Modelling and simulation of a PV

generator for applications on dis-

tributed generation systems

Paraskevi Breza
Master of Science Thesis

Intelligent Electrical Power Grids

 Modelling and simulation of a PV
generator for applications on
distributed generation systems

Master of Science Thesis

For the degree of Master of Science in Sustainable Energy Technology

at Delft University of Technology

Paraskevi Breza
4184122

November 15, 2013

Faculty of Applied Sciences (TNW) · Delft University of Technology

Copyright © Electrical Engineering Mathematics and Computer Science (EEMCS)
All rights reserved.
Cover photo by: Denver International Airport (2008). 2 MW PV system at the Denver
International Airport. [photograph]. Retrieved November 14, 2013, from:
http://images.nrel.gov/viewphoto.php?imageId=6311939
 Delft University of Technology
Faculty of
Electrical Engineering Mathematics and Computer Science (EEMCS)

The undersigned hereby certify that they have read and recommend to the Faculty of
Applied Sciences (TNW) for acceptance a thesis entitled
Modelling and simulation of a PV generator for applications on
distributed generation systems
by
Paraskevi Breza
in partial fulfillment of the requirements for the degree of
Master of Science Sustainable Energy Technology

Dated: November 15, 2013

Supervisor(s):
Dr.Ir. Marjan Popov

Reader(s):
Prof.Dr.Ir. Miro Zeman

Dr.Ir. Armando Rodrigo Mor

 Abstract

Solar energy is expected to play a great role in the grid’s future infrastructure as a distributed
source, due to the fact that it is an easily available renewable source of energy. The expected
high penetration of Photovoltaic (PV) power into the existing electricity grid demands a more
detailed study and analysis of the interaction between PV systems and the grid, to enable safe
and reliable operation. Until recently, in the case that an electromagnetic transient software
package was used for a PV system study, a PV generator model had to be developed by
simulating a complex block-circuit. Therefore, the simulation of a less complex PV generator
model implemented in an electromagnetic transient program would be of great significance.
In this project, a simplified electrical model of PV generator, simulated in Electromagnetic
Transients Program (EMTP), is proposed. The equivalent circuit of the generator is based
on a linearisation process using the Newton-Raphson algorithm, in order to uncouple its
current and voltage quantities. This scheme has been developed explicitly for association
with electromagnetic transient packages for power system studies. The simplified model is
represented by a dependent current source in parallel with a variable conductance.
The association of the proposed model with RLC circuits is simulated and the results are
compared with measurement results. The systems show good transient response and settle to
the steady state quite fast, so the model can successfully predict the nonlinear performance
of a PV generator. An optimisation scheme is also proposed, in order for the PV generator to
follow the changes of atmospheric conditions. When solar irradiance and temperature change
the generator shows good response and reproduces already validated values.
The model is also combined with a single-phase grid-connected system, of which all the
components are developed in EMTP. The evaluation and modelling of the PV system is
performed with data from real PV modules. Results of the modelling and PV characteristics
are described, showing good agreement with initial nonlinear models. An important remark
is that the harmonic content of injected current into the grid is found significantly high,
due to the assumptions made in this project and the limitations of the system. For the
system’s improvement a Maximum Power Point Tracking (MPPT) algorithm is developed
and integrated, based on the concept of appropriately firing the inverter’s switches. The
algorithm corresponds successfully to variable initial conditions and the system response is
validated by numerical calculations.

Master of Science Thesis Paraskevi Breza

ii

For the three-phase grid-connected case, an already established hybrid PV/battery model,
simulated in MATLAB/Simulink, is used. The circuit parameters of the PV generator model
are estimated from data of the output characteristic of the EMTP-based model, extracted in
Excel and inserted in the Simulink model in terms of a system function. Case studies simulated
prove the expandability of the proposed PV generator model in three-phase systems and the
adaptation potential in large-scale systems.
Keywords
electromagnetic transients, photovoltaic generators, photovoltaic generation systems, tran-
sients, nonlinear circuits, hybrid systems, distributed generation

Paraskevi Breza Master of Science Thesis

 Table of Contents

Acknowledgements xiii

1 Introduction 1
1-1 Project objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1-2 Project contribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1-3 Project outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2 Theory behind photovoltaic solar energy 7

2-1 Solar energy conversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2-1-1 Solar radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2-1-2 PV cell operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2-2 Modelling of PV devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2-2-1 Ideal PV cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2-2-2 Modelling of modules and arrays . . . . . . . . . . . . . . . . . . . . . . 11
2-2-3 Factors affecting PV characteristic curves . . . . . . . . . . . . . . . . . 13

3 PV Systems 15
3-1 Types of PV systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3-1-1 Stand-alone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3-1-2 Hybrid systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3-1-3 Grid-connected systems . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3-2 Photovoltaic converters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3-2-1 DC/DC converters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3-2-2 DC/AC converters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3-3 Maximum power point tracking (MPPT) . . . . . . . . . . . . . . . . . . . . . . 25
3-3-1 MPPT techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

Master of Science Thesis Paraskevi Breza

iv Table of Contents

4 Modelling of a photovoltaic generator in EMTP 27

4-1 ENRC of a PV generator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4-1-1 Existing PV generator model . . . . . . . . . . . . . . . . . . . . . . . . 27
4-1-2 Derivation of the ENRC of a PV generator . . . . . . . . . . . . . . . . . 29
4-2 The EMTP-based PV generator model . . . . . . . . . . . . . . . . . . . . . . . 30
4-2-1 EMPT/MODELS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4-2-2 Modelling of the photovoltaic generator . . . . . . . . . . . . . . . . . . 31
4-3 Application and results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4-3-1 Analysis of RLC circuits empowered by a PV generator . . . . . . . . . . 32
4-4 Model optimisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4-4-1 Temperature and irradiance dependence . . . . . . . . . . . . . . . . . . 44
4-4-2 Simulation and results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

5 Modelling of a single-phase grid-connected PV system 49

5-1 System configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5-1-1 Single-phase bridge inverter . . . . . . . . . . . . . . . . . . . . . . . . . 50
5-2 System and component models in EMTP . . . . . . . . . . . . . . . . . . . . . 52
5-2-1 PV generator model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5-2-2 Filter/Inverter models . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5-2-3 Isolation transformer/utility grid models . . . . . . . . . . . . . . . . . . 53
5-3 Application and results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5-4 Maximum Power Point Tracking (MPPT) . . . . . . . . . . . . . . . . . . . . . 60
5-4-1 MPPT Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5-5 System and component models in EMTP . . . . . . . . . . . . . . . . . . . . . 63
5-6 Application and results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5-6-1 Simulation results with MPPT for 0.40kW/m2 irradiance level . . . . . 64
5-6-2 Simulation results with MPPT for 0.55kW/m2 irradiance level . . . . . 69
5-7 Model optimisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
5-7-1 Simulation and results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

6 Case studies using hybrid PV/battery grid-connected system 89

6-1 System configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
6-2 System and component models in MATLAB/Simulink . . . . . . . . . . . . . . . 90
6-2-1 PV generator and MPPT model . . . . . . . . . . . . . . . . . . . . . . 90
6-2-2 DC/DC boost converter model . . . . . . . . . . . . . . . . . . . . . . . 90
6-2-3 DC/AC inverter model . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
6-2-4 Battery system/Load models . . . . . . . . . . . . . . . . . . . . . . . . 92
6-2-5 Transformer/utility grid models . . . . . . . . . . . . . . . . . . . . . . . 93
6-3 Simulation and results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93

Paraskevi Breza Master of Science Thesis

Table of Contents v

7 Conclusions and future work 99

7-1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
7-2 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100

A The Newton Raphson algorithm 101

Bibliography 103

Glossary 107
List of Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

Master of Science Thesis Paraskevi Breza

vi Table of Contents

Paraskevi Breza Master of Science Thesis

 List of Figures

1-1 Evolution of global PV annual installation 2000-2012 . . . . . . . . . . . . . . . 2

1-2 Evolution of new grid-connected PV systems in Europe (MW) . . . . . . . . . . 3
1-3 PV contribution to the energy demand in the EU 27 in 2012 based on cumulative
installed capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2-1 Spectral distribution of the black body radiation and the Sun radiation . . . . . . 8
2-2 Illustration if the AM1.5 path and the direct-normal and global incident radiations
on a sun-facing surface tilted 37o . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2-3 Processes in an irradiated solar cell . . . . . . . . . . . . . . . . . . . . . . . . . 9
2-4 Single-diode model of ideal PV cell and equivalent circuit of practical PV device . 10
2-5 Characteristic I-V curve of a PV cell . . . . . . . . . . . . . . . . . . . . . . . . 11
2-6 Construction of PV array . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2-7 Characteristic I-V and P-V curves of practical PV device . . . . . . . . . . . . . 13
2-8 Irradiance effect on the PV characteristics . . . . . . . . . . . . . . . . . . . . . 14
2-9 Temperature effect on the PV characteristics . . . . . . . . . . . . . . . . . . . 14

3-1 Classification of PV systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

3-2 Stand-alone PV system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3-3 Hybrid PV system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3-4 Grid-connected PV system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3-5 A basic PV system with DC/DC converter . . . . . . . . . . . . . . . . . . . . . 20
3-6 A buck converter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3-7 A boost converter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3-8 A buck-boost converter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3-9 System architectures employed in PV systems . . . . . . . . . . . . . . . . . . . 22
3-10 Team concept of inverters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3-11 Power versus voltage at a PV module . . . . . . . . . . . . . . . . . . . . . . . 25

Master of Science Thesis Paraskevi Breza

viii List of Figures

4-1 ENRC of a PV generator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

4-2 Equivalent scheme of a nonlinear inductance . . . . . . . . . . . . . . . . . . . . 32
4-3 Parallel R,L,C loads energised by the PV generator . . . . . . . . . . . . . . . . 33
4-4 Equivalent circuits where the PV generator is described by the ENRC . . . . . . 33
4-5 The component dialog box of MODEL object . . . . . . . . . . . . . . . . . . . 35
4-6 PV//R//C EMTP-based circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4-7 PV//R//L EMTP-based circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4-8 Transient response of the PV//R///C circuit . . . . . . . . . . . . . . . . . . . 37
4-9 Transient response of the PV//R///L circuit . . . . . . . . . . . . . . . . . . . 38
4-10 RLC circuit energised by the PV generator . . . . . . . . . . . . . . . . . . . . . 39
4-11 PV-RLC EMTP-based circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4-12 Transient response of the PV-RLC circuit . . . . . . . . . . . . . . . . . . . . . 41
4-13 ppv − vpv curves for different levels of irradiance and temperature of the PV-RLC
circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4-14 Current response of a series-parallel RLC circuit supplied by a PV generator or a
constant voltage Voc source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4-15 Capacitor voltage response of a series-parallel RLC circuit supplied by a PV gen-
erator or a constant voltage Voc source . . . . . . . . . . . . . . . . . . . . . . . 43
4-16 ipv − vpv trajectories for different irradiance and temperature levels . . . . . . . . 46
4-17 Voltage-current characteristics of the SIEMENS SM 100-24 PV module . . . . . 47

5-1 A line-commutated single-phase bridge inverter modelled with ideal switches . . . 51

5-2 A typical configuration of single-phase grid-connected PV system . . . . . . . . . 51
5-3 System topology when 1, 2 are closed and 3, 4 are open . . . . . . . . . . . . . . 51
5-4 Single-phase grid-connected PV system simulated with ATP/MODELS . . . . . . 52
5-5 The voltage vpv of the PV generator for irradiance levels of 0.40kW/m2 and
0.55kW/m2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5-6 The current injected into the utility grid for irradiance levels of 0.40kW/m2 and
0.55kW/m2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5-7 The current ipv of the PV generator for irradiance levels of 0.40kW/m2 and
0.55kW/m2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5-8 The trajectory point of the PV generator for the time period (0, tB ) for irradiance
levels of 0.40kW/m2 and 0.55kW/m2 . . . . . . . . . . . . . . . . . . . . . . . 56
5-9 Harmonic content of the injected current for two levels of irradiance . . . . . . . 57
5-10 PV generator current ipv for different capacitor Cf sizes, between 2mF and 340mF 59
5-11 Simulation results for the single-phase grid-connected system, using a capacitor
55mF for irradiance level of 0.40kW/m2 . . . . . . . . . . . . . . . . . . . . . 59
dp
5-12 Operating point according to the sign of dv on the power characteristic . . . . . 61
5-13 Flowchart of the MPPT algorithm . . . . . . . . . . . . . . . . . . . . . . . . . 62
5-14 Power-voltage derivative oscillation limits . . . . . . . . . . . . . . . . . . . . . 62

Paraskevi Breza Master of Science Thesis

List of Figures ix

5-15 System configuration of a single-phase grid-connected PV system with MPP tracker 63

5-16 PV generator’s voltage vpv for α = 137.4o and irradiance level 0.40kW/m2 . . . 64
5-17 PV generator’s current ipv for α = 137.4o and irradiance level 0.40kW/m2 . . . 65
5-18 PV generator’s power ppv for α = 137.4o and irradiance level 0.40kW/m2 . . . . 65
5-19 Firing angle tracking for αo = 125o and irradiance level 0.40kW/m2 . . . . . . . 66
5-20 The PV generator’s voltage vpv , for αo = 125o and irradiance level 0.40kW/m2 . 67
5-21 The PV generator’s current ipv , for αo = 125o and irradiance level 0.40kW/m2 . 67
5-22 The PV generator’s power ppv , for αo = 125o and irradiance level 0.40kW/m2 . 68
5-23 The current (rms) injected into the utility grid iac , for αo = 125o
and irradiance
level 0.40kW/m2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
5-24 Firing angle tracking for αo = 145o and irradiance level 0.40kW/m2 . . . . . . . 69
5-25 The PV generator voltage vpv , for αo = 145o and irradiance level 0.40kW/m2 . . 70
5-26 The PV generator’s current ipv , for αo = 145o and irradiance level 0.40kW/m2 . 70
5-27 The PV generator’s power ppv , for αo = 145o and irradiance level 0.40kW/m2 . 71
5-28 The current (rms) injected into the utility grid iac , for αo = 145o and irradiance
level 0.40kW/m2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5-29 PV generator’s voltage vpv for α = 138.5o and irradiance level 0.55kW/m2 . . . 72
5-30 PV generator’s current ipv for α = 138.5o and irradiance level 0.55kW/m2 . . . 73
5-31 PV generator’s power ppv for α = 138.5o and irradiance level 0.55kW/m2 . . . . 73
5-32 Firing angle tracking for αo = 130o and irradiance level 0.55kW/m2 . . . . . . . 74
5-33 The PV generator’s voltage vpv , for αo = 130o and irradiance level 0.55kW/m2 . 74
5-34 The PV generator’s current ipv , for αo = 130o and irradiance level 0.55kW/m2 . 75
5-35 The PV generator’s power ppv , for αo = 130o and irradiance level 0.55kW/m2 . 76
5-36 The current (rms) injected into the utility grid iac , for αo = 130o and irradiance
level 0.55kW/m2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5-37 Firing angle tracking for αo = 150o and irradiance level 0.55kW/m2 . . . . . . . 77
5-38 The PV generator’s voltage vpv , for αo = 150o and irradiance level 0.55kW/m2 . 78
5-39 The PV generator’s current ipv , for αo = 150o and irradiance level 0.55kW/m2 . 78
5-40 The PV generator’s power ppv , for αo = 140o and irradiance level 0.55kW/m2 . 79
5-41 The current (rms) injected into the utility grid iac , for αo = 150o
and irradiance
2
level 0.55kW/m . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5-42 The choice box in the PV generator model . . . . . . . . . . . . . . . . . . . . . 81
5-43 Trapezoidal irradiance profile for the dynamic MPPT efficiency test . . . . . . . 81
5-44 Trapezoidal irradiance profile test results . . . . . . . . . . . . . . . . . . . . . . 82
5-45 Temperature profile for the dynamic MPPT efficiency test . . . . . . . . . . . . 84
5-46 Temperature profile test results . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

Master of Science Thesis Paraskevi Breza

x List of Figures

5-47 Irradiance and temperature profile test results . . . . . . . . . . . . . . . . . . . 86

6-1 Hybrid PV/battery system configuration . . . . . . . . . . . . . . . . . . . . . . 90

6-2 Three-phase grid-connected hybrid PV/battery system simulated with Simulink . 91
6-3 New PV model modelled in Simulink . . . . . . . . . . . . . . . . . . . . . . . . 92
6-4 System power simulation results of case study 1 . . . . . . . . . . . . . . . . . . 96
6-5 System power simulation results of case study 2 . . . . . . . . . . . . . . . . . . 96
6-6 System power simulation results of case study 3 . . . . . . . . . . . . . . . . . . 97
6-7 System power simulation results of case study 4 . . . . . . . . . . . . . . . . . . 97

Paraskevi Breza Master of Science Thesis

 List of Tables

3-1 Distortion limits for six-pulse converters . . . . . . . . . . . . . . . . . . . . . . 19

4-1 Data for the PV//R//L and PV//R//C circuits simulations . . . . . . . . . . . 34

4-2 Data for the simulation of the PV-RLC circuit . . . . . . . . . . . . . . . . . . . 40
4-3 Irradiance and temperature correction factors and functions . . . . . . . . . . . . 45

5-1 Data for the single-phase grid-connected PV system . . . . . . . . . . . . . . . . 53

5-2 Data for the PV generator for different irradiance levels . . . . . . . . . . . . . . 54

6-1 Typical electrical characteristics of Solarex MSX 60 PV module . . . . . . . . . . 93

6-2 Data for the three-phase grid connected PV array . . . . . . . . . . . . . . . . . 94

Master of Science Thesis Paraskevi Breza

xii List of Tables

Paraskevi Breza Master of Science Thesis

 Acknowledgements

First of all, I would like to express my profound gratitude and deep regards to my supervisor
Dr.Ir. M. Popov for his guidance, monitoring and constant encouragement throughout the
making of this thesis. Furthermore, I would like to thank Dr. A. Theocharis for his supervision
and constant counsel during the first period of this project. I also take this opportunity to
express a deep sense of gratitude to Prof.Dr.Ir M. Zeman and Dr.Ir. A. Rodrigo Mor, who
graciously agreed to serve on my committee.
Special thanks also go to all my friends I met during these two years. I thank you for your
friendship and patience. Although, I cannot mention you all, do know that everyone had an
impact on my life for which I am thankful.
Last but not least, I would like to thank my parents for their endless support and for allowing
me to realize my own potential. You were always there for me, ready to sacrifice a lot,
especially during these hard times, so I could gain this amazing experience.
As closing remark, I would like to thank Nick for the endless courage and support he gave
me when I needed it. Thank you for the smile on my face every time I see you.

Delft, University of Technology Paraskevi Breza

November 15, 2013

Master of Science Thesis Paraskevi Breza

xiv Acknowledgements

Paraskevi Breza Master of Science Thesis

“What you get by achieving your goals is not as important as what you become
by achieving your goals.”
— Henry David Thoreau
 Chapter 1

Introduction

Since global warming and environmental pollution are nowadays two of the main worldwide
concerns, renewable energy can play a significant role on the reduction of environmental
problems and delay of fossil fuel depletion [1]. The world is running out of non-renewable
energy resources, consequently the need of using green energy sources gains more importance.
Wind energy, solar energy and biomass are considered to be the most popular renewable
energy sources. Research and development, focused on each of these areas, is being carried
out globally, with solar energy playing a leading role given the fact that is one of the cleanest
and least expensive sources of energy [1].
Even though a few years ago the penetration of solar energy into the electricity market was
negligible, recent statistics show a significant change in this tendency. According to the
European Photovoltaic Industry Association (EPIA), the Photovoltaic (PV) market in 66
sunbelt countries potentially could have 250GW installed PV capacity by 2020 and 1.1TW
capacity by 2030 [1]. The PV industry is experiencing rapid growth and solar energy is
highly promoted. Conforming with the technical report [1], there was an increase of 74% in
PV plants installation in 2011, while in 2012 more than 100GW of PV systems were installed
internationally. Europe is the leader in terms of cumulative capacity, with more than 70GW as
of 2012, with China also being a strong player in the solar energy sector, followed by the USA
and Japan. Europe’s market has developed rapidly over the past decade, driven mostly by the
"evolution" of PV installations in Italy and Germany [2]. Worldwide PV system installations
between 2000 and 2012 are shown in Figure 1-1.
PV market reports published by various organizations, show that the installation of large
scale grid-connected PV plants is currently the major trend. While more people are getting
acquainted with solar energy, an increase of small scale installations is expected as well. Such
a raise can result in high penetration of large amount of PV energy into the electricity grid.
Europe accounts for the predominant share of the global PV market in this case as well, even
though 17.2GW of PV capacity was connected to the grid in Europe in 2012, compared to
22.4GW in 2011. Germany was the top market in 2012, with 7.6GW of newly connected
systems, followed by China with an estimated 5GW and Italy with 3.4GW. According to
the technical report [2], PV was the main new source of electricity generation installed in

Master of Science Thesis Paraskevi Breza

2 Introduction

35000

31095
30391
30000

25000
Global PV installation [MW]

20000

17064

15000

10000
7376
6708

5000
2575
1133 1411 1582
303 365 471 584
0
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

ROW 88 56 80 77 29 10 105 42 76 80 284 508 *

MEA n/a n/a n/a n/a 1 n/a n/a 1 1 22 46 121 410
China 19 5 19 10 10 8 10 20 40 160 500 2500 5000
Americas 24 32 47 66 104 106 150 213 346 543 1029 2179 3758
APAC 117 140 191 230 282 303 324 271 535 742 1583 2672 4769
Europe 56 133 135 202 707 984 992 2028 5710 5830 13622 22411 17159
Total 303 365 471 584 1133 1411 1582 2527 6708 7376 17064 30391 31095

Figure 1-1: Evolution of global PV annual installation 2000-2012 (MW) (ROW: Rest of the
World, MEA: Middle East and Africa, APAC: Asia Pacific) [2]

Europe. Figure 1-2 shows the increase of grid-connected PV systems in Europe until 2011,
while a small reduction in 2012 is obvious, mainly as a consequence of the financial crisis.
On the other hand, in Europe the grid-connected capacity is extremely high compared to the
off-grid PV capacity, which accounts for less than 1% of the installed capacity. Nevertheless,
in other countries, such as the USA, Australia and Korea the off-grid capacities installed every
year account for many megawatts and thus are considered to play a significant role in the
penetration of solar energy into the total energy market [2].

