J (3)
stator overheating, shaft fatigue, excitation system damage and Where Pas is the average component o f real power generated due to
mechanical damage to the windings. It is therefore recommended, rotor slip, s, Xd is the D axis synchronous reactance, Xd" is the D
that when a generator pole slips, it is dis-connected as soon as axis sub-transient reactance, Xq” is the Q axis sub transient
possible[4]. reactance, Td* is the D axis short circuit transient time constant, Td"
is the D axis sub-transient time constant, and Tq" is the Q axis sub
Loss o f excitation like pole slipping, results in the synchronous transient time constant
generator operating asynchronously. However, since the generator
Xd-Xd' (Td')2 Xd'-Xd" (Td")2
has very little, if any excitation during loss o f excitation, the resultant Qas - s2 —
2 I Xd.Xd' ld (sT d ')2 Xd'.Xd" 1+ (sTd")2
fluctuations in machine power and torque are less. As with pole
. Xq-Xq" (Tq")2 1
slipping, loss o f excitation imposes a very high reactive power
Xq.Xq" 1+ (sTq")2 j (4)
demand on the system, which may depress system voltage and cause
problems for adjacent machines and other consumers. The generator Where Qas is the average component o f reactive power due to rotor
stator winding can become over loaded by the high reactive currents, slip. Note that a negative value o f slip is produced when the rotor is
and may overheat due to the induced currents in the field winding travelling above synchronous speed. The machine reactances are in
and rotor body. The permissible time before damage occurs varies per unit the time constants in radians, ( t ^ j = ®0t[»ec]) the slip
according to the type o f machine, prime mover governor has no units, it is relative ( 1% slip => 0.01).
characteristics, type o f excitation loss, and system conditions. The
maximum permitted time before damage may be as low as 10 Equation (3) shows that a significant amount o f real power
seconds [2]. Another consideration is when the loss o f excitation generation is possible during asynchronous operation.
condition is due to a failure o f one o f the components o f the
excitation system. Prompt tripping o f the generator and any 2.1 Pole Slipping Characteristics.
associated equipment will ensure that any additional damage to other
components and adjacent equipment will be minimised[5]. During pole slipping the dominant characteristics are determined by
(1) and (2). The real and reactive power both take on sinusoidal
Loss o f excitation may be caused by a short in the excitation system, characteristics with increase in load angle during the pole slip. These

M.J.Checksfield/PhDAJniversity of Bath 1996 Appendix D, Page 379

sinusoidal characteristics make the apparent impedance locus track generators and hydro sets are two examples. With salient pole
from right to left across the RX plane, and it is this characteristic machines, the loss o f excitation characteristics can be different to
which the majority o f impedance based pole slipping protection round rotor characteristics. At low loads it is possible that the
schemes rely upon to detect pole slipping. generator may function as a reluctance generator. The maximum
amount o f power which the machine can provide due to reluctance
Asynchronous power is also generated during pole slipping as shown action, Pr(max) is given by (5);
by (3) and (4). Equation (3) shows that an extra component o f real
power will be generated, and this can significantly change the real PKm«) - [-1 - -L] T . (5)
(Xq Xdj 2
power characteristic during pole slipping in some machines. It is
possible that during times o f low excitation, where Eq' is small, the Typical values for Xd and Xq are 2.2S and 1.1 pu respectively, a
contribution to generator output power by asynchronous operation reluctance power output o f 23 % pu may therefore be achieved at
results in a power output characteristic which does not reverse during 1 pu terminal voltage. If the generator functions as a reluctance
pole slipping[8]. This will increase the difficulty in setting some generator, synchronous operation will be maintained and no
impedance based relays which rely on the apparent impedance overheating due to induced currents will occur. The generator should
crossing from right to left across the impedance plane. still be tripped however, as a fault wiil still exist in the excitation
system. If the initial generator output power is greater than the
Equations (2) and (4) show that the reactive power demand that the maximum reluctance torque, then asynchronous operation will occur
generator places on the system during pole slipping can be very high. once that the generator field flux has decayed. This condition will
The worst point occurs when the load angle travels from 180 to appear to be more severe than with a round rotor machine due to the
270°. During this period the synchronous terms o f (2) will cause the rotor saliency. The 2S terms o f (1) and (2) will cause oscillations in
generator to import reactive power. In addition the slip o f the the generator real and reactive power at twice slip frequency. These
generator will be approaching its highest value since the generator variations will superimpose oscillations onto the terminal voltage. If
will be motoring, and (4) shows that asynchronous reactive power the embedded generator is connected to a weak source, the voltage
demand varies approximately with the square o f slip. The amount o f oscillations in addition to the voltage depression caused by the loss
reactive power imported during pole slipping is important when o f excitation will compromise the quality o f supply o f adjacent
considering the stability o f pole slipping and loss o f excitation consumers. It is therefore essential that the generator is tripped as
protection schemes during recoverable power swings. The 'ideal' quickly as possible.
reactive power demand outlined in (2) can become highly distorted
by the induction machine action described in (4). The high
asynchronous reactive power demand occurs for positive (motoring) 3.0 PROTECTION ALGORITHMS FOR DETECTING
as well as negative (generating) slips. The overall effect on generator POLE SLIPPING AND LOSS OF EXCITATION.
reactive power demand during recoverable power swings is therefore
to increase it This brings the apparent impedance locus closer to the The algorithm used to detect pole slipping and loss o f excitation
operating characteristics o f impedance based relays. from high initial powers uses the Critical Stability Point (CSP)[3,9J.
Once the critical stability point has been past, stability can not be
2.2 Loss o f Excitation Characteristics. maintained and pole slipping, or asynchronous running will occur.
The algorithm predicts that pole slipping will occur by detecting if
When a generator loses its field, the Eq' terms in (1) and (2) the load angle is still increasing once the CSP has been exceeded.
gradually fall to zero in a time determined by the open circuit In designing an algorithm for detecting generator asynchronous
transient time constant, Tdo'. Following the loss o f excitation, some conditions, it must be able to predict when a generator is about to,
of the generator real power output will be provided by the second or actually is running asynchronously. It must also remain stable for
term in (1). This is the transient saliency' term, and for the duration all other power system conditions, including faults and recoverable
o f the period where the generator is losing its main field, this term power swings.
will help maintain real output power and prevent load angle increase.
After the transient period o f the generator has died away, the Xd' The algorithm uses reactive power, real power, and rate o f change of
term in (1) becomes an Xd term. real power to detect operation past the CSP. The algorithm is;

Most o f the literature concerning loss o f excitation deals with Q < Qtrip } Continuously
characteristics for turbo • generator units. These are round rotor TRIP and }
IF P < Pt } For 1.5 Power (6)
machines and therefore have Xd =>Xq. A round rotor machine will
and }
begin to accelerate at this point since it can no longer produce (AP/At) < (AP/At)mi, } System Cycles
synchronising forces to keep the machine load angle at a constant
value. As the generator slip begins to increase, the machine will Where Q is the generator three phase reactive power, Qtrip is the
begin to function as an induction generator and it is possible for the reactive power trip setting, based on the Q axis synchronous
prime mover governor characteristics to intersect the asynchronous reactance. P is the generator real power output, Pt is the real power
torque curve. When this occurs the generator operates as a stable trip setting which is derived by the algorithm. (AP/At) is the rate o f
asynchronous generator. Some loss o f excitation schemes may fail change o f real power, (AP/At)min is the rate o f change o f power trip
to detect this condition due to the impedance locus settling to the setting which is derived by the algorithm. The 1.5 power system
right o f their protection characteristic. This is more likely to occur cycle time constraint is to ensure the algorithm does not operate for
for loss o f excitation conditions which occur at low generator faults. The majority o f faults will not satisfy all o f the criteria o f
loadings. At full load stable asynchronous operation may not occur equation (6), the few that do only cause the criteria to be satisfied
and the generator may be forced to operate past the peak of its power during fault clearance for less than one power system cycle.
slip curve. This is a more onerous condition since the machine may
overheat quickly due to the high level o f induced currents in the 3.1 Condition Monitoring Techniques.
rotor structure. The generator therefore needs disconnecting as
quickly as possible when a loss o f excitation occurs at full load. The algorithm is designed fora microprocessor based relaying, where
the only information available to the relay is 2 phase voltages and 3
In contrast to the commonly discussed round rotor generators, current signals provided by the relay's input transducers. Since
embedded generators are often salient pole machines. Diesel generators may be operated over a range o f output powers, a method

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 380

of estimating the prime mover input power, Pm is required so that 3.3 Detection of Loss of Excitation at Low Powers.
the real power trip setting, Pt can be derived. The input power is
estimated by observing the power output of the generator and using The above algorithm will successfully detect pole slipping and will
the following algorithm; also detect loss of excitation from high initial powers. It may fail to
detect loss of excitation from low initial powers. For a round rotor
if IP. - P . . * I and |P - P,_(2. ll8) | < Ptol ; machine, the drop in real power and resultant size of the (AP/At)
then Pt - Pn* Pfact signal may not be sufficient to cause it to trip. The small amount of
saliency also means that once asynchronous operation occurs, the
where Pn is the present real power sample value, Pn . lag is the pulsations in power will not be sufficient to cause it to operate.
sample value o f 'lag' samples ago, Pn . (2«iag) is the sample value of Salient pole machines will cause the algorithm to trip if they operate
'2* lag' samples ago, Ptol is the tolerance band, chosen as nominally asynchronously following loss o f excitation, due to the pulsations in
5% o f machine rating, and Pfact is a coefficient to take into account real and reactive power. If the machine remains in synchronism due
the losses and inaccuracies inherent in measuring machine input to reluctance torque action, no power pulsation will result and the
power from output power. The algorithm processing rate is 4 times algorithm will not trip.
per power system cycle, the value o f 'lag' typically chosen is 200
which equates to 1 second, Pfact is nominally set to 0.9. To provide full loss o f excitation protection at these lower power
ranges, a real power limited VAr relay is used. Figure 1 shows its
3.2 Adaptive Trip Setting (AP/At)min. operating characteristic on the underexcited portion o f a salient pole
generator operating chart[10]. The VAr limit setting, Qset is set
The margin for error, (AP/At)min used in the (AP/At) criterion is based according to V2/Xd, whilst the Watt setting, Pset is based on the
on the theoretical value o f (AP/At) at the CSP[3,9], it is given by; saliency term V2 (1/Xq - 1/Xd). This introduces a margin for error
since the generator can only operate as a reluctance generator at half
(AP/At) — s* Pmax* cos 1180 - sin- 1f—— ] ] W this value.
I i Pmax J J
The value o f slip chosen will not be the exact value at which the If the generator is a round rotor machine, Pset is set to half machine
relay trips, due to effects not taken into account in the derivation of VA rating. The purpose o f Pset is to ensure the scheme does not
(8), such as damping power. Equation (8) is used by the trip nuisance trip during recoverable power swings. During such swings
algorithm to continuously update the trip setting (AP/At)min the reactive power may fall below Qset, but this only occurs when
depending on what the variables Pmax and Pt dictate. the power is at high values, i.e above Pset. The 1 second time delay
enhances algorithm security. The operating characteristic is located
Pmax is a measure of the maximum output power that the generator just at the edge of the practical stability limit. It is very unlikely that
has produced in the last second. It is used so that the algorithm can an embedded generator would be operated in this region since this
adjust its setting according to whether transient or steady state mode of operation is only required if the local load is predominantly
conditions are prevailing. With loss of excitation or steady state pole capacitive, or the system voltage is too high. These conditions occur
slips, a gradual loss o f synchronism occurs, and the resulting power very rarely. In addition, operation in this area would further decrease
output waveforms change gently to begin with. This produces low the machines stability margins increasing the chance o f instability
values o f (AP/At), a sensitive trip setting is therefore required. If the and pole slipping. If the generator was required to operate in this
generator has been transiently disturbed, its output during the first area from, the reactive setting, Qset would have to scaled up by a
swing in load angle, be it stable or unstable, will be much greater safety factor of 1.2.
than its nominal output This results in a greater magnitude in TSL - Theoretical Stability Limit a.
(AP/At) at the CSP, the algorithm therefore needs to adjust its setting
(AP/At)min accordingly. The algorithm adjusts the setting, (AP/At)min Pset = V 2(l/Xq-1/Xd)
according to (9);

If < 0.6 then; (AP/At)mil - (AP/At)miB ♦ (AP/At)tr_ (9)

Pmax
where (AP/At)lrin is an external relay setting based on the degree of
transient saliency in the generator. Transient saliency arises due to
differences in the values o f Xd' and Xq'. (AP/At)lrin is given by; restrain

(A P/a.U C pu)- ( j f c - A , ) <'°> Time Delay of 1 s

For a steady state pole slip or loss o f excitation, no peak will occur Pset
in the power output before the pole slip, the algorithm trip setting
Locus of Synchronous
should therefore be at its most sensitive level to detect the pole slip
Operation for If = 0
since low values o f (AP/At) will result.
Qset = V /Xd
The value of Pmax used by equations (8) and (9) is nominally set to Q(pu)
1.4 * Pt, so that the ratio Pt/Pmax equals 0.71, (AP/At)tna is therefore V2/X(
not included in the calculation o f (AP/At),„,„ during steady state
conditions. If the generator’s power output rises above the existing Figure 1: Under-Excited Section o f Salient Pole Generator
value o f Pmax for greater than one power system cycle, then the Operating Chart and Loss o f Excitation Var Relay Characteristic.
value of Pmax is updated to the new maximum for a duration of one
second. The one second limit is used so that the new value of Pmax 4.0 ALGORITHM VALIDATION STUDIES.
is only used for the period where a pole slip or recoverable power
swing associated with that value is likely to occur. The one cycle The algorithms were tested using field trials on a 625 kVA industrial
constraint is necessary so that incorrect values of Pmax do not result diesel generator connected to the utility network via a 415/11,000
from spurious spikes on the power waveform. volt unit transformer and two parallel 2 km 11 kV cables. The fault
capacity at the generator terminals was 30 MVA. 500 kW o f local

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D. Page 381

load was connected to the generator. The field trial tests were waveform o f Figure 3 shows that the algorithm tripped before the
performed in order to validate the operation o f the new power based generator operated asynchronously. This would have prevented any
pole slipping algorithm within a commercial relaying platform. It was damage due to overheating. The VAr relay did not operate for this
not possible to place faults on the power system to induce pole as the real power did not stay below Pset for 1 s.
slipping, and hence a pole slip was induced by increasing the diesel
engine's output torque with the generator under-excited. This scenario References [3,9] provide details of the rate of change of power
could happen due to operator error, or if the diesel engine control algorithm's response to pole slipping. Further facilities could be
system mal-functioned. Loss of excitation tests were also performed placed into the algorithms to enable the system to diagnosis what
at 20, 50 and 100 % load at under-excited and over-excited initial caused the trip conditions. The VAr relay will only trip for loss of
excitations. excitation, unless a very slow pole slip at a rate o f 0.5 Hz occurs.
If the maximum real power monitor, Pmax is greater than its nominal
value of 1.4 * Pt, before the pole slip occurs, then the pole slip
5.0 FIELD TRIAL RESULTS. would have been caused by a transient disturbance, rather than
generator control system failure. Another approach would be to
The combination o f algorithms successfully detected all of the pole monitor the level o f AP/At, if the amplitude of this signai exceeds a
slipping and loss o f excitation conditions. The rate of change of level before the before the algorithm trips, then the pole slip was
power algorithm detected all o f the high power loss o f excitation probably due to a transient disturbance. Inspection o f the disturbance
whilst the power restrained V Ar relay detected all o f the low and recording also provides another means of diagnosis, all of which help
medium power loss o f excitations. the protection engineer to ascertain what if anything is wrong with
the generator.
RESTRAIN

6.0 CONCLUSIONS.
50 % initial load
Pole slipping and Loss o f excitation are two concerns which need to
be addressed when considering the protection of embedded
generators. Machine damage and system problems may result if
20 % initial load
protection is not provided.

I; The power based algorithms presented successfully detect pole

50J
-625 -500 -375 -250 -125 0 125 250 slipping and loss o f excitation, whilst restraining for all other power
Reactive Power, Q (kVAr) system conditions. The power restrained VAr relay is used to detect
Figure 2: PQ plane Showing Power restrained VAr relay low power loss of excitations in both salient pole and round rotor
Operating Characteristicand Two Loss o f Excitation Loci. generators, whilst the other power based algorithm detect loss of
excitation from high powers and all types o f pole slipping.
Figure 2 shows two PQ loci for loss of excitations from 20 % and 50
% load. For the 20 % load loss o f excitation, the VAr relay tripped Additional features can be added to the algorithms to provide the
1.7 s after the field was short circuited, for the 50 % loss of protection engineer with diagnostic information. This information
excitation, the VAr relay tripped 1.6 s after short circuiting the field. will help to ascertain whether the algorithms tripped due to problems
Due to the time delay, the protection did not operate the instant the with the generator or problems with the power system, such as faults.
PQ locus entered the tripping zone. Note that in both cases the
generator had not lost synchronism, if it had, there would be 7.0 REFERENCES.
significant pulsations in the real and reactive power.
[1] Checksfield M J & Redfem M A, "Assessment of
■Trip boundary ■Pmax
0 Embedded Generator Stability When Connected to a HV Utility
Generator Reactive Power, Q Distribution System", SO**1UPEC, 1996, Greenwich, pp 219 - 222.
^ -0.3
I -0 6 [2] Clark H K & Feltes J W, "Industrial and Co-generation
Protection Problems Requiring Simulation," IEEE Trans. Industry
1 - 0 .9
Applications, Vol.IA-25, No.4, July 89, pp 766 - 775.
O' -1.2
[3] Redfem M A & Checksfield M J, "A New Pole Slipping
-1.5
Protection Algorithm for Dispersed Storage and Generation Using the
1.5 Generator Real Power, P
Equal Area Criterion," IEEE Trans. Power Delivery, Vol. 10, No.l,
^ 10 TRIP-------- Jan 1995, pp 194 - 202.
0.5 [4] J.A. Imhof (Chairman) et al, "Out of Step Relaying for
eu 0 Generators - Working Group Report," IEEE Trans. Power Apparatus
-0.5 and Systems, Vol. PAS-96, No.5, Sept 1977, pp 1556 - 1564.
3.0 [5] Mackenzie W F, "Loss o f Field Relay Operation During
Generator Real Power, AP/At |
0 1 | System Disturbances", IEEE Trans, Vol PAS-94, No.5, Sept 1975,
p p 1464 - 1472.
§ -3.0
< -6.0 grpgh limited for KI [6] CONCORDIA C. 1951, Synchronous Machines: London:
reasons of clarity l J Wiley & Sons, Inc, Ch 2 & 3.
< -9.0
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 [7] VENIKOV V A. 1964. Transient Phenomena in Electrical
Time(s) Power Systems: Permagon Press.
Figure 3: Rate o f Change o f Power Algorithm Response to [8] GUILE A E. Electrical Power Systems,Vol 2.PP106 -141.
Underexcited Full Load Loss of excitation. [9] Checksfield M J & Redfem M A, "A Condition
Monitoring Pole Slipping Protection Algorithm Based on the Equal
Figure 3 shows the rate o f change of power algorithm measurands Area Criterion", 29* UPEC, Galway, 1994, pp 747 - 750.
for a loss of excitation from 100 % initial load. The algorithm [10] Walker J H, "Operating Characteristics o f Salient Pole
tripped 2.85 s after the field was short circuited. The real power Machines", IEE. Proc. 1953, Vol 100 Pt 2, pp 13 - 23.

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 382

 PAPER NINE

"A REVIEW OF POLE SLIPPING PROTECTION”

by M A Redfem and M J Checksfield

Published in
IEE Colloquim Digest 1996/265 - Generator Protection,
October 1996, Newcastle, UK, pp 6/1-6/9.

MJ.Checksfield/PhD/University of Bath 1996 Appendix D, Page 383

 A Re v ie w of P o l e S l ip p in g P r o t e c t io n .

M A Redfem and M J Checksfield

School of Electronic and Electrical Engineering
University of Bath, BATH, UK.

1. In t r o d u c t io n .

Pole slipping protection, also referred to as loss o f synchronism protection, is required when
a generator loses synchronism with the utility's main source o f power and is subjected to high
fluctuations in the current passing through it. When a machine pole slips, it is desirable to
disconnect it from the utility supply as quickly as possible, thus preventing possible damage
to the generator, disturbance to the local power system, and system instability.

generator &
generator transformer ^ne source infinite
impedance impedance impedance bus

— /wywi—
'" U

Eg/Es > 1

Eg/Es = 1

Eg/Es < 1

pole slipping
loci

FIGURE 1. THE CLASSICAL POLE SLIPPING IMPEDANCE LOCI.

Pole slipping o f a synchronous generator connected to a power supply system can be caused
by short circuit faults, general disturbances to the power supply network, or problems with
the generator's prime mover or control system. Although system performance criteria for
transmission systems have become more stringent over the last few decades and as such the
probability o f pole slipping on main generation units is very low, the connection of embedded
generation to sub-transmission and distribution systems has raised new interest in this type
of protection. Some o f the factors which promote pole slipping are long fault clearance times,
low system voltages, reduced excitation, high impedances between the generator and the
supply network, and system switching operations. All o f these can be found in
sub-transmission and distribution systems.

M.J.Checksfi el d/PhD /University o f Bath 1996 Appendix D, Page 384

2. T r a d it io n a l T e c h n i q u e s for D e t e c t in g P o l e S l ip p in g in Generato rs.

The traditional technique for detecting pole slipping is to examine the variations in the
apparent impedance of the generator as seen from its terminals. Several schemes are
commercially available which are based on distance type relays1,2,3, and use combinations of
mho and linear characteristics.

The variations in the impedance seen at the terminals o f the generator during pole slipping4
are shown in figure 1. These impedance loci are derived using the assumptions that the ratio
o f generator to source impedances, Eg/Es remains constant, generator saliency is neglected,
generator damper effects are neglected, transient fault impedance effects have decayed, shunt
loads and shunt capacitance effects are ignored, effects o f automatic voltage regulators and
governors are neglected, and the source voltages behind their equivalent impedances are
sinusoidal and at fundamental frequency1,4,5,7.

For the case where the ratio Eg/Es = 1, the impedance locus will be a straight line, PQ, which
is the perpendicular bisector o f the system impedance line GS. When a pole slip occurs, the
load angle 5 increases causing the impedance to travel from right to left across the R-X plane.
When the load angle reaches 180°, the loss o f synchronism point has been reached and the
impedance will lie on the system impedance line GS. The point on the system impedance line
where the impedance locus appears for 5 = 180° is known as the system centre, at this point
the impedance gives the appearance of a three phase fault. As the pole slip cycle is completed
and 8 approaches an in-phase value of 360° the impedance locus will travel to the left o f the
system impedance line before returning to the load area of the R-X plane.

Figure 1 also shows impedance loci for the cases where the ratio Eg/Es < 1 and Eg/Es > 1.
The effect is to cause the impedance locus to become a circular arc. These curves show that
as the generator internal voltage is increased, the system centre is pushed away from the
generator. AVR field forcing action will have this effect.

For the majority o f situations, the combined impedance o f the generator and its transformer
will be larger than the combination of line and equivalent source impedances. This will result
in the system centre appearing inside the generator or its transformer during pole slipping.

The simplest type o f impedance based pole slipping protection is the single MHO relay
scheme1,5. Its main advantages are that tripping can occur before the load angle 8 = 180°, and
that it provides a degree of back up protection for faults occurring in the transformer and
generator. Its main disadvantage is that very careful selection o f the mho reach is necessary
to avoid tripping during power swings for which the pole slipping protection should remain
stable.

The relay is normally set to cover the generator transient reactance, Xd' and most of the
generator transformer. This may produce a characteristic which is too large and hence be
prone to nuisance tripping. Another recommendation is that the angle of swing where the
impedance locus enters the characteristic is set to be at least 115°.

The single blinder scheme15 is the most secure of the impedance based techniques for
detecting pole slipping. The mho supervisory element restricts operation to swings appearing
inside the generator and its transformer. For the scheme to operate, the impedance locus must

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 385

enter the mho circle and its trajectory must take longer than a specified time to cross from
one blinder unit to the next. The time constraint is normally set to two power system cycles.

The scheme has the advantage over the mho scheme that the diameter of the supervisory mho
element can be increased in order to provide better sensitivity to pole slipping conditions
without the scheme tripping for stable swings.

The double blinder scheme1,5 also uses a supervisory mho characteristic to stop the scheme
tripping for stable swings. For the scheme to operate, the mho element must first pick up, then
the locus must then stay between the inner and outer blinders for a preset time. This will
satisfy the first stage of the scheme logic. It then must stay between the inner blinder
elements for a short time, and then when leaving the inner characteristics, it must again stay
between the inner and outer blinder elements for greater than a preset time. This satisfies the
final stage of the scheme logic, a trip signal is then issued when the supervisory mho resets.

The main difference between the single and double blinder schemes is that with the double
blinder scheme, the locus can enter and leave the characteristic from any direction. With the
single blinder scheme the locus must enter from one side and leave from the other for a trip
to occur. The inner elements of the double blinder scheme must therefore be set so that they
will only pick up for non recoverable swings.

Unfortunately, the impedance of a generator is not as predicable as would be expected and

it depends on the prevailing situation. Hence the concept of the sub-transient, transient and
dynamic impedance. It is therefore generally recommended1,5,6 that transient stability studies
are performed so that the location of the impedance loci are known and the most appropriate
relaying scheme can be selected. These simulations are necessary to enable correct setting of
conventional impedance relays, and they can be time consuming and expensive.

3. P o w e r B a s e d T e c h n iq u e for D e t e c t in g P o l e S l ip p in g in Gen erato rs.

Pole slipping is a power based phenomena and an alternative approach10 to detecting pole
slipping has been developed which uses the Equal Area diagram as a basis to assess the
stability of the machine and determine when it is committed to a pole slip. This approach has
the potential of tripping before the pole slip actually occurs and hence offers the best prospect
of preventing damage to the generator and the adjacent power system.

The technique takes advantage of the processing capabilities of modem multi-function

microprocessor relaying platforms and has been designed to share sub-functions with other
protection functions required for the protection of embedded generation.

The Equal Area Criterion demonstrates the conditions where synchronism is lost and pole
slipping occurs. Figure 2 shows the power/load angle relationship and the rate of change of
load angle (slip, s) for a generator losing synchronism with the utility supply to which it is
connected following a loss of power transfer capability due to a disturbance on a double
circuit line8. For this scenario, the generator is connected to a double circuit line and the
power system disturbance is caused by switching one of the lines out of service for a short
period. Removing the electrical load from the generator, while keeping the mechanical power
constant, causes the generator to accelerate and eventually pole slip.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 386

The Equal Area Diagram, figure 2, illustrates that for a stable swing, the machine operating
point cannot exceed point 5, the critical stability point. This corresponds to the point where
the electrical output of the machine, P, equals the mechanical input from the prime mover,
Pnr When the machine becomes insrable, it moves beyond the critical stability point, which
can be identified in terms of real power as the condition when:-

MECHANICAL
POWER
POWER INPUT.

ASYNCHRONOUS
POWER OUTPUT.

o 360
LOAD
ANGLE

M-------------►

PERIOD OF
\
SYNCHRONOUS SYNCHRONOUS
OSCILLATIONS POWER OUTPUT

FIGURE 2 EQUAL AREA DIAGRAM FOR GENERATOR

LOSS OF SYNCHRONISM

Since this criterion can also be satisfied for machine operating points between points 0 and
1 shown on figure 2, the reactive power measurement is used to differentiate between the two.
The steady state reactive power/load angle relationship for a round rotor machine is given by:-

\
( l + l] V 2 i l V 2
Q - ~7~(cos( 6 )) - +
X * X , 2 X , 2
\ <7 d ) k d )

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 387

This shows that for load angles between 90° and 270°, the reactive power will always be less
than Qtnp. Since point 5 of figure 2 occurs for load angles greater than 90°, then if
_y 2
Q * ~ r < - <?*)
$

the machine must be operating at point 5 and not point 1.

The transient reactive power characteristics vary in magnitude from the above steady state
values. This does not corrupt the above method since the value of reactive power
corresponding to a 90° load angle is identical to the steady state value. The transient reactive
power characteristics above and below this load angle are such that in steady state terms it
appears as if the load angle is correspondingly higher or lower.

Finally, it must be ascertained whether or not the load angle is still advancing. To determine
this, rate of change of real power is used. It can be seen from the synchronous power output
characteristic of figure 2, that if the load angle is still increasing when point 5 is reached, the
machine output power will be decreasing, i.e. the rate of change of real power will be
negative.

In practice a margin for error is allowed for in this rate expression (AP/At), to ensure the
algorithm does not mal-operate. A minimum negative value, (AP/At),^ is designated, based
on a minimum value of slip and an assumed sinusoidal power/load angle relationship. The
conditions which determine that a generator is about to pole slip are therefore

P < Pt , where Pt <* Pm

Q < Q ^iP

(AP/At) < ( A P /A t ) ^

These conditions are used to give the trip criteria for the algorithm. The real power trip level,
Pt is derived from the generator's monitored power output and is proportional to the
mechanical power input Pni. This setting is adjusted periodically. The rate of change of power
trip setting, (APVAt)^ is also continuously adjusted by the algorithm according to the
generator operating point.

Short circuit faults cause added complications since they cause dramatic changes in the relay
measurands. These transitions produce negative values of (AP/At) which are a potential source
of instability to the algorithm. Fortunately, the faults generally introduce sinusoidal terms of
twice the power system frequency into the power measurements. Since pole slipping is a
relatively slow process in comparison to faults, then an imposed minimum tripping time of
one and a half power system cycles will inhibit fault tripping.

MJ.Checksfield/PhD/University of Bath 1996 Appendix D, Page 388

4. P e r f o r m a n c e T e st s

The performance of the various pole slipping algorithms, including the power based algorithm,
have been tested using computer simulation, laboratory machines and field tests. The field
tests were the most enlightening and witnessing the pole slipping of a 'real' machine was itself
an education.

In the most dramatic of the tests using a 625 KVA diesel generator, the power based pole
slipping algorithm successfully detected both of the pole slips, whereas, analysis showed that
the impedance based schemes all failed to trip.

-TRIP BOUNDARY Generator Reactive Power

s -500,000
i
- 1,000,000

-1,500,000
Generator Real Power

2 ,000,000
Pmax
w, 1,000,000

TRIP, TRIP
- 1,000,000
Generator Rate o f Change o f Real Power
20 ,000,000

^ - 20,000,000
pt
-40,000,000

50 550 1050 1552

Time (ms)
F i g u r e 3. P o w er B a se d A lg o r ith m R esp o n se t o
625 k V A D i e s e l G e n e r a t o r P o l e S lip .

Figure 3 shows the power based pole slipping algorithm's response to this test. The real power
plot shows where the algorithm tripped, which for the first pole slip occurred just before
1000 ms on the record. Allowing for a circuit breaker operating time of 100 ms, this would
have isolated the generator before the pole slip. The real power plot also reveals the
potentially damaging effects of pole slipping, since the peak in power output after the first
pole slip reaches 2.7 MW, five times rated power. By tripping before the pole slip occurred,

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 389

the new algorithm would have avoided this dangerous surge in the power output. Due to the
finite disturbance recording time, the transition from normal to underexcited generator
operation was not captured.

The conventional impedance based schemes were set so that they would operate for swings
appearing on the generator side of the transformer high voltage terminals for a range o f values
o f E g / E s from 0.66 to 1.5. The value of X d ’ for the generator was 0.22 p.u., the transformer
reactance was 0.03 p.u. The MHO scheme was set so that the swing locus for a value of
E g / E s of 1 entered the characteristic at an angle o f 120°. If the characteristic had been set to

reach from the transformer high voltage terminals all the way into the generator, the
theoretical locus would have entered at a value o f 102°, making the scheme liable to operate
for recoverable power swings.

NO TRIP left right

blinder blinder

0.44
Xd

-0.5
2
3 0.62 0 .2 9
*X
of
o
8

0.55

1 -0.5 0 0.5
Resistance, R (p.u.)

F ig u r e 4. S in g l e B l i n d e r P r o t e c t io n S c h e m e
C h a r a c t e r i s t ic fo r 625 kV A G e n e r a t o r P o l e S l ip T e s t .

The MHO scheme's failure to trip further highlights the difficulty in using it. Covering the
whole o f the generator X d ’ can lead to too large a MHO characteristic, and this makes the
scheme prone to nuisance tripping. Adding a time delay increases the stability of the scheme
to recoverable swings, but as the time delay is increased, the scheme may fail to detect fast
pole slips whose locus travels through the characteristic quickly. If the circle size is decreased
in order to increase stability, the scheme can fail to detect pole slips whose impedance does
not take the classical trajectory.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 390

The single blinder scheme was set using the recommended blinder settings o f 90° and 270°.
The supervisory MHO circle was set so that theoretical swings appearing in the transformer
and generator could be detected. The scheme characteristics and the impedance locus of the
pole slipping generator are shown in figure 4. It must be noted that the time records are not
the same as used in figure 3 since the disturbance recorders were not synchronised.

The first pole slip occurred between 0.2 and 0.3 s, and the second occurred between 0.5 and
0.6 s. The first pole slip was not detected because the impedance trajectory did not enter the
supervisory MHO characteristic. However, even if the supervisory MHO had been made
larger, the scheme would still have failed to trip since the trajectory did not cross the left
hand blinder. For the second pole slip, the swing impedance travelled across the impedance
plane from right to left as predicted by theory, but again it did not enter the supervisory MHO
characteristic. This was due to the speed at which the generator pole slipped. If the generator
pole slips at a slow speed, then the impedance o f the generator is much larger. At a slip of
0.33 %, the effective impedance of the generator becomes 2 * X d ' n .

The double blinder scheme suffered similar problems to the other impedance based techniques
and failed to trip.

A loss o f field protection included in the relay did however trip. Its MHO characteristic was
set to a value o f X d + X d ' / 2 , which for this generator was -j 2.26 pu. This relay had a 0.5 s
delay.