In conclusion, PV systems can currently provide approximately 2.6% of the electricity demand
in Europe, up from 2% at the end of 2011. A clear overview can be drawn from Figure 1-
3. PV systems can actually produce 5.2% of the peak electricity demand in the European
Union (EU) 27, a result achieved in just a few years, proving how the development of PV
energy in Europe is occurring at a rate faster than expected [2]. On the other hand, even
though by the end of 2011 the contribution of PV towards the global electricity demand
was 0.5% and only 1% towards the peak power demand, the future looks quite promising
[3]. With the PV cell manufacturing costs decreasing every year, the use of solar energy is
becoming more appealing. With organised support schemes by governing bodies along with
better R&D, this renewable energy source can become a fair player in the world energy market
[2].

Paraskevi Breza Master of Science Thesis

 3

25000
European new grid-connected PV capacities [MW]

22411

20000

17159

15000
13622

10000

5710 5830
5000

2028
707 984 992
56 133 135 202
0
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

Italy Czech Republic Spain Germany Total Europe

Bulgaria United Kingdom Greece France Belgium
Rest of Europe Slovenia Netherlands Austria Denmark

Figure 1-2: Evolution of new grid-connected PV systems in Europe (MW) [2]

8
PV contribution to the energy demand [%]

7 6.7

6 5.62

5
4.26
4
3.4
3.11
3 2.9 2.84

1.99
2 1.7
1.09 0.91
1 0.89 0.74
0.62 0.6 0.44
0.31 0.21 0.06 0.05
0.01 0.01 0.01 0.01 0 0 0
0

Figure 1-3: PV contribution to the energy demand in the EU 27 in 2012 based on cumulative
installed capacity [2]

Master of Science Thesis Paraskevi Breza

4 Introduction

1-1 Project objective

The expected high penetration of PV power into the existing electricity grid demands a more
detailed study and analysis of the interaction between PV systems and the grid, to enable safe
and reliable operation. Concerning performance studies on PV systems, two ways are mainly
used; measurements on operating installations and digital simulations. Generally digital
simulations, while compared to measurements, are considered faster, low cost and appropriate
for sensitivity analysis on different design parameters [4]. For the digital simulation studies,
the interest is focused not only on the improvement of the models of the individual components
of the PV systems, but also on the control side of the PV systems, in order to enable its
operation in high efficiency and under maximum power conditions [4]. Digital simulations are
executed either by using differential equations or electromagnetic transient software programs,
based on nodal analysis on equivalent resistive circuits, such as EMTP, ATP, SPICE etc.
Although many nonlinear equations have been successfully used to describe PV generators,
representing in that way their behaviour, they cannot directly be used with models based
on differential equations or with electromagnetic transient software packages. When the PV
system modelling is based on differential equations the most common problems that occur
are [4]:

• an iterative procedure for the solution of the non-linear equation of the PV generator
should be used, so that the operational point on the plane is designated;

• the PV system limitations, regarding its description by using differential equations, lead
to restrictions to employ methods of circuit analysis;

• advanced algorithms should be used in order to avoid numerical instability problems.

It is clear that the aforementioned limitations have an impact on the derivations of the system
equations, increasing the computational load at the same time. In order to overcome these
problems, a linearised equivalent circuit for the representation of the PV generator, based on
differential equations in standard form, has been developed and shown in Theocharis et al.[4].
In the case that an electromagnetic transient program is used for a PV system study, a PV
generator model has to be developed by simulating a block-circuit using various elements and
control demands. This task is quite complex, since there is no incorporated element suitable
for the representation of a PV generator and all components included in the study have to be
described with regard to linear and algebraic node equations [4]. Therefore, the simulation
of a simple PV generator model implemented in an electromagnetic transient program would
be of great significance.
Aiming to discuss and analyse some of the above discussed issues, such as the analytical
implementation of the linearised equivalent electrical circuit of a PV generator proposed in
Theocharis et al.[4], simulations in the electromagnetic transient software EMTP/ATPDraw
are performed. The developed model is then used for utility grid-connection studies. Conse-
quently, a PV system analysis is implemented. In case of a single-phase system, the analysis
covers the modelling of the PV generator, the inverter controller design, that can also be
termed as the AC side of the PV system, and the execution of the Maximum Power Point

Paraskevi Breza Master of Science Thesis

1-2 Project contribution 5

Tracking (MPPT) algorithm which ensures maximum system’s output power. In the three-
phase case, a hybrid PV/battery system, presented in a previous graduation project and
built in MATLAB/Simulink, is used combined with the EMTP-based model of the PV gen-
erator. The analysis then covers issues such as the DC side design of a PV system and the
inverter controller design, which includes many other control mechanisms such as DC voltage
regulation, active power control, etc.

1-2 Project contribution

The technical and scientific contribution of this thesis project has been formulated as follows:

X Modelling of a PV generator using EMTP/ATPDraw as the simulation tool, also suit-

able for implementation in other electromagnetic transient programs.

X Analysis of detailed PV system’s elements design in EMTP, simplified with the proposed
schemes.

X Development of comprehensive EMTP-based PV system models suitable for utility grid-

connection studies.

X Combination of the EMTP-based PV generator model, with established Simulink grid-

connected model.

1-3 Project outline

The thesis consists of seven chapters. Chapters 1 to 3 review the literature, to obtain the
knowledge of solar energy and the state of art of individual components required for the
proposed PV generator and PV system interfaces. Chapters 4 to 6 present the design, imple-
mentation and results of the proposed systems. Finally, the conclusions are presented in the
last chapter.
Chapter 2 presents the theory behind the photovoltaic solar energy and provides information,
such as the basics of a PV cell, to facilitate the understanding of the project. Widely used
electrical models of PV cells are also presented.
Chapter 3 includes a review of PV systems. The definition and configuration of typical PV
systems are presented. The principles of operation of PV converter topologies and a review
of maximum power point tracking methods are discussed briefly at the end of the chapter.
The development of the proposed simulation model of a PV generator is presented in Chapter
4. Specifically, a well-behaved simulation model of a PV generator with known limitations is
developed using EMTP/ATPDraw software package. Different applications and their results
are discussed for validation of the proposed PV model, along with a proposed optimisation
scheme.
A single-phase grid-connected PV system is also designed, simulated and analysed. Chapter
5 presents the design and implementation of all the system’s elements. Furthermore, the
maximum power point algorithm and the behaviour of the converter used are discussed in

Master of Science Thesis Paraskevi Breza

6 Introduction

particularly. Simulation results are analysed for different case studies and a comparative
analysis with experimental results is presented.
A validation analysis of the proposed PV generator model is performed by testing it as a part
of an established three-phase grid-connected model and presented in Chapter 6. Moreover,
the proposed PV system’s behaviour and outputs under load power and solar irradiance
variations are discussed.
Chapter 7 provides an overview of the work performed in this project. It narrows down the
discussion to the main findings and finally potential future work and recommendations for
researchers are presented.

Paraskevi Breza Master of Science Thesis

 Chapter 2

Theory behind photovoltaic solar

energy

The Sun is one of the most significant sources of renewable energy. In one hour the Earth
receives enough energy from the Sun to meet its needs for nearly a year. A Photovoltaic (PV)
cell is a semiconductor device that directly converts the energy of solar radiation into electric
energy. In general, an element that converts sunlight into electricity is called a PV device.
The fundamental PV device is the PV cell, while a set of connected cells form a panel or
module. As an array either a module or a set of modules can be considered [5].
The purpose of this chapter is to provide an introduction into the photovoltaic solar energy
and to present a brief introduction to the behaviour and functioning of the PV devices, without
the intention of providing in-depth analysis of the PV phenomenon and the semiconductor
physics.

2-1 Solar energy conversion

Photovoltaic (PV) energy conversion is often described as the direct conversion of solar ra-
diation into electricity, by means of the photovoltaic effect. Generally, the term photovoltaic
effect refers to the generation of a potential difference at the junction of two different ma-
terials in response to visible or other radiation. Thus, the broad study area of solar energy
conversion into electric energy is denoted as photovoltaics [5].

2-1-1 Solar radiation

As explained above, the basic process of solar cell operation is the generation of the electron-
hole pairs as a result of the absorption of visible or other electromagnetic radiation by a
semiconductor material [5]. The Sun is a light source with a radiation spectrum that can be
compared to the spectrum of a black body at a temperature of nearly 6000K. A black body

Master of Science Thesis Paraskevi Breza

8 Theory behind photovoltaic solar energy

2.5

2.0
Black body 6000K
Energy distribution [kW/m²μm]

AM0 radiation
1.5
AM1.5 radiation

1.0

0.5

0
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

Wavelength [μm]

Figure 2-1: Spectral distribution of the black body radiation and the Sun radiation in the
extraterrestrial space (AM0) and on Earth’s surface (AM1.5) [6]

absorbs and emits electromagnetic radiation in all wavelengths and its theoretical distribution
of wavelengths can be described by Planck’s law. In Figure 2-1 the spectral distribution of
the black body radiation compared to the extraterrestrial and terrestrial solar radiations is
shown.
The spectrum of the sunlight on the surface of the Earth is influenced by different factors, like
the variation of temperature on the solar disc and the influence of the atmosphere, making
the study of the effect of the solar radiation on PV devices quite complicated. At an average
distance between the Earth and the Sun, the irradiance, the flux of solar radiation incident
on the surface outside of the atmosphere is about 1.373kW/m2 , while on the Earth’s surface
is in the order of 1kW/m2 [7]. According to the American Society for Testing and Materials
(ASTM) there are two standard terrestrial spectral distributions; the direct normal and the
global AM1.5. The perpendicular radiation that reaches the sun-facing surface directly from
the Sun corresponds to the direct-normal standard condition, where the global standard
coincides with the spectrum of the direct and diffuse radiations. The atmospheric haze and
the reflection on the Earth’s surface are the factors forcing radiation to diffuse. The AM1.5
standards are defined for a PV device with a surface 37o tilted, while facing the Sun [8].
The AM represents the air mass, more specifically the mass of air between a surface and the
Sun, that influences the spectral distribution and intensity of the sunlight. The AMx number
represents the length of the path of the solar radiation through the atmosphere. Longer paths
lead to more light deviation and absorption. The above phenomena have an impact on the
spectral distribution of the light received by the PV device. The x coefficient of AMx indicates
the length of the path of the sunlight and is defined as:
1
x= (2-1)
cosθz
where θz is the angle of the Sun with reference to the zenith [8], as can be seen in Figure 2-2.

Paraskevi Breza Master of Science Thesis

2-1 Solar energy conversion 9

Sun θz = θz =
zenith
48.19º 48.19º

reflection and
diffusion
AM1.5
37º 37º
Earth
direct-normal global

Figure 2-2: Illustration if the AM1.5 path and the direct-normal and global incident radiations
on a sun-facing surface tilted 37o [8]

Reflection
Front
contact -

n-region I
+-
+- Charge
Charge p-region separation
separation +-
Recombination +
Rear contact
Transmission

Figure 2-3: Processes in an irradiated solar cell [3]

The bigger the x the longer the path and the greater the air mass between the Sun and
the surface of the PV device. The standard AM1.5 distributions correlates with the solar
radiation with a solar angle θz = 48.190o . In Figure 2-2 the definitions of the AM1.5 path and
the direct-normal and global radiations can be seen.
Depending on the geographic position, time, day of the year, climate conditions, composition
of the atmosphere, altitude, etc. the intensity and the spectral distribution of the solar radia-
tion can change. As a result of all the aspects that can affect the solar radiation, the AM1.5
spectral distributions can only be considered as average estimations that can support the eval-
uation and comparison of different PV devices. In the PV industry the AM1.5 distribution
serve as standard and PV modules’ datasheets bring information about the characteristics
and performance of PV devices with respect to the Standard Test Conditions (STC), which
indicate an irradiance of 1000W/m2 with an AM1.5 spectrum at 25o C [8].

2-1-2 PV cell operation

A PV cell is a semiconductor p-n junction photodiode, that can generate electrical power
when exposed to light [9]. There are several types of semiconductor materials used for PV
cells production. The most common types known commercially are mono-crystalline, poly-
crystalline and amorphous silicon (Si) [9]. The principal operation of a PV cell is based on
the phenomenon termed as the photovoltaic effect, depicted in Figure 2-3. This effect can be
defined as a phenomenon in which an electron gets ejected from the conduction band, as a
consequence of the absorption of sunlight of a certain wavelength by a material either metallic,

Master of Science Thesis Paraskevi Breza

10 Theory behind photovoltaic solar energy

practical PV device

ideal PV cell I

Ipv Ipv Rs V
Rp

Figure 2-4: Single-diode model of ideal PV cell and equivalent circuit of practical PV device

non-metallic, solid, liquid or gas. Thus, when light strikes the surface of a PV cell a proportion
of the solar energy is absorbed in the semiconductor material. In case the absorbed energy
is greater that the bandgap energy of the semiconductor, the electron from the valence band
jumps to the conduction band. Therefore, pairs of hole-electron are created in the illuminated
region of the semiconductor. That way the electrons created in the conduction band are able
to move freely. The free electrons have to move in a particular direction by the action of the
electric field present in the PV cells. These flowing electrons compose a current, which can be
drawn from external use by connecting a metal plate on top and bottom of the cells. Finally,
current and voltage, created because of its built-in electric field, generate electric power [9].
The flux of incident light along with the capacity of absorption of the semiconductor affects
mainly the rate of generation of electric carriers. The performance of the cell relies on various
factors, such as the semiconductor bandgap, the reflectance of the cell surface , the intrinsic
concentration of carriers of the semiconductor, the electronic mobility, the recombination rate
and the temperature [8].
The study of the physics behind the PV cells is noticeably complicated and is out of the
scope of this project. In order to study electronic converters for PV systems, it is sufficient
to know the electric characteristics of the PV device. Manufacturers of PV devices provide
either a set of empirical data that can be used to obtain the mathematical equation of the
current-voltage (I-V) curve of the device or provide I-V curves retrieved experimentally for
various operating conditions [8].

2-2 Modelling of PV devices

2-2-1 Ideal PV cell

To simplify the analysis of PV cells in electrical circuits, an electrical model of a PV cell is

introduced. Figure 2-4 shows the equivalent circuit of an ideal PV cell [8]. It consists of a
current source driven by sunlight and an ideal diode. The equation based on the theory of
semiconductors that mathematically describes the I-V characteristics of an ideal PV cell is
the following equation:
qV
   
I = Ipv,cell − I0,cell exp −1 (2-2)
αkT

Paraskevi Breza Master of Science Thesis

2-2 Modelling of PV devices 11

I pv Id I

- =

V V V

Figure 2-5: Characteristic I-V curve of a PV cell [8]

Where [8]:

• Ipv,cell is the current generated by the incident light and is directly proportional to the
solar irradiance [A]

• I0,cell is the leakage current of the diode [A]

• q is the electron charge (1.602x10−19 C)

• k is the Boltzmann constant (1.3806x10−23 J/K)

• T is the temperature of the p-n junction [K]

• α is the diode ideality constant.

The second part of the equation (2-2) is equal to the Shockley diode current Id . In Figure 2-5
the I-V curve of a PV cell can be seen, where the net cell current is calculated as the sum of
the light generated current Ipv and the diode current Id , as the equation (2-2) shows.

2-2-2 Modelling of modules and arrays

Since a single PV cell is generating relatively small output voltage and a relatively high
current, multiple cells are connected in series and enclosed in a common frame to form a
photovoltaic panel or module. By connecting many cells in series, the voltage of the PV
module is increased and the conduction losses in the cables are minimised. A PV module
represents the basic building block for large scale PV power production. Multiple PV modules
can be stacked in series forming strings of modules. As a result, the voltage increases. By
connecting multiple strings in parallel, PV arrays are formed. For an array to perform well
all the modules must not be shaded. Otherwise it will act as a load resulting in heat, which
may damage the solar cell. Bypass diodes are used to avoid damage, however resulting in a
cost increase. Integration of bypass diodes in some large modules during manufacturing is
not uncommon and reduces the extra wiring required [8]. The typical power level of a PV
string can range from a few hundred watts up to 5 kW. For PV arrays, power ratings can
range from a few hundred watts up to hundreds megawatts in case of very large scale PV
plants [4]. In Figure 2-6 the procedure of how PV cells are connected in modules and how
modules can be connected in PV arrays is presented.

Master of Science Thesis Paraskevi Breza

12 Theory behind photovoltaic solar energy

Cell Module Array

Figure 2-6: Construction of PV array

However, the basic equation of the ideal PV cell does not correspond to the I-V characteristic
of a practical PV array. The observation of the characteristics at the terminal of the PV array
requires also additional parameters shown in the equation (2-3) below:

V + Rs I V + Rs I
   
I = Ipv − I0 exp −1 − (2-3)
Vt α Rp

Where [8]:

• Ipv is the PV current [A]

• I0 is the saturation current [A]

• Vt is the thermal voltage of the array [V ]

• Rs is the equivalent series resistance of the array [Ω]

• Rp is the equivalent parallel resistance of the array [Ω]

• α is the diode ideality constant.

Equation (2-3) describes the single-diode model of a PV cell. However more accurate models
have been proposed, such as the double exponential model with an extra diode, to represent
the effect of the recombination carriers, or with three-diodes, to include the influence of effects
that are not considered in the previous models [8]. However, the single-diode model is widely
used in literature, since it offers a good compromise between simplicity and accuracy [9].
In the case that Np parallel connection of cells compose an array the saturation current is
equal to Ipv = Ipv,cell ·Np while respectively the saturation current is I0 = I0,cell ·Np.
From equation (2-3) the I-V and P-V curves of a practical PV device can be produced, as
show in Figure 2-7. Some remarkable points of the curves should be highlighted; the short
circuit point (0, Isc ) where the power is zero, the open circuit point (Voc ,0) where the power
is also zero, and the Maximum Power Point (MPP) (Vmp , Imp ), where the power is maximum
[7]. Similarly to a single PV cell, the maximum power point exists also in case of an array
of modules. The maximum power point is the operating point at which a PV panel or array
of panels delivers maximum output power (Pmp ) at a specific level of solar irradiance. In

Paraskevi Breza Master of Science Thesis

2-2 Modelling of PV devices 13

Figure 2-7: Characteristic I-V and P-V curves of practical PV device [5]

addition, the fill factor of a PV module or array describes how square the I-V curve is, and
is defined as the ratio of two areas by the I-V curve, as illustrated in Figure 2-7 [7]. The fill
factor accounts for the ideality of the PV module; any impairment that reduces the fill factor
will reduce the output power. The fill factor is represented by the following equation [7]:

Pmp Vmp · Imp

FF = = (2-4)
Voc · Isc Voc · Isc
It is quite common that PV modules’ manufacturers provide only some data about thermal
and electrical characteristics, instead of the I-V equation. So, the basic information found
in datasheets are: the nominal open circuit voltage (Voc,n ), the nominal short circuit current
(Isc,n ), the voltage at the MPP (Vmp ), the current at the MPP (Imp ), the open circuit volt-
age/temperature coefficient (Kv), the short circuit current/temperature coefficient (KI) and
the maximum experimental peak output power (Pmax,e ). This information is always provided
with reference to STC of temperature and solar irradiance. Most of the time, manufactur-
ers provide I-V curves for several irradiance and temperature conditions, which makes the
adjustment of the mathematical I-V equation easier [8].

2-2-3 Factors affecting PV characteristic curves

The electrical characteristics of the PV cell represented by the current-voltage and the power-
voltage curves, as presented in Figure 2-7, are totally dependent on the particular climatic
conditions. Specifically, the I-V and P-V characteristics depend on the irradiance incident
on the PV device and the environmental temperature. Figure 2-8 highlights the effect of
different levels of irradiance on the characteristic curves of a PV panel. It is clear that the
short circuit current Isc is semi-linearly dependent on irradiance, while the change in the
open circuit voltage Voc is minor. Thus, the output power at the maximum power point Pmpp
increases as the irradiance increases. More accurately, the power increases in a faster rate
than the irradiance and can it be concluded that the efficiency is higher for high irradiance.
Normally, the characteristics are given for STC, but in practice the irradiance on a PV device
is lower, when light concentration is absent. So, the efficiency is usually lower than the rated
value [7].

Master of Science Thesis Paraskevi Breza

14 Theory behind photovoltaic solar energy

1000W/m²
1000W/m²
800W/m²
800W/m²
Current [A]

Power [W]
600W/m²
600W/m²
400W/m²
400W/m²

Voltage [V] Voltage [V]

Figure 2-8: Irradiance effect on the PV characteristics [7]

Tj = 0ºC
Tj = 0ºC Tj = 25ºC
Tj = 25ºC
Current [A]

Tj = 50ºC
Power [W]

Tj = 50ºC
Tj = 75ºC Tj = 75ºC

Voltage [V] Voltage [V]

Figure 2-9: Temperature effect on the PV characteristics [7]

Temperature also has an effect on the PV characteristics. With an increase in the internal
temperature T j, the short circuit current Isc marginally increases, as a result of exceptional
light absorption. On the other hand, the open circuit voltage Voc is affected strongly by the
temperature and decreases significantly. In real life situations, under sunshine conditions,
the internal temperatures are often higher than the one under STC, and consequently the
efficiency of the PV device is lower [7]. The characteristics for different levels of temperature
can be seen in Figure 2-9.

Paraskevi Breza Master of Science Thesis

 Chapter 3

PV Systems

Photovoltaic (PV) systems are composed of interconnected components designed to accom-

plish specific goals, ranging from powering a small device to feeding electricity into the main
distribution grid. More specifically, PV devices convert sunlight into DC electricity. Such en-
ergy is transferred to the load or to the utility grid by means of a subsystem, that is generally
known as the balance of the system BOS [10]. It involves all the components of a PV system,
other than the PV devices, and includes the following [11]:

• supporting structures for mounting PV modules

• power conditioning units, that adjust and convert the produced DC power to AC power
of required frequency and magnitude

• cables and protection devices, that allow a safe passage for current

• storage devices that store PV generated electricity, to be used when generation is not
sufficient.

The structures are usually weather proof and are placed so that the modules’ orientation
ensures the highest energy production. As for most of the PV systems the main goal is
to harvest all the available energy from the sun constantly, therefore power conditioning
units are needed. The power conditioning units not only maximise the energy harvested but
perform many other functions, like the conversion of DC to AC power or the provision of
galvanic isolation between the PV field and the AC grid. In off-grid applications the power
conditioning equipment is also responsible for managing the energy storage equipment. Due
to solar energy’s intermittent nature, storage systems are usually included into the BOS.
They are mostly adopted in off-grid systems, whenever the extra energy produced by the PV
array must be used during times there is power deficiency (such as during cloudy days or
during night). Batteries are the most common devices used to store energy, although their
environmental impact is an issue, since they use heavy metals and their lifetime is considerably
shorter than that of the PV array [5].

Master of Science Thesis Paraskevi Breza

16 PV Systems

Photovoltaic
Systems

Stand-alone Grid-connected
systems systems

Directly
Without storage With storage Hybrid systems connected to
public grid

Connected to
Appliances With wind turbine public grid via
house grid

With diesel
Small applications
generator

With cogeneration
AC stand-alone
engine

DC stand-alone

Figure 3-1: Classification of PV systems

3-1 Types of PV systems

Photovoltaic systems are classified according to the diagram in Figure 3-1. The two main
classifications as depicted in the figure are the stand-alone and the grid-connected systems,
however the hybrid systems are of significant value as well [12]. The main distinguishing
factor between these two categories is that in stand-alone systems the solar energy output
is matched with the load demand. When a PV system is interconnected with the utility
grid, it might deliver excess PV energy to the grid or use the grid as a backup system, in
case of insufficient PV generation. In these systems a suitable interfacing circuitry has to
be incorporated, so that the PV system will be disconnected from the grid in case of a grid
failure [13].

3-1-1 Stand-alone

Historically the first cost-effective applications of PVs were stand-alone systems in remote
areas, where a connection with the utility grid was not feasible. An example of a stand-alone
system can be seen in Figure 3-2. Even though stand-alone systems are mostly used in the
cases of rural electrification, they can also be used in mobile equipment, communication and
water pumping systems. Typically, a stand-alone system is comprised of the solar device, the
power-conditioning and control units, the storage equipment and the load [12].

Paraskevi Breza Master of Science Thesis

3-1 Types of PV systems 17

I pv

Vpv MPPT control

Ga Vdc DC Bus
Switched mode Vac
Ta Inverter Load
power converter I ac

PV
generator DC/DC Boost
converter
(Bi-directional)

Battery tank

Figure 3-2: Stand-alone PV system

3-1-2 Hybrid systems

When the energy from PV devices can not be supplied in an economically and practically
feasible way, other means are used. In many cases the PV system is used in combination with
a diesel generator. In such a system the energy demands are ensured to be met, while fully
utilising the PV supply [12]. A typical hybrid system can be seen in Figure 3-3.

3-1-3 Grid-connected systems

Grid-connected PV systems offer the opportunity to their users to be self-sufficient in terms

of energy and sustainable in terms of protecting the environment. As mentioned in the
introduction part of this report, the installation of grid-connected systems has increased
considerably in the last years. A further increase is expected in the upcoming years, since
more governments are introducing legislations to promote the use of renewable energy and PV
systems’ costs continue to go down. A great advantage of the grid-connected systems is that
they offer the opportunity of generating significant amounts of energy near the consumption
point, minimising that way the transmission and distribution losses. The operation of such
systems is in parallel with existing electricity grids, so that the exchange of electricity from
and to the grid is allowed [12]. As shown in Figure 3-1, grid-connected PV systems can be
classified in decentralised and central grid-connected systems [7].
In decentralised PV systems, the energy storage can be committed, since the power provided
by the solar radiation in the residencies can be injected into the utility grid in case of surplus.
The central grid-connected PV systems, can be in a range up to MW and can be connected
to a medium or a high voltage grid [7].
With the Institute of Electrical and Electronics Engineers (IEEE) adopting standard 929-2000
in 2000, most of the technical issues regarding the connection of a PV system to the utility

Master of Science Thesis Paraskevi Breza

18 PV Systems

I pv

Vpv MPPT control

Ga Vdc DC Bus
Switched mode Inverter/
Vac
Ta Load
power converter Charger I ac

DC/DC Boost
converter
PV (Bi-directional)
generator
Battery tank

2500 W
Gen

Figure 3-3: Hybrid PV system

I pv Wavelength [μm]
Energy distribution [kW/m²μm]

MPPT control
Vpv

DSP DSP Vac and I ac measurements

Controller Wavelength [μm] Controller
Gating Gating
Signals Signals
+ -

+
Transformer Infinite
-
Bus

DC/DC
+ -

+
Local
- Utility
Transformer
Nonlinear Photovoltaic Avarage Inverter Model
Behavioral Model Behavioral Model
Co
Current
Sensor

Rc+Rco
0V
+ Rd+Rdo
Vcb Battery
Rp

Voltage
-

Nonlinear Battery
Behavioral Model

Figure 3-4: Grid-connected PV system

Paraskevi Breza Master of Science Thesis

3-2 Photovoltaic converters 19

Table 3-1: Distortion limits as recommended in IEEE Std 519-1992 for six-pulse converters

Odd Distortion
harmonics limit
3rd -9th <4%
11th -15th <2%
17th -21st <1.5%
23rd -33rd <0.6%
above 33rd <0.3%

grid have been determined [12]. In this standard, the PV systems’ integration is divided in
two main categories; safety and power quality. In the IEEE Std 929 it is declared that the
limits of the total harmonic distortion, caused by the PV system at the point of common
coupling (PCC), must comply with Clause 10 of IEEE Std 519-1992 [14]. These limitations
are shown in Table 3-1. Regarding safety, a main issue that has extensively been studied is the
issue of islanding. Islanding refers to the condition in which a distributed generator continues
to power a location, even though the power from the utility grid is no longer present [15]. In
this case, the inverter is forced to automatically shut down, given that the source of power is
disconnected from the network. Another issue that has to be taken into consideration is the
Radio Frequency Suppression, that demands proper filtering and shielding [14].

3-2 Photovoltaic converters

A main technology linked to PV systems is the technology of power electronic converters. PV

power has to be converted from DC to AC, in order to feed the power grid or AC loads in
general [13]. An ideal PV converter should be able to supply the maximum power, drawn
from the PV device, to the load side. For grid-connected systems, this has to be done with
the minimum harmonic content in the current and at a power factor close to one. In case
of stand-alone PV systems the output voltage must also be regulated to the desired value.
In this section a short analysis of various topologies often associated with PV systems is
presented. All the semiconductors switches used below are considered to be ideal [13].

3-2-1 DC/DC converters

In order for someone to be able to understand the switch-mode power conversion, the simple
idea behind DC/DC power converters has to be explained first. There are three basic DC/DC
topologies: buck, boost and buck/boost [13]. A simple PV system with a DC/DC converter
is presented in Figure 3-5. DC/DC converters are supposed to control the voltage across the
load, even in cases in which the input voltage varies [4].

Buck converter

The simplest version of a buck DC/DC converter is shown in Figure 3-6. During the on state
of the switch, the input voltage is applied to the load. When the switch is off, the voltage

Master of Science Thesis Paraskevi Breza

20 PV Systems

DC
Storage
PV Load
Array DC

Figure 3-5: A basic PV system with DC/DC converter

id

iL
io
L +
Vd
+ VL - Load
Vo
C
-
-

Figure 3-6: A buck converter

across the load is zero [16].

The average output voltage can be found from the unfiltered voltage as [16]:

ZTs
1 1 ton
V0 = vo (t)dt = (ton · Vd + tof f · 0) = · Vd (3-1)
Ts Ts Ts
0

In order to make the understanding more simple we will now define a new term, the duty
cycle as:
ton
=D (3-2)
Ts
and therefore
Vo = D · Vd (3-3)

In general this output voltage contains undesirable high harmonics, thus filtering is required
[16].

Boost converter

This converter boosts the output DC voltage to a value higher than the input DC voltage.
Applying the inductor volt-second balance we get [16]:

Vd · ton + (Vd − V0 ) · tof f = 0 (3-4)

Paraskevi Breza Master of Science Thesis

3-2 Photovoltaic converters 21

io
iL
L +

Load
+ + VL - Vo

Vd C

- -

Figure 3-7: A boost converter

id

iL
-
Vd
iL
+
Load
VL Vo

- C
- L +
io

Figure 3-8: A buck-boost converter

Equation (3-4) can be solved in order to obtain the ratio for the input to output voltage [16]:

V0 1
= (3-5)
Vd 1−D

Equation (3-5) is valid in the continuous conduction mode, where the output current never
goes to zero between switching cycles. The principle behind it is that energy stored in the
inductor, during switch on operation, is later released against higher voltage V0 . That way
the energy is transferred from lower to higher voltage [16].

Buck-boost converter

In a buck-boost converter the output voltage can be either higher or lower than the input
voltage [16]. By using inductor volt second balance, we obtain:

Vd · ton + (−V0 ) · tof f = 0 (3-6)

and
V0 D
= (3-7)
Vd 1−D
in the continuous conduction mode [16].

Master of Science Thesis Paraskevi Breza

22 PV Systems

PV Strings PV Strings
PV Strings

PV modules

String Module
Central inverter inverter
inverter Multi-string
inverter

AC bus AC bus AC bus AC bus

Figure 3-9: System architectures employed in PV systems [17]

3-2-2 DC/AC converters

System configurations

In this part a general overview of different system architectures, in block diagram form, will
be presented. The system architecture regulates how PV modules are interconnected to the
grid or AC loads via a power converter. Employment of these architectures in PV plants
depends on different factors, such as the environment of the plant, scalability and costs [13].
In Figure 3-9 an overview of types of system architectures is given [13]. Each of these system
configurations will be briefly explained in the following paragraphs, focusing on the main
advantages and disadvantages.
Even though each of these systems has specific properties, there is a main line that all have
to follow. Some of the general characteristics are [18]:

• designed for high efficiency

• capable of special safety features, such as islanding detection
• low harmonic limits of the line currents
• designed for high ambient power
• designed for over 20 years operation under extreme environmental conditions
• operate silently.

Central inverters

A system with central inverter is the most simple architecture employed in PV systems. The
PV modules are connected in strings, hereby increasing the system’s voltage, while strings

Paraskevi Breza Master of Science Thesis

3-2 Photovoltaic converters 23

are connected in parallel forming arrays of modules. The entire array is then connected to
one central inverter which performs maximum power point tracking and power conversion,
as can be seen in Figure 3-9. This configuration is mainly used in very large scale PV power
production, with the central inverter often being DC to three phase [17]. Centralised config-
uration offers the lowest specific cost (cost per kWp of installed power). These configurations
are employed in large scale PV power plants, as mentioned before, where high reliability is
required. The easiest way to achieve that is by using a minimal number of components as in
the central inverter [17].
Despite of their advantages, central inverters suffer from a significant number of drawbacks
[17].

1. Due to the system’s design, a large amount of power is transferred over noticeable
distances using DC wiring. This can lead to safety issues, since faulty DC currents are
difficult to interrupt.

2. The most significant disadvantage is the mismatch losses in the modules, due to the
fact that all strings operate at single maximum power point. The overall system output
can therefore be reduced in case mismatches exist between sections.

3. The configuration of this system is low in flexibility and expandability, since it is de-
signed as a unit.

4. In order to prevent current circulation in the strings, diodes have to be put in series
with each string causing extra power losses.

Module integrated or module oriented inverters

Unlike the central inverters, module integrated inverters, as shown in Figure 3-9, can operate
directly on one or several PV modules. The power rating for these inverters is a couple
hundred of watts and often require a two stage power conversion, due to the low voltage
rating of the PV module. The first stage boosts the voltage to the required value, while
the second stage inverts DC voltage to AC. In order to ensure full galvanic isolation, quite
often a high frequency transformer is incorporated, enhancing system flexibility even further.
These inverters are integrated with the PV panels, contributing to the great flexibility and
expandability of the system. In this configuration the mismatch losses, due to mismatch
in maximum power point between different PV modules, are minimised [13]. However all
the aforementioned advantages come at certain costs. Inverters need to operate in harsh
environmental conditions, which puts strains in the design and reliability of the system,
increasing additionally the specific costs [4].

String inverters

String inverters combine the advantages of central and module integrated inverter concepts.
PV modules connected in series forming a string, with power rating up to 5kWp can be
connected to the utility grid by using a number of smaller inverters [13]. However in this
topology the high DC voltage requires significant consideration. Due to the fact the string

Master of Science Thesis Paraskevi Breza

24 PV Systems

String 1 String 2 String 3 String 4

Team
Connection

String-
Inverters

AC-Connection
Communication

Figure 3-10: Team concept of inverters

inverters are commonly installed on dwellings or office buildings, the protection of the system
requires special attention, especially the use of proper DC cabling [13]. Each string can operate
at its MPP and there is no need of using series diodes, because strings are not connected in
parallel. This fact reduces the losses, although there is always the risk of a hot-spot to occur,
because of the unequal current and power sharing inside the string [13].

Multi string inverters

The multi string inverter concept, as shown in Figure 3-9, is combining the advantage of
higher energy yield of a string inverter with the lower costs of a central inverter. Every
individual string is connected to a lower power DC-DC converter, having independent MPP
tracker optimising the energy output from each string. The system can easily be expanded,
within a certain power range, by simply adding a new string with its DC/DC converter. Then
all the DC/DC converters are connected via a DC bus through a central inverter to the utility
grid. This central inverter is a Pulse Width Modulation (PWM) inverter, using the robust
and relatively cheap IGBT technology [19].

Team concept

Of course many other concepts have been presented in the literature the past few years.
However, the concepts explained above are the most common ones. One of the alternative
configurations, is the team concept. In this concept the string technology is combined with
the master-slave concept [20]. A combination of several string inverters working with the team
concept is shown in Figure 3-10 [20]. At very low solar irradiance conditions, the whole PV
array is connected to a single inverter. That way the overall losses are reduced. By increasing
irradiance more inverters are connected, dividing the PV array into smaller units, until every
string inverter operates close to its rated power. Every string operates separately with its

Paraskevi Breza Master of Science Thesis

3-3 Maximum power point tracking (MPPT) 25

Pmpp

Vmpp or I mpp Vpv or Ipv

Figure 3-11: Power versus voltage (or current) at a PV module

own MPP controller. At low irradiance the inverters are being controlled, in the so-called
master-slave way [20].

3-3 Maximum power point tracking (MPPT)

Another essential part of the PV system is tracking the maximum power point of a PV array.
The maximum power point tracking (MPPT) control by definition allows the PV array, usually
combined with a DC/DC converter, to produce the maximum continuous power, for specific
meteorological conditions [7]. There are many methods used, that differ in complexity, the
convergence speed, cost, popularity etc. The power output from a PV panel is related to the
voltage (or current) at which power is drawn from. In Figure 3-11, the relationship between
delivered power and current (or voltage), applicable for a single PV module with uniform
irradiance throughout the module’s surface, is shown.
The idea behind MPPT is to automatically find the maximum voltage Vmp or current Imp ,
that the PV array has to operate in, in order to obtain the maximum power output Pmpp ,
under given temperature and irradiance conditions. The most difficult part in this operation
happens under partial shading conditions, when is possible to have multiple maxima and it
is more difficult to define the absolute maximum power point.

3-3-1 MPPT techniques

There are many different techniques proposed for maximum power point tracking. In the sur-
vey report as presented in Esram et al.[21], nineteen distinct methods are described. The most
famous techniques are the perturb and observe method, the incremental conductance method,
the fractional open-circuit voltage, the fractional short-circuit current, the ripple correlation
control, the DC-link capacitor droop control, the load current or voltage maximisation and
the dP/dV or dP/dI feedback control.
The MPPT technique used in this project, will be analytically explained in the following
chapters.

Master of Science Thesis Paraskevi Breza

26 PV Systems

Paraskevi Breza Master of Science Thesis

 Chapter 4

Modelling of a photovoltaic generator

in EMTP

A photovoltaic (PV) generator is a nonlinear device having irradiance-dependent volt-ampere

characteristics [22]. More specifically, a PV generator is a collection of interconnected solar
cells and other components. The entire array is assumed to be treated as a system constituting
N strings in parallel, having M cells connected in series per string, with all the cells being
identical. Many electrical models of photovoltaic generators were developed to represent the
nonlinear behaviour of the device, like the single-diode model presented in Chapter 2. In
the following parts the electrical model of the PV generator, developed in this project, will
be explained and will be followed by its digital simulation in Electromagnetic Transients
Program (EMTP) and the results of different applications.

4-1 ENRC of a PV generator

In this part, the theory behind the method of modelling a PV generator as a linearised
equivalent circuit model, appropriate for implementation in electromagnetic transient pro-
grams, is presented. The mathematical model is based on linearisation of the initial nonlinear
ipv (vpv ) PV characteristic, using the fundamental Newton-Raphson algorithm (Appendix A).
This uncoupling of the electrical quantities, in each timestep ∆t of the digital simulation,
can be represented by an equivalent linearised current source in parallel with a linearised
ohmic conductance. This equivalent electrical circuit is called "Equivalent Newton-Raphson
Circuit"(ENRC) [4]. The procedure obtaining the ENRC is based on an already published
ipv − vpv characteristic nonlinear equation shown in Lalouni et al.[23]. Therefore the develop-
ment of a new characteristic is not necessary.

4-1-1 Existing PV generator model

A nonlinear model, called "Four-Parameter Model", has been extensively used in different
software programs. This model uses four parameters under STC and can successfully predict

Master of Science Thesis Paraskevi Breza

28 Modelling of a photovoltaic generator in EMTP

the performance of mono and poly-crystalline PV arrays [23]. The equivalent circuit current
of the generator ipv based on that model, can be expressed as a function of the generator’s
voltage vpv : n o
m
ipv = Isc 1 − k1 [exp(k2 vpv ) − 1] (4-1)
where the coefficients k1 , k2 , k3 , k4 and m are defined as:

k1 = 0.01175 (4-2)

k4
k2 = (4-3)
Vocm

Isc (1 + k1 ) − Impp
 
k3 = ln (4-4)
k1 Isc

1 + k1
 
k4 = ln (4-5)
k1

ln(k3 /k4 )
m= (4-6)
ln(Vmpp /Voc )

where Vmpp is the maximum power point voltage, Voc is the open circuit voltage, Impp is the
maximum power point current and Isc is the short circuit current [23].
The ipv -vpv curve is affected by the variation of the solar irradiance and the temperature,
as shown in the previous chapter. The adaptation for different levels of solar irradiance and
temperature in this case, can be done by using the curve that corresponds to equation (4-1)
as a reference curve. The new characteristic curve ipv,new -vpv,new then is represented by the
following equations:
∆Tα = Tα − Tr (4-7)

G G
   
∆ipv = αr ∆Tα + − 1 Iscr (4-8)
Gr Gr

∆vpv = −βoc ∆Tα − rs ∆ipv (4-9)

where G corresponds to the solar irradiance, Tα to the temperature, Gr is the solar irradi-
ance under reference conditions, αr is the current temperature coefficient, β oc is the voltage
temperature coefficient, rs is the series resistance and Tr is the temperature and Isc is the
short circuit current under reference conditions [23].
So, the new values of the PV generator’s voltage and current are given by:

vpv,new = vpv + ∆vpv (4-10)

ipv,new = ipv + ∆ipv (4-11)

Paraskevi Breza Master of Science Thesis

4-1 ENRC of a PV generator 29

i pv (t)

+
vpv (t) gpv0

I0 gpv0 vpv (t)

Figure 4-1: ENRC of a PV generator at the time instant t=t0 +∆t [4]

4-1-2 Derivation of the ENRC of a PV generator

Using the model presented previously, we will see in this section in-depth the derivation of
the ENRC of a PV generator as proposed in Theocharis et al.[4]. Assuming that at time t0 ,
the operation point of the PV generator is [vpv (t0 ),ipv (t0 )], in the next time instant t=t0 +∆t
the operation point becomes respectively [vpv (t),ipv (t)]. ∆t is the timestep and depends on
the time constant of the system, that the PV generator is a part of. The value of the timestep
should be appropriate, so that the (4-1) can be linearised. By choosing an initial ipv (vpv )
relation, expanding it in a Taylor series in the space [vpv (t0 ),ipv (t0 )], and by evaluating the
expression at [t,vpv (t),ipv (t)], the following equation is derived, by keeping only the first two
terms:
dipv
ipv (t) = ipv (t0 ) + [vpv (t) − vpv (t0 )] (4-12)
dvpv [ipv (t0 ),vpv (t0 )]

From the ipv (vpv ) characteristic curve it is clear that:

dipv
<0 (4-13)
dvpv [ipv (t0 ),vpv (t0 )]

and the incremental conductance of the system at the point [t0 ,vpv (t0 ),ipv (t0 )] can be defined
as:
dipv
gpv0 = − (4-14)
dvpv [ipv (t0 ),vpv (t0 )]

By substituting equation (4-14) into (4-12) and by rearranging the terms we get the following
equation:
ipv (t) = I0 − gpv0 vpv (t0 ) (4-15)

where:
I0 = ipv (t0 ) + gpv0 vpv (t) (4-16)

In conclusion, equation (4-15) gives the operation point of the PV generator at the time
t=t0 +∆t, using values for the previous time instant, via I0 and gpv0 . Finally, equation (4-15)
can be represented by the equivalent circuit as can be seen in Figure 4-1. I0 , updated by using

Master of Science Thesis Paraskevi Breza

30 Modelling of a photovoltaic generator in EMTP

equation (4-16), is represented by a current source, since its value depends on the irradiance
G, the absolute temperature Tα and the position of the operation point at the previous time
instant t0 . That way the trajectory of the operation point of the PV generator, at any time
instant, can be predicted using the tangent at the reference nonlinear curve at the time instant
t0 , by suitably updating the parameters I0 and gpv0 at each time interval ∆t [4].
The parameter gpv0 is updated using a linear algebraic equation, which arises by the substi-
tution of equation (4-14) in (4-1). So we can derive the derivative with respect to the voltage
vpv of the equation (4-1), evaluated at the point [vpv (t0 ),ipv (t0 )]:

" #
dipv d d m d
= (Isc ) − Isc k1 (ek2 vpv ) + (Isc k1 ) (4-17)
dvpv [ipv (t0 ),vpv (t0 )] dvpv dvpv dvpv [ipv (t0 ),vpv (t0 )]

Since in equation (4-17) the first and the last terms are zero, considering the equation (4-14),
gpv0 is given by:
m
gpv0 = (Isc k1 k2 m) [vpv (t0 )]m−1 ek2 [vpv (t0 )] (4-18)

In conclusion, the procedure for updating equation (4-18) is based on the calculated value
of vpv (t0 ), at the time instant t0 . The next step is to calculate ipv (t0 ) from equation (4-1),
so we can follow the tangent in the initial ipv − vpv nonlinear curve. The explained method
can be directly applied on various types of selected initial models, such as the double-diode
model or other based on nonlinear exponential equations. This application will lead to the
ENRC, presented in Figure 4-1, and then, based on the an initial ipv − vpv nonlinear curve,
the interrelated equation for the incremental conductance gpv0 will be obtained [4].