5. C o n c l u s io n s .

Protecting an embedded generator against abnormal operating conditions is rarely as simple

as detecting short circuit faults, but is as important to the safe and secure operation of the
machine and the associated power system.

With the growing number o f small and medium sized synchronous generators embedded in
sub-transmission and distribution systems, the probability of pole slipping is growing and
therefore pole slipping protection is becoming more important. Machine winding damage,
system problems, and shaft failure can all result if adequate protection is not provided.

The availability o f powerful microprocessor based relay hardware has enabled new techniques
to be considered to provide protection against pole slipping. These include a new power based
pole slipping protection algorithm. This has been shown to be effective in detecting the onset
of pole slipping and hence is able to disconnect the generator before the dramatic current
surges can occur.

The pole slip used to illustrate the operation of pole slipping protection was not designed to
be particularly testing for any technique and considerable effort was put into choosing the
most appropriate settings for the impedance based relays so that they provided the effective
protection. The results, however, clearly demonstrated the difficulties o f using impedance
based pole slipping relays and the immediate advantages of the power based algorithm. The
setting data for which is normally supplied with the generator. Using power as the basis for
the detection o f pole slipping proved advantageous.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 391

6. R efer en c es.

1. J. A. Im hof et al, 'Out of Step Relaying for Generators - Working Group Report,1
IEEE Transactions PAS-96, No.5, Sep/Oct 1977, pp 1556-1564.
2. A. Stalewski, J. L. H. Goody and J. A. Downes, 'Pole Slipping Protection,'
Developments in Power System protection, LEE Conf. Publication, No. 185, pp 38-45.
3. S. Shiwen, 'Microcomputer Based Out-of-Step Protection for Large Generator,' IEE
APSCOM Hong Kong, Nov 1991, IEE Conf. Proc. No 348, Nov 1991, pp 839-842.
4. E Clark, 'Impedances Seen by Relays During Power Swings with and without Faults.'
AIEE Trans Vol 64, 1945, pp 372-384.
5. D.W. Smaha, 'Out-of-Step Relay Protection o f Generators.' IEEE Tutorial on the
Protection o f Synchronous Generators, Section 8, 95-TP-102, Power System Relaying
Committee 1995.
6. H.K. Clark & J.W. Feltes, 'Industrial and Cogeneration Protection Problems Requiring
Simulation.' IEEE Trans. Industry Applications, Vol. IA-25, No.4, July 89, pp 766-
775.
7. E.W. Kimbark, 'Power System Stability - Power Circuit Breakers and Protective
Relays.' Vol - II, New York, Wiley & Sons, 1964, Chapter X.
8. V. A. Venikov, 'Transient Phenomena in Electrical Power Systems', Pergamon Press,
1964.
9. T. H. Mason, P. D. Aylett and F. H. Birch, 'Turbo-Generator Performance Under
Exceptional Operating Conditions,' IEE Conf. Proc. Vol. 106, Jan 1959, pp 357-373.
10. M A Redfem and M J Checksfield, 'A New Pole Slipping Protection Algorithm for
Dispersed Storage and Generation using the Equal Area Criterion.' IEEE Transactions
on Power Delivery PWRD Vol 10, No 1, Jan 1995, pp 194-202.
11. GEC Measurements, 'Protective Relay Applications Guide.' GEC Measurements,
Stafford.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 392

 PAPER TEN

"POLE SLIPPING PROTECTION: PROBLEMS AND SOLUTIONS"

M A Redfem, M J Checksfield, J Gosalia and G K Clough

Presented at
23rd Western Protection Relay Conference
Spokane, USA, October 1996, pp5/l-5/5.

MJ.Checksfield/PhD/University of Bath 1996 Appendix D, Page 393

 POLE SLIPPIN G PROTECTION:
PROBLEM S AND SOLUTIONS.

M. A. Redfem, M. J. Checksfield J Gosalia and G K Clough

School of Electronic and Electrical Engineering GEC ALSTHOM T&D Protection and Control
University of Bath, BATH, BA2 7AY, UK. Hawthorne, New York, USA.

1. In t r o d u c t io n . Pole slipping is a power based phenomena and an alternative

approach to detecting pole slipping has been developed which
With the growing number o f Dispersed Storage and Generation uses the Equal Area diagram as a basis to assess the stability
units (DSG) being installed to operate in parallel with utility o f the machine and determine when it is committed to a pole
distribution systems, synchronous generator instability is slip. The settings for this technique rely on data which is
increasingly being recognized as an area of concern. Generator generally available from the generator's manufacturer and
instability, pole slipping, can damage the machine and disturb stability studies are not required. Since this approach [6] is
the local power system. Unfortunately, DSG generators able to recognise the conditions where a pole slip is inevitable,
generally have high per unit impedances, low inertia and small it therefore offers the prospect o f tripping before the actual
transient time constants. These together with the long fault pole slip and hence disconnecting the machine from the system
clearance times associated with distribution networks all before it can be damaged and before there is a major
contribute to the increased probability of pole slipping. disturbance to the adjacent power system.

Pole slipping of a synchronous generator connected to a supply The technique takes advantage o f the processing capabilities
system is the process by which an imbalance between the of modem multi-function microprocessor numeric relaying
mechanical power input to the machine and its electrical power platforms and has been designed to share sub-functions with
output causes the rotor to accelerate and eventually slip with other protection functions required for the protection of
respect to the power system frequency. This can be caused by Dispersed Storage and Generation units.
short circuit faults, general disturbances on the power system,
or problems with the generator's prime mover, its excitation or
its control system. 2. P o w e r b a sed T e c h n iq u e fo r D e t e c t in g P o le S lip pin g .

Traditional techniques for detecting pole slipping examine the The Equal Area Criterion demonstrates the conditions where
variations in the apparent impedance o f the generator as seen synchronism is lost and pole slipping occurs. Figure 1 shows
from its terminals. Several schemes are commercially available the power/load angle relationship and the rate of change of
which are based on distance type relays [1,2,3,4], and use load angle (slip, s) for a generator losing synchronism with the
combinations of mho and linear characteristics. Some of these utility supply to which it is connected following a loss of
schemes are able to trip before the pole slip occurs, whereas power transfer capability due to a disturbance on a double
others rely on the pole slip having taken place before they will circuit line [7]. For this scenario, the generator is connected to
disconnect the generator from the network. a double circuit line and the power system disturbance is
caused by switching one of the lines out of service for a short
Unfortunately, the impedance o f a generator is not as period. Removing the electrical load from the generator, while
predicable as would be expected and depends on the type of keeping the mechanical power constant, causes the generator
disturbance. Hence the concept o f the sub-transient, transient to accelerate and eventually pole slip.
and dynamic impedance. It is therefore generally recommended
[1,4,5] that transient stability studies are performed so that the The most severe disturbance for a generator is the close-up
location of the impedance loci are known and the most three phase fault, for which all o f the generator's load is lost
appropriate relaying scheme and its settings can be selected. and therefore all o f the prime mover power is used to
These simulations can be time consuming and expensive, and accelerate the rotor. For less severe short circuit faults, a
due to the vagaries of pole slipping do not guarantee reliable greater fault duration is required to cause pole slipping.
tripping or the absence o f nuisance tripping.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 394

The Equal Area Diagram, figure 1, illustrates that for a stable the machine must be operating at point 5 and not point 1
swing, the machine operating point cannot exceed point 5, the
critical stability point. This corresponds to the point at which The transient reactive power characteristics vary in magnitude
the electrical output of the machine, P, equals the mechanical from the above steady state values. This does not corrupt the
input from the prime mover, Pm. Instability occurs if the above method since the value of reactive power corresponding
machine moves beyond the critical stability point. This point to a 90° load angle is identical to the steady state value. The
can be identified in terms o f real power as the condition transient reactive power characteristics above and below this
when:- load angle are such that in steady state terms, it appears as if
the load angle is correspondingly higher or lower
P < P
Finally, it must be ascertained whether or not the load angle
is still advancing. To determine this, the rate of change of real
power is used. It can be seen from the synchronous power
output characteristic of figure I, that if the load angle is still
increasing when point 5 is reached, the machine output power
POWER
MECHANICAL will be decreasing, i.e. the rate o f change o f real power will be
POWER INPUT. negative.

SLIP
In practice a margin for error is allowed for in this rate
expression (AP/At), to ensure the algorithm does not mal-
operate. A minimum negative value, (AP/At)mm is designated,
based on a minimum value of slip and an assumed sinusoidal
ASYNCHRONOUS
POWER O U W T .
power/load angle relationship. The conditions which determine
that a generator is about to pole slip are therefore

180 270 360

P < P, , where P, « Pm
LOAD
ANGLE
Q< Qtrip
(AP/At) < (AP/At)

SYNCHRONOUS
POWER OUTPUT
These conditions are used to give the trip criteria for the
algorithm. The real power trip level, P, , is derived from the
FIGURE 1 EQUAL AREA DIAGRAM FOR generator's monitored power output and is proportional to the
GENERATOR LOSS OF SYNCHRONISM mechanical power input Pm. This setting is automatically
adjusted periodically. The rate of change of power trip setting,
Since this criterion can also be satisfied for machine operating (AP\At)min is also continuously adjusted by the algorithm
between points 0 and 1 shown on figure 1, the reactive power according to the generator operating point.
measurement is used to differentiate between the two. The
steady state reactive power/load angle relationship for a round Short circuit faults cause added complications since they cause
rotor machine is given by> dramatic changes in the relay measurands. These transitions
produce negative values o f (AlVAt) which are a potential
V2 source of instability to the algorithm. Fortunately, the faults
E *V
<? = — , o s ( / 6 *))ss -
- (c I 1 + —1
— generally introduce sinusoidal terms of twice the power system
Ad I Aq Ad 2
frequency into the power measurements. Since pole slipping is
a relatively slow process in comparison to faults, then an
This shows that for load angles between 90° and 270°, the imposed minimum tripping time of one and a half power
reactive power will always be less than Qlnp. Since point 5 of system cycles will inhibit fault tripping.
figure 1 occurs for load angles greater than 90°, then if

-v2
0 < ( Q = Qtrip)
X.

M J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 395

3. S im u l a t io n S t u d ie s . the generator transformer term inals o f su fficien t duration to
cause the generator rotor to sw in g up to the critical stability
The perform ance o f the p ole slipping algorithm w as tested point and back down to a stable operating area.
using a laboratory dispersed generation m odel, com puter based
dynam ic sim ulation and field tests. To test the algorithm for d etectin g a p ole slip, the sim ulated
generator w as initially set to run at m axim um output power
The laboratory dispersed generation m odel used tw o 5 kV A The input pow er w as then reduced to h a lf rating to test the
synchronous generators driven by 8 horse pow er dc motors. adaptive settin g section o f the algorithm . A three phase fault
T hese w ere con n ected to a 'local' load and a 200 V , 3 phase o f sligh tly longer duration than the one used to trigger the
laboratory 'Utility' supply. T his w as useful for testing the pow er sw in g, w as then used to cau se a pole slip.
algorithm in a real life situation, sin ce harm onics and heavily
alternating loads nearby m ade the supply far from ideal.
4. S im u l a t io n T e st R e s u l t s .

Figure 2 sh ow s the algorithm variations for the laboratory

system w eak field pole slipping test. Inspection o f plots (a)
a) Generator Reactive Power and (b ) sh ow that the generator w as in itially operating
overexcited at very low power. A t a tim e o f Is on the record
> -1500 the p ow er input to the generator w as increased so that the
Qtrip m achine operated at approxim ately 2.5 kW. O bservation o f the
(AP/A t)mm curve in plot (c) sh ow s that because P t w as not
e- -4500
updated during this period, the trip setting (AP/A t)mm gradually
2.0 4.0 5.0 decreased , resulting in a d e-sen sitisin g o f the algorithm . This
Time (s) is b en eficial sin ce the lik elihood o f a p ole slip is higher during
b) Generator Real Power adjustm ent to a higher output level.
12 3000
At a tim e o f 3s on the record, the adaptive setting part o f the
I 2(XX)
algorithm updates the value o f P t to the new operating level.
TRIP
1000 This caused a drop in m agnitude o f (AP/A t)mm , but it did not
drop back to the initial valu e, due to Pt being at a higher
2.0 3.0 4.0 5.0 level. T his is the desired effect sin ce at higher input pow ers,
Time (s) more severe p ole slip can occur and m ore negative valu es o f
c) Generator Rate of Change (AP/A t) are produced.
of Power
3000
Shortly after the update, the resistor w as con n ected in parallel
5 1500
with the field , cau sin g the p ole slip to occur. Inspection o f the
reactive p ow er plot sh ow s that this criterion w as satisfied at
%-1500 (AP/At> 3.75 s, w h en the reactive p ow er fell b elow the trip setting,
-3000
Q trip, in d icatin g that the load angle had increased to a value
Tune (s) above 90°. Inspection o f the real pow er plot sh ow s that after
insertion o f the parallel resistor, the generator's field w as
FIG U R E 2 - 5 kVA G E N E R A T O R PO LE SLIP w eak en ed su fficien tly so that the generator could not maintain
the lev el o f output required, the real pow er criterion w as
therefore satisfied on inserting the resistor.
The laboratory m odel w as also used to test the algorithm
during p ow er system fault conditions. A ll o f the fault types The final criterion (AP/At) , can be seen to fall b elow its trip
possible w ere placed tem porarily on the local load busbar by settin g (A P /A t)mm at a tim e o f 4.3s. This m eant that all three
sw itch in g in 'fault' resistance. criteria w ere satisfied for the one and a h a lf cy c le required and
a trip sign al w as therefore produced. The trip signal can be
The com puter based dynam ic sim ulation package w as used to found on the real pow er plot, it can be seen that the trip was
test the algorithm 's perform ance against pow er sw in gs and p o le issued a sign ifican t tim e before the p ole slip occurred. The
slipping. The m odel used w as based on the sim ple one usual point w here pole slip is said to have occurred is when
m achine m odel (5 8 8 M V A ) con n ected to an infinite bus m odel the m achine g o e s from generator to m otor action, i.e w hen the
(5 8 .8 G V A ). A stable sw in g w as caused by placing a fault on pow er output o f the m achine g o e s n egative. Unfortunately, in

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 396

this case the whole pole slip could not be recorded, due to Inspection of the (AP At)mm curve shows that until the update,
limitations in the data acquisition system. it stayed at its high power level, resulting in the algorithm
being de-sensitised for a small duration. This is unavoidable,
Figure 3 shows the algorithm's response to a two phase to but is of no great concern since in the event of a pole slip it
earth fault. Of all the fault types tested, this and the three would just result in the algorithm producing a trip signal
phase faults provided the toughest conditions for testing delayed by a small time. At 4.75 s on the record the fault was
algorithm stability. The reactive power criterion was satisfied placed on transformer bus, and removed again at 5.2s.
for a small period on removal o f the fault, at 1.4 s on the
record. During this time the real power criterion is satisfied for
a) Generator React K c Power
a very short amount of time, before the real power output of
„ 500
the generator jumps above the trip setting Pt. It is this short | 250
interval where algorithm mal-operation is most likely. 0
However, inspection of the (AP/At) plot shows that during % -250
Qtrip
this period, the (AP/At) signal jumps to a negative value off ^ -500 ,V.
1.0 2 .0 3.0 4.0 5.0 6.0
the scale for a very short time, before escalating to a very high Time (s)
positive value for a short time. These oscillations took less
b) Generator Real Power
than one power system cycle, and so the algorithm was -----n n a
1000
---- Pt A_____ ,
continually restrained. 500 “ t ... ....J\ ‘
i. 0 — r ft f
* T R ffW
s -500
CL Hi
.in n n .................. n .
Fault on Fault off a) Generator Reactive Power 3.0 4.0 5.0 6.0
Time (s)
e) Generator Rate o f Change
i ° of Power
5 -1500 Q trip
/ H .................................
^ -3000 (AP/At)
L S -3000

T im e ( ®:-4500
5.0 6.0
b) Generator Real Power
Time (s)

FIGURE 4 - 588 MVA GENERATOR POLE

SLIPPING DUE TO A THREE PHASE FAULT.

6L 2000 NO TRIP

0.5 1.0 1.5 2.0 2.5 3.0

Time (s)
On removal of the fault the generator's reactive power was less
c) Generator Rate of Change
of Power
than Qtrip, satisfying the reactive criterion indicating that the
2000 load angle was above 90°. At the same time, the real power
I 0
1
I
. output is much greater than the nominal value before the fault
^ -2000 I (AP/At)m
5;-tooo □
0 0.5 1.0 1.5 2.0 2.5 3.0
This invoked the adaptive Pmwc part of the algorithm, which
tracked the real power output up to its maximum value, and
Time (s)
then held this value for 1 second before returning back to
1.25*Pt.
FIGURE 3 - 5 kVA GENER ATOR 2 PHASE TO
EARTH FAULT TEST
This action resulted in (AP/At)mi„ falling to a lower level, thus
desensitising the algorithm to the pending transient pole slip
The algorithm also successfully restrained for all of the other The (AP/At) signal still fell below the trip setting (AI’/At),,,,,,
fault types tested. for one cycle, and during this time the other two criteria were
satisfied. A trip signal therefore resulted.
Figure 4 shows the results to a test using the dynamic
simulator. Initially the generator was operated at full power, it The algorithm successfully restrained for the power swing test,
was then reduced to half rated operation. The condition which produced a severe swing in load angle up to a value of
monitoring part of the algorithm updated at a time of 3.75s on 160°.
the record.

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 397

4. F ie l d T e s t s . 5. C o n c l u s io n s .

A series of field tests were conducted using a 625 kVA diesel During laboratoiy model, computer simulation and field tests,
generator connected via a transformer to the 11 kV network the power based pole slipping algorithm has proven to be very
supply. In the most dramatic of the tests, a pole slip was successful in detecting pole slips while remaining stable during
induced by quickly ramping the diesel power at a rate faster periods of other types of power system disturbances.
than the generator controls could respond.
It offers the immediate advantage of detecting when the
synchronous Dispersed Storage and Generation unit is
committed to a pole slip rather than has actually slipped and
therefore offers the opportunity of disconnecting the machine
'TRIP BOUNDARY Generator Reactive Power from the network before possible damage or major disturbance
0 to the power system.
S' -500,000
The relay's settings are determined by readily available
I
I -1,000,000 generator data and thus the need for simulation studies is
CoL. avoided.
-1.500,000
Generator Real Power
6. Refer en ce s .
2.000,000
§ Pmax
^ 1,000,000
1. J. A. Imhof et al, 'Out of Step Relaying for
»• Generators - Working Group Report,' IEEE
s
Transactions PAS-96, No.5, Sep/Oct 1977, pp 1556-
1564.
- 1,000,000 2. A. Stalewski, J. L. H. Goody and J. A. Downes, 'Pole
Generator Rate of Change of Real Power Slipping Protection,' Developments in Power System
20 , 000,000
protection, IEE Conf. Publication, No. 185, pp 38-45.
^ 0 3. S. Shiwen, 'Microcomputer Based Out-of-Step
t Protection for Large Generator,' IEE APSCOM Hong
P h - 2 0 ,0 0 0 ,0 0 0

Kong, Nov 1991, IEE Conf. Proc. No 348, Nov 1991,

-4 0 ,0 0 0 ,0 0 0
pp 839-842.
50 550 1050 1552 4. D.W. Smaha, 'Out-of-Step Relay Protection of
Time (ms) Generators.' IEEE Tutorial on the Protection of
F ig ure 5. P o w e r B a s e d A l g o r it h m R espo n se Synchronous Generators, Section 8, 95-TP-102,
to 625 k V A D iesel G e n e r a t o r P ole S l ip . Power System Relaying Committee 1995.
5. H.K. Clark & J.W. Feltes, 'Industrial and
Cogeneration Protection Problems Requiring
Simulation.' IEEE Trans. Industry Applications, Vol.
IA-25, No.4, July 89, pp 766- 775.
Figure 5 shows the power based pole slipping algorithm's 6. M A Redfem and M J Checksfield, 'A New Pole
response to this test. The real power plot shows where the Slipping Protection Algorithm for Dispersed Storage
algorithm tripped, which for the first pole slip occurred just and Generation using the Equal Area Criterion.' IEEE
before 1000 ms on the record. Allowing for a circuit breaker Transactions on Power Delivery PWRD Vol 10, No
operating time of 100 ms, this would have isolated the 1, Jan 1995, pp 194-202.
generator before the actual pole slip. The real power plot also 7. V. A. Venikov, 'Transient Phenomena in Electrical
reveals the potentially damaging effects of pole slipping, since Power Systems', Pergamon Press, 1964.
the peak in power output after the first pole slip reaches 2.7
MW, five times rated power. By tripping before the pole slip
occurred, the new algorithm would have avoided this
dangerous surge in the power output. Due to the finite
disturbance recording time, the transition from normal to
underexcited generator operation was not captured.

MJ.Checksfield/PhD/University of Bath 1996 Appendix D, Page 398

 PAPER ELEVEN

”A NEW POLE SLIPPING PROTECTION

FOR NUMERICAL RELAYING"

by M A Redfem and M J Checksfield

Presented at
11th CEPSI 96, Paper 12-29
Kuala Lumpur, Malaysia,
November 1996, ppl2.29/l-12.29/10.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 399

 A NEW POLE SLIPPING PROTECTION FOR NUMERICAL RELAYING.

Miles A R edfem M atthew J Checksfield

University of Bath
Claverton Down
Bath
B A 27A Y
United Kingdom

ABSTRACT

W ith the increasing num ber o f synchronous generators at risk o f damage from pole slipping, pole slipping
protection is a concern. M achine and system problem s can result i f a generator is p erm itted to pole slip
continuously. The present impedance based m ethods can be d ifficu lt to set, and detailed com puter based
sim ulations are required to derive reliable relay settings, due to the unpredictable nature o f the apparent
impedance locus during pole slipping. This paper presents a new technique which sim plifies the setting
procedure fo r pole slipping protection whilst offering im proved perform ance over conventional schemes. R esults
fro m fie ld trials on a 625 kV A industrial diesel generator and sim ulations using the A TP programme are used
to compare the perform ance o f the new algorithm with the existing impedance based schemes.

1.0 IN T R O D U C T IO N

G enerator pole slipping protection is necessary because when a generator looses synchronism w ith the utility
supply to w hich it is connected, machine damage and system problem s can result. The trend to low
transm ission system reactances compared to the high reactances o f synchronous generators increases the
likelihood o f the electrical system centre appearing inside the generator during pole slipping rather than out in
the transm ission system. This requires generator pole slipping protection relaying to detect this out o f step
situation, rather than transm ission line relaying [1], Generators are also being operated closer to their stability
limits, especially during light load periods. They are therefore more prone to instability and hence pole
slipping[2].

Pole slipping is a particular concern for embedded generation units installed to operate in parallel w ith utility
distribution systems. The electrical centre almost always appears inside the generator or its transform er due to
the small rating o f the generator compared to the utility system to which it is connected. The probability o f pole
slipping on distribution systems is increased since they often have long fault clearance tim es due to the IDMT
overcurrent protection commonly employed [3] . Due to grading considerations, fault clearance tim es o f up
to one second are possible. If the generator protection scheme is not configured to disconnect the generator for
all fault conditions that occur, then the probability o f pole slipping is very high [4],

Pole slipping protection has existed for many years. The conventional impedance based forms o f protection can
be difficult to set [1,2,4]. M ost schemes therefore require detailed power system sim ulations if they are to be
set properly. U nfortunately most techniques for calculating the impedance locus for pole slipping and stable
swing conditions do not take into account effects such as generator damping, non-constant flux linkage, and
control system operation. All of these complicate the situation. The process o f perform ing a detailed com puter
sim ulation can be time consum ing and expensive. It can therefore be a constraint on the choice o f pole slipping
protection.

1.1 Pole Slipping

Pole slipping occurs due to an imbalance between the mechanical input power and the electrical output
pow er o f a synchronous machine, which causes its rotor to accelerate. If the rotor accelerates too far,
the machine will not be able to produce the synchronising forces necessary to keep its rotor in
synchronism with the stator magnetic flux, and the excess m echanical forces w ill cause a pole slip to
occur.

MJ.Checksfield/PhD/University of Bath 1996 Appendix D, Page 400

 The power imbalance can be caused by an external disturbance on the system such as a slow fault
clearance, line switching operations which increase the system's transfer impedance, or a faulty
generator control system. The loss o f synchronism point at which pole slipping occurs is defined as the
point where the generator's real power output becom es zero. Once this point has been reached the
generator will briefly accelerate into a m otoring condition, where it will absorb pow er from the
electrical system. It will then go back into a generating condition. If it cannot produce the
synchronising forces necessary to pull it back into synchronous operation, it w ill repeat the pole
slipping cycle again and again.

Pole slipping can cause generator dam age in numerous ways. Generators can experience currents in
excess o f three phase fault levels for periods m uch longer than the sub-transient fault period. M achine
windings are generally designed to withstand three phase fault levels. The m echanical forces produced
by the higher current levels experienced during pole slipping can therefore dam age the windings. The
unit transformer windings may also be exposed to high mechanical stresses because o f these high
current levels. If prolonged pole slipping is perm itted, then the high current levels in the stator, and
resultant induced slip frequency currents in the rotor can cause over-heating. AC/DC excitation
systems can suffer insulation breakdown due to the excitation rectifiers blocking the field current as
it tries to reverse during the pole slip cycle. H igh transient shaft torques also result, especially in steam
turbine units. If the slip frequency o f the generator coincides w ith a torsional natural frequency in the
shaft system, the resultant torques can break the shaft if it is not suitably over-rated.

System voltage problems can also occur during pole slipping. In extreme cases the voltage depressions
produced can cause plant outages, and even a cascading loss o f synchronism o f adjacent m achines [5].
The w orst voltage fluctuations occur at the system centre. The system centre is the point where the
contribution from the generator equals that from the rest o f the system. At the system centre, the
voltage will be low w hilst the current will be high, giving the appearance o f a three phase short circuit
fault. D istance protection, under voltage and over current protection may all operate during pole
slipping.

2.0 CONVENTIONAL POLE SLIPPING PROTECTION

There are num erous impedance based schemes for detecting pole slipping[6]. The loss o f field, mho and the
single blinder schemes are the most commonly used techniquesfl],

2.1 Generator Pole Slipping Impedance Characteristics.

The majority o f conventional methods for detecting pole slipping use impedance relays to sense the
variations in apparent impedance at the generator or high voltage transformer term inals during the pole
slip. Using non com puter based simulation techniques, the impedance loci can be derived using the
assumption that the ratio o f generator to source impedances, Eg/Es remains constant. G enerator saliency
and generator dam per effects are also neglected, transient fault impedance effects are assumed to have
decayed, shunt loads and shunt capacitance effects are ignored, effects o f automatic voltage regulators
and governors are neglected, and the source voltages behind their equivalent impedances are assumed
to be sinusoidal and at fundamental frequency [1,2,7]. Once the theoretical loci have been calculated,
the setting o f the impedance schemes can be derived. Unfortunately, deriving the settings based on
theoretical impedance loci can result in protection schemes which either fail to detect all pole slipping
conditions, or nuisance trip for recoverable pow er swings or faults. Further inform ation can be found
in literature [1,2,6,8,9],

It is therefore strongly recommended that com puter based simulation studies are perform ed for each
specific generator so that improved settings can be chosen [1,2,6,8],

2.2 Generator Loss o f Field Protection Scheme

Loss o f field relaying inherently provides a degree o f pole slipping protection. Due to the time delays
em ployed to prevent nuisance tripping, loss o f field schemes may fail to detect some pole slips because

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 401

 the time delay may prevent the scheme tripping for pole slips whose impedance locus travels through
the characteristic faster than the time delay. M odem loss o f field relays have the facility to adjust the
reset delay, and this helps to improve the range o f pole slipping frequencies w hich can be detected.
The problem is that several pole slips may occur before the relay operates.

2.3 The Mho Element Scheme

This is the simplest type o f impedance based pole slipping protection, it uses a mho characteristic
w hich reaches into the transformer and generator transient reactance. The scheme's main advantages
are that tripping can occur before the pole slip has occurred, and that it provides a degree o f back up
protection for faults occurring in the transformer and generator. Its m ain disadvantage is that very
careful selection o f the mho reach is necessary to avoid tripping during power swings for w hich the
pole slipping protection should remain stable. The mho circle reach is often set so that the impedance
locus enters the mho circle at a system load angle o f 120°. If this can not be achieved, a time delay
is often included to increase the scheme's stability [2].

2.4 H ie Single Blinder Scheme

There are many possible variations o f this scheme [6,8,10]. The basic scheme consists o f two blinder
elements which track the pole slipping impedance locus as it moves across the RX plane from right
to left. To restrict the operation o f the scheme to swings appearing in the generator or transform er, and
to prevent operation for recoverable swings that pass through both blinder elem en ts[l,l 1], the relay is
supervised by a mho element. This also stops tripping for oscillations in reactive flow after
synchronising[ 1, 9,10],

The blinder units are set parallel to the generator and transform er impedance. This scheme operates
when the impedance locus enters the relay characteristic from one side o f the R-X plane, and leaves
from the other, taking longer than a preset time. The recom m ended settings for the blinders are 90° and
270°, which relieves the circuit breaker from a high out o f phase arc interruption duty. Although, the
scheme is inherently more stable against recoverable pow er swings and faults, detailed com puter based
sim ulations should still be perform ed in order to find the correct blinder, mho, and tim er settings [1,2,
4,11], Details o f the scheme logic used in this paper are given by Goody [8].

3.0 THE NEW POWER BASED POLE SLIPPING ALGORITHM

The algorithm predicts that pole slipping will occur by detecting if the load angle is still increasing once the
Critical Stability Point (CSP) has been exceeded [12,13]. If the generator operates past the CSP, stability cannot
be m aintained and a pole slip is inevitable. In designing a pole slipping algorithm, it m ust be able to predict
when a generator is committed to a pole slip, or detect when a pole slip is actually occurring. It m ust also
remain stable for all other power system conditions, including faults and recoverable pow er swings.

The algorithm uses reactive power, real power, and rate o f change o f real power to detect operation past the
CSP. The algorithm is shown in equation (1).
Q < Qtrip * Continuously
TRIP and )
if: P < Pt } For 1-5 Power (1)
and }
(A P /A t) < (A P /A t)„ {„ } System Cycles

Where Q is the generator three phase reactive power, Qtrip is the reactive pow er trip setting, based on the
quadrature axis synchronous reactance. P is the generator real pow er output, Pt is the real power trip setting
which is derived by the algorithm. (AP/At) is the rate o f change o f real power, (AP/At)mill is the rate o f change
o f power trip setting which is derived by the algorithm. The 1.5 power system cycle time constraint is to ensure
that the algorithm remains stable during short circuit faults. The majority o f faults will not satisfy all o f the
criteria o f equation (1), the few that do only cause the criteria to be satisfied for less than one pow er system
cycle.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 402

3.1 Condition Monitoring Techniques

The algorithm is designed to be included into a m icroprocessor based relaying environment. The only
information available to the relay is 2 phase voltages and 3 current signals provided by the relay's
transducers. Since generators may be operated over a range of output powers, a method o f estim ating
the prime mover input power, Pm is required so that the real power trip setting, Pt can be derived. The
microprocessor relaying platform has a limited amount o f processing power available to achieve this
goal. The input power, Pm is found by observing the power output o f the generator, and if this
remains within a pre-defined tolerance band for a sufficient amount o f tim e, then Pm must
approximately equal this value, the only discrepancy being produced by the losses in the generator[ 13],

3.2 Adaptive Trip Setting (AP/At)min

Using an assumed sinusoidal power load angle curve[12], a value for (AP/At) at the CSP can be
estimated for a given value o f slip at a given value o f Pt by;

(Ap /At)@csp = - s ♦Pm ax ♦ cos (180 - s in - '( - ^ L _ ) )= (AP/At)min (2)

Pmax
Note that the value o f slip chosen will not be the exact value at which the relay produces a trip output,
due to effects not taken into account in the derivation o f (2), such as damping power. Equation (2) is
used by the algorithm to continuously update the trip setting (AP/At)min depending on the variables
Pmax and Pt.

Pmax is a measure o f the maximum output power that the generator has produced in the last second.
It is used so that the algorithm can adjust its setting according to whether a transient or steady state
pole slip will occur. A steady state pole slip occurs due to control system m al-function, w hereas a
transient pole slip occurs after the generator has been severely disturbed, after a fault for example. If
the generator has been transiently disturbed, its output pow er during the first swing in load angle, be
it stable or unstable, will be much greater than its nominal output. This results in a greater magnitude
in (AP/At) at the CSP, the algorithm therefore needs to adjust its setting (AP/At)min accordingly. The
algorithm adjusts the setting, (AP/At)min according to (3);

If -fL . < 0.6 then; (AP/At) ^ - (AP/At).* ♦ (AP/At)*. (3)

Pmax
where (A P /A t)^ is an external relay setting based on the degree o f transient saliency in the generator.
Transient saliency arises due to differences in the values o f Xd' and Xq. (A P /A t)^ is given by;

(AP/At). = ( _L - _ L ) ♦ Seen (4)

Xq X d' 8
Where SgCT is the generator rating. For a steady state pole slip, no peak will occur in the pow er output
before the pole slip, the algorithm trip setting should therefore be at its most sensitive level to detect
the pole slip since low values o f (AP/At) will result.

During steady state conditions the value o f Pmax used by equations (2) and (3) is nominally set to
1.4 * Pt. The ratio Pt/Pmax equals 0.71, (A P /A t)^ is therefore not included in the calculation of
(AP/At)min during steady state conditions. If the generator's power output rises above the existing value
o f Pmax for greater than one power system cycle, then the value o f Pmax is updated to the new
maximum for a duration of one second. The one second limit is used so that the new value o f Pmax
is only used for the period where a pole slip or power swing associated with that value is likely to
occur. The one cycle constraint is necessary so that incorrect values o f Pmax do not result from
spurious spikes on the power waveform.