4-2 The EMTP-based PV generator model

4-2-1 EMPT/MODELS

The big increase of installed PV systems, either connected to the grid or as stand-alone
systems, has also increased the power quality problems. For that reason detailed analysis of
all the PV system’s parameters is required. Tools have been developed to estimate the energy
output characteristics of new systems and new models, based on parametric analysis, have
been introduced for this purpose [24].
PV system analysis, including system and field condition analysis, can be performed with a
transient phenomenon analysis software for electric power systems. As it has been mentioned
in the introduction part of this report, in order for a PV system to be developed, either
laboratory tests or digital simulation are required. Digital time domain simulation programs,
such as Electromagnetic Transients Program (EMTP), have been a key part in the design
and analysis of power apparatus and systems [24]. EMTP can be applied in practically every
problem requiring time domain simulation and is mainly used through ATPDraw. The latter
is a graphical preprocessor of the Alternative Transients Program (ATP) version of EMPT
on the MS-Windows platform [25]. ATP has extensive modelling capabilities and additional
important features besides the computation of transients. One of the integrated simulation

Paraskevi Breza Master of Science Thesis

4-2 The EMTP-based PV generator model 31

modules, the recently introduced MODELS enables the control of the interaction between
the power system and the protective system operations. MODELS is a simulation language,
supported by a set of simulation tools for studying time-variant systems. The description
of the model is self-documenting, so it can be used as the description document, used for
representing the system and as the data used in the actual simulation. With MODELS, the
monitoring and controllability parameters of the power system and algebraic operations for
programming can be calculated. Using MODELS, the user can represent the power system
by describing the physical constants and/or the physical subsystems of the systems to be
examined [25].
In ATPDraw there is also a number of embedded models that can be directly used in a
defined network. These models can be transformer models, machine models or various types
of sources and switches etc. All these components can be sufficient for building a power system
network and apply both steady-state and transient analysis. However, for some studies these
embedded components are not enough, and user defined models should be applied. This
can be done by using either Transient Analysis of Control Systems (TACS) or MODELS,
interfaced to ATPDraw. TACS and MODELS can read a variable from a network in time
domain and use it for the computation of another parameter, and can export parameters back
to the network, such as controlled resistances, current, voltage, etc. [26].
In this section the technique of modelling the EMTP-based generator will be presented. All
the procedures of the components of the PV system are in an EMTP formatted file.

4-2-2 Modelling of the photovoltaic generator

So far a DC source has been used instead of a PV generator in EMTP simulations since there
is no such a component embedded in the program. Due to all the restrictions of using a
PV model compatible with electromagnetic transient programs, explained in the introduction
part, an introduction to a PV generator model, ideal for implementation in EMTP and based
on the ENRC procedure, has been given in this project.
As we saw in Figure 4-1 the ENRC of the PV generator results in the representation of the
generator by a dependent current source I0 in parallel with a variable conductance gpv0 , where
both are updated in each time interval ∆t by using simple linear algebraic equations. For the
computation of the different parameters and equations that lead to the ENRC some of the
already embedded TACS will be used, while some new parameters imported in MODELS will
be introduced.
The idea of setting up the ENRC concept in ATPDraw was based on Popov[26], where a
nonlinear inductance has been modelled by means of a variable resistance type-91 in a parallel
combination with a current controlled type-60 source, as presented in Figure 4-2. The concept
behind this representation, is that in electromagnetic transients programs, the fundamental
differential equations of inductors and capacitors are converted into algebraic form by using
the trapezoidal rule. So by applying the trapezoidal rule in the equation that governs the
behaviour of an inductor:
di(t)
v(t) = L (4-19)
dt
and by rearranging the terms, the current i(t) becomes:
i(t) = G(t) (v(t) + v(t − ∆t)) + i(t − ∆t) (4-20)

Master of Science Thesis Paraskevi Breza

32 Modelling of a photovoltaic generator in EMTP

R(t)
i (t)

u1(t) i hist (t) u2(t)

Figure 4-2: Equivalent scheme of a nonlinear inductance

or
i(t) = G(t)v(t) + ihist (t) (4-21)
where
ihist (t) = G(t)v(t − ∆t) + i(t − ∆t) (4-22)
ihist (t) is the history of the current and G(t) is the variable conductance. The variable
resistance type-91 is in this case:
1
R(t) = (4-23)
G(t)

So, in order to build the PV generator in this project, one of the default MODELS in ATPDraw
will be used. In this new PV generator model the inputs should be the Voc , Isc , Vmpp and the
Impp , in order to build the equations explained in the procedure of obtaining the ENRC. With
the coefficient k1 constant and the inputs known, the first step in the designing procedure
is to calculate the coefficients k2 , k3 , k4 and m. Our PV model will then start the iteration
procedure of calculating all the parameters by a starting point t0 , in which we set an initial
vpv (t0 ) value. With all the above values calculated, the model developed will calculate gpv0
by using equation (4-18) and then from equation (4-15) the ipv (t0 ) will also be calculated. In
the following step the model will calculate I0 by using equation (4-16). The new introduced
MODEL will have as outputs the ihist that will go to the current controlled type-60 source
and the G(t0 ) conductance value. And then the generator is ready to proceed to time instant
t.

4-3 Application and results

Since a PV generator constitutes a special power source, due to its transcendental current-
voltage characteristic, the known performance of linear electrical circuits and systems ener-
gised by conventional sources has to be re-examined. So, in this section the PV generator is
associated with equivalent RLC circuits.

4-3-1 Analysis of RLC circuits empowered by a PV generator

In general, the necessary background for studying the transient performance of complex elec-
trical systems, is provided through studying circuits involving either one or two independent
energy-storing elements. The same practice in the case of a photovoltaic system could be an

Paraskevi Breza Master of Science Thesis

4-3 Application and results 33

i pv i pv

iR iL iR iC
+ +
PV vpv R L PV vpv R C

- -

(a) Equivalent circuit of a PV//R//L circuit (b) Equivalent circuit of a PV//R//C circuit

Figure 4-3: Parallel R,L,C loads energised by the PV generator

i pv i pv

iR iL iR iC
igpv0 + igpv0 +
I0 gpv0 vpv R L I0 gpv0 vpv R C

- -

(a) PV//R//L circuit (b) PV//R//C circuit

Figure 4-4: Equivalent circuits where the PV generator is described by the ENRC

intriguing aspect for an engineer. So, the dynamic behaviour of first-order circuits (circuits
that contain two types of passive elements, namely resistors and one capacitor C or just one
inductor L, along with a network of DC sources and switches, characterised by first-order
differential equations) powered by a PV generator can be the starting point towards that
direction [27]. However, the transient behaviour of circuits involving R,L,C elements in dif-
ferent combinations, especially that of the linear second-order RLC circuits (where both a
capacitor and a inductor are present simultaneously), can also be of particular interest for
many reasons [27], and will be explained in the following sections.
The method to prove that the proposed PV generator model is functional in EMTP, is to
combine it with equivalent electrical circuits involving R,L,C elements and provide a transient
response analysis of these circuits. As it is known, in case a circuit contains one or more storage
elements, the circuit response to a sudden change will go through a transition period prior to
settling down to a steady state value [27]. Roughly speaking, the time varying current and
voltage resulting from a sudden application of sources are called transients, while the steady
state value is the magnitude of voltage or current, after the circuit has reached stability. It is
this transition period that shows how fast the circuit responds to changes and is the significant
part of a transient analysis. So, two first-order, parallel PV//R//L and PV//R//C, and a
second-order, series-parallel PV-RLC, circuits will be examined. In addition, the predicted
results from the simulations will be compared with results retrieved from measurements.
In Figure 4-3a and in Figure 4-3b the parallel PV//R//L and PV//R//C are shown respec-
tively. These two circuits have the same resistive circuit. In Figure 4-4a and in Figure 4-4b
the equivalent circuits can be seen, where the PV generator is represented by the ENRC.
Table 4-1 shows the PV panel’s data under STC for a SIEMENS SM 110-24 module [28],

Master of Science Thesis Paraskevi Breza

34 Modelling of a photovoltaic generator in EMTP

Table 4-1: Data for the PV//R//L and PV//R//C circuits simulation

PV generator PV//R//L PV//R//C

STC Ppv = 110W
Impp = 3.15A
V = 35V
Pmpp = 110W
Isc = 3.45A
Voc = 43.5V R = 22.59Ω R = 60.45Ω
αr = 1.4mA/o C L = 0.3H C = 10−3 F
boc = −152mV /o C ∆t = 10−5 s ∆t = 10−5 s
Measurement conditions G = 900W/m2
Tα = 35o C

which is used for the simulations. The experimental characteristics were obtained by Lalouni
et al.[23] and have been also used in the measurements’ explanation part of Theocharis et
al.[4]. For the experimental characteristics a prototype of the PV system, as described in
Lalouni et al.[23], was designed and implemented at the university of Bejaia (Algeria).
In order to start the implementation in EMTP all the components of the aforementioned
circuits have to be simulated in the ATPDraw interface. As explained before the new MODEL,
representing the PV generator, will have as inputs the Voc , Isc , Vmpp and the Impp . In the
component dialog box of the new model object, where the DATA and NODES attributes
are shown, the values from Table 4-1 are inserted, as can be seen in Figure 4-5. On the
Library page the link to the original file on disk is given and the Reload option becomes
available. The input and outputs of the MODEL and its interface with the rest of the circuit
are automatically managed by ATPDraw. The model descriptions are written directly in the
ATP file [25].
Figure 4-6 shows a diagram of the PV//R//C circuit, which is similar to the diagram of
the PV//R//L circuit. The PV generator is modelled using ATP/MODELS and TACS
components, as explained previously. Each of the presented systems has to be energised by
setting an initial voltage value vpv (t0 ) for the time instant t0 , depending on the structure
and on the values of the system’s components. The transient response of the PV//R//C and
PV//R//L circuits can be shown in Figure 4-8a and Figure 4-9a respectively. In Figure 4-8b
and Figure 4-9b one can see the trajectories of the operation points from the time instant
tA = 0s until the steady state on the ipv − vpv plane. In both graphs the point A corresponds
to the time instant tA = 0s, while the point B corresponds to the steady state of the PV
generator.
From the graphs it is obvious that an initially uncharged capacitor, for tA = 0s behaves as a
short circuit and the system starts to operate at the point A, while for the steady state the
capacitor behaves as an open circuit, so the operating point is B, as it is shown in Figure 4-8b.
Additionally, Figure 4-8a points out several important things. The initial current through the
PV//R//C circuit, in which the capacitor is originally uncharged, is at a maximum. It is not
until charge begins to accumulate that the charge flow begins to diminish. The graphs also
identify a particular point in time, which is of great importance, the time constant τ = RC.
This constant gives an indication of how rapidly the system responds to sudden changes and

Paraskevi Breza Master of Science Thesis

4-3 Application and results 35

Figure 4-5: The component dialog box of MODEL object

PV generator
V V
U

type-60
type-91

R C
MODEL
pvgen

Figure 4-6: PV//R//C EMTP-based circuit

Master of Science Thesis Paraskevi Breza

36 Modelling of a photovoltaic generator in EMTP

V V

U
type-60
type-91

R L

MODEL
pvgen
Figure 4-7: PV//R//L EMTP-based circuit

determines when it settles to its final values. In conclusion, it is clear that the response of
the circuit has a transient which dies out eventually and after the response time, a period of
time almost of t = 0.025s, the circuit reaches and remains on its the final steady state value.
Similarly, an initially uncharged inductor for tA = 0s behaves as an open circuit, while in
steady state behaves as a short circuit. So, the dynamic route starts from the point A and
terminates on the point B, as shown in Figure 4-9b. As for the transient response, it is clear
that the ipv current initially starts from a point that is defined by the values of Voc and R,
while the current at the inductor iL , as the transient effect dies, it approaches the Isc current.
The time constant in this case is defined as τ = L/R and the response time is again about
t = 0.025s, until the system settles at its steady state.
To conclude, the response of the system for first-order circuits is fixed by the time constant,
which provides information about the speed of response of the system. The smaller the τ
the more rapid the change, while on the other hand a system with a larger time constant
provides a slow response, since it takes longer to reach the steady state. The most important
point however is that in both cases the predicted values show perfect matching with the
measurements, as presented in Lalouni et al.[23]. Apart from that, the simulated curves
are also in perfect match with the predicted curves obtained by using the ENRC of the PV
generator and the same R,L,C elements as presented in Theocharis et al.[4].
As a second step, in order to verify the EMTP-based model in a second-order case, a series-
parallel PV-RLC circuit is used. The second-order circuits also exhibit a transient response,
which is of great interest, since these types of circuits frequently take place in PV engineering.
However, as these circuits are more complex, their response varies depending on the respec-
tive values of R,L,C. Furthermore, the study of such a circuit is really valuable, because the
behaviour of many higher-order PV systems are often described in terms of an equivalent
second-order circuit. The equivalent PV-RLC circuit shown in Figure 4-10a, can be trans-
formed in the one shown in Figure 4-10b using the equivalent circuit of the ENRC of the
PV generator as presented previously. Figure 4-11 shows the diagram of the test system in
EMTP used in this project.
The PV module’s data are the same as the ones presented in Table 4-1, since once more
the SIEMENS SM 110-24 module was used for both the simulations and for retrieving the
experimental data. The new data of the R,L,C elements of the series-parallel PV-RLC circuit

Paraskevi Breza Master of Science Thesis

4-3 Application and results 37

0.8

Vpv/Voc
0.6 Ipv/Isc
IC/Isc
IR /Isc

0.4

0.2

0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
t [s]

(a) Transient behaviour

3.5

2.5

2
Ipv [A]

1.5

Predicted values
1 Measured values
B

0.5

0
0 5 10
) 15
φ tanφ =1/R
20 25 30 35 40 45
Vpv [V]

(b) ipv − vpv trajectories

Figure 4-8: Transient response of the PV//R///C circuit

Master of Science Thesis Paraskevi Breza

38 Modelling of a photovoltaic generator in EMTP

0.8

Vpv/Voc
0.6 Ipv/Isc
IL/Isc
IR /Isc

0.4

0.2

0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
t [s]

(a) Transient behaviour

3.5

2.5

2
A
Ipv [A]

1.5

Predicted values
1 Measured values

0.5

0
) φ tanφ =1/R
0 5 10 15 20 25 30 35 40 45
Vpv [V]

(b) ipv − vpv trajectories

Figure 4-9: Transient response of the PV//R///L circuit

Paraskevi Breza Master of Science Thesis

4-3 Application and results 39

i pv

iC iRL
+
vpv R
PV C

- L

(a) RLC equivalent circuit using lumped L and C elements

i pv

iC iRL
igpv0 +
vpv R
I0 gpv0 C

- L

(b) Equivalent circuit where the PV generator is described by the ENRC

Figure 4-10: RLC circuit energised by the PV generator

V V
U

type-60
type-91

C R
MODEL
pvgen

Figure 4-11: PV-RLC EMTP-based circuit

Master of Science Thesis Paraskevi Breza

40 Modelling of a photovoltaic generator in EMTP

Table 4-2: Data for the simulation of the PV-RLC circuit

RLC
R = 122.592Ω
L = 10mH
C = 0.1mF
∆t = 10−5 s

can be seen in Table 4-2, as shown also in Theocharis et al.[29].

Figure 4-12a shows the characteristic transient responses of the circuit to the PV generator
applied at t = 0s, while the trajectories for the time interval from tA = 0s to the steady state
on ipv − vpv and ppv − vpv planes can be seen in Figure 4-12b and Figure 4-13, respectively.
The values chosen for R,L,C determine the magnitude of the damping factor of the circuit’s
response.
The curves in Figure 4-12b and Figure 4-13 are carried out for three levels of irradiance and
temperature. In either case, the circuit response starts from the short circuit point and termi-
nates on the point determined by R. The experimental characteristics obtained are compared
to the simulation characteristics for the same operation conditions (G = 450W/m2 , Tc = 25o C;
G = 650W/m2 , Tc = 33o C; G = 900W/m2 , Tc = 35o C). As one can see, the predicted ipv −vpv
and ppv −vpv curves, using the ENRC of the PV generator implemented in EMTP, are in excel-
lent agreement both with the measured values, shown in Lalouni et al.[23] and in Theocharis
et al.[29], and the predicted values obtained by following the node equations in the equivalent
resistive circuit of the PV-RLC circuit, as shown in Theocharis et al.[29]. As also expected, Isc
increases, while Voc slightly changes with increasing irradiance. Additionally, the maximum
power Pmpp increases highly with irradiance, while slightly decreases with temperature. All
the curves presented are notable. Indeed, the circuit examined in this subsection is frequently
encountered in PV applications. The solution of a PV system with an embedded power con-
verter connected to a RL load with a capacitor filter can be obtained via analogous circuits
and its response may be comprised out of a succession of segments of similar curves to those
shown in this section [27].
Since all the above simulations have been conducted with a timestep equal to ∆t = 10−5 s,
the considerable relation of the linearised circuit with the timestep should also be mentioned.
After conducting many simulation trials in EMTP, based also on the results of Theocharis et
al.[29], the conclusion reached is that the timestep strongly affects the predicted curves. In
the case when the chosen timestep is used, a very good agreement with the measured values is
observed. Additionally, in this case the digital simulation results in EMTP of the PV generator
model present great numerical stability. So, at different studies excellent agreement can be
achieved by choosing the appropriate timestep.
For comparison purposes, the response of the same circuit, driven by a constant voltage source,
is also presented. This constant voltage has been taken equal to the open circuit voltage (Voc )
of the PV generator. As an example, the circuit current and capacitor voltage versus time are
plotted for an initially de-energised series-parallel RLC circuit. In Figure 4-14 the transient
response of the current of the PV-RLC circuit in time can be seen, along with the current

Paraskevi Breza Master of Science Thesis

4-3 Application and results 41

0.8

Vpv/Voc
0.6 Ipv/Isc
IC/Isc
IL/Isc

0.4

0.2

0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
t [s] −3
x 10

(a) Transient behaviour from t = 0 to steady state

3.5
Predicted values
900W/m2 35oC Measured values

2.5

2
Ipv [A]

650W/m2 33oC
1.5

450W/m2 25oC
1

0.5

0
0 5 10 15 20 25 30 35 40 45
Vpv [V]

(b) ipv − vpv trajectories from t = 0 to steady state for different levels of irradiance
and temperature

Figure 4-12: Transient response of the PV-RLC circuit

Master of Science Thesis Paraskevi Breza

42 Modelling of a photovoltaic generator in EMTP

100
Predicted values 900W/m2 35oC
Measured values
90

80

70

60
650W/m2 33oC
Ipv [A]

50

40

30

20 450W/m2 25oC

10

0
0 5 10 15 20 25 30 35 40 45
Vpv [V]

Figure 4-13: ppv − vpv curves for different levels of irradiance and temperature of the PV-RLC
circuit

response of the RLC circuit, while a DC voltage source is used as an input. Comparing, the
two circuits we can easily see quite some differences in their transient behaviour. It is obvious
that the PV-RLC circuit reaches the steady state faster. More specifically, in Figure 4-15
the capacitor voltage response of the two RLC circuits is presented. Analogous to the above
mentioned cases, it is clear that for the situation where the PV drives the system the circuit
reaches the steady state faster. The difference in the voltage limits for the two cases is due
to the voltage drop along a series source resistance used while connecting the DC voltage
source when designing the circuit, this is due to EMTP limitations. In Figure 4-14 the curves
indicate that the PV system shows a more abrupt current decay, compared to the DC circuit,
and settles down quickly to the steady state.
In conclusion, although the transient performance of second-order circuits powered by a PV
generator exhibits the general characteristics of their input response, it presents a variety
of unique features which are not met in conventionally powered circuits. Furthermore, the
performance characteristics presented provide a rough approximation of the general behaviour
expected for higher-order circuits. The most important conclusion of our comparison is that
the same approach seems to apply for any circuit involving all three basic circuit elements
in all possible combinations, a fact that makes it extremely important for the study of more
realistic solar electrical systems.
After verifying the PV generator model by studying two parallel PV//R//C and PV//R//L
circuits and a series-parallel PV-RLC circuit, we can conclude that the ENRC implemented
in EMTP, can successfully predict the nonlinear performance of the PV generator and thus
the proposed EMTP-based model is a strong tool. Also, as it has been proved, the PV

Paraskevi Breza Master of Science Thesis

4-3 Application and results 43

PV generator
constant voltage Voc source
2.5

2
Current [A]

1.5

0.5

0
0 1 2 3 4 5 6
t [sec] x 10
−3

Figure 4-14: Current response of a series-parallel RLC circuit supplied by a PV generator or a

constant voltage Voc source

50

45

40

35
PV generator
constant voltage Voc source
30
Voltage [V]

25

20

15

10

0
0 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009 0.01
t [sec]

Figure 4-15: Capacitor voltage response of a series-parallel RLC circuit supplied by a PV gener-
ator or a constant voltage Voc source

Master of Science Thesis Paraskevi Breza

44 Modelling of a photovoltaic generator in EMTP

generator may introduce responses unattainable in circuits driven by conventional sources

due to principal reasons.

4-4 Model optimisation

4-4-1 Temperature and irradiance dependence

As presented in a previous section, the equations used to describe how the parameters of the
PV model change with respect to the irradiance and temperature changes, are (4-8), (4-9),
(4-10) and (4-11). However, one can see that these equations depend on the series resistance
rs of the single-diode model, that has been described in Chapter 2. A point to note is that
the PV generator model, presented in this project, has been built in EMTP in a different way
than the single-diode model. It is only applicable at one particular irradiance level G and
cell temperature Tc per simulation and the absence of rs inhibit the use of the afore-stated
equations. Thus, in order to expand and optimise the model suitable equations had to be
found, therefore a reference I-V curve was translated regarding the irradiance and PV module
temperature changing conditions.
In Marion et al.[30], the Module Energy Ratings (MER) methodology is presented, which uses
indoor tests to characterise the electrical performance of the PV module and to determine
factors to correct the nonlinear performance, when the irradiance and the temperature vary.
Based on this study, the PV generator’s temperature and irradiance correction factors and
functions are determined from a matrix of short circuit current (Isc ) and open circuit voltage
(Voc ), as resulting from the I-V curve measurements over a range of six irradiances and six
operating temperatures. From this procedure three correction factors and functions are de-
termined. The (Isc ) correction factor for temperature, α; the (Voc ) correction for temperature
as a function of irradiance G, β(G); and the (Voc ) correction for irradiance as a function of
the PV module temperature T, δ(T ).
As found in Marion et al.[30], using the incident irradiance and the PV module’s temperature,
Isc and Voc are calculated and a reference I-V curve is translated to determine the current
at a steady voltage and thus at the maximum power. These procedures were based on
modifications to ASTM E1036-96 (Standard Test Methods for Electrical Performance of Non-
concentrator Terrestrial Photovoltaic Modules and Arrays Using Reference Cells) and use the
following equations to calculate Isc and Voc :
G
Isc = Isc [1 + α(T − Tr )] (4-24)
Gr r

Voc = Vocr [1 + β(Gr )(T − Tr )] [1 + δ(T ) ln(G/Gr )] (4-25)

where the subscript r refers to the reference curve.
In order to determine the I-V curve for the desired conditions, a reference I-V curve is selected
to be translated. The reference curve can be the one measured under the irradiance and
temperature conditions closest to those desired. That way the errors caused by the I-V
translation, without taking into account the changes in fill factor due to the changes in
temperature and irradiance, can be minimised.

Paraskevi Breza Master of Science Thesis

4-4 Model optimisation 45

Table 4-3: Irradiance and temperature correction factors and functions

PV Module α β (E) = mE + b δ(T) = mT + b

(°C-1) (°C-1) (dimensionless)
β(E0) m b m b
a-Si/a-Si/a-Si:Ge, S/N 1736 8.50e-4 -3.97e-3 1.17e-6 -5.14e-3 5.20e-4 4.72e-2
CdS/CulnGaSSe, S/N 5165 -1.32e-4 -3.75e-3 1.34e-6 -5.09e-3 6.07e-4 4.97e-2
CIS, S/N 114 1.94e-4 -4.94e-3 2.22e-6 -7.16e-3 8.39e-4 7.39e-2
Mono-Crystal Si, S/N 0442 3.60e-4 -3.63e-3 0.98e-6 -4.61e-3 3.21e-4 4.15e-2
Multi-Crystal Si, S/N581836 2.58e-4 -3.57e-3 1.02e-6 -4.59e-3 4.80e-4 3.55e-2
a-Si/a-Si:Ge, S/N SYS49 8.36e-4 -3.52e-3 1.56e-6 -5.08e-3 5.36e-4 5.53e-2
CdS/CdTe, S/N 14407 0.60e-4 -2.38e-3 1.21e-6 -3.59e-3 6.00e-4 1.61e-2

The I-V data pairs of the reference curve can be then translated into the desired conditions
by using equations (4-26) and (4-27):
Isc
I = Ir (4-26)
Iscr

Voc
V = Vr (4-27)
Vocr

By using this methodology the fill factor does not change, thus the reference I-V curve data
pair for maximum power becomes the translated I-V curve data pair for maximum power.
In order to determine the current at a specific voltage, the current has to be interpolated
by using the two adjacent I-V curve data pairs from the translated I-V curve, with voltages
above and below the desired voltage.

4-4-2 Simulation and results

For the simulation in ATPDraw the same series-parallel PV-RLC circuits shown in the pre-
vious section are used. As reference values are taken the data from Table 4-1 under STC,
since the SIEMENS SM 110-24 module is used again for the simulations. As for the irradi-
ance and temperature correction factors and functions needed for the equations shown above,
Table 4-3, as given in Marion et al.[30], is used. According to this table, a group of 7 PV
modules representing different technologies were used in order to determine the irradiance
and temperature correction factors and functions, as explained in the previous section. To
use equation (4-25), the table gives values of β evaluated for Gr .
The SIEMENS SM 110-24 PV module is made out of mono-crystalline Si cells, thus the
highlighted row of correction factors, as shown in Table 4-3, is being used as data in the
ATP/MODELS of the generator, along with the STC data of the module and the desired
temperature and irradiance. The timestep in these simulations is again ∆t = 10−5 s. The
results of the simulation can be seen in Figure 4-16. At this point, it has to be mentioned that
the simulations use as a reference curve the one corresponding to the PV data under STC,
shown in Table 4-1. This curve has also been successfully reproduced using the aforemen-
tioned procedure. Based on the reference curve two more new curves have been successfully

Master of Science Thesis Paraskevi Breza

46 Modelling of a photovoltaic generator in EMTP

3.5

2.5

1000W/m2 25oC
Ipv [A]

2 800W/m2 25oC
900W/m2 35oC

1.5

0.5

0
0 5 10 15 20 25 30 35 40 45
Vpv [V]

Figure 4-16: ipv − vpv trajectories for different irradiance and temperature levels

predicted. One for the irradiance and temperature level of 900W/m2 and 35o C and the
second one of 800W/m2 and 25o C respectively. The 100W/m2 -25o C curve along with the
800W/m2 -25o C curve, closely approach the voltage-current characteristic curve as shown in
the SIEMENS SM 110-24 module’s manual. The curves based on the technical data of the
module can be seen in Figure 4-17 [28]. Additionally, the curve of 900W/m2 -35o C perfectly
matches with the predicted curve shown in Figure 4-12b.
After many trials the 650W/m2 -33o C and 450W/m2 -25o C curves, shown also in Figure 4-
12b could not be obtained. That can be a result of using the specific correction factors of
Table 4-3, in the procedure explained earlier. It is clear that these data represent some PV
technologies, by using the data of 7 specific PV modules, and are proposed as suitable for
application in modules based in the same technologies. However, the proposed matching
between modules with the same technology but with different technical characteristics might
not be 100% correct. We should also keep in mind that the PV technology has made a
significant progress through the last years, thus the data in Table 4-3 should be updated in
order to follow the developments. In addition, in Marion et al.[30] it is stated that the reference
I-V curve should be under conditions of irradiance and temperature close to the wanted ones.
For our simulations the reference curve is the one based on the data referred to STC of the
chosen PV module. The 650W/m2 -33o C and 450W/m2 -25o C curves can be considered not
to be close to the initial data, a fact that might affect the outcome of the simulations. For
all these reasons, future in-depth work has to be done, regarding the proposed PV generator
model, with respect to the irradiance and temperature effect.

Paraskevi Breza Master of Science Thesis

4-4 Model optimisation 47

Voltage-current characteristic 24V

4.0

3.5

3.0

SM110: 1000 W/m², 25˚C

2.5
SM100: 1000 W/m², 25˚C
SM110: 800 W/m², 25˚C
Amps

2.0 SM100: 800 W/m², 25˚C

1.5

1.0

0.5

0.0
0 10 20 30 40 50

Volts

Figure 4-17: Voltage-current characteristics of the SIEMENS SM 100-24 PV module

Master of Science Thesis Paraskevi Breza

48 Modelling of a photovoltaic generator in EMTP

Paraskevi Breza Master of Science Thesis

 Chapter 5

Modelling of a single-phase
grid-connected PV system

In this chapter, the way in which the EMTP-based model of the PV generator can be con-
nected with grid-connected systems in the general theoretical framework will be described.
This will be done in terms of modelling a single-phase grid-connected photovoltaic (PV) sys-
tem. Nowadays, grid-connected PV systems are an emerging technology. A valuable insight
into the electrical behaviour of this kind of systems can be obtained through analytical sim-
ulation studies. In this case, existing simulation software packages are already available and
quite reliable. For a large category of studies, such as the modelling of the power converter,
most of them are well applicable. However, if someone wants to focus on the power system
analysis level, the existing software packages can not simultaneously handle, in a computa-
tionally effective way, all the distinctive features of the grid-connected PV systems [31]. Due
to the fact that, once more, in such studies a DC voltage source is mainly used instead of the
PV generator, verifies the problem.
In the following sections, each one of the basic building blocks of a single-phase grid-connected
PV system will be presented and the particularities with respect to modelling will be discussed.
Next, appropriate models will be developed in ATPDraw and will be combined with the ENRC
of the PV generator. The validity of the approach chosen, will be verified by comparing the
simulation results with published measurements. A case study will be also performed in order
to obtain current and voltage waveforms and finally harmonic distortion levels.

5-1 System configuration

In this project it was decided to use a simplified configuration of a single-phase PV system; a

system comprising out of the PV source connected through a DC/AC inverter to the utility
grid. The configuration of the inverter chosen is the central inverter mainly due to its simplic-
ity. Of course, the possibility of designing the system as a unit was an important reason that
lead towards this direction, since it offers the freedom to incorporate the MPPT controller in

Master of Science Thesis Paraskevi Breza

50 Modelling of a single-phase grid-connected PV system

the inverter in a later stage. The system has been initially built, implemented and tested in
Phoenix, Arizona, with the results presented and explained in McNeil et al.[32] and the same
configuration has also been used for simulation studies in Theocharis et al.[4]. To specify,
the system is composed of the PV generator, a DC filter, a DC/AC inverter, an isolation
transformer and the utility grid. The PV generator model is the one described in Chapter 4,
the transformer is referred to the PV side and is represented by a linear equivalent inductance
2 for both primary and secondary sides and a simplified representation of a single-phase
Ltr /αtr
bridge inverter is used. In conclusion, this functional model of a line-commutated PV system
has been chosen, since it can be easily incorporated into a transient stability program for
dynamic analysis.

5-1-1 Single-phase bridge inverter

One important application of the PV based generation, is to feed the generated power (DC)
into the utility grid (AC). For that reason, the last years PWM inverters are used, based
on gate commutated devices such as IGBT, MOSFET and GTO. However, these devices
suffer from high switching losses and their power handling capability and reliability are low,
compared to thyristor/Silicon-Controlled Rectifier (SCR) [33]. For applications in which DC
voltage has to be controlled, thyristors are mainly used, where conduction happens not only
when exceeding threshold voltage in case of forward biased, but also when a triggering signal
is applied at the gate [34].
In the case of a fully controlled single-phase bridge converter two mode operations can take
place; rectification or inversion mode. Inversion mode is equivalent to a firing angle α between
90o and 180o , while rectification mode is said to happen when the firing angle is between 0o
and 90o . The firing angle is the interval in electrical angular measure, through which the firing
pulse of the switching device is delayed by phase control, compared to natural operation that
would exist without controller elements [35]. Thus, the same converter can operate both as a
rectifier and an inverter, depending on the value of α and the polarity of the DC source. In the
case of the inversion mode, the line voltage is used for commutation, and is therefore called
line-commutated inverter [34]. Basically, a line-commutated inverter is a phase-controlled
inverter. Most commonly, the firing angle is varied up to 165o , in order to facilitate the
inversion mode for SCR. However, the most important drawback of the conventional line-
commutated inverter is the quite high value of Total Harmonic Distortion (THD), 48.3% for
square wave line current [33].
So, an appropriate presentation of a line-commutated single-phase bridge inverter has been
used, where all the switching devices had been taken into consideration as ideal switches. An
ideal switch is a device with zero resistance when closed and with infinite resistance when
opened. A scheme for a single-phase bridge inverter can be seen in Figure 5-1. A line-
commutated inverter is based on the principle of natural commutation process. When the
current in the switching device goes through zero, the device is turned off [35]. Now, according
to digital simulations, two operation modes are considered for this inverter, when none or two
switches are conducting. In every timestep of the simulation the mode of the operation has
to be tracked. This implies that in each timestep the current of any closed switch and the
voltage of any open switch are checked. In the case where the current approaches zero value,
the corresponding switch is considered turned off for the next timestep. Where, in the case

Paraskevi Breza Master of Science Thesis

5-1 System configuration 51

- -
Switch 1 Switch 3
+ +

- -
Switch 4 Switch 2
+ +

Figure 5-1: A line-commutated single-phase bridge inverter modelled with ideal switches

2 Lf
Ltr /αtr

1 3 - vpv gpv0

eg Cf vpv gpv0 I0

4 2 + i pv

AC isolation (ideal) inverter DC filter PV generator

grid transformer

Figure 5-2: A typical configuration of single-phase grid-connected PV system

that the voltage is positive and the angle of the voltage approaches the selected firing angle,
the corresponding switch starts conducting [29].

The main components of the PV system that have been mentioned previously, with the ana-
lytical representation of the ENRC of the PV generator and of the line-commutated inverter
are shown in a system model presented in Figure 5-2, as introduced also in Theocharis et
al.[29]. An example of the operation mode, in which switches 1,2 are conducting and 3, 4 are
not, is shown in Figure 5-3.

2
L t = L tr /αtr + L f

- vpv gpv0

e g / α tr Cf vpv gpv0 I0

+ i pv

Figure 5-3: System topology when 1, 2 are closed and 3, 4 are open

Master of Science Thesis Paraskevi Breza

52 Modelling of a single-phase grid-connected PV system

volt1 fireg1
volt2 MODEL v1c v3c
volt3 inverter
volt4 fireg2 fireg1 fireg2

U
I

I
v1a v3a

U(0)
+
V n:1

I
P S

T
v4c v2c
fireg2 fireg1

MODEL
pvgen
U

U
I

I
v4a v2a

Figure 5-4: Single-phase grid-connected PV system simulated with ATP/MODELS

5-2 System and component models in EMTP

In order to start the simulation of the EMTP-based system, all the components included have
to be designed. The diagram of the test system used can be shown in Figure 5-4.

5-2-1 PV generator model

The PV generator has the same structure as showed in Chapter 4 and will not be further
explained in this section.

5-2-2 Filter/Inverter models

The filter consists simply of an inductance and a capacitor. The filter connecting the generator
to the inverter serves to smooth the DC harmonics as well as to limit the rate of change of
DC current. As for the inverter, it is just two pairs of thyristors, conducting based on
the line-commutated inverter scheme presented before. Specifically, for the ATP model the
TACS-controlled TYPE 11 switches were chosen to act as thyristors, while a MODELS was
used for the control of the firing of the thyristors. The main idea behind the inverter’s
simulation is to measure the voltage difference across the thyristors and give it as input to
the MODELS, which in turn will give the firing signals as an output, which will control the
TACS thyristors. As it was mentioned before there are two conduction modes of this inverter,
so the control strategy behind it is that thyristor pair 1,2 is triggered simultaneously at a
switching angle of α, while the pair 3,4 is triggered at a delay of π radians in each cycle. The

Paraskevi Breza Master of Science Thesis

5-3 Application and results 53

Table 5-1: Data for the single-phase grid-connected PV system [32]

AC Grid
F requency = 60Hz
Vrms = 240V
Transformer
Ltr = 320µH
αtr = 1
DC Filter
Lf = 25mH
Cf = 3.3mF
PV Generator
vpv = 190V
rs = 2.15mΩ
Iph = 23.562A, G = 0.55kW/m2
Iph = 17.136A, G = 0.40kW/m2
ID = 339.5µA
W = 0.54932V −1

TYPE 11 switch in case of thyristor mode starts conducting if the voltage across it exceeds
an ignition voltage level, which in this case is set to be zero and stops conducting if the
current through it is approaching zero. In addition, the thyristors are assumed to start/stop
conducting immediately, a fact determined by a switching function on the MODELS code.
The resistances put in parallel to the thyristors are needed for the numerical stability of the
system.

5-2-3 Isolation transformer/utility grid models

The isolation transformer used is an ideal TYPE 18 transformer with unit ratio and includes
only an equivalent inductance for both primary and secondary sides. As for the utility grid,
it is represented by an ideal AC voltage source.

5-3 Application and results

The data used in the simulated system, explained above, can be seen in Table 5-1 and were
obtained from McNeil et al.[32] and also used in Theocharis et al.[4]. So, based on the data,
the performance of the system will be examined for two irradiance levels, 0.40kW/m2 and
0.55kW/m2 . One can realise that the data given in Table 5-1 do not match with the model
of the PV generator built, which takes as inputs the Voc , Isc , Impp and Vmpp of the generator.
For that reason the equations characterizing the single-diode equivalent PV model, as shown
in García et al.[36], were used in such a way that the values needed for the model used in
this project could be obtained. The equations are similar to the ones explained in Chapter 2.
More specifically, the mathematical equation describing the PV current under irradiance G,
and temperature Tα is: h i
ipv = Iph − ID eW (vpv +ipv rs ) − 1 (5-1)

Master of Science Thesis Paraskevi Breza

54 Modelling of a single-phase grid-connected PV system

Table 5-2: Data for the PV generator for different irradiance levels

PV Generator Data
G = 0.40kW/m2
Voc = 197V
Isc = 17A
Vmpp = 158V
Impp = 15A
G = 0.55kW/m2
Voc = 202V
Isc = 23A
Vmpp = 162V
Impp = 21A

where, Iph is the light-generated current and ID is the diode saturation current. W is a term
given by the following equation:
q
W = (5-2)
wKTα
where, q is the electron charge, w us a curve fitting constant and K is the Boltzmann’s
constant. By rearranging the terms of equation (5-1) a relation between ipv and vpv can be
reached, from which after some simulation trials the needed data for the EMTP-based model
of the PV generator were obtained, using also the given values of the generator found on
Table 5-1. The new data can be seen in Table 5-2. The timestep of the simulations has been
set at ∆t = 10−7 s. In order for both irradiance levels the PV voltage to settle at about 190V,
suitable calculations of the firing angle of the inverter have been made. For the irradiance
level of 0.40kW/m2 the firing angle was found to be α = 168o , where for 0.55kW/m2 was
α = 150.5o . The result can be seen in Figure 5-5. As can be noticed from Figure 5-5,
when the irradiance level changes from 0.55kW/m2 to 0.40kW/m2 the system operates under
non-continuous conduction mode, verified also in Figure 5-6. This happens as a result of
the decrease in the PV generator current, which can be seen in Figure 5-7 along with the
minimum value of the current observed at time moment tB .
In Figure 5-8 the trajectory of the operating point on the ipv − vpv plane is presented for the
period [0, tB ], where tB is the moment at which the minimum value of ipv is detected. The
point A corresponds to tA = 0s while the point B corresponds to tB , where the minimum value
of the current ipv is observed, exactly as in Figure 5-7. As can be seen, the predicted curves
using the ENRC model of the PV generator match the ipv − vpv curves given by the PV’s
manufacturer, that were calculated using equation (5-1). A small mismatch can be observed,
since the aforementioned procedure is an approximation and can not be accurate used for
extrapolating equation (5-1) to the manufacturer’s curves. Despite the small limitations, it
can be concluded that the linearised approximation of the PV generator is in good agreement
with the nonlinear model, described by equation (5-1). Figure 5-5, Figure 5-6, Figure 5-7,
Figure 5-8 are also in good agreement with the corresponding figures presented in Theocharis
et al.[4], a fact that further validates the EMTP-based model used in this section.
In Figure 5-9 the measured and predicted values of the harmonic content of the injected
current iac into the utility grid are shown. The measured values where retrieved from McNeil

Paraskevi Breza Master of Science Thesis

5-3 Application and results 55

200

180

160

140

120

0.55 kW/m2
Vpv [V]

100
0.40 kW/m2

80

60

40

20

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t [s]

Figure 5-5: The voltage vpv of the PV generator for irradiance levels of 0.40kW/m2 and
0.55kW/m2

30
0.55 kW/m2
0.40 kW/m2

20

10
Iac [A]

−10

−20

−30
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
t [s]

Figure 5-6: The current injected into the utility grid for irradiance levels of 0.40kW/m2 and
0.55kW/m2

Master of Science Thesis Paraskevi Breza

56 Modelling of a single-phase grid-connected PV system

25
0.55 kW/m2
0.40 kW/m2

20

15
Ipv [A]

10
8.777

2.309

0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
t [s]
tB=0.0493 tB=0.08262

Figure 5-7: The current ipv of the PV generator for irradiance levels of 0.40kW/m2 and
0.55kW/m2

25

Irradiation 0.55 kW/m2

20 A

15
Irradiation 0.40 kW/m2
Ipv [A]

10

8.777 A
B 195.4 V

Manufacturer’s curve
Simulation curve 2.039 A
Manufacturer’s curve 194.9 V
Simulation curve
0
0 20 40 60 80 100 120 140 160 180 200
Vpv [V]

Figure 5-8: The trajectory point of the PV generator for the time period (0, tB ) for irradiance
levels of 0.40kW/m2 and 0.55kW/m2

Paraskevi Breza Master of Science Thesis

5-3 Application and results 57

60
Measured values
Predicted values

50

40
% of fundamental

30

20

10

0
3 5 7 9 11 13
harmonic order

(a) Harmonic content for 0.40kW/m2

16
Measured values
Predicted values

14

12

10
% of fundamental

0
3 5 7 9 11 13
harmonic order

(b) Harmonic content for 0.55kW/m2

Figure 5-9: Harmonic content of the injected current for two levels of irradiance

Master of Science Thesis Paraskevi Breza

58 Modelling of a single-phase grid-connected PV system

et al.[32] and were also used in Theocharis et al.[4]. As it can be observed, the harmonic
content for both irradiance levels is extremely high. In both measured and predicted spectra
the dominance of the third harmonic is clear and for the case of irradiance level of 0.40kW/m2
the spectra show a ripple, in which the seventh harmonic is larger than the fifth harmonic.
For 0.40kW/m2 the Total Harmonic Distortion (THD) of the injected current to the AC
grid is 0.47, while for 0.55kW/m2 the THD is about 0.15. This is mainly due to the fact
that the inverter operates in non-continuous conduction mode. The predicted values are
in a good agreement with the predicted values as presented in Theocharis et al.[4]. More
specifically, comparing the results of the dominant third harmonic, the error between the
measured and predicted values in Theocharis et al.[4] is about 22.7% for 0.55kW/m2 and
12.68% for 0.40kW/m2 , while using the EMTP-based model the differences are almost the
same for both irradiance levels. These errors are observed since the inverter is considered
to be ideal and the transformer to be linear, for simplicity reasons as initially explained. Of
course, a better agreement, between the measured and predicted values, can be achieved if
more precise models for the inverter and the transformer are used.