4.0 ALGORITHM VALIDATION

The pole slipping algorithm was validated and tested against the impedance based techniques detailed in section
two using results from field trials on a 625 kVA industrial diesel generator, and from com puter sim ulations on

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 403

the A lternative Transients Program (ATP) [14].

4.1 625 kVA Industrial Diesel Gene tutor Field Trials

The 625 kVA generator is an industrial diesel generator connected to the utility network via a
415/11,000 volt unit transformer and two parallel 2 km 11 kV cables. The fault capacity at the
generator term inals is 30 MVA. During normal service, there is 500 kW o f local load connected to the
generator. The field trials used a com mercial m icroprocessor relaying platform. It was not possible to
place faults on the power system to induce pole slipping, and hence a pole slip was induced by
increasing the diesel engine's output torque with the generator under-excited. This scenario could
happen due to operator error, or due to a mal-function o f the diesel engine control system.

4.2 ATP Computer Based Simulation Studies.

The ATP was used to simulate both recoverable pow er swings and pole slips. The sim ulations were
based on the pow er system described above. Pole slips and recoverable power swings were sim ulated
by placing a fault in the middle o f one o f the 2 km inter-connection cables. The fault was cleared by
disconnecting the faulty line. The generator AVR and governor system were included in the model.

5.0 SIMULATION STUDIES

5.1 625 KVA D iesel Generator Field Trial Pole Slip T est

The generator was permitted to complete two entire pole slipping cycles before the diesel engine torque
was reduced, and the generator allowed to pull back into synchronism. The new pow er based pole
slipping algorithm successfully detected both o f the pole slips, however, both the mho and single
blinder schem es failed to detect them. The loss o f field scheme did detect the pole slip.

The single blinder scheme characteristic, and the impedance locus o f the pole slipping generator are
show n in figure 1. The first pole slip occurred betw een 0.2 and 0.3 s, whilst the second occurred
betw een 0.55 and 0.65 s. The first pole slip was not detected due to the impedance falling outside the
supervisory mho relay. If the supervisory mho characteristic had been made larger, the scheme would
still have failed to detect the first pole slip as it did not cross the left hand blinder.

Left Blinder Right Blinder

0.44
supervisory
- 0.2 mho relay

equivalent
-0.4 system ^
a impedance
line
- 0.6
0.29

- 0.8 pole slip loci

0.55

0.65

-1.4
-0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8
Resistance, R (pu)
Figu re 1: : Single B linder Pole S lip p in g P rotection S chem e for
Industrial 625 kVA D iesel G en erator F ield T rial Pole Slip.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 404

 For the second pole slip, the swing impedance travelled across the impedance plane from right to left
as predicted by theory, but was again significantly below the supervisory mho characteristic. This was
due to the impedance of the generator being larger than the value o f Xd'. In fact the impedance o f the
generator appeared to be closer to the synchronous value Xd. This observation is consistent w ith slow
pole slip rates. At a slip rate of 0.1 % the generator reactance becomes 0.75 Xd [8].

The swing impedance o f figure 1 also shows why the mho scheme failed to detect the pole slip. The
size o f the mho characteristic is smaller than that o f the single blinder scheme supervisory mho circle,
and so both o f the pole slip impedance loci travelled under the relay characteristic. The mho scheme's
failure to trip highlights one of the constraints o f the mho scheme. Covering the whole range o f
possible pole slip generator reactances, from Xd' to 0.75 Xd can lead to too large a mho characteristic,
which is prone to nuisance tripping during recoverable pow er swings. Adding a time delay increases
the stability o f the scheme to recoverable swings, but as the time delay is increased, the scheme may
fail to detect fast pole slips whose locus travels through the characteristic too quickly. If the circle size
is decreased in order to increase stability, the scheme may fail to detect pole slips whose impedance
does not take the classical trajectory. The problem may be alleviated to some extent by co-ordinating
the scheme with an impedance based loss o f field scheme.

All o f the purpose designed pole slipping detection impedance based schemes failed to detect the pole
slip, due to the impedance falling below the mho relay characteristics. However, the loss o f field relay
used did detect the pole slip, since its mho characteristic was set to a value o f X d + Xd'/2, w hich for
the diesel generator equals a value o f -j 2.26 pu. Figure 1 highlights the degree to w hich the impedance
locus produced by pole slipping can vary from the locus predicted by theory. If conventional schemes
are set using theoretical impedance loci, then a high degree o f confidence in the settings cannot be
justified.

Criterion Trip Boundary Generator Reactive Power

0

O /T"S -0.5
* <2
- 1.0

o< 1.5
- 2.0 Generator Real Power
3.0 Pmax

o 2.0
Ph ALGORITHM TRIP
1.0
04
0

1.0
Generator Rate o f Change o f Power (RCP)

0
P.
graph limited for
-10
reasons o f clarity

-20
0 500.740 1001.462 1502.203 2002.790
Time (ms)
Figure 2: Power Based Pole Slipping Algorithm Response
to 625 kVA Diesel Generator Field Trial Pole Slip.

Figure 2 shows the response o f the power based pole slipping algorithm to the same test. N ote that the
times used in this figure are not the same as in figure 1 due to a different disturbance recorder being
used. The real power plot shows where the algorithm tripped, which for the first pole slip occurred just
before 1000 ms on the record. Allowing for a circuit breaker operating time o f 100 ms, the protection
would still have disconnected the generator before the pole slip.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 405

 The real power plot shows the potentially damaging effects o f pole slipping. The peak in power output
after the first pole slip reaches 2.7 MW, over five times rated power. By detecting the pole slip before
it occurred, the new algorithm would have avoided this dangerous deviation in power output.

5.2 ATP Computer Based 625 KVA D iesel Generator Simulation Results.

The simulation o f fault induced pole slipping o f the diesel generator showed that the mho and power
based pole slipping algorithms both tripped for the pole slipping simulation, whilst the single blinder
schem e failed to trip. The mho scheme tripped at a time o f 386 ms, and the power based scheme
tripped at a time o f 410 ms. The fault occurred at 100 ms and was cleared at 340 ms.

The mho schem e incorrectly tripped for the recoverable power swing simulation, at a time o f 376 ms,
for this simulation, the fault was cleared at 330 ms. The power based and single blinder pole slip
schem es both remained stable during the recoverable power swing. The loss o f field scheme did not
trip for any o f the simulations performed due to the time delay employed in the scheme o f 0.5 s.

Figure 3 shows the single blinder scheme characteristic, and the solid impedance locus shows the pole
slip impedance for the simulated pole slip. At fault clearance, the impedance locus changes rapidly
due to the effects o f pole scatter. After this, the impedance locus m oves downwards rather than right
to left across the RX plane as would be expected from a classical pole slip. The downwards movement
causes the mho elem ent to drop out before the impedance locus crosses the left hand blinder unit, the
schem e therefore fails to detect the pole slip.

The predominant downwards movement o f the impedance locus is due to the short time constant, Tdo'
o f the generator. Another simulation was performed with Tdo' increased from 2.0 to 6.0 seconds, and
the resultant impedance locus is shown as the dotted line in figure 3. This impedance locus takes a
more conventional trajectory. A short time constant means that the flux linkage o f the generator is not
constant during the pole slip. The fault causes the flux linkage to decay, which reduces the effective
generator voltage so the impedance locus produced m oves from a curve with a high value o f the ratio
Eg/Es to a curve with a low ratio o f Eg/Es. Where Eg is the effective generator emf, and Es is the
effective system source emf. Embedded generators w ill often have values o f Tdo' o f around 2 seconds
due to their higher per unit resistances.

0.2
Fault point

Pole
-0.1 scatter
3 Generator
a. effects
T do'= 6.0 s
X
-0.2

0 -0.3
1
-0.4

impedance locus
-0.5 during pole slip
f Gcnerator-
Tdo' = 2.0 s
-0.6
-0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3
Resistance, R (pu)
Figure 3: Single Blinder Pole Slipping Protection Scheme for Industrial
625 kVA D iesel Generator Simulated Pole Slip.

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 406

 Criterion Trip Boundary Generator Reactive Power

cx

« -4

Generator Real Power

Pmax ALGORITHM TRIP

_ -0.5

Generator Rate of Change of Power (RCP)

graph limited
Oh - 1 0 for reasons of
clarity
-20
0.20 0.40 0.60 0.80 1.00
Time(s)
Figure 4: Power Based Algorithm Response to Simulated Pole Slip
on 625 kVA Industrial Generator.

Figure 4 shows the response o f the power based algorithm for the simulated pole slip. The algorithm
tripped at a time o f 0.41 s on the record, which corresponds to the point where the power output o f
the generator is zero. Note the non-sinusoidal nature o f the generator real power waveform follow ing
fault clearance, which is due to the generator flux decaying. The algorithm does not reset until the
reactive criterion is reset, thus preventing relay output chatter. The algorithm function Pmax fails to
update to the peak in power output follow ing fault clearance due to the spiky nature o f the power
waveform. The effect on the rate o f change o f power trip criterion o f an update in Pmax to a greater
than nominal value can be seen at a time o f 0.73 s on the record. This action enhances algorithm
security towards recoverable power swings, whilst still enabling fast tripping during pole slipping.

Mho Pole Slipping

0.34 s
Characteristic

3
O.
X -0.1
$

-0.3

-0.2 -0.1 0 0.1 0.2

Resistance, R (pu)
Figure 5: Single Mho Pole Slipping Protection Scheme for
Industrial 625 kVA Diesel Generator Recoverable Pow er Swing.

Figure 5 shows the impedance locus for the recoverable power sw ing simulation, and the characteristic
for the mho circle scheme. The impedance locus was inside the characteristic for 60 ms, and since the
timer on the scheme was set to 40 ms an incorrect trip condition resulted. If the impedance locus had

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 407

 taken the conventional trajectory, as predicted by basic theory, then it would have left the circle moving
from left to right and a trip would not have occurred. The reason for the downwards component in
the impedance locus is again due to the short time constant o f the generator rendering the constant flux
linkage assumption used in deriving the impedance loci for setting purposes invalid.

Figure 6 show s the power based algorithm measurands for the simulated recoverable power sw ing test.
During the time when the reactive criterion is satisfied, between 0.35 and 0.5 s, the real power criterion
is not satisfied for long enough to cause tripping. The fault used in this simulation was just shorter than
the critical clearance time, and the resultant recoverable power swing is therefore the most severe
possible.

Criterion Trip Boundary Generator Reactive Power

1
0
•1
•2
■3
-4
■5 Generator Real Power
3
NO TRIP
2
1
0
06
1
Generator Rate o f Change of Power (RCP)
0
mr
'I® graph limited
3-20 for reasons
£ r 30 o f clarity
06 -40
-50
0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time (s)
Figure 6: Power Based Algorithm Response to Simulated
Recoverable Power Swing on 625 kVA Industrial Generator.

6.0 CONCLUSIONS

M icroprocessor relaying technology has meant that new adaptive pole slipping algorithms can be developed
which adjust their settings according to the generator operating point. The power based schem e discussed in this
paper successfully detected all o f the pole slipping conditions whilst restraining for stable swings and faults. In
addition to the performance o f the scheme, another advantage compared with conventional techniques is the
scheme's sim plicity to set. The new approach only requires the generator transient reactance Xd', synchronous
reactance, Xq and generator rating to be set. The rest o f the trip points are derived internally by the algorithm.

Conventional impedance based pole slipping protection requires detailed simulations o f the system to be
protected to derive reliable settings. Unfortunately, pole slipping impedance loci predicted from theory are
som etim es not accurate enough to provide reliable setting o f impedance based pole slipping schemes.

The mho schem e can mal-operate for stable swings if set too large, and can fail to detect some pole slips if time
delays are used in order to prevent nuisance tripping during stable swings.

The single blinder schem e is the most secure o f the conventional schemes. The price for this extra security is
that the blinders and supervisory mho settings must be carefully set to detect all pole slips. To confidently set
such a schem e, detailed computer based simulations o f the system to be protected are required, which cost the
applications engineer time and also money.

Embedded generators are very prone to instability and hence pole slipping, and do not have very predictable

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 408

impedance loci. Their short time constants and nearby plant can radically alter the path of the apparent
impedance locus predicted by theory. Computer based simulations are therefore the only guaranteed way of
finding settings which will work for the impedance based protection.

7.0 REFERENCES

[1] J.A. Imhof (Chairman) et al, "Out of Step Relaying for Generators - Working Group Report," IEEE
Trans. Power Apparatus and Systems, Vol. PAS-96, No.5, Sept 1977, pp 1556 - 1564.
[2] D.W. Smaha, "Out-of-Step Relay Protection of Generators," IEEE Tutorial on the Protection of
Synchronous Generators, Section 8, 95-TP-102, Power System Relaying Committee 1995.
[3] M.M.Elkateb & G Fielding, "Coordinating Protection and Control of Dispersed Generators",Tenth
International Conf. on Electricity Distribution, Parti, 5 - Consumers Installations, 1989, ppl31 - 135.
[4] H.K. Clark & J.W. Feltes, "Industrial and Cogeneration Protection Problems Requiring Simulation,"
IEEE Trans. Industry Applications, Vol.IA-25, No.4, July 89, pp 766 - 775.
[5] L J Powell, "An Industrial View Of Utility Cogeneration Protection Requirements", IEEE Trans, Vol
IA-23, No.5, Sep/Oct 1987, pp777-785.
[6] P Kundur, P o w e r System Sta b ility a n d C o n tro l, , EPRI Power System Engineering Series, McGraw-Hill
Inc, London 1993.
[7] E.W. Kimbark, P o w e r System S ta b ility - P o w er C ircu it B reakers a n d P ro tective R elays, Vol - II, New
York: Wiley & Sons, 1964, Chapter 10.
[8] J.L.H. Goody, "Overcoming Problems Associated With Impedance Measurement in Pole Slipping
Protection for Dinorwig", IEE Proc. Vol 133, Pt.C, N o.l, Jan 1986, pp 44-48.
[9] Westinghouse Electric Corporation, A p p lie d P ro tective R ela yin g , 1979, Ch. 19, Silent Sentinels
Publication.
[10] W.C.Morris, "One Slip Cycle Out of Step Relay" AIEE Trans, 1949, Vol 68 pp 1246 - 1249.
[11] A. Stalewski, J.L.H.Goody, & J.A.Downes, "Pole Slipping Protection", Developments in Power System
Protection, IEE Conf. Publication No. 185, pp 38 - 45.
[12] M.A. Redfem & M.J. Checksfield, "A New Pole Slipping Protection Algorithm for Dispersed Storage
and Generation Using the Equal Area Criterion," IEEE Trans. Power Delivery, Vol. 10, N o.l, Jan 1995,
pp 194 - 202.
[13] M.J. Checksfield & M.A.Redfem, "A Power Based Condition Monitoring Algorithm for Detecting
Dispersed Generator Loss of Synchronism", IPEC 95, Singapore, Feb 1995. Vol 1, pp 328 - 333.
[14] A lte rn a tiv e T ransients P rogram (ATP) R ule B o o k , July 87 Ed, Leuven EMTP Centre (LEC), Belgium

8.0 ACKNOWLEDGEMENTS

The authors would like to thank the University of Bath for the use of facilities. Special thanks is also given
to the Engineering and Physical Sciences Research Council (EPSRC) and GEC Alsthom T & D Protection and
Control Limited for their support of this project.

9.0 BIOGRAPHY

Miles A Redfem (M'79) received his BSc degree from Nottingham University and PhD degree from Cambridge
University in 1970 and 1976 respectively. In 1970, he joined British Railways Research, and in 1975, moved
to GEC measurements where he held various posts including Head of Research and Long Term Development
and Overseas Sales Manager. In 1986, he joined The University of Bath with interests in Power Systems
Protection and Management. He is chairman of the organising committee for the 1997 IEE conference
"Developments in Power System Protection", and is a member of IEE Professional Group PI 1, Power System
Measurement, Protection and Control.

Matt J Checksfield received his B Eng degree from the University of Bath in 1993. He is currently studying
for his PhD degree at the University of Bath and is sponsored by GEC ALSTHOM Protection and Control. His
interests are in embedded generation, renewable energy, and protection systems.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 409

 PAPER TWELVE

"FIELD TRIALS TO DEMONSTRATE THE PERFORMANCE

OF A NEW POLE SLIPPING PROTECTION"

M A Redfem, M J Checksfield and H T Yip

To be presented at >
6th International IEE Conference on 'Developments in Power System Protection',
25 - 27 March 1997, Nottingham, UK.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 410

FIELD TRIALS TO DEM ONSTRATE THE PERFORM ANCE OF A NEW POLE SLIPPING PROTECTION.

M A R edfem , M J Checksfield, H T Yip

University o f Bath, UK GEC ALSTHOM T & D Protection and Control Ltd, UK.

ABSTRACT For these reasons together w ith the opportunities offered

by numeric relays, research has been carried out into a
This paper presents a new approach to pole slipping new pole slipping algorithm. The new approach reduces
protection w hich elim inates the need for simulations, the complexity o f setting, can provide a trip signal
thus greatly sim plifying the setting process. The new before the pole slip actually occurs and can be included
adaptive algorithm uses power measurements to in a m odem m icroprocessor based protection scheme.
ascertain if the generator is com mitted to a pole slip. In
doing so it offers the prospect o f detecting the pole slip
before it occurs. The paper shows that this new POLE SLIPPING PROTECTION
approach works successfully by using results from field
trials perform ed on an industrial diesel generator. Pole slipping protection is required to reduce the
damage caused to the generator by the pole slipping
condition, and to ensure that the quality of supply for
INTRODUCTION consumers connected electrically close to the effected
generator remains acceptable. The severity of the
The majority o f conventional pole slipping relays use problems produced by pole slipping depends upon the
the variation in 'apparent' im pedance as seen at the relative rating o f the bus to w hich the generator is
generator's term inals to detect pole slipping, as described connected, the speed with w hich the generator pole
by ImJhof et al (1). If the impedance crosses two slips, and the overcapacity rating o f the generator.
measurement characteristics on an impedance plane
w ithin a specified tim e lim it, a trip signal is issued. If the generator is connected to an infinite bus, the
currents experienced during pole slipping can be in
There are two problem s w ith this approach. W ith the excess o f those experienced during three phase faults.
more secure o f the im pedance relay schemes, such as Generator windings are normally designed to withstand
the single blinder scheme, the relay trips after observing three phase fault levels, the higher current levels
a complete pole slip. This means that an unstable experienced can therefore damage the windings (2).
generator may already have been dam aged and the pole
slip could have caused voltage disturbances before the Turbine generators may suffer shaft damage since the
generator was disconnected from the system. Also, pulsating torques experienced, can torsionally excite
additional problem s could result if the pole slipping resonant sections o f the turbine shaft. M asrur et al (3)
impedance locus does not take the 'ideal' trajectory found that the generator-exciter section was most at risk
across the impedance plane and the relay fails to detect after the fir st torsional oscillation produced by a pole
the phenomena. Small em bedded generators and slip. The cum ulative effects o f sm aller oscillations may
industrial generators located close to induction motors also lead to significant shaft damage. It can therefore be
can suffer from this problem , Clarke & Feltes (2). concluded that if shaft damage due to pole slipping is to
be minimised, the generator must be disconnected from
Choosing the correct trip boundaries and relay timer the power system before the pole slip occurs, in the
settings for the im pedance based schemes can therefore hope that the torsional impulse will be avoided.
be a difficult process. If the w rong relay settings are
chosen, the pole slip may go undetected. If the settings Synchronous m achines are not designed to operate
are too sensitive, the relay may mal-operate due to a asynchronously, the resulting induced currents
power swing or a pow er system fault. experienced during pole slipping can cause stator and
rotor overheating. Induced field w inding currents can
The MHO schem e is a sim pler schem e and can trip stress the exciter insulation and semiconductor devices
follow ing the start o f the pole slip thus reducing the due to the high voltages they produce. The
potential for both dam age to the generator and semiconductor devices block reverse field current flow,
disturbance to the system . H ow ever it is a less secure a high voltage therefore results when the field current
schem e, still suffering from the above setting dilemmas. cannot reverse during the pole slip.

M.J.Checksfi el d/PhD/University of Bath 1996 Appendix D, Page 411

The effect o f pole slipping on other consumers Generator Pre-Fault
connected close to the effected generator is most Power . , power cu™e Critical
pronounced if the generator is connected to a weak
Stability
system. Prime
mover Point
input (CSP)
The voltage fluctuations can cause plant outages, and in
power,
extreme cases, cause a cascading loss o f synchronism o f
Pm
nearby machines, including both generators and motors, Post Fault
Powell (4). If the voltage depressions last for too long power curve
a period, then induction motors may stall depressing
Fault-Duration
system voltage further.
power curve

Transmission line and other feeder protection relays may

83 180 Load Angle
also mal-operate during pole slipping, since at one
location in the system, a pole slip appears as a three
phase fault, i.e a low voltage and high current. The Figure 1: Equal Area Criterion Diagram.
pole slipping condition can also cause other protection
relays to mal-operate, such as under-voltage and over­ The algorithm uses reactive power, real power, and rate
current relays. o f change o f real power to detect operation past the
CSP. The algorithm is shown in equation (1).

T H E EQUAL AREA C R IT E R IO N TRIP IF :-

Q < Qtrip All criteria

The Equal Area Criterion is one o f the simplest methods
for assessing generator stability for a given transient and must be
disturbance. If the power/load angle curves for all states satisfied
o f power system operation are known, then stability and P < Pt for ( 1)
hence the likelihood o f a pole slip can be estimated for 1.5 power
and
a given disturbance. The disturbances most likely to system cycles
cause a pole slip are severe faults, large load switching (AP/At) < (AP/At) continuously
operations, and generator control system faults.

Figure 1 shows the Equal Area Diagram for clearance o f Where Q is the generator reactive power, Q trip is the
a three phase fault on one side o f a double circuit line. reactive power trip setting, P is the generator real power
The fault was removed by disconnecting the faulty output, P t is the real power trip setting, which is derived
circuit, leaving the healthy side o f the line to transfer by the algorithm. (AP/At) is the generator rate o f change
power at a reduced capacity. o f real power, (AP/At)min is the rate o f change o f power
trip setting which is also derived by the algorithm The
It can be seen from Figure 1 that the limiting point for 1.5 power system cycle time constraint is to ensure that
stability occurs when the generators output power falls the algorithm remains stable during short circuit faults.
below its input power, this point is referred to as the The majority o f faults w ill not satisfy all o f the criteria
C ritical Stability Point (C SP). If the generator operates o f equation (1), the few that do only cause the criteria
past this point, then stability cannot be maintained, and to be satisfied for less than one power system cycle
a pole slip will occur.
The reactive power trip criterion, Q trip, is calculated
using the generator quadrature axis synchronous
TH E NEW PO LE SL IP PIN G A L G O R IT H M reactance,.Xq. The reactive power output characteristics
o f a generator are such that at a load angle o f 90°, the
The algorithm predicts that pole slipping will occur by reactive power demand is given by;
detecting if the load angle is still increasing once the
CSP has been exceeded, R edfem & Checksfield (5). If r 2
= Q trip (2)
the generator operates past the C SP, stability cannot be
maintained and a pole slip is inevitable In designing a
pole slipping algorithm, it must be able to predict when Figure 1 shows that the CSP occurs at load angles
a generator is committed to a pole slip, or detect when greater than 90°. The reactive power criterion is
a pole slip is actually occurring. It must also remain therefore used to indicate when the generator is
stable for all other power system conditions, including operating in a potentially unstable region
faults and recoverable power swings.

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 412

Condition M onitoring Techniques the degree o f transient saliency in the generator.
Transient saliency arises due to differences in the values
The algorithm is designed to be included into a o f X d ' and Xq. (AP/At)^ is given by;
m icroprocessor based relaying environment. The only
inform ation available to the relay is the voltage and (AP/AtY = ( — *S (5)
current signals provided by the relay's transducers. Xq Xd'
Since generators may be operated over a range o f output W here S is the generator rating. For a steady state
powers, a method o f estim ating the prime m over input pole slip, no peak will occur in the pow er output before
power, Pm is required so that the real pow er trip setting, the pole slip, the algorithm trip setting should therefore
Pt can be derived. The m icroprocessor relaying platform be at its most sensitive level to detect the pole slip since
has a limited amount o f processing pow er available to low values o f (AP/At) will result.
achieve this goal. The input power, Pm is found by
observing the power output of the generator, and if this D uring steady state conditions the value o f Pmax used
remains within a pre-defined tolerance band for a by equations (3) and (4) is nom inally set to 1.4 * Pt.
sufficient amount o f tim e, then Pm must approxim ately The ratio Pt/Pm ax equals 0.71, (AP/At)^ is therefore
equal this value, the only discrepancy being produced by not included in the calculation o f (AP/At)min during
the losses in the generator. A safety factor is also steady state conditions. If the generator's power output
included, by making Pt equal to 90 % o f the estimated rises above the existing value o f Pm ax for greater than
value o f Pm. one power system cycle, then the value o f Pmax is
updated to the new maximum for a duration o f one
second. The one second limit is used so that the new
Adaptive Trip Setting (AP/At)niin value o f Pm ax is only used for the period where a pole
slip or power swing associated w ith that value is likely
Using an assumed sinusoidal pow er load angle curve to occur. The one cycle constraint is necessary so that
(5), a value for (AP/At) at the CSP can be estim ated for incorrect values o f Pmax do not result from spurious
a given value o f slip, s at a given value o f Pt by; spikes on the power waveform.

(AP/At) . =- s * Pm ax * cos (8 )
(3) ALGORITHM VALIDATION
Pt
where 5 = 180 - sin_1( ) )
Pmax
The pole slipping algorithm was validated and tested
Note that the value o f slip chosen w ill not be the exact using results from com puter sim ulation, laboratory tests
value at w hich the relay produces a trip output, due to and field trials on a 625 k VA industrial diesel generator.
effects not taken into account in the derivation of (3), The industrial generator was connected to the utility
such as dam ping power. Equation (3) is used by the network via a 415/11,000 volt unit transform er and two
algorithm to continuously update the trip setting parallel 2 km 11 kV cables. The fault capacity at the
(AP/At)min depending on the variables Pmax and Pt. generator term inals is 30 MVA. During normal service,
500 kW o f local load is connected to the generator.
Pmax is a measure o f the maximum output pow er that
the generator has produced in the last second. It is used The series o f tests included large local load changes,
so that the algorithm can adjust its setting according to synchronisation, generator run up, and one very
whether a transient or steady state pole slip w ill occur. interesting pole slip scenario.
A steady state pole slip may occur due to control system
mal-function, w hereas a transient pole slip occurs after It was not possible to place faults on the pow er system
the generator has been exposed to a disturbance such as to induce pole slipping, and hence pole slips were
a fault. If the generator has been transiently disturbed, induced by increasing the diesel engine's output torque
its output pow er during the first swing in load angle, be with the generator under-excited. This scenario could
it stable or unstable, will generally be m uch greater than happen due to operator error, or due to a mal-function
its nominal output. This results in a greater magnitude of the diesel engine control system. A range of pole
in (AP-'At) at the CSP, the algorithm therefore needs to slips and loss o f excitations were performed in order to
adjust its setting (AP/At)mjn accordingly. The algorithm fully test the algorithm.
adjusts the setting, (AP/At)min according to;
Diesel Generator Field Trial Test Results.
I f ———
7>___ < 0.6 then,
Pmax
(4) The new algorithm successfully restrained for the load
(AP/At) = (AP/At) (AP/At)ln change, run up and synchronisation tests. The generator
was permitted to complete two entire pole slipping
where (AP/At)tran is an external relay setting based on cycles before the diesel engine torque was reduced, and

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 413

the generator allowed to pull back into synchronism. Results from field trial tests on a 625 kVA industrial
The new power based pole slipping algorithm diesel generator have proved that the algorithm
successfully detected both o f the pole slips. successfully detects pole slipping before the actual pole
slip occurs. The new approach also has the added
Figure 2 shows the response o f the pow er based pole advantage that it is simple to set, the generator transient
slipping algorithm to a pole slipping test. The real reactance, the quadrature axis synchronous reactance,
power plot shows w here the algorithm tripped, which and the generator rating are required to derive settings
for the first pole slip was at 1000 ms on the record. which will detect the whole range o f pole slipping
Allowing for a circuit breaker operating time of 100 ms, conditions.
the new algorithm w ould still have disconnected the
generator before the actual pole slip occurred. Criterion Trip Boundary Generator Reactive Power

The reactive pow er plot shows that when the disturbance

| -0.5
recorder was triggered, the reactive power trip criterion
had already been satisfied. This indicates that the - 1.0
generator was already operating w ith a potentially
unstable load angle o f greater than 90°. The real power
plot shows that the generator was at its maximum power - 2 . 0'
Generator Real Power
output o f 550 kW w hen the disturbance recorder was 3.0'
triggered. Pmax
2.0
ALGO]
At a time o f 1000 ms on the record, the real power
begins to decrease, indicating the onset of instability.
This is a steady state loss o f stability, brought on by
I
PL,

- 1.0
having too much input pow er for the level of excitation. Generator Rate of Change of Real Power
The low level o f excitation m eans that the generator was
unable to transfer the m echanical input power to
electrical output power. The m ism atch in power caused
the generator rotor to accelerate and the machine to -10 graph limited for
eventually lose synchronism and pole slip. reasons of clarity
-20
501 1002 1502 2003
Shortly after the real pow er begins to decrease, the Time (ms)
algorithm detects that the generator is operating past its
CSP and therefore trips. A llow ing 100 ms for circuit Figure 2: Pole Slipping Algorithm Response to 625 kVA
breaker operation w ould result in the generator being Diesel Generator Field Trial Pole Slip.
disconnected ju st before the real power went negative.
This would place a high sw itching duty on the generator REFERENCES
circuit breaker since it w ould have to disconnect as the
voltages across its poles approached 180 ° out of phase. 1. Im hof J A et al, "Out o f Step Relaying for
Generators - W orking Group Report". IEEE Transactions
The real pow er plot shows the potentially damaging in Power Apparatus and Systems. Vol PA S-96. No.5,
effects o f pole slipping. The peak in pow er output after September 1977, pp 1556 - 1564.
the first pole slip reaches 2.7 M W , over five times rated 2. Clark H K, and Feltes J W ," Industrial and Co­
power. By detecting the pole slip before it occurred, the generation Protection Problems Requiring Simulation",
new algorithm would have avoided this dangerous IEEE Transactions in Industrial A pplications, Vol IA25,
deviation in power output.. No.4, July 89, pp 766 - 775
3. Masrur M A, Ayoub A K, and Tielking J T,
"Studies on Asynchronous O peration o f Synchronous
CONCLUSIONS M achines and Related Shaft Torsional Stresses", IEE
Conference Proc. Part-C, Vol. 138, Jan 91, pp 47-56
M icroprocessor technology has m eant that adaptive, 4. Powell L J, "An Industrial View of Utility Co­
condition m onitoring pole slipping algorithms can be generation Protection Requirements", IEEE Trans. Ind.
developed which adjust their settings according to the App.. Vol IA-23, No.5, Sep 87, pp 777-785.
generator operating point. The approach presented in 5. Redfem M A and Checksfield M J, "A New
this paper can use sensitive settings to detect pole slips Pole Slipping Protection Algorithm for Dispersed
which occur due to a loss o f steady state stability, whilst Storage and Generation Using the Equal Area Criterion",
automatically selecting more secure settings when IEEE Trans. Power D elivery, Vol. 10, N o.l,
recoverable power swings occur. January 1995, pp 194-202.

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 414

 PAPER THIRTEEN

’TRANSIENT STABILITY CONSIDERATIONS ASSOCIATED

WITH CONNECTING EMBEDDED GENERATION
TO UTILITY NETWORKS”

M A Redfem and M J Checksfield

Proceedings of the second ERA International Conference

"Protecting Electrical Networks and Quality of Supply"
London, January 1997, pp 6.3.0-6.3.10

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 415

 1

TRANSIENT STABILITY CONSIDERATIONS ASSOCIATED

WITH CONNECTING EMBEDDED GENERATION
TO UTILITY NETWORKS.

M. A. Redfem and M. J. Checksfield

School of Electronic and Electrical Engineering

University of Bath, BATH, UK.

1. In t r o d u c t io n .

With the growing number of embedded generators being installed to operate in parallel with
utility distribution systems, transient instability is increasingly becoming recognized as an area
of concern. Generator instability, can damage the embedded generator and compromise the
quality of supply of the local power system. Unfortunately, embedded generators generally
have high per unit impedances, low inertia and short transient time constants. These together
with the long fault clearance times associated with the types of protection used on distribution
networks all contribute to the increased probability of transient instability.

Transient instability of an embedded generator results in loss of synchronism between the

embedded generator and the supply to which it is connected. This causes the embedded
generator to pole slip with respect to the utility network's grid supply. This pole slipping is
accompanied by dramatic fluctuations in the currents both absorbed and provided by the
embedded generator.

The classical cause of transient instability is a three phase short circuit fault close to the
terminals of the embedded generator. The typical clearance time for this type of fault on
distribution networks is governed by the response of inverse definite minimum time delay
overcurrent relaying, IDMT, and can be of the order of one second or more. Coupled with the
critical clearance time for a typical embedded generator being of the order of 300
milliseconds, transient instability can be expected following such a fault. Fortunately, three
phase faults are not prevalent on distribution systems, and the critical clearance times for an
embedded generator increases as the distance between the generator and the fault increases.
The generator may also be disconnected from the grid supply as a result of the circuit breaker
tripping.

It is not feasible to design generators which have a greater tolerance to transient instability
problems. These problems could be avoided by replacing the distribution network's IDMT
relays with high speed protection, but the costs involved for existing distribution systems can
rarely be justified. An alternative solution is to use pole slipping protection for the embedded
generator, such that when transient instability is detected, the embedded generator is
disconnected from the network before it is damaged or the system is disturbed.