The system designed based on the Table 5-1 and shown also in McNeil et al.[32] and Theocharis
et al.[4] is a simplified system that does not correlate with a commercial grid-connected PV
system, since both the harmonic content of the injected current into the utility grid and
the fluctuation of the current of the PV generator are very high. A study regarding the
performance of the system has been conducted in Theocharis et al.[4] for variable values of
the inductor Lf size, the capacitor Cf size and the PV voltage vpv . Along with a sensitivity
investigation of the system, conducted in McNeil et al.[32], the previous study showed that
the capacitor size does not affect the THD of the injected current, since the capacitor affects
only voltage waveforms. As for the inductor size, the influence on the THD is remarkable.
For the irradiance case of 0.55kW/m2 the THD can be significantly reduced by increasing the
inductor, since the inverter goes from below to beyond continuous conduction mode and the
current wave becomes more sinusoidal against the tendency to become a square wave [32].
This increase will also increase the power factor. So, in order to improve the THD and the
power factor a variable inductance can be used in the filter, depending on the irradiance level.
The idea behind this is to maintain the inverter as close to the continuous mode as possible
[32]. As for the fluctuation of the PV generator current ipv , it is affected by the inductor’s
current but mainly it is affected by the voltage of the PV generator vpv , which is the same
with the capacitor’s voltage. Simulation results, shown in Theocharis et al.[4], prove that
for an inductor Lf sized between 10mH and 300mH with a capacitor size Cf = 3.3mF and
vpv = 190V the big fluctuation of the current ipv can be avoided. More specifically, the ipv
fluctuation can be reduced by reducing the fluctuation of the voltage vpv , through applying
an appropriate capacitor Cf size. After simulation trials, by using various capacitor sizes,
it has been concluded that above 55mF the fluctuation is insignificant. In Figure 5-10 the
fluctuation of the current ipv can be seen, for three different sizes of the capacitor Cf . From
this graph it is clear that while varying the capacitor size between 2mF and 340mF , the
current fluctuation becomes negligible for sizes above 55mF . The results of simulating the
same system under irradiance of 0.40kW/m2 , using the 55mF capacitor and an inductor
of 25mH, with an appropriate firing angle that corresponds to a PV generator’s voltage of
vpv = 190V , can be seen in Figure 5-11. It is clear, that the voltage of the PV generator can
be considered typically constant, so indeed the fluctuation of the ipv current is negligible.

Paraskevi Breza Master of Science Thesis

5-3 Application and results 59

0.3

0.25

0.2

0.15

0.1

0.05 Ipv/Isc, Cf=3.3mF

Ipv/Isc, Cf=10mF
Ipv/Isc, Cf=55mF

0
1.5 1.51 1.52 1.53 1.54 1.55 1.56 1.57 1.58
t [s]

Figure 5-10: PV generator current ipv for different capacitor Cf sizes, between 2mF and 340mF

0.8

0.6

0.4

0.2

−0.2

−0.4

−0.6
Vpv/ Voc [V]

−0.8 Ipv/ Isc [A]

Iac / I sc [A]

−1
2.5 2.55 2.6 2.65 2.7 2.75 2.8 2.85 2.9 2.95 3
t [s]

Figure 5-11: Simulation results for the single-phase grid-connected system, using a capacitor
55mF for irradiance level of 0.40kW/m2

Master of Science Thesis Paraskevi Breza

60 Modelling of a single-phase grid-connected PV system

5-4 Maximum Power Point Tracking (MPPT)

As shown in previous sections, the output current and voltage characteristics prove the non-
linear nature of the PV generator. For a particular cell temperature and solar irradiance there
is only one optimal operating point, called the maximum power point (MPP). Under temper-
ature and irradiance variations the MPP would differ. So, the output of the PV generator
is seldom maximum and the operating point is not optimal. Thus, in order to compensate
this characteristic exhibited by the PV generator, a MPPT controller has to be incorporated,
to force the system to always operate at the maximum power point and improve the pho-
tovoltaic generation system’s efficiency. In this section, a control method for the MPPT, of
the photovoltaic system introduced in the previous section, will be presented. The steps of
designing the controller will be explained together with its simulation. The simulation results
will then be discussed and compared to the ones obtained by operating the system without
the MPPT but for maximum power operation.

5-4-1 MPPT Algorithm

As stated above the characteristics of a PV system vary with respect to the temperature and
solar irradiance. Thus, the MPPT controller is responsible to track the new modified max-
imum power point in every temperature and irradiance variation. In this project, a method
based on the slope of the power-voltage (p-v) curve has been developed, that constantly tracks
the MPP through iterative checks of the output current and voltage of the PV generator. The
MPPT can determine on which side of the power-voltage characteristic the current operat-
ing point is, in that way the current injection is indirectly controlled. In Figure 5-12 it is
clear that by analysing the power-voltage derivative one can easily determine whether the PV
generator is operating at the MPPT or far from it since:

dp
• dv > 0 for vpv < Vmpp
dp
• dv = 0 for vpv = Vmpp
dp
• dv < 0 for vpv > Vmpp

More analytically, when the operating point is moving from the left side towards the MPP
the derivative is decreasing, while in the case that it is moving from the MPP towards the left
side the derivative it is increasing. Accordingly, on the right hand side, while moving from the
MPP to the right, the derivative is decreasing, while in the opposite direction is increasing.
So, the method proposed on this project consists of climbing the operating point along the
generator characteristic to its maximum. Based on equation (5-3)

2Vm
V0 = cosα (5-3)
π
and the linear relation between the voltage and the firing angle, the operating point vpv is
adjusted with every MPPT cycle though the adjustment of its firing angle, providing the
one that corresponds to the maximum power point [34]. Furthermore, an extra feature in
the algorithm is used for monitoring the maximum and minimum values of the power-voltage

Paraskevi Breza Master of Science Thesis

5-4 Maximum Power Point Tracking (MPPT) 61

Pmax (dp/dv=0)

Power
P(dp/dv>0) P(dp/dv<0)

Voltage

dp
Figure 5-12: Operating point according to the sign of dv on the power characteristic

derivative oscillation on the PV side. Since a capacitor and inductance are used in the circuit,
in the case that the operating point is close to MPP, the influence of the ripple of voltage and
current are significant [37]. So, by simply detecting the voltage or current, the estimation
of the change in the firing angle may be inaccurate. In order to overcome this problem, a
method to control the change of the firing angle through power detection was chosen. When
the system operates in the area around the MPP the power ripple of the PV side is minimised
[37]. This characteristic can be used in order to detect in which part of the power-voltage
characteristic the system operates. So, the power oscillation is of significant value, in order to
find how close to MPP the current operating point is, and thus slow down the increment of
the firing angle so that the MPP is not crossed. A flowchart, shown in Figure 5-13, presents
the multilevel variable-step MPPT developed. It shows how the firing angle of the inverter
voltage changes, so that the operating point remains as close to the MPP as possible.
The algorithm uses the output generator voltage vpv and the output current ipv at the time
instant k and k − 1 as the input values and gives the firing angle of the inverter voltage
that corresponds to the desired MPPT. As can be seen in the flowchart, vpv and ipv are
used to calculate the current generator output power ppv (k). This value is then used to
determine the difference between ppv (k) and the value obtained from the last measurement
ppv (k − 1). Respectively, the difference between vpv (k) and vpv (k − 1) is found. The derivative
of this differences is then calculated and declared as D. Thus, the MPP can be tracked
by comparing the derivative of the power and voltage instantaneous difference and so judge
whether the system works at the MPP or at the left or the right side of the power-voltage
angle. Additionally, by analysing the derivative the operating point can be tested and checked
to determine if the generator is operating at its MPP or how far from it. The measure in
the derivative oscillation on the PV side is used to qualify the increment size, denoted as
incr, which determines the movement toward the MPP and ensures that the firing angle will
rapidly converge into the vicinity around the MPP. As soon as the operating point is close
to the MPP, the derivative should ideally be zero, although in a more pragmatic approach
should not exceed a threshold limit, as shown in Figure 5-14. If the derivative is above the
threshold limit, an adjustment has to occur by increasing the voltage, thus by increasing the
firing angle with the appropriate increment size. In the case that the derivative is below the
limit the adjustment has to occur by decreasing the voltage correspondingly. However, this
might cause a power loss which also depends on the increment size of each adjustment. If the

Master of Science Thesis Paraskevi Breza

62 Modelling of a single-phase grid-connected PV system

Begin

Read
Ipv(k-1),Ipv(k)
Vpv(k-1),Vpv(k)
Ppv(k-1),Ppv(k)

DV=Vpv(k)-Vpv(k-1)
DP=Ppv(k)-Ppv(k-1)

D=DP/DV

|D|<ε D>ε D<-ε |D|>15 |D|∈[10,15) |D|<10

incr =10 incr =1 incr =0.1

α(k+1)=α(k)+incr α(k+1)=α(k)-incr

Return

Figure 5-13: Flowchart of the MPPT algorithm

dp/dv

+ε

-ε Voltage

Figure 5-14: Power-voltage derivative oscillation limits

Paraskevi Breza Master of Science Thesis

5-5 System and component models in EMTP 63

PV generator DC filter MPPT + Transformer

Inverter

=
AC

Figure 5-15: System configuration of a single-phase grid-connected PV system with MPP tracker

operating point is very far from the threshold limits of the derivative the increment step is big
and varies regarding how far from the limits it is. The value of the increment size is system
dependent and had to be determined experimentally. The value of the oscillation limit ±ε is
system dependent too and had to be determined regarding the frequency of the system and
the timestep of the digital simulation.

5-5 System and component models in EMTP

The system configuration in the case with the MPPT remained the same, as the one pre-
sented in Figure 5-2. The slight difference in this case is that the MPPT algorithm has been
embedded in the control of the line-commutated inverter. The configuration of the single-
phase grid-connected system with MPP tracker can be seen in Figure 5-15. Regarding the
implementation of the system in ATP nothing has changed and the same one as shown in
Figure 5-4 was used. Additionally, the MPPT algorithm, as presented before, has been added
in the inverter MODEL and the TACS device, DEVICE66, has been added in order to obtain
rms characteristics of the measured values.

5-6 Application and results

Although the PV system analysed above has been designed for specific values of Voc , Isc ,
Vmpp , Impp , that correspond to unique values of irradiance and temperature one can claim
that the use of the MPPT is not required. Although, the use of the MPPT is a good proof
that the system can be successfully adjusted in new data and above all that it can be easily
expanded in a more complete system, that takes into consideration irradiance and temperature
variations. In order to verify the performance of the PV system, for each case of irradiance
level mentioned previously, the simulation results using different initial values for the firing
angle α will be presented. The validation will follow by comparing the results with the ones
obtained by simulating the PV system without the MPPT and a firing angle equal to the one
corresponding to the maximum power.
The EMTP-based PV system, shown in Figure 5-4, was used once more for the following
simulations, although with the MPPT algorithm added in the inverter control. The corre-
sponding data used are the ones used in the prototype system of the same configuration,

Master of Science Thesis Paraskevi Breza

64 Modelling of a single-phase grid-connected PV system

180

160

140

120

100
Vpv [A]

0.40 kW/m2, α=137.4o

80

60

40

20

0
0 0.5 1 1.5 2 2.5 3
t [s]

Figure 5-16: PV generator’s voltage vpv for α = 137.4o and irradiance level 0.40kW/m2

shown in Table 5-1 and Table 5-2, however with a capacitor Cf = 55mF , for irradiance levels
of 0.40kW/m2 and 0.55kW/m2 and different initial firing angle values. The timestep of the
simulations was chosen to be ∆t = 10−6 s. In addition, the MPPT algorithm is not energised
from t > 0s of the simulation and is updated every 5 cycles, in order to provide the system
some stability.

5-6-1 Simulation results with MPPT for 0.40kW/m2 irradiance level

As explained before, the first step in order to prove the validity of the proposed MPPT
algorithm is to compare the simulation results with the ones obtained from the system without
using the MPPT and for an appropriate firing angle value, that corresponds to the maximum
power of the system. Hence, after running various simulations and numerical calculations,
the firing angle corresponding to the maximum PV generator’s power Pmax of the system was
detected. Following, the system mentioned above, for irradiance level 0.40kW/m2 and a firing
angle α = 137.4o corresponding to the maximum power, is simulated. In Figure 5-16 the PV
generator’s voltage vpv is presented. As it can be seen the maximum voltage value obtained
from this simulation is about Vmax = 158.3V . In Figure 5-17 the PV generator’s current ipv
can be seen, the maximum value in this case is about Imax = 15A. In Figure 5-18 the PV
generator’s power, that corresponds to a firing angle α = 137.4o , is shown. The maximum
value is about Pmax = 2384W . The aforementioned digital simulation results are in very good
agreement with the values collected from the numerical calculations. As known, the value of
the maximum PV power Pmax should be equal to Pmax = Vmax ∗ Imax .
The next step is to run simulations for different initial values of the firing angle, smaller and
greater than the one corresponding to the MPP, in order to check the system’s behaviour and

Paraskevi Breza Master of Science Thesis

5-6 Application and results 65

20

18

16

14

12
Ipv [A]

10 0.40 kW/m2, α=137.4o

0
0 0.5 1 1.5 2 2.5 3
t [s]

Figure 5-17: PV generator’s current ipv for α = 137.4o and irradiance level 0.40kW/m2

2500

2000

1500
Ppv [W]

0.40 kW/m2, α=137.4o

1000

500

0
0 0.5 1 1.5 2 2.5 3
t [s]

Figure 5-18: PV generator’s power ppv for α = 137.4o and irradiance level 0.40kW/m2

Master of Science Thesis Paraskevi Breza

66 Modelling of a single-phase grid-connected PV system

140

135
α [degrees]

0.40 kW/m2, αo=125o

130

125
0 1 2 3 4 5 6
t [s]

Figure 5-19: Firing angle tracking for αo = 125o and irradiance level 0.40kW/m2

the effectiveness to reach the desired MPP. For that reason the system configuration, explained
in Figure 5-15, is used along with the MPPT algorithm, as presented in Figure 5-13.

The first set of simulation results with respect to an initial firing angle αo = 125o . In Figure 5-
19 the optimal firing angle tracking in respect with time is presented. As it can be seen the
tracking changes accordingly to the MPPT algorithm with changes of 0.1, 1 and 10 degrees for
every iteration, depending on how far of the MPP the current operating point is. The optimal
firing angle value of the inverter voltage, that corresponds to the maximum PV power Pmax ,
tracked by the system simulation is α = 137.4o , which is exactly the value that was expected
by the previous simulations and calculations. In Figure 5-20 the PV generator’s voltage vpv
can be seen. The maximum value obtained when the system reaches the steady state, after
almost t = 4.45s, is about Vmpp = 158V , a value that is in very good agreement with the one
obtained using the same system without the MPPT for optimum firing angle. Accordingly to
the vpv changes, the PV generator’s current ipv simulation results can be seen in Figure 5-21.
It is obvious that the current is decreasing in the time periods that the voltage is increasing
and accordingly is reaching the steady state after almost t = 4.45s. The maximum value
obtained after this moment is about Impp = 15.09A, value again in very good agreement with
the one obtained from the same system without the MPPT, running for maximum power.
Combining the data of the PV generator’s voltage vpv and current ipv , the PV generator’s
power curve was obtained and shown in Figure 5-22. The maximum value in this case was
found at Pmax = 2384.01W , as expected. The rms value of the current injected into the utility
grid iacrms , is displayed in Figure 5-23. The big fluctuations of the graphs, seen for t > 2s
that the MPPT is energised, are due to the fact that the system is initially found far from
the MPPT. From the graphs it is clear that before t = 2s the system is trying to approach a
steady state but with the MPPT energised a period for calculating the new values is passing
until the system settles to the optimal steady state.

Paraskevi Breza Master of Science Thesis

5-6 Application and results 67

180

160

140

120

100
Vpv [A]

80 0.40 kW/m2, αo=125o

60

40

20

0
0 1 2 3 4 5 6
t [s]

Figure 5-20: The PV generator’s voltage vpv , for αo = 125o and irradiance level 0.40kW/m2

20

18

16

14

12
0.40 kW/m2, αo=125o
ipv [A]

10

0
0 1 2 3 4 5 6
t [s]

Figure 5-21: The PV generator’s current ipv , for αo = 125o and irradiance level 0.40kW/m2

Master of Science Thesis Paraskevi Breza

68 Modelling of a single-phase grid-connected PV system

2500

2000

1500
0.40 kW/m2, αo=125o
Ppv [W]

1000

500

0
0 1 2 3 4 5 6
t [s]

Figure 5-22: The PV generator power’s ppv , for αo = 125o and irradiance level 0.40kW/m2

20

18

16

14

12
Iacrms [A]

10
0.40 kW/m2, αo=125o

0
0 1 2 3 4 5 6
t [sec]

Figure 5-23: The current (rms) injected into the utility grid iac , for αo = 125o and irradiance
level 0.40kW/m2

Paraskevi Breza Master of Science Thesis

5-6 Application and results 69

146

145

144

143

0.40 kW/m2, αo=145o

142
α [degrees]

141

140

139

138

137

136

135
0 1 2 3 4 5 6
t [s]

Figure 5-24: Firing angle tracking for αo = 145o and irradiance level 0.40kW/m2

The next step is to approach the MPP with an initial firing angle greater than the one
expected. So, the value is set at αo = 145o . In Figure 5-24 the optimal firing angle tracking
with respect to time is presented for this case. The tracking is following precisely the MPPT
algorithm. The optimal firing angle, that corresponds to the maximum PV power Pmax ,
tracked by the system is found at α = 137.4o , once more the value as expected. In Figure 5-
25 the PV generator’s voltage vpv can be seen. The maximum value obtained when the
system reaches the steady state, after almost t = 3.5s, is about Vmpp = 158.4V , once more
value that is in very good agreement with the one obtained from the same system without
the MPPT and for maximum power. Following, the PV generator’s current ipv curve can be
seen in Figure 5-26. The maximum value, obtained after this point, is about Impp = 15.05A,
a value really close to the one obtained from the same system without the MPPT. The PV
generator’s power curve is shown in Figure 5-27. The maximum value in this case is found
to be Pmax = 2383.95W , a value which agrees with the one expected. The rms value of the
current injected into the utility grid iacrms , is displayed for this case in Figure 5-28.
Here, the MPPT algorithm is energised for t > 2s as well and the firing angle drops 10
degrees exactly after the MPP starts the tracking. That means that the system is initially
found operating at a point far from the MPPT. This causes the spikes seen in the graphs,
especially in Figure 5-27 and in Figure 5-28 at the time instant after t = 2s.