6.3.1

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 416

 2

2. POLE SLIPPING PROTECTION.

Pole slipping of a synchronous generator connected to a supply system is the process by

which an imbalance between the mechanical power input to the machine and its electrical
power output causes the rotor to accelerate and eventually slip with respect to the power
system frequency. This can be caused by short circuit faults, general disturbances on the
power system, or problems with the generator's prime mover, its excitation or its control
system.

Traditional techniques for detecting pole slipping examine the variations in the apparent
impedance of the generator as seen from its terminals. Several schemes are commercially
available which are based on distance type relays [1,2,3,4], and use combinations of mho and
linear characteristics. Some of these schemes are able to trip before the pole slip occurs,
whereas others rely on the pole slip having taken place before they will disconnect the
generator from the network.

Unfortunately, the impedance of a generator is not as predictable as would be expected and

it depends on the type of disturbance to which the generator is subjected. Hence the concept
of the sub-transient, transient and synchronous impedance. It is therefore generally
recommended [1,4,5] that transient stability studies are performed so that the trajectory of the
impedance loci due to the disturbance is known and the most appropriate relaying scheme and
its settings can be selected. These simulations can be time consuming and expensive, and due
to the vagaries of pole slipping do not guarantee reliable tripping or the absence of nuisance
tripping.

Pole slipping is a power based phenomena and an alternative approach for detecting pole
slipping has been developed which uses the Equal Area diagram as a basis to assess the
stability of the machine and determine when it is committed to a pole slip. The settings for
this technique rely on data which is generally available from the generator's manufacturer and
therefore stability studies are not required. Since this approach [6] is able to recognise the
conditions where a pole slip is inevitable, it offers the prospect of tripping before the actual
pole slip and hence disconnecting the machine from the system before it can be damaged and
before there is a major disturbance to the quality of supply to the adjacent power system.

The technique takes advantage of the processing capabilities of modem multi-function

microprocessor numeric relaying platforms and has been designed to share sub-functions with
other protection functions required for the protection of embedded generation.

3. P ower based T echniq ue for D etecting P ole S lipping .

The Equal Area Criterion demonstrates the conditions where synchronism is lost and pole
slipping occurs. Figure 1 shows the power/load angle relationship and the rate of change of
load angle (slip, 5 ) for a generator losing synchronism with the utility supply to which it is
connected following a loss of power transfer capability due to a disturbance on a double

6.3.2

M.J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 417

 3

circuit line [7], For this scenario, the generator is connected to a double circuit line and the
power system disturbance is caused by switching one of the lines out of service for a short
period. Removing the electrical load from the generator, while keeping the mechanical power
constant, causes the generator rotor to accelerate and eventually pole slip.

The most severe disturbance for a generator is the close-up three phase fault, for which all
of the generator's load is lost and therefore all of the prime mover power is used to accelerate
the rotor. For less severe short circuit faults, a greater fault duration is required to cause pole
slipping.

A
MECHANICAL
POWER
POWER INPUT.

ASYNCHRONOUS
POWER OUTPUT.

i, 1, l l - L - ______

360
J

LOAD
ANGLE

M--------- ►

PERIOD OF
SYNCHRONOUS SYNCHRONOUS
OSCILLATIONS POWER OUTPUT

F IG U R E 1. E Q U A L A R E A D IA G R A M F O R G E N E R A T O R L O S S OF
S Y N C H R O N IS M .

The Equal Area Diagram, figure 1, illustrates that for a stable swing, the machine operating
point cannot exceed point 5, the critical stability point. This corresponds to the point at which
the electrical output o f the machine, P, equals the mechanical input from the prime mover,
Pm . Instability occurs if the machine moves beyond the critical stability point, which is point
5 of figure 1. This condition can be identified in terms of real power as when:-

P < P.

6 .3 .3

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 418

 4

Since this criterion can also be satisfied for machine operating between points 0 and 1 shown
on figure 1, a reactive power measurement is used to differentiate between the two. The
steady state reactive power/load angle relationship for a synchronous machine operating
connected to an infinite bus is given by:-

^ -H co s( 6 )) - (__ 0
E* V 1 + __ V2 + i i)
Q= - y - (cos( 2 6 ) )
\ x„9 XA
dJ T k

where E is the machine internal voltage and V is its terminal voltage.

This shows that for load angles between 90° and 270°, the reactive power will always be less
than QMp, where Qtfip is given by the equation below. Since point 5 of figure 1 occurs for load
angles greater than 90°, then if

Q< (<3 ■= )

the machine must be operating at point 5 and not point 1.

The transient reactive power characteristics vary in magnitude from the above steady state
values. However, this does not corrupt the above method since the value of reactive power
corresponding to a 90° load angle is identical to the steady state value. The transient reactive
power characteristics above and below this load angle are such that in steady state terms, it
appears as if the load angle is correspondingly higher or lower.

Finally, it must be ascertained whether or not the load angle is still advancing. To determine
this, the rate of change of real power is used. It can be seen from the synchronous power
output characteristic of figure 1, that if the load angle is still increasing when point 5 is
reached, the machine output power will be decreasing, i.e. the rate of change of real power
will be negative.

In practice a margin for error is allowed for in this rate expression (AP/At), to ensure the
algorithm does not mal-operate. A minimum negative value, (AP/At)min is designated, based
on a minimum value of slip and an assumed sinusoidal power/load angle relationship. The
conditions which determine that a generator is about to pole slip are therefore :-

P < Pt , where Pt « Pm

Q < Qtrip

(APJAt) < (A P /A t)^

6.3.4

M.J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 419

 5

These conditions are used to give the trip criteria for the algorithm. The real power trip level,
P, , is derived from the generator's monitored power output and is proportional to the
mechanical power input Pm. This setting is automatically adjusted periodically. The rate of
change of power trip setting, (AP\At)min is also continuously adjusted by the algorithm
according to the generator operating point.

Short circuit faults cause added complications since they cause dramatic changes in the relay
measurands. These transitions produce negative values of (AP/At) which are a potential source
of instability to the algorithm. Fortunately, the faults generally introduce sinusoidal terms of
twice the power system frequency into the power measurements. Since pole slipping is a
relatively slow process in comparison to faults, an imposed minimum tripping time of one and
a half power system cycles will inhibit fault tripping.

4. S im ula tio n S t u d ie s .

The performance of the pole slipping algorithm was tested using a laboratory generator,
computer based dynamic simulation and field tests using an industrial diesel generator.

The laboratory generator system used two 5 kVA synchronous generators driven by 8 horse
power dc motors. These were connected to a 'local' load and a 200 V, 3 phase laboratory
'Utility' supply. This was useful for testing the algorithm in a real life situation. The system
was subjected to harmonics from other research projects and to heavily alternating loads
nearby which made it far from an ideal text book system.

Pole slipping tests were conducted by inserting a resistance in parallel with the generator's
field winding and therefore weakening the excitation. The laboratory model was also used to
test the algorithm during power system fault conditions, and all of the fault types possible
were placed on the local load busbar by switching in 'fault' resistance.

The computer based dynamic simulation package was used to test the algorithm's performance
against power swings and pole slipping. The model used was based on the simple one
machine model using a 588 MVA generator connected to an 'infinite bus' model rated at 58.8
GVA. A stable swing was caused by placing a fault on the generator transformer terminals
of sufficient duration to cause the generator rotor to swing up to the critical stability point and
back down to a stable operating area. Transient instability was caused by using a fault of
duration greater than the critical clearance time.

Other tests to demonstrate the algorithm's ability to detect a pole slip were simulated by
initially setting it to run at maximum output power, and then reducing the input power to half
rated power. This tested the adaptive setting section of the algorithm.

5. S im ula tio n T est R esu lts .

Figure 2 shows the algorithm variations for the laboratory system weak field pole slipping
test. Inspection of plots (a) and (b) show that the generator was initially operating overexcited
at very low power. After one second on the record, the power input to the generator was

6.3.5

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 420

 6

increased so that the machine operated at approximately 2.5 kW. Observation of the (AP/At)
curve in plot (c) shows that because P, was not updated during this period, the trip setting
(AP/At)^n gradually decreased to a lower negative value, resulting in a de-sensitising of the
algorithm.

a) Generator R eactive Power

^ 0
%-1500
J
| -3000 Q trip

2 -4500

-6000
0 1.0 2.0 3.0 4.0 5.0
Tim e (s)
b) Generator Real Power

3000 j ___

J 2000
II 1000
Cl,

2.0 3.0 4.0 5.0

Tim e (s)
c) Generator Rate o f Change
o f Power
3000
£ 1500

3
% -1500
(A P/A t)r
-3000
2.0 3.0 4.0
Tim e (s)

F IG U R E 2 - 5 kV A G E N E R A T O R P O L E SLIP.

At a time of three seconds on the record, the adaptive setting part of the algorithm updates
the value of P, to the new operating level. This caused a drop in magnitude of (AP/At) min, but
it did not drop back to the initial value, due to P, being at a higher level. This is the desired
effect since at higher input powers, more severe pole slips can occur and therefore more
negative values of (AP/At) are produced.

Shortly after the update, the resistor was connected in parallel with the field, causing the pole
slip to occur. Inspection of the reactive power plot shows that this criterion was satisfied at
3.75 seconds, when the reactive power fell below the trip setting, Q,rip> indicating that the load
angle had increased to a value above 90°. Inspection of the real power plot shows that after
insertion of the parallel resistor, the generator's field was weakened sufficiently so that the
generator could not maintain the level of output required to match the mechanical power
input. The real power criterion was therefore satisfied on inserting the resistor.

The final criterion (AP/At) , can be seen to fall below its trip setting (AP/At)min at a time of

6 .3 .6

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 421

 7

4.3 seconds. All three criteria were then satisfied for the one and a half cycle required and
a trip signal was produced. The trip signal can be found on the real power plot and it can be
seen that the trip signal was issued a significant time before the pole slip occurred. The usual
point where pole slip is said to have occurred is when the machine goes from generator to
motor action, i.e when the power output of the machine goes negative. Unfortunately, in this
case the whole pole slip could not be recorded, due to limitations in the data acquisition
system.

F au lt on F a u lt o f f a) Generator R eactive Power

I 0L J ^ i

g -1500 Qtrip
© iT T ._____________
^ -3000
1
0.5 1.0 1.5 2.0 2.5 3.0
Time (s)

b) Generator Real Power

Pmr
5000 Pt
4000 [

fc 3000

NO TRIP

0 0.5 1.0 1.5 2.0 2.5 3.0

Time (s)
c) Generator Rate o f Change
o f Power

~ -2 0 0 0 (A P /A tW n
-lorn

2.5 3.0
lim e (s)

F IG U R E 3 - 5 kV A G E N E R A T O R 2 P H A S E T O
EARTH FAULT T E ST

Figure 3 shows the algorithm's response to a two phase to earth fault. Of all the fault types
tested, this and the three phase faults provided the toughest conditions for testing algorithm
stability. The reactive power criterion was satisfied for a small period after removal of the
fault at 1.4 seconds on the record. During this time the real power criterion was satisfied for
a very short time, before the real power output of the generator jumps above the trip setting
P, . However, inspection of the (AP/At) plot shows that during this period, the (AP/At) signal
jumps to a negative value off the scale for a very short time, before escalating to a very high
positive value for a short time. These oscillations took less than one and a half power system
cycles, and so the algorithm was continually restrained.

The algorithm also successfully restrained for all of the other fault types tested.

Figure 4 shows the results to a test using the dynamic simulator. Initially the generator was
operated at full power, it was then reduced to half rated operation. The condition monitoring

6 .3 .7

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 422

part o f the algorithm updated at a time o f 3.75 seconds 011 the record.

a) G enerator R eactive Power

500
250
0
-250
Qtrip
-500 \
1.0 2.0 3.0 4.0 5.0 6.0
Tim e (s)

b) G enerator Real Power

Pmax
1000

500

TRIP
o -500

-1000
2.0 3.0 4.0 5.0 6.0
Tim e (s)
c) G enerator R ate o f C hange
o f Power

(A P / A tW n

-3000
-4500

lim e (s)

F IG U R E 4 - 5 8 8 M V A G E N E R A T O R P O L E
S L IP P IN G D U E T O A T H R E E P H A S E FAULT.

Inspection of the (AP/At)min curve shows that until the update, it stayed at its high power
level, resulting in the algorithm being de-sensitised for a small duration. This is unavoidable,
but is of no great concern since in the event of a pole slip it would just result in the algorithm
producing a trip signal which is subject to a short time delay. The fault was applied to the
transformer busbar at 4.75 seconds and removed again at 5.2 seconds.

On removal of the fault, the generator's reactive power was less than Q,rip , satisfying the
reactive criterion indicating that the load angle was above 90°. At the same time, the real
power output is much greater than the nominal value before the fault. This invoked the
adaptive Pmax part of the algorithm, which tracked the real power output up to its maximum
value, and then held this value for 1 second before returning back to its default value of
1.25*P, .

This action resulted in (AP/At)min falling to a lower level, thus desensitising the algorithm to
the pending transient pole slip. The (AP/At) signal still fell below the trip setting (AP/At)mm
for one and a half cycles, and during this time the other two criteria were satisfied. A trip
signal therefore resulted.

Tlie algorithm successfully restrained for the power swing test, which produced a severe
swing in load angle up to a value of 160° before recovering to the stable condition.

6 .3 .8

M .J.Checksfield/PhD/University o f Bath 1996 Appendix D, Page 423

 9

6. F ield T e s t s .

A series of field tests were conducted using a 625 kVA diesel generator connected via a
transformer to the 11 kV network supply. In the most dramatic of the tests, a pole slip was
induced by quickly ramping the diesel engine's output power at a rate faster than the generator
controls could respond.

’TRIP BOUNDARY Generator Reactive Power

-500.000
I
^ - 1,000.000
o
CL,
-1,500,000
Generator Real Power

2 ,000,000
Pmax
1,000,000

TRIP, TRIP
- 1,000,000
Generator Rate o f Change o f Real Power
20,000,000

-20,000,000
U
&
-40,000,000

50 550 1050 1552

Time (ms)

F igure 5. P ower B a se d A lgorithm R espo n se to 625 kV A

D iesel G enerato r P ole S lip .

Figure 5 shows the power based pole slipping algorithm's response to this test. The real power
plot shows where the algorithm tripped, which for the first pole slip occurred just before
1000 ms on the record. Allowing for a circuit breaker operating time o f 100 ms, this would
have isolated the generator before the actual pole slip. The real power plot also reveals the
potentially damaging effects of pole slipping, since the peak in power output after the first
pole slip reaches 2.7 MW, five times rated power. By tripping before the pole slip occurred,
the new algorithm would have avoided this dangerous surge in the power output. Due to the
finite disturbance recording time, the transition from normal to underexcited generator
operation was not captured.

7. C o n c l u sio n s .

Transient instability is a point of concern for embedded generation connected to utility

distribution networks. Typical short circuit fault clearance times obtained with the protection

6.3.9

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 424

 10

used on these systems are in excess of the critical stability times associated with typical
embedded generation units. Loss of stability can weaken or damage the embedded generator
and the disturbance to the local supply network can compromise the quality of supply.

It has to be accepted that it is not feasible to design generators which are immune to transient
instability and therefore there is an increased role for pole slipping protection. The traditional
approach is to use impedance type relaying, however, there are complications in the choice
of settings and power system simulation studies are recommended.

These complications can be avoided by using a power based technique to detect when the
embedded generator is committed to pole slipping. Relay settings for this algorithm are
determined from general machine data provided by manufacturers and the trip levels are
automatically adjusted to the current loading on the machine.

This new technique is suited to modem numeric relay platforms and can be integrated into
a multi-function generator protection package.

Laboratory tests, computer simulation and field tests have all demonstrated that the power
based pole slipping algorithm is successful in detecting pole slips while remaining stable
during periods of other types of power system disturbances.

It offers the immediate advantage of detecting when the synchronous embedded generator is
committed to a pole slip rather than has actually slipped and therefore offers the opportunity
of disconnecting the machine from the network before possible damage or major disturbance
to the power system.

8. R e fe r e n c e s .

1. J. A. Imhof et al, 'Out of Step Relaying for Generators - Working Group Report,'
IEEE Transactions PAS-96, No.5, Sep/Oct 1977, pp 1556-1564.
2. A. Stalewski, J. L. H. Goody and J. A. Downes, 'Pole Slipping Protection,'
Developments in Power System protection, IEE Conf. Publication, No. 185, pp 38-45.
3. S. Shiwen, 'Microcomputer Based Out-of-Step Protection for Large Generator,' IEE
APSCOM Hong Kong, Nov 1991, IEE Conf. Proc. No 348, Nov 1991, pp 839-842.
4. D.W. Smaha, 'Out-of-Step Relay Protection of Generators.' IEEE Tutorial on the
Protection of Synchronous Generators, Section 8, 95-TP-102, Power System Relaying
Committee 1995.
5. H.K. Clark & J.W. Feltes, 'Industrial and Cogeneration Protection Problems Requiring
Simulation.' IEEE Trans. Industry Applications, Vol. IA-25, No.4, July 89, pp 766-
775.
6. M A Redfem and M J Checksfield, 'A New Pole Slipping Protection Algorithm for
Dispersed Storage and Generation using the Equal Area Criterion.' IEEE Transactions
on Power Delivery PWRD Vol 10, No 1, Jan 1995, pp 194-202.
7. V. A. Venikov, 'Transient Phenomena in Electrical Power Systems', Pergamon Press,
1964.

6 .3 .1 0

M.J.Checksfield/PhD/University of Bath 1996 Appendix D, Page 425

 Appendix E

THE PRACTICAL IMPLEMENTATION OF THE

POWER BASED POLE SLIPPING ALGORITHM
INTO A COMMERCIAL MICROPROCESSOR RELAY

The following report contains the information which was used to implement the power
based pole slipping algorithm in a commercially available microprocessor relay. The
symbols used in this report refer only to this report, and are defined in the report's
symbols list.

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 426

 Page 1 o f 39
Issue PC
10 May 1996

INCLUSION OF THE POWER BASED

POLE SLIPPING ALGORITHM INTO THE
LGPG 111 GENERATOR PROTECTION RELAY.

Author M Checksfield (University of Bath)

Project Leader H T Yip (GEC Alsthom T & D, Protection & Control Ltd)

Supervisor M A Redfem (University of Bath)

M.J.Checksfield/PhD/University o f Bath 1996 Appendix E, Page 427

 Page 2 o f 39
Issue PC
10 May 1996

Table of Contents.

LIST OF SYMBOLS ............................................................................................................................ 4

1.0 INTRODUCTION..................................................................................................................... 7

2.0 THE POLE SLIPPING ALGORITHM..................................................................................... 7

Including Hysteresis in the Pole Slipping Algorithm.;
Algorithm Initialisation Values
2.1 THE REACTIVE POWER TRIP CRITERION........................................................... 8
2.1.1 Calculation of Reactive Power Measurand, Q.................................................. 8
2.1.2 Calculation of the Half Cycle Moving Average Filter Values.......................... 9
2.1.3 Calculation of Reactive Power Trip Level, Qtrip............................................. 9
2.2 THE REAL POWER TRIP CRITERION.................................................................... 9
2.2.1 Calculation of Real Power, P........................................................................... 9
2.2.2 Calculation of Real Power Trip level, Pt......................................................... 10
External Settings of'Ptol' and 'lag'.
2.3 THE RATE OF CHANGE OF POWER TRIP CRITERION.................................... 10
2.3.1 Calculation of (AP/At) Signal.......................................................................... 10
2.3.2 Calculation of Rate of Change of Power Trip Level, (AP/At)^..................... 10
Minimum Setting Value for(AP/At)mm,(AP/At)fact;
Calculation of Pmax; Calculation of (AP/At)mm;
The Externally Set Value for Slip.;
The Externally Set Value (AP/At)tran

3.0 INCLUSION OF POLE SLIPPING ALGORITHM IN LGPG..................................................13

3 .1 REACTIVE POWER VARIABLES - powerQ, Qtrip........................................ 13
3.1.1 Calculation of Reactive Power, powerQ......................................................... 13
3.1.2 Setting ' Qtrip'.....................................................................................................14
3.2 REAL POWER VARIABLES - powerP, Pfact, pt, pmax, Ptol....................... 14
3.2.1 Calculation of Real Power, P..............................................................................14
3.2.2 Setting 'Pfact'..................................................................................................... 15
3.2.3 Calculation of pt................................................................................................. 15
3.2.4 Calculation of pmax........................................................................................... 16
3.2.5 Setting ' Ptol'.......................................................................................................16

3.3 RATE OF CHANGE OF POWER VARIABLES

Dmin, deriv, Dfact, Dtran..................................................................................................... 17
3.3.1 Calculation of Rate of Change of Power Setting Dmin................................... 17
Initialisation of the Look Up Table.;

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 428

 Page 3 o f 39
Issue PC
10 May 1996

Numerical Overflow;
Calculation of lookup val;
Minimum Setting Value for Dmin
3.3.2 Setting Dfact............................................................................................. 20
3.3.3 Calculation of deriv Inside the Algorithm....................................................21
3.3.4 Setting Dtran............................................................................................ 21
3.4 OTHER SETTINGS - lag, slip............................................................................ 22
3.4.1 Setting'lag'............................................................................................... 22
3.4.2 Setting 'slip'..............................................................................................22
3.5 INITIALISATION OF VALUES...........................................................................23
3.5.1 Initialising the cos (5C) Look Up Table........................................................ 23
3.5.2 Initialising Other Variables.........................................................................23
Variables 'PU_POLE_SLIP_???' etc etc;
'Pt', & Variable 'P2' ;
Variable 'P I';
Variable 'Pmax';
Variables 'last_p_val' & 'pen__p_val';
Variables 'last_powP' & 'last_powQ';
Counter Variables 'trip_count', 'cnt_pmax_ofF, 'cnt_pmax', & 'count';
Variables 'Dmin', 'deriv', 'powerP' & 'powerQ'
3.6 NUMERICAL OVERFLOW CONSIDERATIONS & RESOLUTION..................... 25
Reactive power, Q;
Real power, P;
Real Power Trip Level, P t;
Real Power Maximum Output Level, Pmax ;
Rate of Change of Power (AP/At);
Rate of Change of Power trip level (AP/At)^
3.7 DISTURBANCE RECORDING......................................................................... 27
3.7.1 Disturbance Recorder Requirements.................................................................. 27
3.7.2 Scaling from 32 Bit Precision to 16 Bit Precision.............................................. 28
4 .0 SUMMARY OF ALL SCALING FACTORS IN ALGORITHM................................................ 32

FIGURE 1 - Algorithm Flow Chart.............................................................................................................. 37

FIGURE 2 - Pt Calculation Function........................................................................................................... 38
FIGURE 3 - Pmax Calculation Function......................................................................................................39

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 429

 Page 4 o f 39
Issue PC
10 May 1996

LIST OF SYMBOLS

(AP/At)tran generator transient saliency component used in the calculation of (A P /A t)mjn (W/s)

(AP/At)fact maximum value for rate of change of power trip level, (AP/At)min(W/s)
(A P /A t)min rate of change of power trip level (W/s)
(AP/At) rate of change of power measurand (W/s)
5C angle used in calculating (AP/At),™ (degrees)
dt length of time between the two samples used in calculating (AP/At) (seconds)
ia instantaneous phase value of current (cosine fourier filtered A-Phase current) (Amps)
ib instantaneous phase value of current (cosine fourier filtered B-Phase current) (Amps)
ic instantaneous phase value of current (cosine fourier filtered C-Phase current) (Amps)
In relay secondary nominal current (rms value) (Amps)
lag time between calculations of the real power trip setting, Pt (seconds)
N number of samples used in half cycle moving average window (2)
P real power (W)
Pfact scaling factor used in Pt calculation to introduce a margin for error in the estimation
Pmax generator maximum output power in the last 1 second (W)
Pn present sample value of moving average filtered real power (W)
Pn-2 previous but one (half a cycle ago) value of moving average filtered real power (W)
Pt real power trip setting (W)
Ptol tolerance band used in calculation of Pt (W)
Q reactive power (VAr)
Qtrip reactive power trip level
Sgen 3 phase VA rating of generator (VA)
slip external setting used in calculation of (AP/At)™, (%)
Sn relay secondary rating of generator = V3 * Vn * In (W)
V ab instantaneous line value of voltage (cosine fourier filtered AB voltage) (Volts)
Vbc instantaneous line value of voltage (cosine fourier filtered BC voltage) (Volts)
Vn relay secondary nominal voltage (rms value) (Volts)
X(„) present sample input value to moving average filter
Xq generator quadrature axis synchronous reactance (per unit value)
y(n) present sample output value from moving average filter.

Variables Used in Pole Slipping Algorithm Computer Code.

cnt_pmax counter used to check if powerP >= pmax for one power system cycle (4 counts)
cnt_pmax_off counter used to check if pmax should be reset to nominal value (reset after 1
second)
COSDELTASCALE scaling value used gor cos (8c) term in Dmin calculation.
count counter used to see if pt calculation should be performed (every 'lag' seconds)

M.J.Checksfield/PhD/University o f Bath 1996 Appendix E, Page 430

 Page 5 o f 39
Issue PC
10 May 1996

deriv (AP/At)
DERJV_OF number of counts that powerP - last_p_val is limited to in deriv calculation
DERTV_MULT multiplier used in deriv calculation (multiplier at 50 Hz)
DMIN SCALE internal scaling factor from Dmin calculation
DMINMAXPMAX maximum value that pmax is allowed to be in Dmin calculation
Dfact (AP/At)facl
Dmin (A P /A t)min

Dtran (AP/At)tran

Iaoc.Fcos ia
Iboc.Fcos ib
Icoc.Fcos ic
LAG SCALE scaling used for ‘lag’ term used in Pt calculation.
last_p_val Pn-2 - stored moving average filter value of real power - used to calculate deriv
last_powP previous sample of 'raw' real power required for 2 point (half cycle) moving
average calculation
lastjpowQ previous sample of 'raw' reactive power required for 2 point (half cycle) moving
average calculation
lookup[i] look up table equal to cos(Sc) * slip, i , the look up index ranges, from 0 to 7
lookup_val value looked up from lookup table used in calculating Dmin
master_look_up[i] look up table value equal to cos(5c), i , the look up index ranges from 0 to 7
on_off_switch variable used to see whether pole slipping function is enabled within LGPG
P 'raw' value of real power
pi value of moving average filtered real power stored by pt calculation function, this
is the value used by the function 2*lag seconds ago
p2 value of moving average filtered real power stored by pt calculation function, this
is the value used by the function lag seconds ago
pen_p_val Pn-i - stored moving average filter value of real power - used to calculate deriv in
the following sample where it will become Pn-2
Pfact multiplying factor used in pt calculation,
pmax generator maximum power output - pmax.
PMAX_OFl maximum value pt is allowed to be in pmax calculation.
PMAX OF2 maximum value powerP is allowed to be in pmax calculation.
powerP 2 point half cycle moving average filter value of real power
powerQ 2 point half cycle moving average filter value of reactive power
pt real power trip setting, Pt.
pt_pmax variable equal to pt xlOO / pmax - used to find look up index, i
pt_xl00 real power trip setting scaled up by 100 (note Pmax is internally scaled by 10)
PU_I 'per unit' internal value of current = 800 counts / Amp for 1 amp relay
PU POLE SLIP DMIN Dmin scaling factor = 30518 counts/W/s
PU POLE SLIP P 'per unit' internal value of power = PU_V * PU_I = 8000 counts/VA

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 431

 Page 6 o f 39
Issue PC
10 May 1996

PUPOLESLIPPOW powerP, (real power) scaling factor = 4000 counts/W

PUPOLESLIPPMAX pmax scaling factor = 400,000 counts/W/s
PUPOLESLI PPT pt scaling factor = 40,000 counts/W/s
PU POLE SLIP Q 'per unit'internal value of real power = PU_V * PU_I * V3 =8000 counts/VA
PU_V 'per unit' internal value of voltage =10 counts / Volt
Q 'raw' value of reactive power
Qtrip reactive power trip setting.
root3 square root of 3 = 1.732050808
SLIP SCALE scaling used for slip value used in Dmin calculation.
trip boolean flag used to show whether algorithm has tripped
trip count counter used to check that P, Q, (AP/At) trip criteria areall satisfied for one power
system cycle
Vab_l_poleslip.Fcos Vab
Vbc_l.Fcos Vbc

MJ.Checksfield/PhD/University of Bath 1996 Appendix E, Page 432

 Page 7 o f 39
Issue PC
10 May 1996

1.0 INTRODUCTION

The power based pole slipping algorithm measures three different power system quantities in order to
establish if a pole slipping condition exists. These are the Reactive Power, Q, the Real Power, P, and the
Rate of Change of Power, (AP/At).

The trip settings for P and (AP/At) are derived internally, and the algorithm adjusts these settings
according to the operating point of the generator. The reactive power trip setting, Qtrip is entered as an
external setting. The precise nature of the trip conditions will be explained later.

Section Two explains how the algorithm uses P, Q & (AP/At) to determine when a pole slipping condition
exists.

Section Three discusses the scaling values used in implementing the algorithm within the LGPG 111 relay,
it also discusses the intricacies of getting the algorithm to work over the operating range of the relay, as
well as the disturbance recording requirements of the algorithm.

2.0 THE POLE SLIPPING ALGORITHM.

The algorithm produces a trip output if the following conditions exist at the terminals of the generator.

Q < Qtrip }

and } ALL OF THESE NEED TO BE

P < Pt } SATISFIED FOR 1.5

and } POWER SYSTEM CYCLES.

(AP/At) < (AP/At)min }

The algorithm is frequency dis-abled if the system frequency does not lie between 40 and 70 Hz. This
allows for 10 Hz either side of the 50/60 Hz operating frequencies. The algorithm processing rate is
synchronously locked to the power system frequency and is at a rate of 4 samples per power systemcycle,
i.e at a frequency of 50 Hz, the above criteria are checked every 5 ms. The above criteria must thereforebe
satisfied over 6 successive sample instants in order for the algorithm to produce a trip output. The
flowchart for the algorithm can be found in Figure 1.

MJ.Checksfield/PhD/University o f Bath 1996 Appendix E, Page 433

 Page 8 o f 39
Issue PC
10 May 1996

The function which calculates the real power trip setting, Pt is run asynchronously. It is not called at the
same time as the rest of the algorithm. Since it is asynchronous the exact time at which it is calculated may
vary by a small amount. This is due to its priority being lower than other tasks in the multi-tasking
executive priority level. A tolerance of +/- 10 % of the 'lag' setting is acceptable.

Including Hysteresis in the Pole Slipping Algorithm.

In most protection algorithms, a hysteresis facility is provided in order to avoid 'chatter' of the trip output.
Chatter occurs when the algorithm measurand is at the trip level, and is therefore crossing to and fro
between 'trip' and 'no trip'. The problem is usually overcome by reducing the trip level when a trip first
occurs, so that a trip condition exists continuously after the initial trip.

With the pole slipping algorithm, it would be most desirable to stay 'tripped' until the generator is back in a
stable generating condition ( or is dis-connected ), the algorithm would then be reset. This is achieved by
latching the trip signal the first time it occurs, and only resetting it once the reactive power trip criterion is
NOT satisfied. With this arrangement, the algorithm will produce one trip output per slip cycle. This
should work for both pole slipping and loss of excitation. The algorithm structure used to achieve this can
be found at the bottom half of Figure 1.

Algorithm Initialisation Values

Before the algorithm runs, some variables need to be initialised. This is to ensure that the algorithm does
not mal-operate in the first few seconds when it is calculating it's settings according to the generator
operating point. The total time the algorithm takes to initialise is given by '2 * lag' seconds. This is the
time it takes for the real power trip setting Pt to be derived. Initialisation is also necessary to ensure no
maths errors occur due to dividing by zero. Specific details of what each variable will be initialised to can
be found in section 3.5

2.1 THE REACTIVE POWER TRIP CRITERION.

2.1.1 Calculation of Reactive Power Measurand, Q.

The reactive power is calculated using the 2 watt meter method, i.e ;

Q = -j j ( Vab* ( ic - h ) + Vbc* ( ia - h ) ) (units ofVArs) 1

where Vab , Vbc are the instantaneous fourier filtered line voltages, and ia , ib , ic are the instantaneous
fourier filtered phase currents. Note that the sign convention used for reactive power is that reactive power
is negative at the terminals of an under excited generator. This is the opposite of the standard IEE
definition for reactive power.

Once the reactive power has been calculated, a half cycle moving average filter is applied to it. This is
done in order to remove any power system frequency unbalance in the signal.

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 434

 Page 9 o f 39
Issue PC
10 May 1996

2.1.2 Calculation of the Half Cycle Moving Average Filter Values.

The moving average filter is expressed mathematically as;
y N -l

y (n ) ~ 77
jv
2r=0 X ("-r)
2

Where y(n) and X(„) are the n111output and input respectively, N is the window size.
Since the number of samples per cycle to be used is 4, the value of N required for a half cycle moving
average is 2. The above expression therefore simplifies to;

Y(n) = X(n) + X(n-l) / 2 3

Note from this expression that the previous sample value is required for this calculation, this must therefore
be stored between function calls.

2.1.3 Calculation of Reactive Power Trip Level, Qtrip.

The reactive power trip level 'Qtrip' is entered as an external setting. In absolute values, Qtrip is
determined from;

Qtrip = - * Sgen 4
Xq

Where Sgen is the generator three phase VA rating, and Xq is the per unit quadrature axis synchronous
reactance of the generator. The value of 'Xq' is expected to vary between 0.3 and 2.5 PU on generator
rating.

2.2 THE REAL POWER TRIP CRITERION.

2.2.1 Calculation of Real Power, P.

The real power is calculated using the 2 watt meter method, i.e ;

P = Vab* ia - Vbc* ic (units o f Watts) 5

where Vab , VbCare the instantaneous fourier filtered line voltages, and ia , ic are the instantaneous fourier
filtered phase currents. Once the real power has been calculated, a half cycle moving average filter is
applied to it. This is done in order to remove any unbalance in the signal, (see section 2.1.2). As well as
having to store the previous 'raw' value of power for the moving average calculation, the previous two
values of moving average filtered power also need to be stored so that the (AP/At) calculation can be
performed (see section 2.3.1).

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 435

 Page 10 o f 39
Issue PC
10 May 1996

2.2.2 Calculation of Real Power Trip level, Pt.

The real power trip level, Pt is derived internally by the algorithm. The trip setting is an estimate of the
generator mechanical input power. It is estimated using the assumption that in the steady state, the
generator electrical output power is equal to the generator's mechanical input power. Detection of
operation in the steady state is detected by checking that the generator's real output power stays within a
pre-defmed tolerance band ( Ptol) for a specified time of (2*lag) seconds. Both 'lag' and 'Ptol' are entered
as external settings.

A safety margin is included in this estimation which ensures the estimate will always be an under estimate.
The estimation process is performed by comparing the most resent value of real power with two other
previously stored values. The estimation process is performed nominally every 1 second, there is a setting
provided in the algorithm called 'lag' which can change this time span between 0.1 and 10 seconds. The
flowchart for this function can be found in Figure 2.

External Settings of'Ptol' and 'lag'.

Both 'lag' and 'Ptol' are entered as external settings. 'Lag' can be set between 0.1 and 10 seconds, in 0.1
second intervals, whilst 'Ptol' can be entered between 1 % and 20 % of generator rating.

2.3 THE RATE OF CHANGE OF POWER TRIP CRITERION.

2.3.1 Calculation of (AP/At) Signal.

The rate of change of power is calculated using;
(A P /A t) = ( Pn - Pn ) * 1 / dt
- 2 ( units o f Watts / second) ^

Pn is the present moving average filtered sample value of real power. Pn_2 is the previous but one sample,
dt is the time between Pnand Pn_2 , which is every half a power system cycle. This technique is used to
prevent any double frequency terms present in the real power signal being transferred into the (AP/At)
signal.

2.3.2 Calculation of Rate of Change of Power Trip Level, (AP/At)min .

The rate of change of power trip level is calculated from an external setting value of 'slip' and two real
power measurements, namely Pt (see section 2.2.2) and Pmax ( see below). It should be noted that all of
the rate of change of power tripping and blocking levels will be negative.

Minimum Setting Value for (AP/Atkin

In addition to the calculation mentioned above for finding (AP/At)mjn, a check is performed on the calculated
value, if it's magnitude is below a certain value, a minimum setting value is assigned. This minimum

MJ.Checksfield/PhD/University of Bath 1996 Appendix E, Page 436

 Page 11 o f 39
Issue PC
10 May 1996

setting value is entered as the external setting, '(AP/At)fact ' . It can be set between -10 % and -50 % of
generator rating.

Calculation of Pmax
Pmax is a measure of the maximum output power which the generator has produced in the last second. A
default value of 1.4 * Pt exists to ensure a minimum value for the rate of change of power trip level.
Measuring the generator maximum output power allows the algorithm to differentiate between 'transient'
and 'steady state' generator operation. During transient operation, the rate of change of power trip setting
should be very negative, to ensure stable swings in power are blocked. For steady state pole slips, the trip
setting should be at a less negative level in order to provide the earliest detection possible.

It is important to note that the Pmax function should only track slow changes in generator output power. If
the generator produces a high value of output power for a short time (less than one power system cycle)
then this value should be rejected. The function should only track swings in generator output power which
are a consequence of the generator rotor dynamics (i.einthe l-10Hz region).

The flowchart describing how Pmax is obtained can be found in Figure 3.

Calculation of (AP/At)mj„
The trip level (AP/At)mmis calculated using the following formula;

(AP / At ) min = - slip * Pmax * cos (Sc )

if ratio Pt / Pmax <0.6 then

(AP / AUmin = (AP / A + (AP / At),rm 1

where (AP/At)^ is entered as an external setting, derived from the generator quadrature axis synchronous
reactance, Xq, and direct axis transient reactance, Xd\ This term is used to take into account the effects of
transient saliency in the generator. Note that

Sc = 180° - s in '1 (—----- )

P max 8

Trigonometric functions require a relatively large amount of computation, theexpression 'cos 5c' is
therefore found using a look up table. This look up table is split into 8 different levels, asdefined inTable
1.

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 437

 Page 12 of 39
Issue PC
10 May 1996

Table 1 - Lookup Table Used For Finding the Value 'cos (5c)'

lookup index ratio Pt/Pmax ratio Pmax/Pt 5c cos (5c)

0 0-0.1 10 - infinity 180 -1
1 0.1 -0. 2 5 - 10 174.26 -0.99498
2 0 . 2- 0. 3 3.33-5 168.46 -0.97980
3 0.3 - 0. 4 2. 5- 3 . 33 162.54 -0.95394
4 0 . 4- 0. 5 2-2.5 156.42 -0.91652
5 0.5 -0. 6 1. 66- 2 150.00 -0.86603
6 0.6-0.7 1.43-2 143.13 -0.80000
bo
’-'j

(nominal)
1

7 1.25 - 1.43 135.57 -0.71414

0

NOTE: The nominal value for Pmax = 1.4 * Pt, when this condition exists, then the ratio of Pt/Pmax =
1/1.4 = 0.714. This therefore falls into look up index 7. As the ratio of Pmax to Pt becomes greater than
1.4 a more negative value of cos (5c) is selected. The most negative value of cos (5c) is therefore -1 when
the ratio of Pmax to Pt lies between 10 and infinity.

The Externally Set Value for Slip.

The value of slip used in calculating (AP/At)™,, can vary between 0.1 % and 5 %, in 0.1 % intervals. The
nominal setting used is 0.5 %.

The Externally Set Value (AP/At),™

The formula for calculating (A P/A tV an is given below;

(AP / At ) trtm = [ ) / Xq - )/x d ) *Sgen 9

Where Sgen is the generator three phase rating, Xq is the quadrature axis synchronous reactance, and Xd’
is the direct axis transient reactance. The value of Xq is expected to vary between 0.3 and 2.5 PU, whilst
Xd’ is expected to vary between 0.1 and 2 PU on generator base. Note that Xd’ < Xq. (AP/At),™ is
therefore always negative.

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 438

 Page 13 o f 39
Issue PC
10 May 1996

3.0 INCLUSION OF POLE SLIPPING ALGORITHM IN LGPG.

This section details the settings and scaling factors required for the implementation of the power based pole
slipping algorithm in the LGPG111 generator protection relay. These settings have been designed with the
following conversion values from relay secondary volts and amps to internal' counts'.
1 Volt =10 counts;
1 Amp = 800 counts (1 Amp relay), or 5 Amps = 800 counts (5 Amp relay);
1 VA = 8000 counts (1 Amp relay), or 5 VA = 8000 counts ( 5 Amp relay);
PU_POLE_SLIP_P = 800 * 10 = 8000 counts.

In - Nominal input current = 1 A (1 Amp relay) or 5A (5 Amp relay) (RMS phase value);
Vn - Nominal input voltage = 100 to 120 Volts. (RMS line to line value)
Sn - Three phase rated apparent power = Vn * In * V3

The value Sn can be considered as equivalent to the generator rating in terms of relaying secondary
quantities. All variables referred to in section 3 which are used in the code (and are not constants) will
have the prefix ‘pole_slip_prot. ’ since they belong to a data type structure called pole_slip_prot. This
section will refer to all variables by their computer code names. Most variable names are defined in file
‘data_13.h’, however, some are defined in poleslip.c.

The setting and scaling values in the following section have been calculated f o r the 1 Amp relay. The
minimum and maximum setting values have been calculated using the following minimum and maximum
values of Vn and In;
Vn (max) =120 Volts, Vn (min) =100 Volts,
In (min) = 0.5 Amps, In (max) = 1.5 Amps

3.1 REACTIVE POWER VARIABLES - powerQ, Qtrip.

3.1.1 Calculation of Raw Reactive Power, Q.

The reactive power is calculated using the formula below;
10
n _ Vab _ \ pole slip. F cos* (Ic_oc. F c o s —Ib_ oc. F _c o s) + Vbc_\.F cos*{Ia_ oc. F c o s —Ib_ oc. F c o s)

where Vab_l_poleslip.Fcos, Vbc l.Fcos are the relay instantaneous fourier filtered line voltages, which
have been scaled down by the relays internal hardware by a factor of V2, and ia oc.Fcos, ib_oc.Fcos,
ic oc.Fcos are the instantaneous fourier cosine filtered phase currents which have also been scaled down by
the relays internal hardware by a factor of V2.

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 439

 Page 14 o f 39
Issue PC
10 May 1996

NOTE that the factor V3 shown in section 2.1.1 has not been included, this would result in an unnecessary
division. The expression is divided by 2 in order to a) introduce a scaling factor of 1/2 to avoid numerical
overflow, b) avoid having to perform another division by 2 when calculating the half cycle moving average
value. Note that it is effectively divided by an additional factor of 2 due to the relays internal hardware
scaling.

The half cycle moving average calculation of reactive power, powerQ, is therefore reduced to a simple
addition of the present (Q) and previous (last_powQ) raw values of reactive power, i.e the moving average
filtered value, called powerQ is given by CQ + last_powQ\

To scale from internal 'counts' into VAr, Q should be

SCALED BY (PU_POLE_SLIP_P * V3 ) /2 = 6928.2 => 6928 counts/VAr
This variable is referred in the code as PUPOLESLIPQ.

NOTE the factor of V3 has been included since it was not used in the reactive power calculation. The
reactive power, Q was therefore V3 times too big. The factor of 2 comes from the scaling down used in the
calculation of Q.

3.1.2 Setting ' Qtrip'.

Qtrip is given by equation 4, Xq varies between 0.3 PU and 2.5 PU.

The minimum and maximum setting values in terms of relay secondary quantities are therefore;
MINIMUM QTRIP = -1/0.3 * 120 V * 1.5 A * V3 = -1039 VAr =>-1100VAr
MAXIMUM QTRIP = -1/2.5 * 100 V * 0.5 A * V3 = -35 VAr =>-30 VAr
in INCREMENTS OF 1 VAr

To scale into internal 'counts' these input values should be

SCALED BY ‘PUJPOLE_SLIP_Q\ i.e (8000 counts/VA* V3 ) 12 = 6928.2 => 6928 counts/VAr

NOTE the factor of V3 has been included since it was not used in the reactive power calculation. The
reactive power, Q was therefore V3 times too big. The factor of 2 comes from the scaling down used in the
calculation of Q.

3.2 REAL POWER VARIABLES - powerP, Pfact, pt, pmax, Ptol

3.2.1 Calculation of Raw Real Power, P.

The real power is calculated using the formula below;
11

MJ.Checksfield/PhD/University of Bath 1996 Appendix E, Page 440

 Page 15 o f 39
Issue PC
10 May 1996

_ Vab_\_ poleslip.F cos* (Ia _oc. F c o s ) - Vbc _ \.F cos* Ic _ o c .F cos)

2
where Vab_l_poleslip.Fcos, Vbc l .Fcos are the relay instantaneous fourier cosine filtered line voltages,
which have been scaled down by the relays internal hardware by a factor of V2, and ia_oc.Fcos,
icoc.Fcos are the instantaneous fourier filtered phase currents which have also been scaled down by the
relays internal hardware by a factor of V2.

The expression is divided by 2 in order to a) introduce a scaling factor of 1/2 to avoid numerical overflow,
b) avoid having to perform another division by 2 when calculating the half cycle moving average value.
Note that it is effectively divided by an additional factor of 2 due to the relays internal hardware scaling.

The half cycle moving average calculation of real power, powerP, is therefore reduced to a simple addition
of the present (P) and previous (last_powP) raw values of real power, i.e the moving average filtered
value, called powerP is given by ‘P + last_powP\

To scale from counts into Watts, these values should be

SCALED BY PU_POLE_SLIP_P/2 = 8000 counts/VA 12 =4,000 counts/W

This variable is referred in the code as PU POLE SLIP POW = 4000 counts/W.

3.2.2 Setting ’Pfact'.

This setting is used by the algorithm in the pt calculation purely as a means of scaling by a factor which
varies between 0.7 and 1.0. Since floating point maths is not permitted, the actual multiplication by Pfact
in the pt calculation function must be using integers between 7 & 10. The minimum and maximum setting
values for Pfact are;

MINIMUM PFACT =0.7

MAXIMUM PFACT = 1.0
in INCREMENTS OF 0.1

To scale into internal 'counts' these input values should be

SCALED BY 10 counts/unit

3.2.3 Calculation of Real Power Trip Setting, pt.

The value of pt is derived in the algorithm by the formula;
pt = powerP * Pfact 12

Note that the variable Pfact varies between 7 and 10, it should actually vary between 0.7 and 1.0, it has
been scaled up by a factor of 10 to accommodate the integer mathematics requirement, pt is therefore

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 441

 Page 16 of 39
Issue PC
10 May 1996

internally SCALED by 10/2 = 5. The factor of 10 comes from the Pfact scaling, the factor of 2 comes
from the real power scaling.

To scale from counts into W these values should be

SCALED BY PU_POLE_SLIP_P * 10/2 = PU_POLE_SLIP_POW * 10 = 40,000 counts/W
This variable is called PUPOLESLIPPT.

3.2.4 Calculation of pmax.

The value of ‘pmax’ is derived in the algorithm by either one of the following formulae;
pmax = 14 * pt (14 used instead o f 1.4) 13

or

pmax = 10 * (10 * powerP) (10 used instead o f 1.0) 14

however, before multiplying by pt, a numerical overflow check needs to be performed. The number in
counts which ‘pt’ should not exceed to prevent numerical overflow of pmax, denoted PMAXOF1 is given
by;
PMAXOF 1 = 231/14 = 15339 1689

before multiplying by the moving average filtered real power, powerP, a numerical overflow check should
be performed. The number in counts which ‘powerP’ should not exceed to prevent numerical overflow of
pmax, denoted PMAX_OF2 is given by;

PMAXOF2 = 231 / 100 = 21474836

If either of these numerical overflow checks are exceeded, pmax should be set equal to 2147483600 counts,
ie the value PMAX_OF2 * 100.

pmax is internally SCALED by (10*10)/2 = 50. One factor of 10 comes from the pt scaling, the other
comes from using 14 instead of 1.4, the 2 comes from the real power calculation.

To scale from counts into Watts these values should be

SCALED BY PU_POLE_SLIP_PT * 10 = 400,000 counts/W
This variable is called PU_POLE_SLIP_PMAX.

3.2.5 Setting ' Ptol*.

Section 2.2.2 stated that Ptol should be entered between 1% and 20 % of the generator rating.

Minimum and maximum setting values in terms of relay secondary quantities are therefore ;
MINIMUM PTOL = 1% /100 M OO V * 0.5 A * V3 = 0.866 W => 1 W

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 442

 Page 17 of 39
Issue PC
10 May 1996

MAXIMUM PTOL = 20 % / 100 * 120 V * 1.5 A * V3 = 62.35 W => 70 W

in INCREMENTS OF 0.1 W

To scale into internal 'counts' these input values should be scaled by the same factor as the real power,
powerP, i.e they should be SCALED BY
PU POLE SLIP P 12 = PUPOLESLIPPOW = 4000 counts/W

The factor of 2 comes from the scaling down used in the calculation of powerP.

3.3 RATE OF CHANGE OF POWER VARIABLES

Dmin, deriv, Dfact, Dtran

3.3.1 Calculation of Rate of Change of Power Setting Dmin.

Dmin is calculated using equations 7 & 8. The external setting, slip, should vary between 0.1 and 5.0 %,
due to the integer maths requirement this is scaled up by 10 to range from 1 to 50. The slip setting value
has to be converted into actual units. These units need to be referenced to the frequency of the generator,
the % slip setting should therefore be scaled up by;

2 * pi * freq (Hz) / 100

In the actual code the slip setting is scaled up when the lookup table, ‘lookup’ is initialised, the slip is
scaled up by a factor of 2 * pi. The scaling by freq(Hz)/100 is performed later on in the code. As
described in section 3.2.4, pmax is internally scaled by the factor (10*10)/2 = 50. The values for cos (5C)
are not calculated trigonometrically, instead they are stored in an 8 point lookup table (Table 1).

Initialisation of the Look Up Table.

To avoid the slip * Pmax calculation in equation 7, the look up table is actually initialised with values of
cos (5C) * slip *2 * pi. The multiplication by the system frequency, and the division by 100 is performed
when the look up value is taken from the lookup table, just before the Dmin calculation is performed. The
cos (5C) values have been scaled up by 1000, this is referred to as COSDELTASCALE, this preserves
three decimal places of accuracy. The lookup table therefore takes the form shown in Table 2, in the code ,
the initialised look up table is referred to as ‘lookup [pt_pmax]’, where pt_pmax is the lookup index;

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 443

 Page 18 of 39
Issue PC
10 May 1996

Table 2 - Look Up Table Used in the Pole Slipping Algorithm LGPG Code.

lookup index ratio Pt/Pm ax cos (8c) * slip

0 -6283 * slip

■
o
o
to
1 -6 2 5 2 * slip

1
0

o
2 0 .2 - 0 .3 -6 1 5 6 * slip

3 0.3 - 0 .4 -5 9 9 4 * slip

L /i
4 -5 7 5 9 * slip

1
0

o
o'

o’
^O
5 -5441 * slip

1
6 0 .6 - 0 .7 -5 0 2 7 * slip
o'

o’
oo
7 t"- -4 4 8 7 * slip
1

NOTE:
1) W hen calculating the ratio pt/pm ax, pt is first multiplied so that its scaling is ten tim es bigger than
that o f pm ax , this variable is denoted as ‘p tx lO O ’ in the code. The ratio for pt/pm ax then varies between
0 and 7, and can be used as the index to the lookup table, the variable name w hich is equal to
p t_ x l0 0 /p m a x is called ‘pt_pm ax’ in the code. The variable name is called pt xlOO because the pm ax
scalin g is such that it is already 10 times greater than pt, this is w hy pt needs to be m ultiplied b y 100 to
m ake it ten tim es bigger than pmax.

A ny remainder left from the division o f pt xlOO and pm ax is alw ays discarded, the ratio pt/pm ax is
therefore rounded down. This results in the larger value o f cos (5C) alw ays being selected. Several checks
are performed in the code to ensure numerical overflow does not occur. I f pm ax is equal to 0, then the
d ivision pt/pm ax is not performed, the quantity p tjp m a x is set to 0. A further check is performed to see if
the quantity pt_pm ax is between 0 and 7, i f it is not, no look up value w ill exist, and the code m ay fail.
T o overcom e this the lookup index is set to 0.

2) In the actual form ula for Dmin, there is a negative sign in the formula. Since the sign o f co s(8 c) is
also negative, and the slip should technically be negative (it is entered as positive), the sign from the
form ula has been omitted.

The essential requirement is that the overall sign o f Dmin is negative.

D ue to the scaling in pm ax, slip, and cos (8c), the value for Dmin has to be scaled back dow n to prevent
num erical overflow . The value chosen w as 131072 which corresponds to a value 2 17. This value w as
chosen, as w ere all o f the other scaling values which involve a factor o f tw o, to reduce the com putational
burden since a division by 2 can be implemented by one shift right operation. To ensure that a m inimal
loss o f resolution occurs, the look up table value is scaled down by 2, and Pmax is scaled down by 6 5 5 3 6 .

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 444

 Page 19 o f 39
Issue PC
10 May 1996

The procedure used for calculating Dmin is as follows;

1) D m in = (pole_slip_prot.pm ax / 2 16) * (lookup_val/2)

2) i f pt_pm ax < 6 then D m in = Dmin + Dtran
3) i f D m in > D fact then D m in = Dfact.

N um erical O verflow
B efore step 1 in the ab ove calculation is performed, a numerical overflow check is performed. The
numerical overflow checking is performed on pmax. I f this is above a certain magnitude, it is set to that
magnitude in order to lim it the m agnitude o f Dmin. The m axim um permitted m agnitude o f pm ax w hich
w ill not cau se num erical overflow o f Dm in is calculated from the follow ing variables. The other variables
in the D m in calculation have m axim um magnitudes of;
Dtran = -3 0 0 0 W /s = -3 0 0 0 * 30518 counts/W /s = -9 1 5 5 4 0 0 0 , (section 3 .3 .4 )
slip = 5% * 70 H z * 2 * pi * 1 0 /1 0 0 = 2 5 0 counts, (70 H z m axim um operating frequency)

cos(8c) = -1000.

pm ax is divided by 2 16 in the code w hilst the term ‘slip * co s(5 c) ’ or ‘lo o k u p v a l’ in the code is divided by
2, D m in w ill overflow at -2 31 counts , the m axim um value pm ax can be (denoted as
D M I N M A X P M A X ) w ithout the variable Dm in overflow ing is therefore given by;

-2 31 = (-1 0 0 0 * 2 2 0 )/2 * (D M IN _ M A X _ P M A X /2 16) - 91554000

D M I N M A X P M A X = 1 ,2 2 4 ,8 8 5 ,5 0 4 counts. The m agnitude o f pm ax is checked first before the Dmin

calculation, i f it is greater than 1 ,2 2 4 ,8 8 5 ,5 0 4 counts, then step 1 o f the D m in calculation is sim plified to;

1) D m in = 18690 * lookup_val/2

The number 18690 com es from D M I N M A X P M A X / 2 16 . This should be implemented in the code in
this format.

C alculation o f lookup val

The lookup_val is calculated in the code in the follow ing manner;
1) lookup val = lookup [ptjpm ax] (this takes value from lookup table)
2) lookup_val = ((lookup_val * psfreq) + 50 0 0 ) / 10000 (this adjust value according to frequency)

The variable psfreq is the current system frequency (H z) , w hich has been scaled up by a factor o f 100.
5 0 0 0 is added to it here in order to round it up. It is then divided by 10000, in order to a) scale the
frequency down to the correct value, and b) scale the lookup_val down by a factor o f 100, this effectively
scales the slip used in calculating the lookup table ‘lookup’ from % into absolute units, (see 3 .3 .1 )

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 445

 Page 20 of 39
Issue PC
10 May 1996

Including the E ffects o f Transient Saliencv

The second line in the Dm in calculation is so that the algorithm can take into account the effects o f
generator transient saliency in the calculation o f Dmin. Dtran is entered as an external setting. Dtran
should only be used in the calculation o f Dmin when the generator is in a transient, rather than steady state
mode o f operation. Dtran is only added to the Dmin calculation i f the ratio o f pt/pm ax < 0 .6 , since this
im plies that the generator is in a transient state. Under steady state conditions, the ratio o f pt/pm ax is set
by the pm ax = 1.4 * pt criterion, which makes the ratio o f pt/pm ax = 1/1.4 = 0 .7 1 4 , i.e lookup index 7
w ill be used during steady state conditions.

M inim um Setting V alue for Dmin

A s described in section 2 .3 .2 an absolute minimum negative value for D m in is implemented v ia the external
setting D fact. This is step 3 o f the Dm in calculation shown above. I f the value calculated by the above
form ula is greater than the value set by D fact (recall D fact is negative), then the setting D m in is m ade
equal to D fact.

Overall D m in Scaling
The resultant am ount by w hich Dmin is scaled in the code is given by;
S L I P S C A L E * P U P O L E S L I P P M A X * C O S _D E L T A _S C A L E /217
i.e (10 * 50 * 1000) * 8,000 / 131072 = 3 0 5 1 7 .5 7 counts/W /s
= 305 1 8 counts/W /s

This scaling value is called P U P O L E S L I P D M I N = 3 0 5 1 8 counts/W /s

The internal scaling is given by;

SLIP SC A L E * P U PO LE SLIP P M A X * COS D E L T A SC A L E / (PU _P O L E _SL IP _P * 2 17 =
3 .8 1 4 6 9 7 2 7 , this value is denoted as ‘D M I N S C A L E ’

3 .3 .2 S ettin g D fact
Section 2 .3 .2 stated that D fact should be entered between -10% and -50 %o f the generator rating.
M inim um and m axim um setting values in terms o f relay secondary quantities are therefore;

M IN IM U M D fact = -50% / 100 * 120 V * 1.5 A * V3 = -1 5 5 .9 W /s => -1 6 0 W /s

M A X IM U M D fact = -10% /1 0 0 * 100 V * 0.5 A * V3 = -8 .7 W /s => -5 W /s

in IN C R E M E N T S OF 0.1 W /s

T o scale into internal 'counts' these input values should be

SC A L E D B Y P U JP O L E SLIP D M IN = 30518 counts/W s'1

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 446

 Page 21 o f 39
Issue PC
10 May 1996

3 .3 .3 C alcu lation o f d eriv In sid e the A lgorith m .

deriv is calculated using equation 6. W ith a sam pling rate o f 4 sam ples per cycle and a nominal operating
frequency o f 50 H z, dt = 10 ms. Therefore 1/dt = 100 s'1. In order to get the value o f rate o f change o f
pow er to the sam e scaling as Dmin, the follow ing form ula has been used;

deriv = (powerP - last_p_val) * D E R IV _M U L T * (psfreq /5000);

where D E R I V M U L T = D M IN SC A LE * 100 * PU _PO L E _SL IP_P /P U _P O L E _SL IP _P O W =

= 3 .8 1 4 6 9 7 2 7 * 100 * 8 0 0 0 /4 0 0 0 = 763
D M IN S C A L E com es from the scaling in the D m in calculation, the 100 com es from the nom inal value o f
1/dt at 50 H z. The factor o f 2 (8 0 0 0 /4 0 0 0 ) com es from the real power scaling.

The deriv calculation is actually split into three separate steps.

S tep l T he first step is to calculate the quantity ‘ powerP - la st_ p _ v a l\ Once this has been done,
num erical saturation checks are performed on it to ensure that the deriv signal does not overflow .

Step2 The m agnitude o f the quantity ‘powerP - last_p _val’ is checked to see whether it is greater than a
pre-determined value called D E R I V O F , i f it is, then it is set equal to DERJV_OF. D E R I V O F is given

by;

D E R IV OF = 2 31 / M A X DERTV M U L T

where M A X D E R I V M U L T = D E R IV M U L T * 7 0 /5 0 = D M IN SC A LE * 100 * 2 * 7 0 /5 0 =
3 .8 1 4 6 9 7 2 7 * 100 * 2 * 70 /5 0 = 1069 counts. D ER IV _O F = 2 31 / 1069 = 20 0 8 871 counts, this value is
set by the fact that the algorithm does not have to work at frequencies above 70 Hz.

Step3 T he term ‘powerP - last_p _val’ is then multiplied by D E R IV M U L T * (p sffeq /5 0 0 0 ) i f the

frequency is b elow 70 Hz. I f the frequency is above 70 Hz, the term is m ultiplied by
M A X D E R I V M U L T . N ote that the protection is only enabled to provide a trip output betw een 40 and
70 Hz.

3 .3 .4 S ettin g D tran
Dtran is calculated from equation 9, i.e
Dtran = (1/X q - 1/ X d ’) * Sgen

The m axim um and minimum values o f X q and X d ’ are given by;

0.3 < Xq < 2.5 pu

0.1 < Xd’ < 2 pu Xq > X d’

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 447

 Page 22 of 39
Issue PC
10 May 1996

M IN IM U M Dtran = (1/2.5 - 1/0.1 ) * 120 V * 1.5 A * V3

= -2993 W /s => -3 0 0 0 W /s

M A X IM U M Dtran = (1/2 - 1/2) * 100 V * 0.5 A * V3

= 0 W /s
in IN C R E M E N T S OF 1 W /s

T o scale into internal 'counts' these input values should be

SC A L E D B Y PU _PO L E _SL IP_D M IN = 3051 8 counts/W s'1

3.4 OTHER SETTINGS - LAG, SLIP.

3 .4 .1 S ettin g ’L a g 1.
The external setting 'lag' is used in the calculation o f Pt, and can be varied betw een 0.1 and 10 seconds in
0.1 second intervals.

M IN IM U M L A G = 0.1 s
M A X IM U M LA G = 10 s
in IN C R E M E N T S OF 0.1 s

Pt is calculated in it's ow n function 'pole_slip_pt_calc', which is called every 20 m s (asynchronously). In

order to get 'lag' working correctly within this function, a counter is used to count the number o f function
calls, w hen this counter is equal to 'lag' * 50, a new value o f Pt is calculated.

T o scale into internal 'counts' the input value for 'lag' should be
SC A L E D B Y L A G SC A LE , which = 5 0 counts/sec .

3 .4 .2 S ettin g 'Slip'
The external setting 'slip' is used in the calculation o f Dmin

M IN IM U M SLIP = 0.1 %
M A X IM U M SLIP = 5 %
in IN C R E M E N T S OF 0.1 %

D ue to the integer maths requirement, scaling is required, to scale into internal 'counts' the input value for
'slip' should be SC A L E D B Y 10 counts/% . This is referred to as SLIP SCALE.

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 448

 Page 23 of 39
Issue PC
10 May 1996

3.5 INITIALISATION OF VALUES.

3.5 .1 In itialisin g th e cos (5 C) L o o k U p T able.

W hen the algorithm is being initialised, the look up table containing the terms 'slip * cos (8c)' should be
initialised. This is necessary since 'slip' is entered as an external input, and therefore the look up table
values vary according to this setting. The Slip setting is also part converted from percentage to absolute
values at this stage.

3 .5 .2 In itia lisin g O th er V ariab les.

The program m ing language 'C' w ill initialise all static type variables to zero i f no other initial value is
specified. M ost variables therefore need to be initialised in order to prevent divide by zero errors, and
algorithm m al-operation. The algorithm takes ' 2 * lag' seconds to initialise fully.

V ariables 'PU P O LE SLIP ? etc' These variables are stored in the include file 'dconst.h' and are given

by;

D E R IV M U L T = D M IN SC A L E * 100 * PU _PO L E _SL IP_P /P U _P O L E _SL IP _P O W = 763 counts

D E R I V O F = 2 31 / M A X D E R I V M U L T = 20 08871 counts
D M I N M A X P M A X = see section 3.3.1 = 1 ,2 2 4 ,8 8 5 ,5 0 4 counts
D M IN S C A L E =
SLIP SC A L E * P M A X SC A L E * C O S D E L T A S C A L E / (2 17 * PU _PO LE_SLIP_P)
= 3 .8 1 4 6 9 7 2 7
M A X D E R IV M U L T = D E R IV M U L T * 7 0 /5 0 = 1069 counts
P F A C T _SC A L E = 1 0 counts (Pfact setting scaled by 10)
P M A X OF 1 = 2 31 / 14 = 153391689 counts
P M A X O F2 = 2 31 / (P U _P O L E _SL IP_PM A X /P U _P O L E _SL IP_P O W ) = 2 31 / 100
= 2 1 4 7 4 8 3 6 counts
PU PO LE SLIP P = PU _I * P U _V = 800 * 10 = 8 0 0 0 counts/V A
P U P O L E S L IP Q = P U _I * P U _V * root3 / 2 = 800 * 10 * 1 .7 3 2 0 5 0 8 0 8 /2 = 69 2 8 counts/V A r
P U P O L E S L I P P O W = P U POLE SLIP P/2 = 4 ,0 0 0 counts/W
P U P O LE SLIP P T = P U POLE SLIP PO W * P F A C T S C A L E = 4 0 ,0 0 0 counts/W
PU P O LE SLIP P M A X = P U _PO L E _SL IP_PO W * P FA C T SC A LE *10 = 4 0 0 ,0 0 0 counts/W
PU P O LE SLIP D M IN = P U _PO L E _SL IP_P * D M IN SC A LE = 8000 * 3 .8 1 4 6 9 7 2 7
= 30518 counts/W /s
SLIP S C A L E = 1 0 counts/% (slip setting scaled up by 10)
L A G SC A L E = 50 counts/s

'Pt'. & V ariable 'P2' The variable Pt should be initialised to a value o f 1 'count' this m eans in
secondary values, that the algorithm Pt setting is at it's m ost secure value o f zero W atts. This w ill also
occur during generator run up where the machines output power should be practically zero (neglecting
transformer losses).

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 449

 Page 24 of 39
Issue PC
10 May 1996

The variable 'P2' which is used within the Pt estim ation function should also be set to 1 count. The other
variable 'PI' used in this function should be set to a larger value so that the terms |P - P2| & |P - P 11 are
not identical and one w ill at least be greater than Ptol.

Pt = 1 count, P2 = 1 count.

V ariable 'PT A s mentioned above, P I should be set to a high value o f power, so that it's difference
com pared to P2 is alw ays greater than Ptol. This ensures that Pt is not incorrectly estim ated until a true
steady state condition has been reached. The chosen value o f real pow er is tw o tim es nominal rated power
therefore 1 = 2 A m ps, V = 120 Volts;

PI = 120 V * 2 A * PU POLE SLIP Q = 3 ,3 2 5 ,5 3 8 counts

The P U PO LE SLIP Q value w as used since this contains the root3 factor w hich is necessary in
calculating the true generator per unit power.

V ariable 'Pmax' Pm ax should be set to a value equivalent to the generator rating, the ratio Pt/Pm ax
w ill then be in the region 0 to 0.1 thus resulting in a large, secure setting value for (AP/At)min
o f -1 * slip * Pmax.

Pm ax should therefore be initialised to ;

Pm ax = 120V * IA * PU _PO L E _SL IP_Q * P U _PO L E _SL IP_P M A X /P U _P O L E _P

= 8 3 ,1 3 8 ,4 3 9 counts.

The P U POLE SLIP Q value w as used for the sam e reason as in P 1.

V ariables 'last_p val' & 'pen p val' The variables 'last_p_val' and 'pen_p_val' are the stored values o f
m oving average filtered pow er used to calculate (AP/At). I f they are set to 0 counts, then the first tw o

values o f (AP/At) calculated w ill be positive i f power is positive, or w ill be zero i f the pow er is zero.

last_p _val = pen_p_val = 0 counts.

V ariables 'last powP' & 'last powO' These variables are the stored values o f 'raw' real and reactive
pow er used in calculating the respective m oving average values. I f both o f these are set to rated generator
output power, then they w ill make the first value o f real and reactive pow er calculated more positive than it
should be, this results in more secure values from an algorithm stability viewpoint.

last_pow P & last_pow Q = 120V * IA * PU _PO LE_SLIP_Q 12 = 4 1 5 6 8 0 counts.