5-6-2 Simulation results with MPPT for 0.55kW/m2 irradiance level

Following the steps explained in the previous section, for the irradiance level of 0.55kW/m2
the firing angle corresponding to the maximum PV generator power Pmax of the system is
calculated, after running various simulations and numerical calculations. Thus, the same
system as explained previously, for irradiance level 0.55kW/m2 and an optimum firing angle

Master of Science Thesis Paraskevi Breza

70 Modelling of a single-phase grid-connected PV system

180

160

140

120

100 0.40 kW/m2, αo=145o

Vpv [A]

80

60

40

20

0
0 1 2 3 4 5 6
t [s]

Figure 5-25: The PV generator voltage vpv , for αo = 145o and irradiance level 0.40kW/m2

20

18

16

14

12
Ipv [A]

10
0.40 kW/m2, αo=145o
8

0
0 1 2 3 4 5 6
t [s]

Figure 5-26: The PV generator’s current ipv , for αo = 145o and irradiance level 0.40kW/m2

Paraskevi Breza Master of Science Thesis

5-6 Application and results 71

2500

2000

0.40 kW/m2, αo=145o

1500
Ppv [W]

1000

500

0
0 1 2 3 4 5 6
t [s]

Figure 5-27: The PV generator’s power ppv , for αo = 145o and irradiance level 0.40kW/m2

60

40

20
Iacrms [A]

0
0.40 kW/m2, αo=145o

−20

−40

−60
0 1 2 3 4 5 6
t [sec]

Figure 5-28: The current (rms) injected into the utility grid iac , for αo = 145o and irradiance
level 0.40kW/m2

Master of Science Thesis Paraskevi Breza

72 Modelling of a single-phase grid-connected PV system

180

160

140

120 0.55 kW/m2, α=138.5o

100
Vpv [A]

80

60

40

20

0
0 0.5 1 1.5 2 2.5 3
t [s]

Figure 5-29: PV generator’s voltage vpv for α = 138.5o and irradiance level 0.55kW/m2

α = 138.5o , is simulated. In Figure 5-29 the PV generator’s voltage vpv curve is shown. As
it can be seen the maximum voltage value obtained from this simulation is about Vmax =
163.51V . In Figure 5-30 the PV generator’s current ipv is presented, where the maximum
value was detected at about Imax = 20.82A. In Figure 5-31 the PV generator’s power, that
corresponds to a firing angle α = 138.5o , is shown. As can be seen the maximum value
is Pmax = 3405.4W . All the above simulation results are in excellent agreement with the
numerical calculations computed initially.
The next step to follow, with the same settings, is to run simulations for different initial values
of the firing angle, smaller and greater than the one corresponding to the MPP, in order to
check the system’s performance. The same system configuration, explained in Figure 5-15, is
used once more.
The first set of simulation results is obtained by setting an initial firing angle αo = 130o . In
Figure 5-32 the optimal firing angle tracking is presented. As it can be seen, also in this set
of simulations, the tracking is strongly following the MPPT algorithm with changes of 0.1, 1
and 10 degrees for every iteration, depending on how far of the MPP the current operating
point is. The optimal firing angle value of the inverter voltage, that corresponds to the
maximum PV power Pmax , tracked by the system simulation is α = 138.5o , which is exactly
the value that was expected by the previous simulations and calculations. In Figure 5-33
the PV generator’s voltage vpv can be seen. The maximum value obtained when the system
reaches the steady state, after almost t = 3.55s, is about Vmpp = 163.52V , a value that is
in very good agreement with the one obtained from the same system without the MPPT
for maximum power, as explained at the beginning of the section. Respectively, the PV
generator’s current ipv simulation results are presented in Figure 5-34. It is clear that the
current is decreasing in the time periods that the voltage is increasing and accordingly is
reaching the steady state almost after t = 3.55s. The maximum value obtained after this

Paraskevi Breza Master of Science Thesis

5-6 Application and results 73

25

20

15
Ipv [A]

0.55 kW/m2, α=138.5o

10

0
0 0.5 1 1.5 2 2.5 3
t [s]

Figure 5-30: PV generator’s current ipv for α = 138.5o and irradiance level 0.55kW/m2

3500

3000

0.55 kW/m2, α=138.5o

2500

2000
Ppv [W]

1500

1000

500

0
0 0.5 1 1.5 2 2.5 3
t [s]

Figure 5-31: PV generator’s power ppv for α = 138.5o and irradiance level 0.55kW/m2

Master of Science Thesis Paraskevi Breza

74 Modelling of a single-phase grid-connected PV system

140

139

138

137

136
α [degrees]

135
0.55 kW/m2, αo=130o

134

133

132

131

130
0 1 2 3 4 5 6
t [s]

Figure 5-32: Firing angle tracking for αo = 130o and irradiance level 0.55kW/m2

180

160

140

120

100 0.55 kW/m2, αo=130o

Vpv [A]

80

60

40

20

0
0 1 2 3 4 5 6
t [sec]

Figure 5-33: The PV generator’s voltage vpv , for αo = 130o and irradiance level 0.55kW/m2

Paraskevi Breza Master of Science Thesis

5-6 Application and results 75

25

20

15
0.55 kW/m2, αo=130o
Ipv [A]

10

0
0 1 2 3 4 5 6
t [s]

Figure 5-34: The PV generator’s current ipv , for αo = 130o and irradiance level 0.55kW/m2

moment is about Impp = 20.82A, a value in excellent agreement with the one explained at the
beginning of the section. Subsequently, the PV generator’s power curve is obtained and shown
in Figure 5-35. The maximum value in this case is found at Pmax = 3405.5W , as expected.
The rms value of the current injected into the utility grid iacrms , is displayed in Figure 5-36.
In this simulation, the MPPT algorithm is again energised for t > 2s and the firing angle
increases 10 degrees exactly after the MPP starts tracking. So, the system is initially found
operating at a point far from the MPPT, which causes the big spike seen in Figure 5-36 at the
time instant exactly after t = 2s. In addition, for this simulation case the transient regime is
longer while the system is trying to settle in a steady state, for the period of time prior to the
MPPT energisation, compared to the simulation cases of the previous section. During this
period only the inverter controller is active controlling the firing angle. As it is known, every
controller parameter is tuned for one range of values, which in this case is the PV generator’s
power. When the same controller is used for another range of values it may lead to some
overshoot or damping and the settling time may increase or decrease slightly, compared to the
PV power for which it is tuned. In conclusion, the inverter controller is tuned for the values
corresponding to the previous section, causing this transient response prior to the MPPT
energisation.

Similarly to the case shown above, the following action is to approach the MPP with an initial
firing angle greater than the one expected. So, the value is set at αo = 150o . In Figure 5-37
the optimal firing angle tracking in respect with time is presented for this case. The tracking
is following precisely the MPPT algorithm. The optimal firing angle, that corresponds to the
maximum PV power Pmax , tracked by the system is found at α = 138.5o , as expected. The
demonstration of the PV generator’s voltage vpv can be found in Figure 5-38. The maximum
power point, obtained when the system settles in the steady state almost after t = 3.55s of
transient regime, is tracked at Vmpp = 163.6V voltage, once more value that is in very good

Master of Science Thesis Paraskevi Breza

76 Modelling of a single-phase grid-connected PV system

3500

3000

0.55 kW/m2, αo=130o

2500

2000
Ppv [W]

1500

1000

500

0
0 1 2 3 4 5 6
t [s]

Figure 5-35: The PV generator’s power ppv , for αo = 130o and irradiance level 0.55kW/m2

30

25

20
Iacrms [A]

15

10

0.55 kW/m2, α o=130o

0
0 1 2 3 4 5 6
t [sec]

Figure 5-36: The current (rms) injected into the utility grid iac , for αo = 130o and irradiance
level 0.55kW/m2

Paraskevi Breza Master of Science Thesis

5-6 Application and results 77

150

148

146 0.55 kW/m2, αo=150o

α [degrees]

144

142

140

138

0 1 2 3 4 5 6
t [s]

Figure 5-37: Firing angle tracking for αo = 150o and irradiance level 0.55kW/m2

agreement with the one obtained using the same system without the MPPT for maximum
power. The PV generator’s current ipv curve is shown in Figure 5-39. The maximum value,
at which the MPP is tracked after the transient regime, is about Impp = 20.79A. The PV
generator’s power curve is presented in Figure 5-40. The maximum value in this case is found
Pmax = 3405.48W , value in accord with the one expected. The rms value of the current
injected into the utility grid iacrms , is displayed for this case in Figure 5-41. The conclusions
of this simulation are quite similar to the above mentioned case. The MPPT algorithm is
again energised for t > 2s and the firing angle abruptly decreases exactly after the MPP starts
the tracking. So, the system is initially found operating at a point far from the MPPT and
this causes the big spike clearly seen in Figure 5-41 at the time instant exactly after t = 2s.

After running all simulations, explained in the previous sections, we can conclude that the
system responds to the changes of the firing angle successfully. It can predict, with a quite
good accuracy, the value that corresponds to the maximum PV power. All the simulation
results, as previously mentioned, coincide with the results of the numerical calculations but
also are really close to the Voc , Isc , Vmpp , Impp values given as data for each simulated PV
generator. Small differences might occur due to numerical errors at the simulations, which
associate directly with the value of the chosen timestep. In addition, the MPPT algorithm is
built in a way that it can track the optimal operating point that is within between accepted
limits. So, the predicted values can not always be in full agreement with the given data and a
small error is expected. It has to be mentioned again that the implementation of the MPPT
in this system can prove that the user can have great controllability of the system and the
operating point for every time instant can be detected on the ipv − vpv and ppv − vpv curves,
regardless of the circumstances.

Master of Science Thesis Paraskevi Breza

78 Modelling of a single-phase grid-connected PV system

200

180

160

140

120
0.55 kW/m2, α o=150o
Vpv [A]

100

80

60

40

20

0
0 1 2 3 4 5 6
t [s]

Figure 5-38: The PV generator’s voltage vpv , for αo = 150o and irradiance level 0.55kW/m2

25

20

15
Ipv [A]

10
0.55 kW/m2, αo=150o

0
0 1 2 3 4 5 6
t [s]

Figure 5-39: The PV generator’s current ipv , for αo = 150o and irradiance level 0.55kW/m2

Paraskevi Breza Master of Science Thesis

5-6 Application and results 79

3500

3000

2500

2000
0.55 kW/m2, αo=150o
Ppv [W]

1500

1000

500

0
0 1 2 3 4 5 6
t [sec]

Figure 5-40: The PV generator’s power ppv , for αo = 150o and irradiance level 0.55kW/m2

50

45

40

35

30
Iacrms [A]

25

20

15

10
0.55 kW/m2, αo=150o
5

0
0 1 2 3 4 5 6
t [s]

Figure 5-41: The current (rms) injected into the utility grid iac , for αo = 150o and irradiance
level 0.55kW/m2

Master of Science Thesis Paraskevi Breza

80 Modelling of a single-phase grid-connected PV system

5-7 Model optimisation

As we saw in previous sections, the characteristics of the PV generator depend on the varia-
tions of the cell temperature Tc and irradiance G, which actually affect the Isc , the Voc and
consequently change the power ppv of the generator. As explained before, in order to opti-
mise the ENRC based PV generator model and take into consideration the irradiance and
temperature effect, four new equations had to be added in the generator model. Equations
(4-24), (4-25), (4-26) and (4-27) were used, so that the new data of the PV generator could be
determined for any irradiance and temperature change. In this section, the proposed MPPT
method and the optimisation scheme of the PV generator will be investigated and verified
under rapid variations of the solar array temperature and the solar irradiance.

5-7-1 Simulation and results

For the ATP simulation the system, shown in Figure 5-4, was chosen for the reference irra-
diance level of 0.40kW/m2 and an initial firing angle of αo = 110o . The data from Table 5-2
we taken as reference values, since the data for STC were not available in McNeil et al.[32],
where the same system was initially used and presented. For defining the irradiance and
temperature factors and functions that are needed for the equations shown above, the table
given in Marion et al.[30] was again used, with the data corresponding to the specific PV
generator highlighted as also shown in Table 4-3. The system was simulated for a frequency
fr = 60Hz and a timestep ∆t = 10−6 s. The approach used in order to demonstrate the
performance of the system was to make a choice box in the PV generator model, as can be
seen in Figure 5-42. The choice box, called "Scenario", corresponds to 4 possible choices:

• Scenario=1 −→ No variations

• Scenario=2 −→ Irradiance variations

• Scenario=3 −→ Temperature variations

• Scenario=4 −→ Irradiance and temperature variations

To examine the performance of the system all four scenarios were simulated and the results
will be presented and explained in the following section for each case. However, scenario 1
will not be demonstrated, since it coincides with the results presented in previous sections.

Scenario 2

With increasing irradiance, the current Isc increases almost linearly, while the voltage Voc
slightly increases and thus the MPP varies [7]. When MPP changes due to irradiance varia-
tions, the MPP tracking in dynamic conditions is then analysed using staircase or trapezoidal
irradiance profiles. In most of the cases, trapezoidal irradiance profiles are used to test the
inverter in dynamic irradiance conditions [38]. Figure 5-43 shows the trapezoidal irradiance
profile, chosen in this case. The output power, voltage, current and firing angle trajectories
are shown in Figure 5-44. For this simulation the MPPT was energised for t > 0s. The tran-

Paraskevi Breza Master of Science Thesis

5-7 Model optimisation 81

Figure 5-42: The choice box in the PV generator model

1000

900

800
G [W/m2]

700

600

500

400
0 1 2 3 4 5 6 7 8 9
t [s]

Figure 5-43: Trapezoidal irradiance profile for the dynamic MPPT efficiency test

Master of Science Thesis Paraskevi Breza

82 Modelling of a single-phase grid-connected PV system

140

6000

135

5000

130
4000

α [degrees]
Ppv [W]

125
3000

120
2000

1000 115

0 110
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9
t [s] t [s]

(a) PV generator power Ppv (b) Firing angle α

180 45

160 40

140 35

120 30

100 25
Vpv [V]

Ipv [A]

80 20

60 15

40 10

20 5

0 0
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9
t [s] t [s]

(c) PV generator voltage Vpv (d) PV generator current Ipv

Figure 5-44: Trapezoidal irradiance profile test results

Paraskevi Breza Master of Science Thesis

5-7 Model optimisation 83

sient response of the system, clearly shown in the first seconds of the simulation, is mainly due
to numerical instability, since the system data have to change directly regarding the abrupt
irradiance change from 0.40kW/m2 to 1kW/m2 . The system values at the steady state are
found about vpv = 164.3V , ipv = 34A, ppv = 5579.5W while the firing angle settles in about
α = 138.2o .
In order to validate this performance one has to carefully examine the efficiency of a PV
module and how this is defined. In general, the conversion efficiency of a PV module is the
proportion of the sunlight energy, that is received and is converted to electrical energy. More
specifically, is the ratio between the module’s output and incident light power [7]:

Pout Vpv · Ipv

η= = (5-4)
Pin Apv · G

where Apv is the solar module surface and G the irradiance.

To be more accurate the "true" efficiency of the panel is the multiplication of several factors,
such as the junction temperature increase, the power losses by Joule effect in the cables, the
factor due to losses in the inverter and the factor related to the MPPT. Only one of the
factors is the efficiency presented in equation (5-4) and is the one best suited in this system,
since for simplicity reasons the rest of the factors were omitted in this project.
However, in the case examined in this project the surface of the PV generator remains the
same through the whole simulation and the efficiency of the generator is considered to be
constant. Thus, from equation (5-4) and the system characteristics, it is now clear that the
power output ppv of the generator is expected to be proportional to the irradiance G. This
is actually proved and presented in Figure 5-44. So, in the case of a trapezoidal irradiance
variation, that can occur in the case that the sun is temporarily shadowed, the MPPT tracking
is working properly. This validates the good design of the MPPT algorithm and of course the
optimisation scheme proposed.

Scenario 3

While running simulations for this case, the system has to adapt in temperature variations.
With an increase in temperature the rate of photon generation increases, hence the reverse
saturation current increases and this causes a decrease in the band gap. As a result, PV
current slightly changes while this leads to major changes in the voltage. Temperature in
total affects negativity the solar cell performance [7]. Figure 5-45 shows the temperature
profile, chosen in this case. The digital simulation results are shown in Figure 5-46. For this
simulation the MPPT was active for t > 1s. At the steady state the system roughly settles
at vpv = 160.7V , ipv = 15.09A, ppv = 2422.9W and the optimal firing angle at α = 138o . As
it can be seen in the graphs, the settling to the steady state regime, follows the incremental
change of the firing angle towards its final value.
In order to validate the performance of the system, while simulating the given temperature
profile, numerical calculations have been simulated also in Matlab using equations (4-24),
(4-25), (4-26) and (4-27), so that the results could be compared. The outcome of the compar-
ison shows clearly that the digital EMTP-based simulations can successfully deliver the results
expected from the numerical calculations. Unfortunately, the results could not be evaluated

Master of Science Thesis Paraskevi Breza

84 Modelling of a single-phase grid-connected PV system

25

20

T [oC]

15

10
0 1 2 3 4 5 6 7 8 9
t [s]

Figure 5-45: Temperature profile for the dynamic MPPT efficiency test

using the temperature coefficients for power, Voc and Isc normally given in the PV datasheet.
The PV datasheet for this case was not available and the results were not comparable with
the coefficients of the most recently used PV’s, since the data used for this system and found
also in McNeil et al.[32] can be considered outdated.
To conclude, from the graphs presented in this case it can be confirmed that the effect of
the temperature variation in the PV output voltage is quite significant. In addition the
results show excellent agreement with the numerical calculation results. So, also in the case
of temperature variation the MPPT tracking is working properly. This validates the good
design of the MPPT algorithm and of course the optimisation scheme proposed in this case.

Scenario 4

The choice 4 of the scenario box corresponds to both irradiance and temperature variations. In
this case, the profiles used in the previous sections and showed in Figure 5-43 and Figure 5-45
are again chosen.
The digital simulation results are shown in Figure 5-47. For this simulation the MPPT is
activated for t > 0s, so the transient response of the system at the first seconds of the
simulation is quite clear. The system values at the steady state were found roughly at about
vpv = 167.1V , ipv = 33.8A, ppv = 5663.3W and the maximal firing angle at α = 139.4o .
As in the previous case, the validation of the digital simulations came from the good agreement
with the results of numerical calculations in Matlab. Finally, from the graphs presented it can
be confirmed that the effect of the irradiance variation effect plays a significant role compared
to the temperature variation, since it is clear that the output current changes almost linearly
with the change in the irradiance. So, also in the case of irradiance and temperature variation
the MPPT tracking is working properly. This validates once more the MPPT algorithm and
the optimisation scheme.
Since the value of the increment size in the MPPT algorithm is system dependent, in all the
above mentioned cases had to be determined experimentally. The value of the oscillation

Paraskevi Breza Master of Science Thesis

5-7 Model optimisation 85

140
2500

135

2000

130

1500

α [degrees]
Ppv [W]

125

1000
120

500
115

0 110
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9
t [s] t [s]

(a) PV generator power Ppv (b) Firing angle α

180 20

160 18

16
140

14
120

12
100
Vpv [V]

Ipv [A]

10

80
8

60
6

40
4

20 2

0 0
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9
t [s] t [s]

(c) PV generator voltage Vpv (d) PV generator current Ipv

Figure 5-46: Temperature profile test results

Master of Science Thesis Paraskevi Breza

86 Modelling of a single-phase grid-connected PV system

6000
140

5000 135

4000 130

α [degrees]
Ppv [W]

3000 125

2000 120

1000 115

0 110
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9
t [s] t [s]

(a) PV generator power Ppv (b) Firing angle α

180 45

160 40

140 35

120 30

100 25
Vpv [V]

Ipv [A]

80 20

60 15

40 10

20 5

0 0
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9
t [s] t [s]

(c) PV generator voltage Vpv (d) PV generator current Ipv

Figure 5-47: Irradiance and temperature profile test results

Paraskevi Breza Master of Science Thesis

5-7 Model optimisation 87

limits ±ε is also system dependent and had to be determined regarding the frequency of the
system and the timestep of the digital simulation for each case.
At this point, one might object regarding how realistic the abrupt changes in irradiance and
temperature are. For this reason, it should be made clear that the goal of this optimisation
proposal was to prove the controllability and good design of the MPPT algorithm and the
accuracy on delivering correct results of the proposed system.

Master of Science Thesis Paraskevi Breza

88 Modelling of a single-phase grid-connected PV system

Paraskevi Breza Master of Science Thesis

 Chapter 6

Case studies using hybrid PV/battery

grid-connected system

As stated in Chapter 5, the implementation of PV distributed generation systems, such

as PV’s connected to the utility grid, has increased drastically in the last decades mainly
due to the need of supplying the rising demand for electric power worldwide. However the
disadvantage of PVs is their intermittent generation due to big changes in weather conditions.
In order to overcome these disadvantages, energy storage elements can be used. Due to
the mature technology battery energy storage systems are widely used to maintain the PV
generation output stable and reliable, enhancing both steady and dynamic behaviours of the
whole system [39]. Following this line, this chapter presents the way that the EMTP-based
model of the PV generator, presented in previous chapters, can be associated with three-
phase grid-connected systems and more specifically with a hybrid PV/battery three-phase
grid-connected system. For simplicity reasons an established system, built and presented
as part of the past graduation project Alberto[40], simulated using MATLAB/Simulink, is
used so that the EMTP-based models can be tested and validated in a three-phase scenario.
Another reason for choosing this system is that in general MATLAB/Simulink models are
appropriate for studies regarding the performance of hybrid systems. In addition, dynamic
operation and control system strategies can be easily incorporated in an existing model to
facilitate studies on the overall performance of the system [39].
In the following sections the basic building blocks of a three-phase grid-connected PV system
will be presented and analysed. Then, the models developed in MATLAB/Simulink will be
combined with the EMTP-based models, explained previous chapters. In order to verify the
proposed system four different case studies will be performed and the dynamic performance
of the system will be tested and analysed.

6-1 System configuration

The configuration of the system can be seen in Figure 6-1. The system will operate in dispatch
mode. This mode can be used in real life applications for purpose of power contract with

Master of Science Thesis Paraskevi Breza

90 Case studies using hybrid PV/battery grid-connected system

Solar Array

Load

Battery
Electric Grid

Figure 6-1: Hybrid PV/battery system configuration

utility [41]. In this mode power is used to compensate power mismatch between the PV
array and the dispatched amount. In dispatch mode the battery might experience frequent
shifts between charging and discharging mode [41]. As can be seen the battery energy storage
can charge or discharge to help maintain the balance between the PV generation and loads
demand. When the generation exceeds the demand the PV generation will charge the battery
in order to store the excess power, while in case of PV generation shortage the battery will
discharge and provide the stored power to supply the grid.

6-2 System and component models in MATLAB/Simulink

In this section all the components of the system, as modelled in the project shown in
Alberto[40], will be introduced along with the modifications introduced in this project. The
model of the system in Simulink can be seen in Figure 6-2.

6-2-1 PV generator and MPPT model

Since the core idea of this project is the EMTP-based model of the PV generator, the model
that has been explained in the previous chapters is used for this study as well. The results
extracted in an Excel file format are inserted to Simulink and combined with the rest of the
system in terms of a function. In order to extract the values the single-phase grid-connected
model with the MPPT, as analytically explained in the previous chapter, is used.