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 450

 Page 25 of 39
Issue PC
10 May 1996

The division by 2 is due to the algorithm's internal scaling value.

Counter V ariables 'trip count1, 'cnt pm ax o f f . 'cnt pmax'. & 'count' A ll counter variables should be
initialised to zero.

V ariables 'Dmin'. 'deriv'. 'powerP' & 'powerQ' T hese variables can be initialised to a value o f 1 count,
their initialisation is not important since they w ill be calculated in the first tim e step. Since the algorithm
requires 1.5 pow er system cycles to operate, it is unlikely that it w ill nuisance trip on start up anyway.

3 .6 N U M E R IC A L O V E R F L O W C O N S ID E R A T IO N S & R E S O L U T IO N .

The m axim um input values o f voltage and current are 2 0 4 .8 V olts (rms line) and 2 0 .4 A m ps (rms phase).
These figures are given as rms quantities due to the hardware scaling w hich is used in the relay so that the
phasors calculated are in rms quantities. These therefore set the overflow considerations for the pow er
variables. In calculating the m aximum pow er from the m aximum voltage and current inputs, a factor o f V3
is included since the voltage used is a line voltage.

Reactive power. Q The scaling of this value is set to 6,928 countsA/Ar (PU_POLE_SLIP_Q). The
maximum generator power is;
2 0 4 .8 V * 2 0 .4 A * V3 = 7 2 3 6 .3 7 VAr. => 5 0 1 3 3 5 7 0 counts

This u ses log (5 0 1 3 3 5 7 0 ) / log (2) = 25 .6 => 26 bits required for m agnitude.

The resolution for a 1 Amp relay is 1/6928 = 144 .34 p.VAr (721.71 pVAr for 5 Amp)

R eal power. P The scaling o f this value is set to 4 ,0 0 0 counts/W (PU _P O L E _SL IP _P O W ). The
m axim um generator power is ;
20 4 .8 V * 2 0 .4 A * V3 = 7 2 3 6 .3 7 W => 2 8 9 4 5 4 8 0 counts
This uses log (2 8 9 4 5 4 8 0 ) / log (2) = 24.8 => 25 bits required for magnitude.

The resolution for a 1 Amp relay is 1/4000 = 2 5 0 pW ( 1.25 m W for 5 A m p)

Real Pow er Trip Level, Pt The scaling o f this value is set to 4 0 ,0 0 0 counts/W ( P U P O L E S L I P P T ) .
The m axim um generator power is ;
20 4 .8 V * 20 .4 A * V3 = 7 2 3 6 .3 7 W => 2 8 9 4 5 4 8 0 0 counts

This u ses log (2 8 9 4 5 4 8 0 0 ) / log (2) = 28.11 => 29 bits required for magnitude.

The resolution for a 1 Amp relay is 1/40000 = 25 pW ( 125 pW for 5 Am p).

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 451

 Page 26 of 39
Issue PC
10 May 1996

Real Pow er M axim um Output Level. Pmax The scaling o f this value is set to 4 0 0 ,0 0 0 counts/W
(PU PO LE SLIP P M A X ). The maximum value o f pm ax is limited by numerical saturation and is equal
to
P M A X O F 2 * 100 = (2 31 / 100) * 100 = 2 1 4 7 4 8 3 6 0 0 counts

The m axim um value o f pm ax in terms o f secondary watts is therefore given by;

P M A X O F 2 * 100 / PU POLE SLIP P M A X = 2 ,1 4 7 ,4 8 3 ,6 0 0 /4 0 0 ,0 0 0 = 5 3 6 8 .7 W atts

The resolution for a 1 Amp relay is 1/400000 = 2.5 pW ( 1 2 .5 pW for 5 Amp).

Rate o f Change o f Pow er (AP/At) The scaling o f this value is set to 3 0 5 1 8 counts/W /s
(P U P O L E S L IP D M IN ) . The maximum number that this can reach is limited within the algorithm
code, it is equal to;

D E R IV OF * M A X D E R IV M U LT = 2 0 08871 * 1069 = 2 ,1 4 7 ,4 8 3 ,0 9 9 counts

This in terms o f secondary W atts/second is equivalent to;

2 ,1 4 7 ,4 8 3 ,0 9 9 /3 0 5 1 8 = 70 ,3 6 7 .8 W /s

The resolution for a 1 Amp relay is 1/30518 = 3 2 .7 p W /s ( 163.8 p W /s for 5 Amp).

Rate o f Change o f Pow er trip level (A P /A tL in The scaling o f this value is set to 30518 counts/W /s
(P U P O L E S L IP D M IN ). The m aximum number that this can reach is limited in the code by the
variables Pm ax, slip, and cos(5c) (see equation 7).

The m axim um value for lookup val term in the D m in calculation is -1 0 0 0 * 2 2 0 counts, the maximum
value Dtran is -9 1 5 5 4 0 0 0 counts. The maximum value for pm ax is limited in the Dmin calculation to is
D M IN _M A X _P M A X / 2 16 = 18690 counts. R ecall that the lookup val is scaled down by a factor o f tw o
in the D m in calculation.. The m aximum number is therefore ;
((-1 0 0 0 * 2 2 0 )/2 * 1 8 6 9 0 ) -9 1 5 5 4 0 0 0 = -2 ,1 4 7 ,4 5 4 ,0 0 0 counts.
This u ses log (2 ,1 4 7 ,4 5 4 ,0 0 0 ) / log (2) = 3 0 .9 9 9 => 31 bits required for magnitude. This w ill not
overflow even i f Pm ax and all o f the other settings were at their m axim um value. This in terms o f
secondary W att/s is; -2 ,1 4 7 ,4 5 4 ,0 0 0 /3 0 5 1 8 = 7 0 3 6 6 .8 W /s

The resolution for a 1 Am p relay is 1/30518 = 32 .7 pW /s ( 163.8 p W /s for 5 Amp).

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 452

 Page 27 of 39
Issue PC
10 May 1996

3.7 D IS T U R B A N C E R E C O R D IN G .

3 .7 .1 D istu rb an ce R eco rd er R eq uirem ents.

The LG PG 111 is capable o f storing 2 disturbance records in its volatile memory. Once these tw o records have
been filled up, no further disturbance recording w ill be performed until the records in the buffer have been
uploaded to a PC. For the Scottish field trials, interrogation o f the relay can be achieved remotely the University
o f Bath.

The L G PG 111 is capable o f storing 8 channels, each channel having a capacity to store 7 6 8 sam ples,
which at 4 sam ples per cycle produces a disturbance record o f 3 .8 4 seconds at 50 Hz. In addition to these
8 channels the relay also stores all o f the digital inputs and relay status outputs. The variables w hich need
to be recorded are powerP, pt, pm ax, powerQ, deriv, Dmin. In addition to these 6 algorithm variables, if
possib le it is hoped that the spare tw o channels can be used to store one voltage m agnitude and voltage
phase angle.

Within the algorithm code 3 additional ‘relay status’ outputs w ill be formed which w ill flag when each o f the
three algorithm criterions are satisfied ((powerP < pt), (powerQ < Qtrip), & (deriv < Dmin)). This w ill help with
the debugging o f the algorithm. The relay status outputs w ill be named as follows;

1) Pole slipping algorithm trip fimction details name > POLE SLIP
2) Reactive power criterion function_details name > Q T R I P C RITERI ON
3) Real power criterion fimction details name > PT C R H ERION
4) Rate o f Change o f power criterion function details name > D M IN CRITERION

Disturbance Recorder Trigger Conditions.

There are tw o different conditions which should trigger the disturbance recorder. The first trigger is provided by
the pole slipping algorithm relay output, P O L E S L IP . This is the normal w ay that the disturbance recorder is
triggered. The second trigger condition is provided when the pole slipping algorithm’s reactive trip criterion is
satisfied ( powerQ < Qtrip), the relay status QTRIP CR11LRION will provide the trigger. Disturbance records
triggered in this manner should provide an indication o f conditions which m ay potentially cause algorithm mal-
operation if settings are incorrect. The trigger conditions can be selected from the LG PG ’s front end menu
system. An overcurrent trip will also be used for the field trials to provide a higher likelihood o f capturing data
during a disturbance.

In summary the disturbance recorder should be triggered if;

1) Q < Qtrip (within pole slipping algorithm)
or
2) Pole slipping algorithm produces trip output
or
3) Overcurrent algorithm produces a trip.

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 453

 Page 28 of 39
Issue PC
10 May 1996

Disturbance Recorder Pre & Post Fault Time Allocations.

The amount o f pre trigger and post trigger time in the disturbance record can be adjusted from the relays front
end. The amount o f pre-trigger, and post - trigger time should be allocated as follows;

pre-trigger period = 2.34 seconds (117 power system cycles)

post trigger period = 1.50 seconds (75 power system cycles)
total disturbance record = 3.84 seconds

These times w ill ensure that with pole slipping events, the disturbance record will capture the pole slip. For
events such as stable power swings, which should not cause the algorithm to trip, the 1.5 second post trigger
period should be adequate to allow analysis o f the algorithm measurands.

Frequency M easurement.
Since the disturbance record results are frequency locked, it is im possible to derive the frequency
m easurem ent in the usual w ay, (zero crossings). The disturbance record contains the tim e at w hich each
sam ple w as stored however, and this information can therefore be used to calculate the frequency to an
accuracy o f around 0 .0 2 H z.

3 .7 .2 S ca lin g from 3 2 B it P recision to 16 B it P recision.

The disturbance recorder requires that the variables it stores are o f 16 bit precision. Eventually, the w hole
algorithm w ill probably be implemented in 16 bit format, since this w ill im prove the operation speed
greatly. T his conversion w ill be performed when the algorithm is optimised.

N O T E : the valu es calculated below are for the 1 Amp relay

R eactive pow er. powerQ The scaling o f this value is set to P U P O L E S L I P Q , 6 ,9 2 8 counts/VA r. If
the internal 32 bit variable is divided by 29 = 5 1 2 , then the above scaling figure becom es;
6 9 2 8 /5 1 2 = 13.53125 counts/V A r

The resolution is therefore 1/13.53125 = 7 3 .9 0 mVAr.

The reactive pow er disturbance record m ay therefore overflow at a value of;

2 15 cou n ts/ 13.53125 counts/V A r = 24 2 1 .6 5 VAr
Section 3 .6 states that the maximum input level is 7 2 3 6 .4 Var. I f you consider a m achine o f secondary
rating 120 * 1 * V3 = 207 .8 4 V A , then the maximum possible reactive pow er m easurement is
2 4 2 1 .6 5 /2 0 7 .8 4 = 11.7 * rating o f generator, this is very unlikely. This resolution therefore provides
2 0 7 .8 4 V A r/7 3 .9 m V A r = 2812 discrete levels between zero and nominal rated power.

Real pow er. powerP The scaling o f this value is set to P U P O L E S L I P P O W , 4 ,0 0 0 counts/W . If the
internal 32 bit variable is divided by 29 = 512, then the above scaling figure becom es;

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 454

 Page 29 of 39
Issue PC
10 May 1996

4 0 0 0 /5 1 2 = 7.8125 counts/W

T he resolution is therefore 1/7.8125 = 128 mW.

T he real pow er disturbance record m ay therefore overflow at a value of;

2 15 c o u n ts /7.8125 counts/W = 4 1 9 4 .3 W .
Section 3 .6 states that the m axim um input level is 7 2 3 6 .4 W . I f you consider a machine o f secondary rating
120 * 1 * V3 = 2 0 7 .8 4 V A , then the m aximum possible real pow er measurement is 4 1 9 4 .3 /2 0 7 .8 4 = 2 0 .0 *
rating o f generator.

R eal Pow er Trip Setting . pt The scaling o f this value is set to PU _PO L E _SL IP_P T , 4 0 ,0 0 0 counts/W .
I f the internal 32 bit variable is divided by 2 n = 4 0 9 6 , then the above scaling figure becom es;
4 0 0 0 0 /4 0 9 6 = 9 .7 6 5 6 2 5 counts/W

T he resolution is therefore 1 /9.765625 = 102.4 mW .

T he real pow er trip setting disturbance record m ay therefore overflow at a value of;
2 15 c o u n ts /9 .7 6 5 6 2 5 counts/W = 3 3 5 5 .4 4 W .

Section 3 .6 states that the m axim um input level is 7 2 3 6 .4 W . I f you consider a m achine o f secondary
rating 120 * 1 * V3 = 2 0 7 .8 4 V A , then the m axim um p ossib le real pow er trip level m easurement is
3 3 5 5 .4 4 /2 0 7 .8 4 = 16.1 * rating o f generator.

R eal Pow er M axim um Output Monitor, pm ax The scaling of this value is set to
P U _P O L E _S L IP _P M A X , 4 0 0 ,0 0 0 counts/W . I f the internal 32 bit variable is divided by 2 15= 32768,
then the above scalin g figure becom es;
4 0 0 0 0 0 /3 2 7 6 8 = 12.207 counts/W

The resolution is therefore 1 /1 2 .2 0 7 = 81.9 mW.

pm ax is lim ited in the code to 2 ,1 4 7 ,4 8 3 ,6 0 0 counts, this converts to;
2 ,1 4 7 ,4 8 3 ,6 0 0 / 2 15 counts = 6 5 5 3 6 counts. This requires lo g (6 5 5 3 6 ) /lo g (2 ) = 15.99
bits and w ill therefore overflow at 2 15 /1 2 .2 0 7 = 2 6 8 4 .3 6 W atts (secondary)

I f you consider a m achine o f secondary rating 120 * 1 * V3 = 2 0 7 .8 4 V A , then the m axim um p ossible
value o f Pm ax m easurem ent is 2 6 8 4 .3 6 /2 0 7 .8 4 = 12.9 * rating o f generator.

Rate of Change of Real Power Measurement. (AP/At)

The scaling o f this value is set to P U POLE SLIP D M IN , 3 0 ,5 1 8 counts/W /s. The m axim um magnitude
that this variable is permitted to reach by the algorithm code is

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 455

 Page 30 of 39
Issue PC
10 May 1996

D E R IV O F * M A X D E R IV M U L T = 2 ,0 0 8 ,8 7 1 * 1069 = 2 ,1 4 7 ,4 8 3 ,0 9 9 counts

T his needs to be scaled down by 2 15 to achieve 16 bit precision. The above scaling figure therefore
becom es
3 0 ,5 1 8 / 2 15 = 0 .93133545 counts/W /s

The resolution is therefore 1/0. 9 3133545 = 1.0737 W /s.

The (AP/At) measurement has a m aximum of;

2 ,1 4 7 ,4 8 3 ,0 9 9 counts / 2 15 = 6 5 5 3 6 counts,
this requires log (6 5 5 3 6 ) / log (2) = 16 bits. It m ay therefore overflow.

O verflow w ill occur at 2 15 / 0.931 3 3 5 4 5 = 35183 W /s (secondary)

R ate o f Change o f Real Power Trip Setting . (A P /A fL in

The scaling o f this value is the sam e as above ( 30 ,5 1 8 counts/W /s). It w ill therefore require the same
scaling down as the above to get it into 16 bit format, i.e it needs scaling down by 2 15 = 3 2 ,7 6 8 . The
m axim um value o f Dmin is limited in the code to 2 ,1 4 7 ,4 5 4 ,0 0 0 counts. The m axim um value i f therefore
given by
2 ,1 4 7 ,4 5 4 ,0 0 0 /2 15 = 65535 counts this requires lo g (6 5 5 3 5 ) / log(2) = 15.9 bits and m ay
therefore overflow !

O verflow w ill occur at 2 15 / 0.93 1 3 3 5 4 5 = 35183 W /s (secondary)

Sum mary o f Scaling V alues U sed in Converting from 32 bit to 16 bit Format.

Tables 3 and 4 detail the scaling, overflow points and resolution for the 1 and 5 amp relays.

Table 3 - Scaling values for 1 Amp Relay (32 bit to 16 bit conversion for data capture)

Variable Scaling Factor Overflow point Resolution Comment

Q 29 = 512 2421.7 VAr 73.9 mVAr numerical overflow possible
p 29 = 512 4194.3 W 128 mW numerical overflow possible
Pt 2 n = 4096 3355.4 W 102.4 mW numerical overflow possible
Pmax 215 = 32768 2684.4 W 81.9 mW numerical overflow possible
(AP/At) 215 = 32768 35183 W/s 1.0737 W/s numerical overflow possible
(AP/At)min 215 = 32768 35183 W/s 1.0737 W/s numerical overflow possible

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 456

 Page 31 of 39
Issue PC
10 May 1996

Table 4 - Scaling values for 5 Amp Relay (32 bit to 16 bit conversion for data capture)

V ariable Scaling Factor Overflow point Resolution Comment

Q 29 = 512 12108.5 V Ar 369.5 m VAr

p 29 = 512 2 0971.5 W 6 4 0 mW

Pt 2 12 = 4096 16777.2 W 5 12 m W
Pm ax 2 15 = 3 2768 13422W 4 0 9.5 m W should be an unsigned int

(A P /A t) 2 15 = 32768 175915 W /s 5.3 6 8 5 W /s saturation definitely required

( A P / At),nin 2 15 = 3 2 768 175915 W /s 5.3 6 8 5 W /s saturation definitely required

VERSION CONTROL

Issu e A uthor R eason for Change D ate

PA M att Checksfield Original 30 /6 /9 5

PB M att C hecksfield Addition o f disturbance recorder information & Corrections to 14/9/95
scaling values.
PC M att C hecksfield M odification o f algorithm (Dtran etc) and addition o f sym bolic 10/5/96
constants to replace ‘m agic num bers’ in code

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 457

 Page 32of 39 32
Issue PC
10/5/96
M.J.Checksfield/PhD/University of Bath 1996

4.0 SUMMARY OF ALL SCALING FACTORS IN ALGORITHM

The table belows contains a summary of the user interface pole slipping variables for display and setting. The default values given on the 1 Amp tables are
for the SLOY Scottish generator. The dafault values for the 5 Amp tables are 5 times those for the 1 amp.

SETTINGS 1 AMP RELAY

Protection Mapping Affected

Setting Minimum Maximum Step Default Menu Scaling Total Scaling by
xformer
ratios
and In
lVAr = 1 6928 lVAr = 6928
Qtrip -1100 VAr -30 VAr 1 VAr -272 VAr [PU_POLE_SLIP_Q] PU_POLE_SLIP_Q Yes
[(-1/0.7) *110V*1A*V3]
1W =10 400
Ptol 1.0W 70 W 0.1 W 6.9 W [PU_POLE_SLIP_POW/l 0] 1W = 4000 Yes
[5%*138VA] PU POLE SLIP POW
1=10 NONE
Pfact 0.7 1.0 0.1 0.9 1=10 No

lW /s = 10
Dfact -160.0 W/s -5.0 W/s 0.1 W/s -47.6 W/s 3051.8 lW /s = 30518 Yes
[(AP/At)fact] [-0.25s'1*110V*lA*‘\/3] [PU_POLE_SLIP_DMIN/l 0] PU_POLE_SLIP_DMIN
Appendix E, Page

lW /s = 1
Dtran -3000 W/s 0 W/s 1W -572 W/s 30518 lW /s = 30518 Yes
[(AP/At)tran] [-3s'l*110V*lA*V3] [PU_POLE_SLIP_DMIN] PU_POLE_SLIP_DMIN

1%=10
Slip 0.1% 5% 0.1 % 0.5 % NONE 1%=10 No

ls=10
1.0 s 5 ls=50 No
458

Lag 0.1s 10.0 s 0.1 s

(20ms resolution)
 Page 33of39 33

Issue PC
10/5/96
M.J.Checksfield/PhD/University of Bath 1996

SETTINGS 5 AMP RELAY

The table below was derived from the 1 Amp table by multiplying all setting values which are dependent upon current input levels by a factor of 5 , and dividing all scaling values by a factor
of 5

Protection Mapping Affected by

Setting Minimum Maximum Step Default Menu Total Scaling xformer
Scaling ratios and In
lV A r = 1 1386 lV A r= 1386
Qtrip -5500 VAr -150 VAr 5 VAr -1360 VAr [PU_POLE_SLIP_Q/5] PU_POLE_SLIP_Q/5 Yes
[(-1/0.7)*110V*5A*V3]
1W = 10 80
Ptol 5.0 W 350 W 0.5 W 34.5 W [PU POLE SLIP POW/ 1W = 800 Yes
[5%*5*138 VA] (10*5)] PU_POLE_SLIP_POW/5

1=10 NONE
Pfact 0.7 1.0 0.1 0.9 1=10 No

—»
o Sj-
►

II
Dfact -800.0 W/s -25.0 W/s 0.5 W/s -238.0 W/s 610.35 IW /s = 6104 Yes
[(AP/At)fact] [-0.25s'I*n0V*5A*V3] [PU POLE SLIP DMEN/ PU_POLE_SLIP_DMIN/5
(10 * 5)1
IW /s = 1
Dtran
Appendix E, Page

-15,000 W/s OW/s 5W -2858 W/s 6103.5 IW /s = 6104 Yes

[(AP/At)tran] [-3s'1*110V*5A*V3] [PU POLE SLIP DMDST/ PU_POLE_SLIP_DMIN/5
51
1%=10
Slip 0.1% 5% 0.1 % 0.5 % NONE 1%=10 No

ls=10
Lag 0.1s 10.0 s 0.1 s 1.0 s 5 ls=50 No
(20ms resolution)
459
 Page 34of39
Issue PC
10/5/96
M.J.Checksfield/PhD/University of Bath 1996

CALCULATION OF SETTINGS

The table below details the methods and maximum and minimum setting limits

Setting Formula M inim um Maximum setting

setting

Qtrip -1/X q * V n * I n * V 3 X q = 2.5 pu X q = 0.3 pu

Ptol ‘x % ’/100 * V n * In * V3 x =1% x = 20 %

Pfact none 0.7 1.0

Dfact ‘y % V100 * V n * In * V3 y = - 10 % y = - 50 %
(AP/At)fect
Appendix E, Page

Dtran (1/X q - 1/X d ’) * V n *In *V3 X q = 2.5 pu X q = 2 pu

(AP/At)tran Xd* = 0.1 pu X d ’ = 2 pu

shp none 0.1 % 5%

lag none 0.1 seconds 10 seconds

460
 Page 35of39
Issue PC
10/5/96
M.J.Checksfield/PhD/University of Bath 1996

MEASUREMENTS 1 AMP RELAY

Resolution M axim um Affected by

Measurement (1 Amp) M agnitude M enu Scaling Comments xformer ratios
Internally. and In

PU_POLE_SLIP_PT m aximum magnitude due to V & I input

Pt 25 pW 7236.37 W 40,000 counts/W lim its being reached Y es

PU_POLE_SLIP_POW m aximum magnitude due to V & I input

P 250 pW 7236.37 W 4000 counts/W lim its being reached Y es

PU_POLE_SLIP_PM AX m axim um magnitude lim ited by code - max

Pm ax 2.5 pW 5368.7 W 400,000 counts/W m agnitude is PM A X OF2 * 100 Y es

PU_POLE_SLIP_Q m axim um magnitude due to V & I input

Q 144.34 pVAr 7236.37 VAr 6928 counts/W lim its being reached Y es

PU_POLE_SLIP_DM IN m axim um m agnitude lim ited by code - m ax

D m in 32.7 pW /s 7 0 ,366.8 W /s 30518 counts/W /s magnitude lim ited by Y es
Appendix E, Page

[(AP/At)min] D M IN _M AX _PM A X , and slip setting m ax

PU_POLE_SLIP_DM IN m axim um m agnitude lim ited by code - m ax

Deriv 32.7 pW /s 7 0 ,367.8 W /s 30518 counts/W /s magnitude set by D E R IV OF * Y es
[(AP/At)] MAX_DERTV_M ULT
461
 Page 36of39 s6
Issue PC
10/5/96
M.J.Checksfield/PhD/University of Bath 1996

MEASUREMENTS 5 AM P RELAY

Resolution M aximum Affected by

Measurement (5 Amp) Magnitude Menu Scaling Comments xformer ratios and
Internally. In

PU_POLE_SLIP_PT/5 m axim um m agnitude due to V & I input

Pt 125 pW 36181.9 W 8,000 counts/W lim its being reached Y es

PU_POLE_SLIP_POW /5 m axim um magnitude due to V & I input

P 1.25 mW 36181.9 W 800 counts/W lim its being reached Y es

PU_POLE_SLIP_PM AX/5 m axim um magnitude lim ited by code - m ax

Pmax 12.5 pW 26843.5 W 80,000 counts/W m agnitude is PM A X _O F2 * 100 Y es

PU_POLE_SLIP_Q/5 m axim um m agnitude due to V & I input

Q 721.7 pVAr 36181.9 VAr 1386 counts/W lim its being reached Y es

PU_POLE_SLIP_DM IN/5 m axim um magnitude lim ited by code - m ax

D m in 163.5 pW /s 351,834 W /s 6104 counts/W /s magnitude lim ited by D M IN M A X PM A X , Y es
[(AP/At)min] and slip setting m ax
Appendix E, Page

PU_POLE_SLIP_DM IN/5 m axim um magnitude lim ited by code - m ax

Deriv 163.5 pW /s 351,839 W /s 6104 counts/W /s m agnitude set by D ER IV OF * Y es
[(AP/At)] M A X D E R IV M U L T
462
 T h e P o w er B a sed P ole S lip p in g A lg o rith m .

start

Cosine fourier filter V & I's NOTE

@ 4 samples / cycle
1) All o f the functions in this
flowchart are calculated every quarter
Calculate real power & apply half cycle o f a power system cycle. The pole
moving average filter. Store present and slipping function is synchronously
previous tw o values o f filtered power for locked to the power system
(AP/At) calculation. Store previous 'raw' frequency.
value o f power for next moving average
calculation. 2) Pt, the real power trip level is
calculated Asynchronously every
Tag' seconds.
Calculate reactive power & apply half cycle
moving average filter. Store previous *raw'
value o f power for next moving average 3) The algorithm only resets once
calculation. the reactive power criterion is NOT
satisfied. This is to prevent 'chatter'
o f the relay outputs.
Calculate Pmax

Calculate (AP/At)mii

Calculate (AP/At)

yes Are P & yes

Is Qtrip
(AP/At) trip Increment Trip Count
Criterion
Criteria
Satisfied ?
Satisfied ?

no no

yes
Is TRIP'
Reset Trip Count signal NOT Reset Trip Count
TRUE?

no
Trip Count = 6

yes Trip Count = 6

Trip Count TRIP' = TRUE
>=6 ? TRIP SCHEME LOGIC

no
NOTE: flowchart does
TRIP' = FALSE not describe numerical
BLOCK SCHEME saturation techniques or
LOGIC how trip settings are
derived, this can be found
in the main body o f the
report.

Figure 1
Flowchart Showing Operation of The Pole Slipping Algorithm
MJC - 21/11/95 - 9 am Page 37 o f 39
PC

M.J.Checksfi el d/PhD/University of Bath 1996 Appendix E, Page 463

 Pt C a lcu la tio n F u n ctio n .

Start

(check generator has been

operating in the steady state
for the last 2 * lag' seconds)
(Update Pt setting to
P-P2 <Ptol Yes N ew value)

P -P l <Ptol

Yes
set Pt
No
to 10 P t< 0
counts

No
P i = P2
P2 = P
(Ensure Pt and therefore
Pmax is not negative)

(Store previous values)

NOTE
1) The Pt calculation function is called every 'lag' seconds, the most recent value of
moving average filtered real power calculated at the time the function is called is set
to 'P'. This function is separate from the rest of the algorithm and is performed
asynchronously, all of the other algorithm functions are called synchronously every
quarter of a power system cycle.

2) Pi is the value of P used '2*lag' seconds ago, whilst P2 is the value of P used 'lag'
seconds ago.

3) Pfact is the scaling factor used to introduce a safety margin in the Pt estimation.
Nominally, it is set to 0.9 in absolute terms. It can be externally set from 0.7 to 1.0
in 0.1 increments.

4) Ptol is the tolerance band which the generator output power must stay within
over the time period' 2 * lag' seconds for the value of Pt to be updated.

5) The Pt function is actually called every 20 ms. The menu scaling on variable lag
is therefore set so that menu scaled lag * 2 0 ms = actual lag in seconds, i.e the menu
scaling is 50, this is so a counter within the Pt calculation function can count when
Pt should be calculated. The Pt calculation function although called every 20 ms,
therefore only runs through the above flowchart every 'lag' seconds.

Figure 2
The Generator Input Power Estimation Function (Calculates Trip Setting, Pt.)

M JC- 21/11/95 - 9 am Page 38 o f 39

M.J.Checksfield/PhD/University of Bath 1996 Appendix E, Page 464

 P m ax C a lcu la tio n F u n ctio n .

Start

Yes
Yes 'count Pmax' Pmax = P
Increment
Is P > Pmax
'count Pmax'
>*= 4 'count Pmax' = 1
v (1 cycle) j 'count Pmax o f f = 1

No
No
'count Pmax' = 0

This part o f the function sets Pmax to values greater than nominal
if the power output o f the generator rises above 1.4 * Pt
for more than 1 power system cycle.

This part resets Pmax to its nominal value o f 1.4 * Pt

if no update has occured in the last second

Yes
Increment
'count Pmax o ff
' count Pmax o ff
v \=o y

No
Yes
'count Pmax o:
Pmax = 1.4*Pt 'count Pmax o ff = 0
>=200
s. (1 secon d ),

No

NOTE
1) With reference to the boxes which check whether 'count Pmax' is greater than or equal to one cycle,
the criterion is that it should be in terms o f power system cycles, not absolute time. This will therefore
work as intended over the complete frequency operating range.
2) The function tracks any slow changes in power output from the generator which are greater than the
nominal value o f 1.4 * Pt. It will reject any spikes produced by disturbances such as faults, this is why
the power, P has to be greater than Pmax for one power system cycle before an update occurs. Once an
update occurs, the updated value is held' for one second, Pmax is then reset to its nominal value.
3) The algorithm has to successfully detect pole slipping over the operating range o f 4 0 - 70 Hz. Since
the Pmax o ff counter is counted in numbers o f quarter cycles, the time o f one second w ill vary with
frequency. The actual 'Pmax o f f time will vary between 1.25 and 0.714 seconds. The amount o f time
required to reset is approximately inversely proportional to the square o f the system frequency. It is
dictated by the natural frequency o f swinging o f the generator rotor.
4) Flowchart does not contain any details o f the numerical saturation employed, this can be found in the
main body o f the report.

Figure 3
Generator Maximum Power Output Monitor (Pmax Measuring Function).
M J C - 2 1 / 1 1 / 9 5 - 9 am Page 39 o f 39
Issue PC

M.J.Checksfi el d/PhD/University of Bath 1996 Appendix E, Page 465

 Appendix F
EXAMPLES OF COMPUTER FILES USED TO
TEST THE POLE SLIPPING ALGORITHM.

F l.l PPSS Simulation File PSL3GA - Pole Slip From Full Load,
Governor & AVR Included

This is the ATP file used to produce test 'PSL3GA'.

C File PSL3GA.DAT
C Details POLE slipping induced by 2 Phase to earth Fault,
C Generator control details;
C A VR (Y E S )
C GOVERNOR (Y E S )
C INITIAL LOAD 100 % (200 MW)
C
C Source System taken from example given on Page 7-18 of the
C EMTP Vol 3, Wookbook 3, June 1989, a research project
C report made by the EPRI, Poject 2149-6, Vol EL-4651
C Source capacity and transformer impedance both decreased
C
C System :- 200 MVA syncheonous generator using type SM 59 model,
C 13.8 kV (line-line rms), No saturation modelling.
Transformer has 3.5 % PU leakage reactance (200 MVA base).
Infinite bus seperated from machine by 144 Km dist param line.
Load is 235.4 MW 0.9985 power factor lagging (0 132 kV)
When fault cleared, load and feeder are also cleared.
Fault separated from generator by type 1,2,3 line, (unbalanced)
Load seperated from fault by inductance.(transformer)
System frequency changed from 60 to 50 Hz.

The Governor model has been changed to the one given in DC47.DAT (and
also in LEC EMTP Summer Course, Leuven 15 july 91, Rotating machines)
This is faster than the original hydro governor.

M Checksfield 10/8/95

BEGIN NEW DATA CASE

POWER FREQUENCY 50.0 (special request card
C ............. Set up I/O to write to correct file................................
$CLOSE, UNIT=4 STATUS=DELETE
$OPEN, UNIT=4 FILE=C:\atp\gecptl\psl3ga.pl4 FORM=FORMATTED STATUS=UNKNOWN !
C ............................Misc Data..............................................
C <DELTA< TMAX< XOPT< COPT<-EPSILN<-TOLMAT<-Tstart
200.0E-6 6.0
C <-IOUT<— IPLOT<-IDOUBL<-KSSOUT<-MAXOUT<-- IPUN<-MEMSAV< ICAT<-NERERG<-IPRSUP
500 5 1 1

M.J. Checksfi el d/PhD /University of Bath 1996 Appendix F, Page 466

TACS HYBRID
C opName s<— ipl s<— ip2 s<— ip3 s<— ip4 s<— ip5 <-gain<-FxLo<-FxHi<-NmLo<-NmHi
C < — No, Do< N l ,Dl< N2,D2< N 3 ,D3< N 4 ,D4<----N5, D5<---- N6, D6< N 7 ,D7
C Zero_block
VR +V2 1. -2.0 2.0
VF +DVF +UNITY 1.
C Laplace S blocks
1V2 -V I +UNITY -V3 400,

.02
1V1 +VDC
.523598 8
.03
1DVF +VR

.015
1V3 +DVF
0. .03
1. .5
C SPEED GOVERNOR SYSTEM
92VEL1
C CONVERSION OMEGA TO PER UNIT SYSTEM
99TACOM = .0031831 * VELl
C Speed Governor System.
ODIFOME -UNITY +TACOM
OKDIFOM +DIFOME 20.
0PS1 +UNITY -KDIFOM -PGV
0PS2 +PS1 10. -.4 .4
88PS21 = PS2
1PGV +PS21 1.0 0. 1. 1
1.
0.5
1PGV1 +PGV 1.

0.25
1PGV2 +PGV1

10,
1PGV3 +PGV2

0 .5
88PM1 = 0.274 * PGV1
88PM2 = 0.243 * PGV2
88PM3 = 0.483 * PGV3
88PMT = PM1 + PM2 + PM3
88PM = 1.0 * PMT
C NODE_V
90BUS1A 50
90GEN3GA 50
90GEN3GB 50
90GEN3GC 50
C emtp switch current defined as flowing left node to rgt
91GEN3A - 1.0 9999,
91GEN3B - 1.0 9999,

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 467

91GEN3C -1.0 9999.
C RMS_VALUE
88VT_A 66+GEN3GA 50.
88VRMS1A66+BUS1A 50.
88VDC - (ABS(GEN3GA)+ABS(GEN3GB)+ABS(GEN3GC))*SQRT(3/2)/13800
88WREF =2*PI*50
C tacs algebraic & logical fortran expressions for calculating P and Q etc
C OPNAME = < -------------------------------------------------------------------------------------------FR EE FORMAT FORTRAN E X P R E S S IO N
88REAL = (GEN3GA*GEN3A) + (GEN3GB*GEN3B) + (GEN3GC*GEN3C)
88QA = GEN3GA * (GEN3C - GEN3B)
88QB = GEN3GB * (GEN3A - GEN3C)
88QC = GEN3GC * (GEN3B - GEN3A)
88REACT = (QA + QB + QC) *0.57735027
C TAC S_OUT PUT
C <-Name<-Name<-Name<-Name<-Name<-Name<-Name<-Name<-Naine<-Name<-Name
33GEN3GAGEN3GBGEN3GCGEN3A GEN3B GEN3C
C TACS_IC
77VDC 1.732051
77VF 1.
77DVF 0.
77V2 0.
77VR 0.
7 7VI 1.
77V3 0.
C
77VEL1 314.15926
77TACOM 1.0
77PS1 0.0
77PGV 1.0
77PGV1 1.0
77PGV2 1.0
77PGV3 1.0
77PMT 1.0
77PM 1.0
BLANK card terminates TACS data
C .................................. Circuit_Data....................................
C Bus— >Bus — ><---------- ><----R<---------- L<---- R<---------- L<---- R<---------- L
51THEVA BUS7A .0163 2.9638 0
52THEVB BUS7B .0075 4.999 0
53THEVC BUS7C 0
C Bus— >Bus— >Bus— >Bus— >< R< L< C< R< L< C<----R<---- L<---- C
1BKR1A BUS12A 0.728510.209.06562
2BKR1B BUS12B 0.59274.8968-.01510.748410.179.06834
3BKR1C BUS12C 0.57954.0768-.00380.59274.8968-.01510.728610.209.06562
C Bus— >Bus— >Bus— >Bus— >< R'<---L 1<--- C 1<— len 0 0 0<------- Blank--------- >0
-1BUS7A BUS1A 0.3167 3.222.00787 144.4 0 0 0 0
-2BUS7B BUS1B 0.0243 .9238 .0126 144.4 0 0 0 0
-3BUS7C BUS1C 0
C Bus— >Bus— >Bus— >Bus— >< R< L< C O
BUS12ABUS13A 70.16 0
BUS12BBUS13BBUS12ABUS13A 0
BUS12CBUS13CBUS12ABUS13A 0
BUS13A 221.41 38.26 0
BUS13B BUS13A 0

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 468

 BUS13C BUS13A 0
C Saturable_transformer_components.
C ----------- ><-Bus3<----><---- 1<— Phi<BusSt<-Rmag<----------------------------- >0
TRANSFORMER DELTAB 0
C ------- current<------------ flux
9999
C <-Busl<-Bus2<---------- ><---Rk<---Lk<---Nk<------------------------------------>0
1GEN3A GEN3B .0421 13.8 0
2BUS1A 11.69132.79
C Note.- These Leakage values were calculated assuming a 3.5 % p.u reactance at
C 200 MVA for the transformer, divided in half between the two windings.
C <---------- ><-Bus3<----------------- X B u s S t
TRANS FORMER DELTAB DELTBC
C <-Busl<-Bus2
1GEN3B GEN3C
2BUSIB
TRANSFORMER DELTAB DELTCA
1GEN3C GEN3A
2BUS1C
BLANK card terminates circuit data
C .................................... Switch_data...................................
C BUS— >BUS— >< Tclose< Topen<------- Ie 0
BUS1A BKR1A -1. 580.E-3 0 0
BUS1B BKR1B -1. 580.E-3 0 0
BUS1C BKRlC -1. 580.E-3 0 0
C Fault switch, duration is 330 ms
BUS12A 250.E-3 9999. 0
BUS12B 250.E-3 9999. 0
BUS12C 99999.E-3 9999. 0
C Generator circuit breaker switch
GEN 3GAGEN3A -1. 9999. 9999999.
GEN3GBGEN3B -1. 9999. 9999999.
GEN3GCGEN3C -1. 9999. 9999999.
BLANK card terminates switch data
C Source Data..............................
C Bus— ><I<Amplitude<Frequency<— TO |Phio<---0=PhiO< Ignore< Tstart< Tstop
14THEVA 187.79E3 50. 0. 0. -1. 9999
14THEVB 187.79E3 50. -120. 0. -1. 9999
14THEVC 187.79E3 50. 120. 0. -1. 9999
C Dynamic Synchronous Machine
C Terminal connection for phase "a"
C Bus— > < Volt< Freq< angle
59GEN3GA 11267.65 50. -30.43
GEN3GB
GEN3GC
C Machine Parameter Cards (Optional)
c ---------------------- ><----- FM
PARAMETER FITTING 1.
C Electrical Parameters of Machine
C <-<-<-NP< SMOutP< SMOutQ< RMVA<------ RkV<---AGLINE<------- Sl<------- S2
1 1 2 1. 1. 200. 13.8 935.016 1000. 1440.
C Col: (1-2) NuMas, (3-4) KMac, (5-6) Kexc
C Note: AGLine is used to get the real magnitude in AMP of the field current
C in principle any value can be used here.

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 469

C ------- ><------ AD1<------ AD2<------ AQ1<------ AQ2<----- AGLQ<------ S1Q<------ S2Q
BLANK
C If S.M. is not saturable (AGLine >= 0), leave Si -> S2Q Blank
C Manufacturer supplied pu data (if PARAMETER FITTING used)
C When transient data not available for Q-axis, make Xq' = Xq, TqO'=0
C This will eliminate the 'G' winding form the SM model.
C ------ Ra<------- Xl<------- Xd<------- Xq<------ X 'd<------ X 'q<------ X"d<------X"q
0.001096 0.15 1.70 1.64 .238324 1.64 .184690 .185151
C T 'dO< T 1qO< T"dO< T"qO<------- XO<------- Rn<------- Xn<------- Xc
6.194876 0. 0.028716 0.074960 1.40
C Mechanical parameters for the shaft system (mass card) .
c <------ ><---- ExTrs<----- HICO<------ DSR<------ DSM<------ HSP<------DSD
1 1. .181128
C col: (1-2) ML
BLANK card terminates mass data
C output requests
C GA<— ><---Nl< N2<---N3<---N4<---N5<---N6<---N7<--- N8<-- N9<--N10<— N 1 K - - N 1 2
10
21
30
C Col: (3) Group, (4) All
BLANK card terminates synchronous machine output requests
C TACS input cards
C Exciter (EMTP assumes that the exciter is regulating to 1 pu = Vf-TACS = 1)
C Bus— ><---- ><KI
71VF
72PM 1
74VEL1 2
FINISH
BLANK card terminates source data
C ............................ Output Request Data..................................
C Bus— >Bus— >Bus— >Bus— >Bus— >Bus— >Bus— >Bus— >Bus— >Bus— >Bus— >Bus— >Bus— >
BLANK card terminates output requests
$CLOSE, UNIT=4 STATUS=KEEP
BLANK card terminates plot requests
BLANK card terminates EMTP solution-mode

F1.2 Example of BCTRAN Calculation file for 1.5 MVA Delta Star Transformer

BEGIN NEW DATA CASE

C ATP file tr_l_5dy.dat tr-transformer, l_5m - 1.5MVA, dy- delta star
C ---------------------------------------------------------------------------
C transformer model used for CEPSI Paper.
C 11 kV/415 V 1.5 MVA transformer
C GEN1T? high voltage delta winding
C GENl? low voltage star winding
C GEN1N star point - connect earthing resistor here if required
C 25/4/96
C FILE TO DERIVE A LINEAR REPRESENTATION FOR A TRANSFORMER
C --------------------------------------------------------------------------------------------------------------------------------------------------
C 1 2 3 4 5 6 7 8

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 470

C 3456789-123456789-123456789-123456789-123456789-123456789-123456789-123456789-
ACCESS MODULE BCTRAN { request for BCTRAN overlay
$ERASE {Emtpy punch buffer, in case data already there from previous simu
C — EXCITATION TEST DATA
C cols 1-2 no of windings/core leg|| NP - 0 true 3 phase trans, NP-1 singlephase
C let Exlos(kw) = 0.15 % -> 0.15/100 * 1.5 MVA = 2.25 kW (positive sequence)
C let Exlos(kw) = 0.30 % -> 0.30/100 * 1.5 MVA = 4.50 kW (zero sequence)
C <------ POSITIVE SEQUENCE TEST PARAMETERS<— ZERO SEQ TST, D-WINDINGOPEN <-IT— IP
C < Freq<— Iexl (%) <— SN (MVA) <ExLos (kW) <— IexO (%) <— SN (MVA) (ExLos (kW) NP— IW—
2 50.0 1.3 1.500 2.25 100. 1.500 4.5 0 2 2-1
C Rated Line Voltaeges & Node names
C first card - HV winding || second card M V or LV winding... (cols 1-2 wind no)
C ]<-Vrat (kV)<— WindRes <-IN A<-OUTA<-IN B<-OUTB<-IN CC-OUTC
1 11.0 GEN1TAGEN1TBGEN1TBGEN1TCGEN1TCGEN1TA
2 0.2396 GEN1A GEN1N GEN1B GENIN GENIC GENlN
C — SHORT CIRCUIT TEST
C Let short ciruit loss = 2% -> 2/100 * 1.5 MVA = 30 kW
C <-POSITIVE SEQ TEST PARAMETERS<------ZERO SEQ TEST IL
C <J<— LOS (kW) <— Zpos (%) <Spos (MVA) <— Zzer (%) <Szer (MVA) ID—
1 2 30.0 5.5 1.500 4.5 1.500 0 1
BLANK card terminates input of short circuit test data
$PUNCH
BLANK card ends BCTRAN case
BEGIN NEW DATA CASE
BLANK card to end EMTP solution mode

BCTRAN OUTPUT FILE - tr_l_5dy.pch

$VINTAGE, 1,
1GEN1A GENlN 76.628252798683
2GEN1B GENlN 0.0
76.628252798683
3GEN1C GENlN 0.0
0.0
76.628252798683
USE AR
1GEN1TAGEN1TB 27.626520962479 2.42
2GEN 1A GENlN -1268.329426491 0.0
59164.519802098 0011481632
3GEN1TBGEN1TC 2.2886655921248 0.0
-105.0722934615 0.0
27.626520962479 2.42
4GEN1B GENlN -105.0722934615 0.0
5724.2256543802 0.0
-1268.329426491 0.0
59164.519802098 0011481632
5GEN1TCGEN1TA 2.2886655921248 0.0
-105.0722934615 0.0
2.2886655921248 0.0
-105.0722934615 0.0
27.626520962479 2.42
6GEN1C GENlN -105.0722934615 0.0
5724.2256543802 0.0

M.J.Checksfi el d/PhD/University of Bath 1996 Appendix F, Page 471

 -105.0722934615 0.0
5724.2256543802 0.0
-1268.329426491 0.0
59164.519802098 .0011481632
$VINTAGE, 0,
$UNITS, -l.,-l.
USE RL
C ------------------- « case separator > » -----------
$VINTAGE, 1,
1GEN1A GENlN 76.628252798683
2GEN1B GENlN 0.0
76.628252798683
3GEN1C GENlN 0 .0
0.0
76.628252798683
USE RL
$UNITS, 0.50E+02 , 0.
1GEN1TAGEN1TB 2.42 12585.505242951
2GEN1A GENlN 0.0 273.88431155028
.0011481632 5.9656982770407
3GEN1TBGEN1TC 0.0 -6166.886509987
0.0 -134.3068074521
2.42 12585.505242951
4GEN1B GENlN 0.0 -134.3068074521
0.0 -2.925446460501
0.0 273.88431155028
.0011481632 5.9656982770407
5GEN1TCGEN1TA 0.0 -6166.886509987
0.0 -134.3068074521
0.0 -6166.886509987
o
o

-134.3068074521
2.42 12585.505242951
6GEN1C GENlN 0.0 -134.3068074521
0.0 -2.925446460501
0.0 -134.3068074521
0.0 -2.925446460501
0.0 273.88431155028
.0011481632 5.9656982770407
$VINTAGE, 0,
$UNITS, -l.,-l.
USE RL
C ------------------- « case separator » >

F1.3 ATP File Simulation Including a Fast Governor for Test fFG2*

BEGIN NEW DATA CASE

$LISTOFF
POWER FREQUENCY 50.0 {special request card
FIX SOURCE {declaraction of EMTP Load Flow
AVERAGE OUTPUT
$COMMENT
C <DELTA< TMAX< XOPT< COPT<-EPSILN<-TOLMAT<-Tstart

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 472

90.91E-6 1.0 50. 50.
C <- IOUT<— IPLOT<- IDOUBL<-KSSOUT<-MAXOUT<--- 1PUN<-MEMSAV<-- ICAT<-NERERG<-1PRSUP
1500 11 0 0 1
TACS HYBRID
C opName s<— ipl s<— ip2 s<— ip3 s<— ip4 s<— ip5 <-gain<-FxLo<-FxHi<-NmLo<-NmHi
C <— No, Do<----- Nl,Dl<--N2, D2<---- N3, D3<---- N4, D4<----N5,D5<----N6,D6<--- N7, D7
C NOTE-because the Tacs variuable Vf is scaled according to the initial value
C the effective gain of the AVR will vary with loading,the gain may be
C three times or more, the amount at full load compared to no load.
C the _1 refers to PI1, _2 on the variable name refers to PI2
C note the limiter function is set to -1 so that Vf can drop to zero, the
c highest it can be is 3 times initial value (i.e 3times fullload possibly ?)
C Anderson appendix D - typically 3 to 4times full load
c
c AVR parameters for GEN1G - 0.750 MVA generator
C opName s<— ipl s<— ip2 s<— ip3 s<— ip4 s<— ip5 <-gain<-FxLo<-FxHi<-NmLo<-NmHi
C <— No, Do<----- Nl, Dl<--N 2 ,D2<---- N3, D3<---- N 4 ,D4<----N5, D5<---- N6, D6<--- N 7 ,D7
C Zero_block
0VR_1 +V2_1 1. -1.0 2.0
0VF_1 +DVF_1 +UNITY 1.
C S_block
1V2_1 -Vl_l +UNITY -V3_l 700.
1.0
1.0 .10
1V1_1 +VDC_1 1.
1.0
1.0 .03
1DVF_1 +VR_1 1.
1.0
1.0 .50
1V3_1 +DVF_1 1.
0.0 .180
1.0 .50
C Speed Governor System for PIl - Diesel governor first order approximation
C opName s<— ipl s<— ip2 s<— ip3 s<— ip4 s<— ip5 <-gain<-FxLo<-FxHi<-NmLo<-NmHi
C <— No, Do< Nl, Dl< N 2 ,D2<----N3, D3< N 4 ,D4<---- N5, D5<---- N6, D6< N 7 ,D7
92VEL_1
C CONVERSION OMEGA TO PER UNIT SYSTEM ( 4 pole 50 Hz -> w = pi*50)
99W_1 = .0031830988 * VEL_1
C Speed Governor System, value of 25 gives 4 % droop, 12 gives 8 %
0DIF_W1 -UNITY +W_1 1.0
0KDIFW1 +DIF_W1 25.
0PS1_1 +UNITY -KDIFW1 1.0
0PM_1 +PMD_1 +PMDEL1 1.0 0.0 1.10
1PS2_1 +PS1_1 1.0
1.,0
1.,0 0..04
C governor main lag ranges from .04 to 0.4 (ATP wookbook 3, ch8)
c Diesel Engine Firing Delay - 15 ms constant assumed for modelling purposes
C <-nameCDs<— INI s<— IN2 s<--IN3 s<— IN4 s<— IN5 < A< B< C< D< E
98PMD_1 53+PS2_l 0.015
c this block is necesary to fill
C <op-nmXX<-------- A<-------- B<--------- CXXXXXXXXXXXXXXXXXXXX<— T_start<-- T_stop
11PMDEL1 1.0 0. 0.015

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 473

C <-NAME <-------- A < -------- B<-------- C <— T-START<-- T-STOP
90GEN1GA 50.
90GEN1GB 50.
90GEN1GC 50.
C emtp switch current defined as flowing left node to rgt
91GEN1A -1.0
91GEN1B -1.0
91GEN1C -1.0
C tacs algebraic & logical fortran expressions for calculating P and Q etc
C OPNAME =<-----------------------------------------FREE FORMAT FORTRAN EXPRESSION
88REAL = ((GEN1GA*GEN1A) + (GEN1GB*GEN1B) + (GENlGC*GEN1C)) / 750.E3
88REACTA = (GENlGA* (GENIC-GEN1B) )
88REACTB = (GEN1GB* (GEN1A-GEN1C) )
88REACTC = (GENlGC*(GEN1B-GEN1A) )
88REAC = ((REACTA + REACTB + REACTC)*0.57735027) / 750.E3
C per unit impedance calculation
88IRMS = S Q R T ( (GEN1A*GEN1A) + (GEN1B*GEN1B) + (GEN1C*GEN1C) )
88VDC_1 = S Q R T ( (GEN1GA*GEN1GA) + (GEN1GB*GEN1GB) + (GENlGC*GEN1GC) ) / 415.0
88VRMS = VDC_1 * 415.
88R = (REAL * VDC_1 * VDC_1) / ( (REAL * REAL) + (REAC * REAC) )
88X = (REAC * VDC_1 * VDC_1) / ( (REAL * REAL) + (REAC * REAC) )
C TACS_output
c <-NAME<-NAME<-NAME<-NAME<-NAME<-NAME<-NAME<-NAME<-NAME<-NAME<-NAME<-NAME<-NAME
33GEN1GAGEN1GBGEN1GCGEN1A GENIB GENIC
C 33REAC REAL PM_1
C R X VRMS IRMS
C <-NAME <-INIT_VAL
C AVR for GEN1 IC's
77VDC_1 1.0
77V1_1 1.0
77V2_1 0.0
77V3_1 0.0
77VR_1 0.0
7 7 DVF_1 0.0
77VF_1 1.0
C Governor for PIl IC's
77VEL_1 314.15926
77W_1 1.0
77PS1_1 1.0
77PS2_1 1.0
77PM_1 1.0
77PMDEL1 1.0
BLANK card terminates TACS data
C .................................. Circuit_Data....................................
C Thevenin Source Impedance - 70 MVA at 11 kV
C Bus— >Bus — ><------------ ><-- R<----------- L<--- R<---------- L< R<---------- L
51GSPA THVA 1.386 4.942
52GSPB THVB 0.104 1.728
53GSPC THVC
C recall Zll = 1/3 (Z0 + 2 Zl) = 0.333 * (4.942 + 2(1.728)) = 2.79 ohms
C Numerical damping resistors for above Rp = 171.87 * X L (11)
THVA GSPA 380.6
THVB GSPB 380.6
THVC GSPC 380.6

M.J.Checksfi el d/PhD/University of Bath 1996 Appendix F, Page 474

C CABLES FROM GRID SUPPLY POINT TO GENERATOR TRANSFORMER
C Bus— >Bus— >Bus— >Bus— ><----- R<---L< C<---- R<--- L< C< R< L<-
C feeder THVA to BUS1 (2 km of 20 kv 150 mm2 cable )
1BUS3A BUS2A .42 .26 220.
2BUS3B BUS2B .308 .1804 -10. .42 .26 220.
3BUS3C BUS2C .308 .1804 -10. .308 .1804 -10. .42 .26 220.
C Numerical damping resistors for above Rp = 171.87 * X L (11)
BUS2A BUS3A 34.68
BUS2B BUS3B 34.68
BUS2C BUS3C 34.68
C feeder THVA to BUS5 (2 km of 20 kv 150 mm2 cable)
1THVA BUS5A .42 .26 220.
2THVB BUS5B .308 .1804 -10. .42 .26 220.
3THVC BU35C .308 .1804 -10. .308 .1804 -10. .42 .26 220.
C Numerical damping resistors for above Rp = 171.87 * X L (11)
BUS5A THVA. 34.68
BUS5B THVB 34.68
BUS5C THVC 34.68
C Bus— >Bus — >Bus — >Bus— ><--- R<---- L<--- C<----- R<---- L<---- C<----R<---- L<- C
C feeder BUS1 to BUS2 (2 km of 20 kv 150 mm2 cable)
1BUS1A BUS2A .42 .26 220.
2BUSIB BUS2B .308 .1804 -10. .42 .26 220.
3BUS1C BUS2C .308 .1804 -10. .308 .1804 -10. .42 .26 220.
C Numerical damping resistors for above Rp = 171.87 * X L (11)
BUS2A BUS1A 34.68
BUS2B BUSIB 34.68
BUS2C BUS1C 34.68
C Bus— >Bus — >Bus— >Bus — ><----R<---- L<---C<----- R<---- L<---- C<----R<---- L<- C
C feeder BUS5 to GEN1T (2 km of 20 kv 150 mm2 cable)
1BUS5A GEN1TA .42 .26 220.
2BUS5B GEN1TB .308 .1804 -10. .42 .26 220.
3BUS5C GEN1TC .308 .1804 -10. .308 .1804 -10. .42 .26 220.
C Numerical damping resistors for above Rp = 171.87 * X L (11)
GEN1TABUS5A 34.68
GEN1TBBUS5B 34.68
GEN1TCBUS5C 34.68
C
C LOCAL LOAD1 - 500 kW, 0.9 pf 6 11kV RMS lineto line
C Busl->Bus2-> <----R<---- L<---C
GEN1TA 196.16 94.8
GEN1TB 196.16 94.8
GEN1TC 196.16 94.8
C GENERATOR TRANSFORMER - 1.5 MVA 5.5% impedance
C .415/11 kV star delta transformer taken from BCTRAN output tr_l_5dy.pch
$VINTAGE, 1,
C Bus— >Bus— >Bus— >Bus— ><---------------R<---------------L
1GEN1A GENlN 76.628252798683
2GEN1B GENlN 0 .0
76.628252798683
3GEN1C GENlN 0. 0
0.0
76.628252798683
USE AR
1GEN1TAGEN1TB 27.626520962479 2.42

M.J. Checksfiel d/PhD /University of Bath 1996 Appendix F, Page 475

 2GEN1A GENlN -1268.329426491 0.0
59164.519802098 .0011481632
3GEN1TBGEN1TC 2.2886655921248 0.0
-105.0722934615 0.0
27.626520962479 2.42
4GEN1B GENlN -105.0722934615 0.0
5724.2256543802 0.0
-1268.329426491 0.0
59164.519802098 .0011481632
5GEN1TCGEN1TA 2.2886655921248 0.0
-105.0722934615 0.0
2.2886655921248 0.0
-105.0722934615 0.0
27.626520962479 2.42
6GEN1C GENlN -105.0722934615 0.0
5724.2256543802 0.0
-105.0722934615 0.0
5724.2256543802 0.0
-1268.329426491 0.0
59164.519802098 .0011481632
$VINTAGE, 0,
$UNITS, -l.;-l.
USE RL
C Generator Transformer Earthing Point - Solidly Earthed
C Busl->Bus2-> < R< L< C
GENlN 0.1
GENlN 0.1
GENlN 0.1
BLANK card terminates circuit data
C .................................... Switch_data........................... .
C Generator Circuit Breaker
C BUS— >BUS--><--- Tclose<---- Topen<-------Ie
GEN1GAGEN1A -1. 9999. 0.
GEN1GBGEN1B -1. 9999. 0.
GEN1GCGEN1C -1. 9999. 0.
C GEN1T - BUS1 Circuit Breaker
C BUS— >BUS— ><--- Tclose<---- Topen<------- Ie
GEN1TABUS1A -1. 440.E-3 0.
GEN1TBBUS1B -1. 440.E-3 0.
GENlTCBUS1C -1. 440.E-3 0.
C GEN1T - BUS1 Circuit Breaker
C BUS— >BUS— ><--- Tclose<---- Topen<-------Ie
THVA BUS3A -1. 440.E-3 0.
THVB BUS3B -1. 440.E-3 0.
THVC BUS3C -1. 440.E-3 0.
C Fault Switch
BUS2A 100.E-3 470.E-3 0.
BUS2B 100.E-3 470.E-3 0.
BUS2C 100.E-3 470.E-3 0.
BLANK card terminates switch data
C ........................................ Source Data.......................
C Bus— ><I<Amplitude<Frequency<— TO |Phio<--- 0=Phi0< Ignore<-- Tstart<— -Tstop
14GSPA 8981.46 50. 0. 0. -1. 9999.
14GSPB 8981.46 50. -120. 0. -1. 9999.

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 476

14GSPC 8981.46 50. 120. 0. -1. 9999.
C Dynamic Synchronous Machine - "generator”
C Terminal connection for phase "a"
C voltage has to be set differently to get right reactive flow
C Bus— > <----- Volt<----- Freq<---- angle
59GEN1GA 338.846 50. -00.0
GEN1GB
GENlGC
c ---------------------- >< FM
PARAMETER FITTING 1.
C Electrical Parameters of Machine
C <-<-<-NP< SMOutP< SMOutQ< RMVA<------RkV<---AGLINE<------- Sl<------- S2
1 1 2 1. 1. 0.750 0.415 -5.0 5.297 8.472
C ------- ><------ AD1<------ AD2<------ AQ1<------ AQ2<----- A G L Q <------ SlQ<---- — S2Q
- 1.
C ------ Ra<------- Xl<------- Xd<------- Xq<------ X 'd < ------ X 'q<------ x " d < ----- X"q
0.0322 0.10 2.63 1.57 0.21 1.57 0.15 0.18
C T 'd0< T 'q0<-----T"d0<----- T"q0<------- XO<------- Rn<------- Xn<------- Xc
3.0 0.0 0.035 0.028 0.024
C Mechanical parameters for the shaft system (mass card) .
C 0.18E-3 — > an H of 0.5 s
C <------ ><---- ExTrs<----- HICO<------ DSR<------ DSM<------ HSP<------DSD
1 1. 0.18E-3
C col: (1-2) ML
BLANK card terminates mass data
C output requests
C GA<— >< Nl<---- N2<-- N3<-N4<-- N5<--- N6<---N7<--- N8<--- N9<— N10<— Nll<— N12
10
21
30
C Col: (3) Group, (4) All
BLANK card terminates synchronous machine output requests
C Bus— >< ><KI
71VF_1 1
7 2 PM_1 1
7 4VEL_1 2
FINISH
BLANK card terminates source data
C EMTP LOAD FLOW 0- P,Q 1 - P,V 2 - Q, THETA
C <-busl<-bus2<-bus3<------- Pk-or-Qk<------- Qk-or-Vk<-- Vmin<---Vmax<THmin<Thmax
0 GEN1GAGEN1GBGEN1GC 600.E3 450.E3 328. 348.
C NNNOUT NITERA NFLOUT NPRINT RALCHK CFITEV CFITEA VSCALE KTAPER
0 2000 20 0 0.00001 0.1 2.0 2
BLANK card terminates output requests
BLANK card terminates plot requests
BLANK card terminates EMTP solution-mode

F1.4 ATP Simulation File of Slow Loss of Synchronism During Loss of Field

C emtp simulation for thesis based on 625 Diesel gen at GEC Stafford
c The saturable transformer model has been used, instead of BCTRAN.
C Generator loss of field from 225 kW, showing effect of transient saliency.

M.J.Checksfi el d/PhD/University of Bath 1996 Appendix F, Page 477

C generator voltage is 415 V, (rating 625 kVA)
C contains full governor model.
C also contains a DY transformer at genertaor terminals
C Local load - 0.5 MW
C Matt Checksfield 10.12.96
BEGIN NEW DATA CASE
POWER FREQUENCY 50.0 {special request card
FIX SOURCE {declaraction of EMTP Load Flow
AVERAGE OUTPUT
$COMMENT
C <DELTA< TMAX< XOPT< COPT<-EPSILN<-TOLMAT<-Tstart
125.0E-6 15.0 50. 50.
C if Iplot every 5 — > 1600 Hz output (32 samps/cyc).
C Iplot every 9 — > 888.8 Hz output (17.7 samps/cyc).
C Iplot every 13 — > 615.4 Hz output (12.3 samp/cyc).
C <-IOUT<— IPLOT<-IDOUBL<-KSSOUT<-MAXOUT<--- IPUN<-MEMSAV< ICAT<-NERERG<-IPRSUP
4500 81 1 1 1
TACS HYBRID
c this block 'simulates' loss of field by ramping Vf to zero
C opName s<— ipl s<— ip2 s<--ip3 s<— ip4 s<— ip5 <-gain<-FxLo<-FxHi<-NmLo<-NmHi
0 VF_1 +LOE_l +LOE_2 1. 0.0 1.0
C <op-nmXX<-------- A< -------- B<---------CXXXXXXXXXXXXXXXXXXXX<— T_start<-- T_stop
11L0E_1 1.0 -0. 0.3
24LOE_2 -1.0 0.1 0.2 0.3
c Speed Governor System for PI1 - Diesel governor first order approximation
92VEL_1
C CONVERSION OMEGA TO PER UNIT SYSTEM ( 4 pole 50 Hz -> w = p i *50)
99W_1 = .0063662 * VEL_1
C Speed Governor System, value of 25 gives 4 % droop, 12 gives 8 %
C droop settings on governors are notoriously variable (DAB)
0DIF_W1 -UNITY +W_1 1.0
0KDIFW1 +DIF_W1 25.
0PS1_1 +UNITY -KDIFW1 1.0
0PM_1 +PMDEL1 +PMD_1 1.0 0.0 1.10
1PS2_1 +PS1_1 1.0
1.0
' 1.0 0.10
c Diesel Engine Firing Delay - 15 ms constant assumed for modelling purposes
C <-nameCDs<— INI s<— IN2 s<— IN3 s<— IN4 s<--IN5 < A< B< C< D< E
98PMD_1 53+PS2_l 0.015
c this block is necesary to fill in gap from delay
C <op-nmXX<-------- A < -------- B<---------CXXXXXXXXXXXXXXXXXXXX<— T_start<-- T_stop
11PMDEL1 1.0 0. 0.015
C Node_Voltage
C C-NAME <-------- A < -------- B<-------- C <— T-START<-- T-STOP
90GEN1GA 50.
90GEN1GB 50.
90GEN1GC 50.
C emtp switch current defined as flowing left node to rgt
91GEN1A -1.0
91GEN1B -1.0
91GEN1C -1.0
C tacs algebraic & logical fortran expressions for calculating P and Q etc
C OPNAME =<----------------------------------------- FREE FORMAT FORTRAN EXPRESSION

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 478

88REAL = ((GEN1GA*GEN1A) + (GEN1GB*GEN1B) + (GEN1GC*GEN1C))
88REACTA = (GENlGA*(GEN1C-GEN1B))
88REACTB = (GENlGB*(GEN1A-GEN1C))
88REACTC = (GENlGC*(GEN1B-GEN1A))
88REAC = ((REACTA + REACTB + REACTC)*0.57735027)
88VRMSGA66+GEN1GA 50.
88IRMSA 66+GEN1A 50.
C control system input_fcns - normalise voltage to 1 pu using Vline to line RMS
88VDC_1 =SQRT( (GEN1GA*GEN1GA) + (GEN1GB*GEN1GB) + (GENlGC*GEN1GC) ) / 415.0
88VRMS = VDC_1 * 415
C per unit impedance calculation
88R = (REAL * VDC_1 * VDC_1) / ( (REAL * REAL) + (REAC * REAC) )
88X = (REAC * VDC_1 * VDC_1) / ((REAL * REAL) + (REAC * REAC))
C TACS_output
c <-NAME<-NAME<-NAME<-NAME<-NAME<-NAME<-NAME<-NAME<-NAME<-NAME<-NAME<-NAME<-NAME
33REAC REAL VRMS IRMSA
C TACS_I.C .
C <-NAME <-INIT_VAL
C AVR for GEN1 IC's
77VF_1 1.0
C Governor for PI1 IC's
77VEL_1 157.07963
77W_1 1.0
7 7 PS1_1 1.0
77PS2_1 1.0
77PM_1 1.0
BLANK card terminates TACS data
C ................................. Circuit_Data....................................
C Thevenin Source Impedance
C Bus— >Bus — ><---------- ><--- R<----------- L<---- R<---------- L<----- R<---------- L
51GSPA THVA 1.386 4.942
52GSPB THVB 0.104 1.728
53GSPC THVC
C recall Zll = 1/3 (Z0 + 2 Zl) = 0.333 * (4.942 + 2(1.728)) = 2.79 ohms
C Numerical damping resistors for above Rp = 171.87 * X L (11)
THVA GSPA 380.6
THVB GSPB 380.6
THVC GSPC 380.6
C CABLES FROM GRID SUPPLY POINT TO GENERATOR TRANSFORMER
C BUS — >Bus — >Bus— >Bus — ><--- R<----L<----- C<-R < L< C<-R < L< C
C feeder THVA to BUSl (2 km of 150 mm2cable)
1BUS3A BUS2A .42 .26 220.
2BUS3B BUS2B .308 .1804 -10. .42 .26 220.
3BUS3C BUS2C .308 .1804 -10. .308 .1804 -10. .42 .26 220.
C Numerical damping resistors for above Rp = 171.87 * X L (11)
BUS2A BUS3A 34.68
BUS2B BUS3B 34 .68
BUS2C BUS3C 34.68
C feeder THVA to BUS5 (2 km of 150 mm2 cable)
1THVA BUS5A .42 .26 220.
2THVB BUS5B .308 .1804 -10. .42 .26 220.
3THVC BUS5C .308 .1804 -10. .308 .1804 -10. .42 .26 220.
C Numerical damping resistors for above Rp = 171.87 * X L (11)
BUS5A THVA 34.68

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 479

 BUS5B THVB 34.68
BUS5C THVC 34.68
C Bus— >Bus— >Bus— >Bus — ><---- R<---- L< C<---- R<---- L<---- C<---- R<---- L<----C
C feeder BUS1 to BUS2 (2 km of 150 mm2 cable)
1BUS1A BUS2A .42 .26 220.
2BUSIB BUS2B .308 .1804 -10. .42 .26 220.
3BUS1C BUS2C .308 .1804 -10. .308 .1804 -10. .42 .26 220.
C Numerical damping resistors for above Rp = 171.87 * X L (11)
BUS2A BUS1A 34.68
BUS2B BUSIB 34.68
BUS2C BUS1C 34.68
C BUS— >Bus — >Bus— >Bus— ><---- R<---- L< C<---- R<---- L<----C<---- R<---- L<----C
C feeder BUS5 to GEN1T (2 km of 150 mm2 cable)
1BUS5A GENlTA .42 .26 220.
2BUS5B GEN1TB .308 .1804 -10. .42 .26 220.
3BUS5C GEN1TC .308 .1804 -10. .308 .1804 -10. .42 .26 220.
C Numerical damping resistors for above Rp = 171.87 * X L (11)
GENlTABUS 5A 34.68
GEN1TBBUS5B 34.68
GEN1TCBUS5C 34.68
C LOCAL LOADl - 500 kW, 0.9 pf @ 11 kV RMS line to line
C Busl->Bus2-> <---- R<---- L<--- C
GENlTA 196.16 94.8
GEN1TB 196.16 94.8
GEN1TC 196.16 94.8
C Saturable_transformer_components.
C ----------- ><-Bus3<----><---- 1<— Phi<BusSt<-Rmag<----------------------------- >0
TRANSFORMER DELTAB 0
C ------- current<------------ flux
9999
C <-Busl<-Bus2<-----------><---Rk<---Lk<-- Nk<------------------------------------>0
1GEN1A GEN1B .7E-3 .00315 0.415 0
2GEN1TA .198 2.218 6.3508
C Note.- These Leakage values were calculated assuming a 5.5 % p.u reactance at
C 1.5 MVA for the transformer, divided in half between the two windings.
C <--- ;------ ><-Bus3<----------------- X B u s S t
TRANSFORMER DELTAB DELTBC
C <-Busl<-Bus2
1GEN1B GENIC
2GEN1TB
TRANSFORMER DELTAB DELTCA
1GEN1C GEN1A
2GEN1TC
BLANK card terminates circuit data
C .................................... Switch_data...................................
C Generator Circuit Breaker
C BUS— >BUS— >< Tclose< Topen<-------- Ie O
GENIGAGENLA -1. 9999. 0. 0
GEN1GBGEN1B -1. 9999. 0. 0
GEN1GCGEN1C -1. 9999. 0. 0
C GEN1T - BUS1 Circuit Breaker - fault clearing
C BUS— >BUS— >< Tclose< Topen<-------- Ie 0
GENlTABUS 1A -1. 9999. 0. 0
GEN1TBBUS1B -1. 9999. 0. 0

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 480

 GEN1TCBUS1C -1. 9999. 0.
C GEN1T - BUS1 Circuit Breaker - fault clearing
C BUS— >BUS— >< Tclose< Topen<- ■Ie
THVA BUS 3A - 1. 9999. 0.
THVB BUS3B - 1. 9999. 0.
THVC BUS3C -1. 9999. 0.
C Fault Switch
BUS 2A 100 . E-0 -440.E-3 0.
BUS2B 100 . E-0 -440.E-3 0.
BUS2C 1 0 0 . E-0 -440.E-3 0.
BLANK card terminates switch data
C ......................................... Source Da t a ...............................
C Bus— ><I<Amplitude<Frequency<— TO |Phio<---0=PhiO< Ignore< Tstart< Tstop
14GSPA 8981.46 50, 0. 0. - 1. 9999.
14GSPB 8981.46 50, - 120. 0. - 1. 9999.
14GSPC 8981.46 50, 120. 0. - 1. 9999.
C Dynamic Synchronous Machine - "generator"
C Terminal connection for phase "a"
C Bus— > < Volt< Freq< angle
59GEN1GA 338.846 50. -00.0
GEN1GB
GEN1GC
C TOL < EPSUBA< EPOMEG< EPDGEL---------- <---NIOMAX
C TOLERANCES l.E-16 20
C **EPSUBA is the built in damping value see p 8-5 of ref manual
C **EPOMEG is the tolerance for rotor speed iteration, if blank then the
C **default used. Only minor benefits gained from tight convergence Criteria.
C Machine Parameter Cards (Optional)
c ---------------------- ><----- FM
PARAMETER FITTING 1.
C Electrical Parameters of Machine
C <-<-<-NP< SMOutP< SMOutQ< RMVA<------ RkV<---AGLINE<------- Sl<------- S2
1 1 4 1. 1. 0.625 0.415 -5.0 5.297 8.472
C Col: (1-2) NuMas, (3-4) KMac, (5-6) Kexc
C Note: AGLine is used to get the real magnitude in AMP of the field current
C in principle any value can be used here.
C ------- ><------ AD1<------ AD2<------ AQ1<------ AQ2<----- AGLQ<------ SlQ<------ S2Q
- 1.
C If S.M. is not saturable (AGLine >= 0), leave Si -> S2Q Blank
C Manufacturer supplied pu data (if PARAMETER FITTING used)
C When transient data not available for Q-axis, make Xq' = Xq,TqO'=0
C This will eliminate the G winding form the SM model.
C X q * ' made up = 1.5 * X d ' ' , also T q o 1' = 0.75 * T d o ' ' (kimbark III)
C ------ Ra<------- Xl<------- Xd<------- Xq<------ X 1d<------ X 'q<------ X"d<------X"q
0.03 0.10 2.25 1.10 0.22 1.10 0.15 0.18
C T 'dO< T 'qO< T"dO< T"qO<------- XO<------- Rn<------- Xn<------- Xc
4.5 0.0 0.035 0.028 0.051
C Mechanical parameters for the shaft system (mass card) .
C 0.9E-3 — > an H of 0.7 kWs/kVA ish ???
C <------ ><----ExTrs<----- HICCK------ DSR<------ DSM<------ HSP<------DSD
1 1. 0.9E-3
C col: (1-2) ML
BLANK card terminates mass data
C output requests

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 481

C GA<— ><--- Nl<--N2<--- N3<-N4<---- N5<--- N6<---N7<---N8<--- N9<-— N10<— Nll<— N12
10
21
30
C Col: (3) Group, (4) All
BLANK card terminates synchronous machine output requests
C TACS input cards
C Exciter (EMTP assumes that the exciter is regulating to 1 pu = Vf-TACS = 1)
C Bus— >< ><KI
71VF_1 1
7 2 PM_1 1
7 4VEL_1 2
FINISH
BLANK card terminates source data
C EMTP LOAD FLOW 0 - P,Q 1 - P,V 2 - Q, THETA
C <-busl<-bus2<-bus3<------- Pk-or-Qk<------- Qk-or-Vk<-- Vmin<---Vmax<THmin<Thmax
0 GEN1GAGEN1GBGEN1GC 225.E3 -150.E3 328. 348.
C NNNOUT NITERA NFLOUT NPRINT RALCHK CFITEV CFITEA VSCALE KTAPER
0 10000 20 0 0.00001 0.1 2.0 2
C ............................ Output Request Data..................................
C Bus— >Bus — >Bus— >Bus— >Bus— >Bus— >Bus— >Bus— >Bus — >Bus— >Bus— >Bus— >Bus— >
c GSPA GEN 1A BUS1A BUS 2A
BLANK card terminates output requests
BLANK card terminates plot requests
BLANK card terminates EMTP solution-mode

F2.1C Code for PC Based Simulation o f Single Blinder Scheme

int sn g B ld r (v o id )

{
/* trip signal issued if im p c au ses 'left* elem ent to drop out */

/* trip o u tp u t file is 2 for m otor pole slip, 1 for gen pole slip */

extern in t sng_bldr_p, s a m p c y c , per_unit;

extern R eal rating, vbase, sng bldr x offset, sng_bldr_r_offset;

extern R eal sng_bldr_diam , sng_bldr_l_grad, sng_bldr_r_grad;

extern R eal sng_bldr_l_intcpt, sng_bldr_r_intcpt, sng_bldr_del;

R eal rad sqrd, m od_term , r o f f s e t , x_offset, pu_base;

R eal left_val, right_val;

int l_bldr_pick, r_bldr_pick, trip = 0;

static int f l a g a = 0, f l a g b = 0, old l pick = 0, o ld_r_pick = 0;

static R eal t l = 0, t2 = 0;

/*** is function enabled o r n o t ? * * •* /

if (sng b ld r p == 0 ) return -1;

p u b a s e = (vbase* vbase)/ra ting;

/*** scale p e r u n it values accordingly ***/

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 482

 if (per_unit == 0 ) {

x_oflfset = sn g _ b ld r_ x _ o ffset * pu_base;

r offset = sng_ b ld r_ r_ o ffset * p u b a s e ;

sng_bldr_diam = sn g_bldr_diam * pu_base;

s n g b l d r l i n t c p t = sn g _ b ld r_ l_ in tc p t * pu_base;

sn g_bldr_r_intcpt = sn g _ b ld r_ r_ in tc p t * pu base;

}
else {

x _ o ffset = sng_bldr_x_offset;

r_ofifset = s n g b l d r r o f f s e t ;

left_val = ( (valu e(X )/sn g _ b ld r_ l_ g rad ) + sng_bldr_l_intcpt);

right_val = ( (value(X )/sn g _ b ld r_ r_ g rad ) + sng_bldr_r_intcpt);

/* * * •* c he c k w hich blin d ers have picked up * * * * /

i f ( v alue(R ) > left_val) l_b ld r_ p ick = 1;

else l_bldr_pick = 0;

i f ( valu e(R ) < r i g h tv a l ) r_b ld r_ p ick = 1;

else r_bldr_pick = 0;

/ * * * * c h e c k i f m ho has pick ed u p * * •* /

rad_sqrd = (sng_bldr_diam /2)*(sng_bldr_diam /2);

m od_term = ( (v a lue(X )-x_offset) * (v alu e(X )-x _ o flset) ) +

( (value(R )-r_ofFset) * (value(R )*r_offset) );

i f ( m od_term < rad_sqrd ) {

/**** gen pole slip, c ro ss first b lin d e r (right) * * * * * * /

if (old _ r_ p ick == 0 & & r_bldr_pick == 1){

t l = value(T IM E );

f l a g a = 1;

f l a g b = 0;

/ * * * * * m o to r pole slip, cross first blin d er (left) * * * * •/

if (old l p ick == 0 & & l b ld r p ick == 1){

t l = value(T IM E );

f l a g a = 0;

f l a g b = 1;

/ * * * * * * c h e c k i f g e n era to r pole slip has happ p en ed * *****/

if ( old l p ick == 1 & & l bldr p ick == 0 & & flag a == 1 & &

flag b == 0 & & r b ld r p ick == 1){

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 483

 t2 = value(TIM E);

i f ( (t2 - t l ) >= sng_bldr_del){

trip = 1;

f l a g a = f l a g b = 0;

t l = t2 = 0;

}
else {

f l a g a = f l a g b = 0;

t l = t2 = 0;

/* * * •* chegk jf motor pole slip has happened ******/

if ( o l d r p i c k = = 1 && r bldr pick = = 0 && flag a = = 0 & &

flag_b 1 & & l_bldr_pick == 1){

t2 = value(TIME);

i f ( (t2 - t l ) > = sng_bldr_del){

trip = 2;

f l a g a = f l a g b = 0;

tl = t2 = 0;

}
e lse {

f l a g a = f l a g b = 0;

t l = t2 = 0;

}
}
else {

trip = 0;

f l a g a = f l a g b = 0;

t l = t2 = 0;

/* u p d a te stored values o f b lin d er o utputs */

old _ r_ p ick = r_bldr_pick;

old _ l_ p iek = l_bldr_pick;

set (S N G _T , trip);

i f (trip != 0 ) return 1;

else return 0;

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 484

F2.2 C Code for PC Based Simulation of Power Based Pole Slipping Algorithm

/ • --------------P O L E SL IPPIN G FU N C T IO N P R O T O T Y P E S --------------------------- */

P R IV A T E void init_pole_slip_lookup_tab (void);

P R IV A T E void init_pole_sIip_protection (void);

P R IV A T E void pole_slip_pt_calc (void);

/* ------------------------------------------------------------------------------------------------

Pole S lipping Protection D ata Structure

*/
ty p e d e f struct

{
/* Settings */

in t o n_off_sw itch; /* Flag indicates w h e th er the * /

/* function is en ab led o r disabled. */

/* 1 = enabled, 0 = disabled */

R eal Q trip; /** reactive trip level **/

R eal Ptol; /* real pow er tolerance band in calc o f Pt **/

R eal Pfact, /* in p u t po w er estim ation factor in P t calc **/

R eal D fact, /* m axim um size o f D m in **/

R eal D tran; /* transient saliency */

R eal slip; /* u sed in Dm in calc **/

int lag; /* tim e setting used in Pt calc */

/* P rivate */

R eal p i ; /* used in P t calc - previous value o f p o w e r * /

R eal p2; /* used in P t calc - previous value o f p o w e r */

int trip_count; /* * trip counter, trip a t 6 */

int cnt_pm ax_off; /* c o u n ter u se d to reset P m ax to nom inal */

in t cnt_pm ax; /* co u n ter used in pm ax calc */

R eal lookup[8]; /* lookup index used in D m in calc = ratio 10*pt/P m ax */

int co u n t, /* used in P t calc function */

int trip; /* 1 = true */

/* P u b lic M easu rem en ts * /

R eal pt, /* real p o w er trip level */

R eal pm ax; /* m axim um pow er o u tp u t in last second */

R eal pow erP ; /* m oving av erage value o f p o w e r */

R eal pow erQ ; /* m oving av erage value o f p o w er * /

R eal D m in; /* rate o f change o f p o w e r trip level */

R eal deriv; /* rate o f change o f p o w e r */

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 485

} pole_slip_struct;

pole_ slip _ stru ct p o l e s l i p p r o t ;

static float c o n st m aster_look_up[8] = {

/* Idx A n g le C o s(A ngle)*1000 */

/* 0 180.000 */ -1.00000,

/* 1 174.268 */ -0.99498,

/* 2 168.522 */ -0.97980,

/* 3 162.554 */ -0.95394,

/* 4 156.491 */ -0.91652,

/* 5 149.997 */ -0.86603,

/* 6 143.130 */ -0.80000,

/* 7 135.561 */ -0.71414

void in it_ pole_slip_lookup_tab(void)

int i;

/* * initialsie lookup table b a se d on the slip value **/

for(i=0; i < 8; i++ )

pole_slip_prot.lookup[i] = pole_slip_prot.slip * m aster_look_up[i];

void in it_ p ole_slip_protection(void)

extern R eal adm it_trip, lg_ptol, lg_p_fact, d p f a c t , dp_tran;

extern R eal rating, slip, lg_lag;

extern in t lgpg_p, sam p_cyc, p e r u n it ;

if((sam p_cyc != 4 ) & & ( s a m p c y c != 8) & & ( s a m p c y c != 12) &&

(sa m p _ cy c != 16) & & (sam p cyc != 2 0 ) & & (sam p_cyc != 2 4 ) )

{
fatal("sam ples p e r cycle m ust b e a m ultiple o f 4 for p o w er b ased pole slip fu n c tio n ”);

/** m odify trip settings from % rating actual values **/

/* * * * * * * * * * * * * * * C H A N G E V A L U E S A C C O R D IN G T O P E R U N IT SE T T IN G * * * * * * * * * * * * * /

if (p e r u n it == 0 )

{
p o le_slip_prot.Q trip = (-l/a d m it_ trip ) * rating;

p ole_slip_prot.P tol = lg_ptol * rating;

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 486

 p oIe_slip_prot.D fact = dp_ fact * rating;

pole_slip_prot.D tran = d p t r a n * rating;

po le_ slip _ p ro t.p t = 0.01 * rating; /* low so algorithm m o st secure */

p o le_slip_prot.pm ax = 2 * rating;

}
else

{
p o le_slip_prot.Q trip = (-l/ad m it_ trip );

p o le_slip_prot.P tol = lg_ptol;

pole_ slip _ p ro t.D fact = d p f a c t ;

pole_slip_prot.D tran = d p t r a n ;

pole_ slip _ p ro t.p t = 0.01; /* low so algorithm m ost secure */

p ole_ slip _ p ro t.p m ax = 2;

p o le_ slip _ p ro t.P fa c t = l g p f a c t ;

pole_ slip _ p ro t.slip = slip * 3.141593; /* scale slip up into p roper units * /

/* i.e units o f t radians 18/10/95 */

pole_ slip _ p ro t.lag = lg_lag * 50;

/** in itialise all o th er varaibles **/

po le_slip_prot.on_off_sw itch = lgpg_p;

p o le _ slip _ p ro t.p l = rating;

pole_ slip _ p ro t.p 2 = 0; /* need p i and p2 different so P t does not */

/* incorrectly u pdate */

po le_ slip _ p ro t.trip _ co u n t = 0;

p o le_ slip _ p ro t.c n t_ p m a x _ o ff = 0;

pole_ slip _ p ro t.cn t_ p m ax = 0;

pole_ slip _ p ro t.p o w erP = rating;

p o l e s li p p r o t .p o w e r Q = 0;

p ole_slip_prot.D m in = pole_slip_prot.D fact;

pole_ slip _ p ro t.d eriv = 0; /* will be initialised from

** pow erP and last p val */

p o le_ slip _ p ro t.c o u n t = 0;

po le_ slip _ p ro t.trip = 0;

in t lg p g (v o id )

{ /* function n e e d s to be processed 4 tim es p e r cycle */

ex tern in t sam p cyc, lgpg_p, p e r unit;

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 487

 e xtern R eal rating;

R e al pt_xlO ;

in t pt_pm ax;

R eal lookup_val;

static int i = 0, k = 0;

i f (lg p g _ p == 0 ) return -1;

/*** initialise variables * * * /

if (i == 0 ) {

init_pole_slip_protection();

init_pole_slip_lookup_tabO ;

if(i == sam p _ cy c/4 ) /* so th a t function is processed @ 4 tim es/cyc */

{
i = 0; /* reset c o u n ter */

pole_ slip _ p ro t.p o w erP = value(P); /* h a lf cycle m oving average value o f p o w e r */

pole_ slip _ p ro t.p o w erQ = value(Q ); /* h a lf cycle m oving average value o f p o w e r */

/** call Pt calculation function every 20 m s ( 1 cycle )**/

i f (k == 4 )

{
pole_slip_pt_calc();

k = 0;

}
k+ +;

/* C alculate a daptive P m ax every 5 m s (q u arter c y cle) */

i f ((p ole_slip_prot.pow erP ) > pole_slip_prot.pm ax)

{ /* need 1 c y cles w orth o f conditions before pm ax re-calc'ed * /

p ole_slip _prot.cnt_pm ax ++;

i f (p ole_slip_prot.cnt_pm ax > =4)

{
p o le_slip_prot.pm ax = pole_slip_prot.pow erP ;

p ole_slip_prot.cnt_pm ax = 1;

pole_ slip _ p ro t.cn t_ p m ax _ o ff = 1;

}
}
e lse p o le_slip_prot.cnt_pm ax = 0; /* reset c o u n ter */

i f (p o le_ slip _ p ro t.c n t_ p m a x _ o ff != 0)

{
p o le_ slip _ p ro t.c n t_ p m a x _ o ff ++;

i f (po !e_ slip _ p ro t.cn t_ p m ax _ o ff >= 2 0 0 )

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 488

 {
p o l e s l i p p r o t . c n t p m a x o f f = 0;

p ole_slip _ p ro t.p m ax = 1.4 * pole_slip_prot.pt,

}
}
else pole_slip _ p ro t.p m ax = 1.4 * pole_slip_prot.pt;

/* C alculate d p /d t (d t = l/(4 x F sy s) = 5m s for F sys= 50H z) */

pt_xlO = pole_ slip _ p ro t.p t * 10; /* pt*10 to g et lookup index

** (betw een 0 & 7) */

/* pm ax bo m b p ro o fin g i f pm ax = 0 then p t p m a x = 0 **/

/* (0 is m o st secu re trip level) */

if(pole_slip_prot.pm ax — 0)

{ ■

p t p m a x = 0;

printf(”w arn in g function lgpg, P m ax = 0\n");

}
else

{
p t_ p m ax = (intX pt_xlO / pole_slip_prot.pm ax);

/* pt/P m ax bom b p ro o fin g incase pt_ p m ax > 7 o r < 0 */

if((pt_pm ax > 7) || (p t pm ax < 0))

{
p t p m a x = 0;

p rintf("w am ing function lgpg, Pt/Pm ax = % d\n",pt_pm ax);

}
}
lo o k u p v a l = pole_slip_prot.Iookup[pt_pm ax];

if(pt_pm ax < 6 ) /* i f in tran sien t state add tran sie n t saliency * /

pole slip prot.D m in = (p o le_slip_prot.pm ax * lookup_val) +

pole_slip_prot.D tran;

else

pole_slip_prot.D m in = (pole_slip_prot.pm ax * lookup_val);

if(pole_slip_prot.D m in > pole_slip_prot.D fact)

p o le_slip_prot.D m in = pole_slip_prot.D fact,

p o le_slip_prot.deriv = value(D P );

/* T rip D eterm ination h ysteresis has been Im plem ented */

if (pole_slip _ p ro t.o n _ o ff_ sw itch = = 1 )

M.J.Checksfi el d/PhD /University of Bath 1996 Appendix F, Page 489

 if (pole_slip_prot.pow erQ < pole_slip_prot.Q trip)

{ /* if starter satisfied... */

i f ((pole_slip_prot.deriv < pole_sIip_prot.D m in) & &

(pole_slip_prot.pow erP < p ole_slip_prot.pt ))

{ /* if all pole slipping criteria m et... * /

pole_slip_prot.trip_count++ ; /* inc trip c o u n ter */

}
else if (pole_slip_prot.trip != 1)

{ /* else if started, b u t n o t tripped; reset */

pole_slip_prot.trip_count = 0; /* reset c o u n te r */

}
else /* Im plicit code: rem ain tripped w hile Q < Q trip */

{ /* else if started A N D tripped; stay tripped */

pole_slip_prot.trip_count = 6; /* hold c o u n te r a t 6 */

}
}
else

pole_ slip_prot.trip_count = 0; /* reset c o u n te r */

if (pole_slip_prot.trip_count >= 6)

{
pole_slip_prot.trip_count = 6; /* hold c o u n ter a t 6 */

pole_slip_prot.trip = 1;

/** fonction details |= PO L E _SL IP; • * / /* SET O U T P U T */

}
else

{
pole_slip_prot.trip = 0;

/** fonction d etails & = ~P O L E _S L IP ; **/ /* R eset O utput */

}
}
else

{
pole_slip_prot.trip = 0;

p o le_slip _prot.trip_count = 0;

/** function_details & = -P O L E SLIP; * * / /* R eset O utput */

}
}
i++; /** variable i also used to g et function w orking every 5 m s **/

/** w rite to o u tp u t file **/

set (L G _D P M IN , pole_slip_prot.D m in);

set (L G _P T , pole_slip_prot.pt);

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 490

 set (L G _P M A X , pole_slip_prot.pm ax);

if(p er_ u n it == 1) set (L G _T , (pole_slip_prot.trip*0.5) );

e lse se t (L G _T , (pole_slip_prot.trip*0.5*rating) );

return pole_slip_prot.trip;

v o id p o le_ slip _ p t_ calc(v o id )

{ /* d e sig n ed to b e called ev ery 20m s */

ex tern R eal rating;

static in t i = 0;

/** o n ly 3 sec o f test d a ta , initialise Pt & P m ax a fter tw o cycles **/

i f (i == 2 )

{
pole_ slip _ p ro t.p t = pole_slip_prot.pow erP * pole_slip_prot.P fact;

p o le_ slip _ p ro t.p l = pole_slip_prot.p2 = pole_slip_prot.pow erP ;

po le_ slip _ p ro t.p m ax = 1.4 * pole_slip_prot.pt;

}
i++;

i f (pole_ slip _ p ro t.o n _ o ff_ sw itch == 1)

{
i f (+ + p o le_ slip _ p ro t.c o u n t >= pole_slip_prot.lag)

/* P t calculation (d o n e every lag*20m s) */

i f ((A B S (p o le_ slip _ p ro t.p o w erP - p o le_slip_prot.p2) <

pole_slip_prot.P tol) & &

(A B S (pole_slip_prot.pow erP - p o le _ slip _ p ro t.p l) <

pole_slip_prot.P tol))

{
p o le_ slip _ p ro t.p t = pole_slip_prot.pow erP *

pole_slip_prot.P fact;

if(p o le_ slip _ p ro t.p t < 0) pole_slip_prot.pt = 10;

}
p o le_ slip _ p ro t.p l = pole_slip_prot.p2;

p o le_slip_prot.p2 = pole_sIip_prot.pow erP ;

p o le_ slip _ p ro t.c o u n t = 0; /* reset c o u n t */

}
}
else /* p o le_ slip_prot.on_off_sw itch == 0 */

M.J.Checksfi el d/PhD /University of Bath 1996 Appendix F, Page 491

 pole_slip_prot.pl = rating;

pole_slip_prot.p2 = 0;

pole_slip_prot.pt = 0.01 * rating;

pole_slip_prot.count = 0; /* reset count */

}
}

/ * = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = _ E _N . D _ =

2.3 IDMT OverCurrent Relay Simulation 'C* Code

int idm tO c (void)

{ /* qauntity P * In e n tered in config file as PU (idm t_psm ) */

/* this is a v o ltag e con tro lled o /c , un less idm t_vc set to -1 */

/* w hen v o ltag e d ro p s b e lo w i d m t v c setting, then the current setting */

/* (p sm ) is re d u ce d by a facto r o f 0.4 (as in GEC C D V 22 & 6 2) */

extern int idm t_p, idm t_type, sam p cyc, per_unit;

extern R eal i d m t p s m , id m t tm s, idm t vc, rating, vbase;

static int re set tim er a = 0, re set tim er b = 0, reset tim er c = 0;

static R eal integ a = 0, integ_b = 0, integ c = 0;

int trip = 0, trip a = 0 , trip b = 0, trip e = 0;

R eal psm , v o lt c o n , n u m er, tim estep;

d o u b le c u r r a , c u rr b, c u rr c, p o w er of;

if (idm t p = = 0 ) return -1;

i f (p e r_ u n it == 1) {

psm = id m t p s m ;

v olt_con = idm t_vc;

} else {

psm = idm t psm * (rating/( 1.73205*vbase));

v olt_con = id m t vc * (vbase * 0.5773502);

}
tim estep = 1/(50*(R eal)sam p_cyc);

/* se t n u m era to r an d 'to the p o w e r o f e xpressions */

i f (idm t ty p e == 0 ) {

n u m e r = 13.5;

p o w e r o f = 1;

} else i f (id m t_ ty p e == 1) {

n u m er = 0.14;

p o w e r_ o f = 0.02;

} else i f (id m t_ ty p e == 2 ) {

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 492

 n u m er = 80;

p o w e r o f = 2;

} else fatal("Invalid idm t_type '% d' in c onfig file", idm t_type);

/* * * * * * * * * * * * * * * * * * * * P hase A e le m en t ************ * * * * * * * * * * * * * /

/* v oltage control elem en t */

i f (v alu e(V R M S _ A ) < volt co n )

{
/* deriv e cu rren t in m ultiples o f the tap setting *1

c u r r a = value(IR M S _A ) / (0.4 * psm );

}
else

c u rr a = value(IR M S _A ) / psm ;

/* ch ec k w h e th er to start integration p ro c ess (current > 1 ) */

/* re set is done o v e r 5 cycles as an expo n en tial */

if (c u rr a >= 1 ) {

integ a += ( (pow (curr_a, p o w e r_ o f) - 1) * (tim estep/num er) );

r e s e t t i m e r a = 1;

} else {

i f ( r e s e t t i m e r a != 0 ) {

integ_a = in teg _ a * e x p ( -reset_tim er_a/sam p_cyc );

reset_tim er_a+ +;

if (reset tim er a == ((5 * sam p _ cy c) + 2 ) ) {

integ_a = 0;

r e s e t t i m e r a = 0;

}
}
}

/* c h ec k to see w h e th er to trip o r n o t */

if (in te g _ a > = idm t tm s) {

t r i p a = 2;

}
e lse trip a = 0;

/* note: the phase b and c e lem en ts h a v e been rem oved fo r the thesis as they are identical to the p h ase a elem en t */

set (ID M T T , (trip_a * psm ));

i f (trip > = 2 ) return 1;

e lse return 0;

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 493

2.4 Frequency Measuring Algorithm 'C' Code

P R IV A T E R eal c alcF req 0

{
/** this function c alu cla te s frequency using the zero crossing

* m ethod o u tlined in ’M O 17'

* N B : val[0] & v a l[l] are sam ple values eith er side o f zero crossing

* a t b e g in n in g , w h ilst val[2] & val[3] are values either side o f

* z ero c ro ssin g a t end o f m easuring period (h a lf cycle).

* fourier c o sin e c o m p o n e n ts used. 2 C Y C L E C osine fourier

* filter used to sm ooth w aveform

**/

/** v ariables fo r freq m ea su re m e n t * * /

extern int sa m p _ cy c , p e r unit;

extern R eal rcf_set;

static int c c o u n t = 0, tim e r = 0 , reject = 1, i = 0;

static R eal val_c[4] = {0,1,2,3}, z_cc[2] = {1,2}, val[20];

static R eal c_ freq _ o ld = 0, c_ freq _ n ew = 50, old freq = 50, m ovavg;

static R eal rej_lev;

R eal c_p erio d = 0, freq, m vg;

/** v ariables fo r 2 c y cle c o sin e filter */

static R eal v a l_cos[85];

static int x = 0;

int z, n = 1;

R eal tw opiN , lastval = 0, re su lt = 0;

if(sam p_cyc > 4 0 ) fatal ("F unction calcF req, sam p cyc > 40");

tw opiN = (2 * 3 .1 4 1 5 9 )/sam p _ cy c;

lastval = v al_cos[x];

v a l_cos[x] = value(V _A );

z = x - 1;

w hile (z > = 0)

{
re su lt += v al_ co s[z] * cos(tw opiN * n);

z~;

n++;

}
z = (2 * sa m p _ cy c ) - 1;

w hile (z > x )

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 494

 result += val_cos[z] * cos(tw opiN * (R eal)n);

z -;

n+ +;

}
result += ( (val_cos[x]/2) + (lastval/2) );

result /= ((R e a l)s a m p c y c * 0.5);

i f (x == ((2*sam p_cyc) - 1)) x = 0;

else x+ +;

i f (tim er == 0 ) rej lev = rcf_set/( (R e a l)s a m p c y c * 5);

/* * N B m ult by 5 , n o t 50 cos freq only ch an g es every h a lf c y cle **/

/** set n e w sam ples to z _ c c [l] **/

z _ c c [l] = result;

/** ch ec k to see if zero crossing has ju s t happened **/

if(((z_cc[0] <= 0 )& & (z _ cc [l] > 0 » || ((z_cc[0] > = 0 )& & (z _ c c [l] < 0))){

val_c[3] = z _ c c [l];

val_c[2] = z_cc[0];

c period = c _count + 1 - (val_c[0] / (val_c[0] - v a l_ c [l]))

- (val_c[3] / (val_c[3] - val_c[2]));

c_freq_old = c_freq_new ;

c_ freq _ n ew = (R eal)sam p cyc * 12.5 / c_period;

/** set values ready for n e x t h a lf cycle **/

val_c[0] = val_c[2];

v a l_ c [l] = val_c[3];

c _ co u n t = 1;

}
else c_count+ + ;

/** re set zero crossing detecto r values **/

z_cc[0] = z _ c c [l];

tim er++ ;

/** c h eck th at m ax rate o f change has n o t been e x ceed ed * * /

if( tim er > (intX (float)sam p_cyc * 2.5) ) /* to ensure filter settled */

{
freq = c f r e q n e w + c_freq_old;

i f ( A B S (freq - old freq) > = (rej_lev * re je c t/10) )

{
freq = old_freq;

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 495

 reject+ +;

}
else

{
o ld_freq = freq;

reject = 1;

if (p er_ u n it = = 1 ) freq = freq/50;

}
else

{
i f (p e r u n it = = 1 ) freq = c _freq_new = 1;

e lse freq = c _freq_new = 50;

/** h a lf c y cle m oving average filter (for sm oothing, m ainly for R O C O F ) * * /

m o v av g -= val[i];

val[i] = freq;

m o v av g + = val[i];

m vg = m ov av g / ((R eal)sam p_cyc/2);

i f (i == ( (intX sam p_cyc/2) - 1)) i = 0;

e lse i++;

i f (tim er > = (2* sam p _ cy c) ) return m vg;

e lse return c_freq_new ;

M.J.Checksfield/PhD/University of Bath 1996 Appendix F, Page 496