6-2-2 DC/DC boost converter model

The EMTP-model as simulated in the ATPDraw interface includes already the MPPT al-
gorithm. For that reason the MPPT is omitted from the Simulink model of the PV array.
As for the DC/DC converter the boost configuration has been chosen for its better dynamic
performance. The configuration of the boost converter is the same as the one presented in
Figure 3-7. The inductance of the converter is set at 0.85mH, the input capacitance at
11.7mF and the output capacitance at 24mF [40]. The new Simulink model including the

Paraskevi Breza Master of Science Thesis

6-2 System and component models in MATLAB/Simulink 91

PV Plant Vabc_S Vabc

Iabc_S Iabc
pulse
Vdc

0 Qg*
+ +v Q*1
- Grid_control
C1
+

IBGT Inverter1
A

C
g

B
b

c
a
A

C
B

VI_S
aA

cC
bB

Transformer 690/10kV
Aa Aa A
a A
Bb B B N
Bb b C
Cc N C
Cc c
VI_g
VI_w
A

C
B

1e6 Fo=12Hz
Load Pg* Udc Pwsgb_100.mat
Pref A Fo=12Hz
Add Pws Terminator5 To File1
0 Qg* Fo=17Hz
Pg
Q*2 B Fo=12Hz
Pws Pb
+
Fo=20Hz
Qws
DC Voltage Source
C Fo=30Hz
- Qg
Fo=20Hz
DC/AC Qb

Terminator6

Figure 6-2: Three-phase grid-connected hybrid PV/battery system simulated with Simulink

Master of Science Thesis Paraskevi Breza

92 Case studies using hybrid PV/battery grid-connected system

S cope3
S cope1 Pulse
Generator
m 1 V+
+ -i a
+ k + -i
Current Measurement
Diode Current Measurement1
L1

g
1
+
+ v

+
- I deal Switch
Ci Co
Voltage Measurement

2
s

EMTP model -K- 2

From Gain V-
Workspace

S cope2

Figure 6-3: New PV model modelled in Simulink

function representing the EMTP-based PV and MPPT outputs along with the boost converter
is shown in Figure 6-3. The input of the converter is the voltage output of the EMTP model,
the boost converter will change this input into a DC voltage of different value depending also
on the pulse that turns the converter bridge on and off. The desired duty cycle, necessary to
boost the input voltage to the expected value and maintain it as constant as possible, was
determined empirically and the pulses are given in the new model through a pulse generator.

6-2-3 DC/AC inverter model

The output of the PV generator model is connected to a DC/AC inverter with IGBTs, con-
trolled using Space Vector Pulse Width Modulation (SVPWM). The inverter maintains the
voltage at the DC side at about 2000V and controls the reactive power flow to the grid [40].
The d-axis (direct) reference current is actually determined by the DC voltage PI controller,
to control the converter’s output real power. The reactive power is maintained at zero by the
control of the q-axis (quadrature) as explained and presented in Alberto[40].

6-2-4 Battery system/Load models

A three-phase parallel RLC load has been used in a Y grounded configuration. As for the
battery system, even though most of them in literature are modelled accounting the bat-
tery’s internal parameters, such as the internal resistances, capacitances and the State of
Charge (SOC), in Alberto[40], where the system is simulated for a very short time period,
the battery is considered to have very small variations so that the internal characteristics do
not significantly influence the voltage and current of the battery. Based on this assumption
the battery is modelled as a DC constant voltage source of 1500V and the power output is
controlled by a DC/AC inverter, that is connected to the grid [40]. The system simulated
in Alberto[40] is a 2MW/0.5MWh system based on the model of A123 company, composed
by Li-Ion batteries. The system’s response is in the range of 120ms. As for the controller
of the system, it is designed to control the active and reactive power outputs of the system
based on SVPWM, aligning the voltage vector with the d-axis and setting q-axis component

Paraskevi Breza Master of Science Thesis

6-3 Simulation and results 93

to zero. That way, the d-axis current controls the active power, while the reactive power is
controlled by the q-axis current. The reference active power is set at zero and the reference
active power is described as the difference between the desired power that has to be injected
to the grid and the difference between the solar power and the load [40]:

Pbatt = Pset − (Psolar − Pload ) (6-1)

where Pbatt is the battery’s power, Pset is the desired power, Psolar is the power output of the
PV system and Pload is the power consumed by the load.

6-2-5 Transformer/utility grid models

The PV generator and battery systems are both connected to the utility grid through a
RL series branch of 0.5mH and 1mΩ. As for the transformer, a three-phase two winding
690V/10kV model is used. The grid in this case is represented by a 10kV line-to-line constant
voltage source and constant frequency of 50Hz.

6-3 Simulation and results

Based on the above explained models and control methods four case studies, similar to the
ones presented in Alberto[40], are studied in this project, so that a comparative study can be
developed and the behaviour of the EMTP-based PV generator can be tested in a three-phase
system scenario.
As explained before the EMTP-based model of the PV generator is used in order to extract
the values and insert them into the Simulink model. So, for the case studies of this chapter
the data of the PV generator, as used in Alberto[40] are used as inputs for the simulations
in the ATP interface, namely the Solarex MSX 60 datasheet values under STC. This module
consists out of 36 cells in series and its electrical characteristics can be seen in Table 6-1.
For the simulations in Alberto[40] project a number of 800 series and 54 parallel modules
were used in order to build a solar plant of about 2.3MW. The values of Table 6-1 are used
accordingly to the specifications of each case study as inputs for the EMTP-based model, in
order to extract the characteristic values of the PV array and insert them in the Simulink
model. It is clear that for this study the PV generator is working as a PV array.

Table 6-1: Typical electrical characteristics of Solarex MSX 60 PV module [42]

Mamimum power (Pmax) 60W

Vmpp 17.1V
Impp 3.5A
Guaranteed minimum Pmax 58W
Isc 3.8A
Voc 21.1V

Master of Science Thesis Paraskevi Breza

94 Case studies using hybrid PV/battery grid-connected system

Table 6-2: Data for the three-phase grid connected PV array

Vmpp 923.4V
Impp 3040A
Isc 3040A
Voc 1139.4V

Case study 1

For the case study 1, by recalculating the values of the series and parallel modules and by
using the data of Table 6-1, a PV array with an output power of about 3MW and constant
atmospheric conditions is simulated. The load is set to be 1MW while the total power is set
to dispatch 2MW to the grid. The results of the simulation running for t = 6s can be seen in
Figure 6-4. The results are in good agreement with the graph presented in Alberto[40]. The
battery operates in charging mode when there is surplus generation and turns to discharging
mode in case of generation shortage.

Case study 2

The number of modules is once more adjusted in order to result in a PV array power of about
2MW. The load power for this case study remains at 1MW and the design of the system
is carried out for feeding 2MW power to the grid. The results of the simulation running
again for t = 6s can be seen in Figure 6-5. Once more the graph is in agreement with the
corresponding graphs presented in Alberto[40]. In addition, it is clear that the system follows
equation (6-1).

Case study 3

this simulation is running after increasing the load power to 3.5MW and maintaining the
solar power at the same levels as explained in case study 1. The results of the simulation
are presented in Figure 6-6. In this case, it is obvious that the assumption that the SOC of
the battery will not be taken into account for such a small time interval, leads to the battery
system not being able to absorb or provide power of more than 2MW. So, from the graphs we
can see that when the load consumes more power than the PV array can deliver the demand
of the grid cannot be met and the power drops, a case that can cause stability problems in
real life situations.
From the simulation results of case study 3 it can be concluded that for this system to work
properly a very good forecast of the load power and the weather conditions should be made,
so that the active power requested from the battery system does not exceed the amount of
power of 2MW. In addition, it has to be clear that the objective of these case studies is
to keep the grid power constant, an approximation that does not correspond with real life
conditions. Also, the grid has been simulated as a constant voltage source, which stands far
from the reality where the voltage and the frequency of the system can vary significantly.
The system can be improved, by modelling the battery system taking into account the SOC,

Paraskevi Breza Master of Science Thesis

6-3 Simulation and results 95

and by controlling the voltage and frequency of the utility grid by regulating the active and
reactive power injected by the battery system.
At this point it has to be clear that the transient regime lasting the first two seconds is mainly
due to the fact that the MPPT in the EMTP-based model is energised for t > 2s. In addition,
the transient response of the PV array shows similar behaviour as in the previous chapters.
The initial mismatches are mainly due to the fact that the modelling of the EMTP-based PV
generator starts the simulations for Isc causing the voltage to start from zero and thus the
power generation to start from zero. Due to the limitations of the Simulink system the initial
mismatches can be seen in the graphs.

Case study 4

Assuming the case presented in scenario 2 in Section 5-7, where the solar irradiance changes,
a new case study is investigated. As explained in previous chapters, the I-V characteristic of
the PV array is dependent on the solar irradiance changes and changes with these variations.
Accordingly the MPPT controller is following these variations in order to make the PV always
generate the maximum power [39]. The profile of the irradiance is the same as the one
presented in Figure 5-43. During the time period t = 0 − 2s the irradiance level is 1000W/m2 ,
for t = 2 − 4s it decreases at 800W/m2 , for t = 4 − 5s it increases at 900W/m2 and from
t = 6s and on, the irradiance remains constant at 900W/m2 . For this case study the load
power is again set at 1MW. Figure 6-7 shows the active and reactive power flux .
With the power generated from the PV array not always equal to the power demands, the
battery will charge and discharge to compensate this power unbalance. In this case the MPPT
algorithm built in this project is energised for t = 1s and as can be seen only for an irradiance
level of 1000W/m2 the solar power can sufficiently cover the demand of the load and the grid.
For the other two levels the battery system has to discharge and supply the power demanded.
As a conclusion for the aforementioned case studies it can be stated that the system shows
good performance under dispatch operation. The battery operates in charging mode when
there is surplus power and turns to discharging state when there is generation shortage.
Battery’s power generation or absorption equals the amount by which the PV generation is
subtracted from the dispatched power.

Master of Science Thesis Paraskevi Breza

96 Case studies using hybrid PV/battery grid-connected system

6
x 10
6
Psolar−Pload
Pgrid
5
Pbattery

3
Power[W]

−1

−2

−3
0 1 2 3 4 5 6
t[s]

Figure 6-4: System power simulation results of case study 1

6
x 10
3

0
Power[W]

−1

−2

−3 Psolar−Pload
Pgrid
Pbattery

−4
0 1 2 3 4 5 6
t[s]

Figure 6-5: System power simulation results of case study 2

Paraskevi Breza Master of Science Thesis

6-3 Simulation and results 97

6
x 10
3

−1
Power[W]

−2

−3

−4
Psolar−Pload
Pgrid
−5
Pbattery

−6
0 1 2 3 4 5 6
t[s]

Figure 6-6: System power simulation results of case study 3

6
x 10
5
Psolar−Pload
Pgrid
Pbattery
4

2
Power[W]

−1

−2
0 1 2 3 4 5 6 7 8 9
t[s]

Figure 6-7: System power simulation results of case study 4

Master of Science Thesis Paraskevi Breza

98 Case studies using hybrid PV/battery grid-connected system

Paraskevi Breza Master of Science Thesis

 Chapter 7

Conclusions and future work

In this chapter the main conclusions extracted from this study, as well as recommendations
for future researchers are presented.

7-1 Conclusions

As stated in the project objective, a PV generator EMTP-based model is proposed in this

report based on the linearisation of the current-voltage characteristic. A nonlinear character-
istic of current vs. voltage of a PV generator should be modelled by at least two exponential
factors, to be simulated precisely. However electromagnetic transient programs cannot handle
this nonlinearity. With the proposal in this project all the restrictions of using a PV model
within an electromagnetic transient software can be overcome.
In order to study the dynamic behaviour of the proposed PV generator model first a study of
circuits involving R,L,C elements has been implemented. The model successfully reproduces
ipv − vpv and ppv − vpv curves and matches real module data under different scenarios. In
the proposed model of the PV generator the values Isc , Voc , Imp , Vmp can be conveniently
inputted, when they are known under specific atmospheric conditions, while the model is also
expandable in case of irradiance and temperature variations. However, the functionality of
the optimisation scheme proposed has limitations and is not adjustable for every value of
irradiance and temperature.
Following, the model is used for utility grid-connected studies. First a single-phase grid
connected system is developed, with all the included elements analysed and modelled in
EMTP, in order to gain valuable insights of the system’s behaviour for further studies on
a power analysis level. The model shows very good numerical stability and the predicted
waveforms agree with measured ones. Additionally, a new MPPT algorithm is proposed,
based on the specifications of the system, that after simulations shows also good dynamic
behaviour. However, the modelling of the system does not correspond to a well designed
commercial PV system, due to its high harmonic levels and the simple configuration chosen
for the system’s elements. In general though, the association of the PV generator model with

Master of Science Thesis Paraskevi Breza

100 Conclusions and future work

the grid-connected system proved that the proposed model can be successfully used to analyse
the characteristics and problems of distributed PV systems, but most of all is applicable to
both slow and fast simulations.
As a next step, in order to validate the expandability and flexibility of the proposed PV gen-
erator model, an already established three-phase grid-connected hybrid PV/battery system,
presented in a past graduation report, is used and combined with the EMTP-based models
developed in this project. Four case studies are simulated. The results validate the PV gener-
ator model’s behaviour in a three-phase scenario and a real life application approach system,
without providing in-depth insights of the system’s transient response and analysis in the
power system level though.
In conclusion, in this project a new simple model for the representation of PV generators in
a commercial electromagnetic transient program is proposed. It has the potential to be a
useful tool for power engineers dealing with PV systems modelling, while working both as
a PV module and an array depending on the system configuration. The model simplifies
the representation of PV generators in EMTP, avoiding complex block-circuits with several
elements and control commands, while additionally being suitable for implementation in other
electromagnetic transient software packages.

7-2 Future work

During this project, there were topics found to be unanswered or left for further studies.
Improvements on the equations used to determine the parameters of the nonlinear PV gen-
erator model, like k1, k2, k3, k4, m should be made in order to increase the accuracy of
the model. It is really important that these changes are not limiting the simplicity of the
procedure.
The optimisation scheme, that is proposed so that the PV generator can change according
to atmospheric condition variations, can also be improved and more appropriate equations
should be derived, so that the PV generator adjusts better in every condition.
As for the model of the single-phase grid-connected PV system, it should also be improved, by
developing a more concrete inverter and controller system. The results should be verified with
modern commercial PV system outputs while the system’s harmonics should be maintained
at acceptable levels. Fault simulations and detailed transient analysis should be the next
steps to be done.
Nowadays, the integration of solar energy in microgrids is becoming a main concern of the
power system industry. At the same time, the modelling of renewable energy sources for large-
scale power system integration simulation is becoming more and more important, therefore
simulation tools will play a significant role during optimal design and intelligent management
processes [43]. An extensive technical study is required in terms of the impact of these large-
scale systems on the voltage stability, the power quality, the response to faults and short
circuit contributions, etc. [44]. For these studies a complete three-phase PV solar system
is required in an electromagnetic transient software environment such as EMTP. Hybrid
systems should also be a future target, since their control strategies are quite different form
the control in conventional systems [41]. The focus on these systems should be on the battery
charging/discharging control and on the technique of using low-pass filters [41].

Paraskevi Breza Master of Science Thesis

 Appendix A

The Newton Raphson algorithm

Newton Raphson algorithm is a powerful technique for solving equations numerically. Like
many other differential calculus, stands on the simple idea of linear approximation and it
is based on a Taylor series expansion of the function f(x). More analytically, if the f(x) is
expanded around a point α the function becomes:
0 1 00
f (x) ≈ f (a) + (x − a)T f (a) + (x − a)T f (a)(x − a) (A-1)
2
0 00
where f (·) is the gradient vector and f (·) is the hessian matrix of second derivatives. This
creates quadratic approximation for f . In order to maximize this quadratic function the
following equation must be solved:
d 0 1 00 0 00
f (a) + (x − a)T f (a) + (x − a)T f (a)(x − a) = f (a) + (x − a)T f (a) = 0 (A-2)
dx 2

0 00
x = a − [f (a)]T [f (a)]−1 (A-3)

The Newton Raphson process iterates this equation. More specifically, if x0 is the starting
point for the algorithm and x1 , x2 are successive estimates defined recursively through the
equation:
0 00
xi+1 = xi − [f (xi )]T [f (xi )]−1 (A-4)

In case the function f (x) is quadratic, the approximation is exact and the Newton Rapshon
method converges to the maximum in one iteration. If the function is concave, then the
method successfully converges to the right answer. In case the function is convex only for
some values of x, then the algorithm may or may not converge. The method might converge
to a local and not global maximum, a local minimum or it cycle between two points. The
best method for driving convergence to the global maximum is to start the algorithm near
the global maximum [45].

Master of Science Thesis Paraskevi Breza

102 The Newton Raphson algorithm

Paraskevi Breza Master of Science Thesis

 Bibliography

[1] R. E. P. N. for the 21st Century, “Renewables 2012 global status report,” tech. rep.,
Renewable Energy Policy Network for the 21st Century, 2012.

[2] E. P. I. Association, “Global market outlook for photovoltaics 2013-2017,” tech. rep.,
European Photovoltaic Industry Association, 2013.

[3] V. Quaschning, Understanding Renewable Energy Systems [Multimedia Multisupport],

vol. 1. Earthscan, 2005.

[4] A. D. Theocharis, V. P. Charalampakos, A. Drosopoulos, and J. Milias-Argitis, “Equiva-

lent circuit of photovoltaic generator using newton-raphson algorithm,” COMPEL - The
International Journal for Computation and Mathematics in Electrical and Electronic
Engineering, vol. 31, no. 4, pp. 1224–1245, 2012.

[5] M. Zeman, “Photovoltaic basics syllabus.” University Lecture, 2011.

[6] H. J. Möller, Semiconductors for solar cells. Artech House Bostonˆ eMA MA, 1993.

[7] D. Rekioua and E. Matagne, Optimization of Photovoltaic Power Systems: Modelization,

Simulation and Control. Springer, 2012.

[8] M. G. Villalva, J. R. Gazoli, et al., “Comprehensive approach to modeling and simula-

tion of photovoltaic arrays,” Power Electronics, IEEE Transactions on, vol. 24, no. 5,
pp. 1198–1208, 2009.

[9] S. Said, A. Massoud, M. Benammar, and S. Ahmed, “A matlab/simulink-based photo-

voltaic array model employing simpower systems toolbox,” Journal of Energy and Power
Eng, vol. 6, pp. 1965–1975, 2012.

[10] B. WILAMOWSKI and J. Irwin, Power Electronics and Motor Drives (The Industrial
Electronics Handbook). CRC Press Taylor and Francis Group, 2011.

[11] A. Malla and A. Niraula, “Importance of balance of system in a solar pv application,”

Master of Science Thesis Paraskevi Breza

104 Bibliography

[12] J. A. VENTRE et al., Photovoltaic systems engineering. CRC press, 2004.

[13] P. B. Miro Zeman, “Photovoltaic systems reader.” University Lecture, 2013.

[14] IEEE, “Recommended practice for utility interface of photovoltaic (pv) system, ieee std
929-2000,” tech. rep., Institution of Electrical and Electronics Engineers, 2000.

[15] Wikipedia, “Islanding,” 2013.

[16] U. Mohan, “Robins (2003) power electronics: Converters, applications, and design.”

[17] S. V. Araújo, P. Zacharias, and R. Mallwitz, “Highly efficient single-phase transformerless

inverters for grid-connected photovoltaic systems,” Industrial Electronics, IEEE Trans-
actions on, vol. 57, no. 9, pp. 3118–3128, 2010.

[18] A. Falk, M. Meinhardt, and V. Wachenfeld, “Efficiency and grid compatibility of pho-
tovoltaic inverters–state-of-the-art and future trends,” in Proceedings of the European
Conference on Power Conversion and Intelligent Motion (PCIM), Nuremberg, Germany,
pp. 14–20, 2009.

[19] M. Calais, J. Myrzik, T. Spooner, and V. G. Agelidis, “Inverters for single-phase grid con-
nected photovoltaic systems-an overview,” in Power Electronics Specialists Conference,
2002. pesc 02. 2002 IEEE 33rd Annual, vol. 4, pp. 1995–2000, IEEE, 2002.

[20] C. Rodriguez and J. D. Bishop, “Organic architecture for small-to large-scale photovoltaic
power stations,” Industrial Electronics, IEEE Transactions on, vol. 56, no. 11, pp. 4332–
4343, 2009.

[21] T. Esram and P. L. Chapman, “Comparison of photovoltaic array maximum power point
tracking techniques,” Energy conversion, IEEE transactions on, vol. 22, no. 2, pp. 439–
449, 2007.

[22] S. Alghuwainem, “Matching of a dc motor to a photovoltaic generator using a step-up

converter with a current-locked loop,” Energy Conversion, IEEE Transactions on, vol. 9,
no. 1, pp. 192–198, 1994.

[23] S. Lalouni, D. Rekioua, T. Rekioua, and E. Matagne, “Fuzzy logic control of stand-alone
photovoltaic system with battery storage,” Journal of Power Sources, vol. 193, no. 2,
pp. 899–907, 2009.

[24] H. Cho, S. Yeo, C. Cim, V. Terzijia, and Z. Radojevic, “A steady-state model of the pho-
tovoltaic system in emtp,” in The International Conference on Power Systems Transients
(IPST2009), Kyoto, Japan, 2009.

[25] H. Hoidalen, “Atp draw version 5–users manual supplements,” 2007.

[26] M. Popov, “Atpdraw-based models: non-linear elements, surge arresters and circuit
breakers,” in European EMTP-ATP conference, vol. 22, p. 23, 2008.

[27] T. Zacharias, J. Milias-Argitis, and V. Makios, “First-order circuits driven by a photo-

voltaic generator,” Solar cells, vol. 31, no. 1, pp. 57–75, 1991.

Paraskevi Breza Master of Science Thesis

 105

[28] Siemens, “Solar module sm110/sm100,” 2000.

[29] A. D. Theocharis and E. C. Pyrgioti, “Electromagnetic transient programs and the pho-
tovoltaic input,” in Environment and Electrical Engineering (EEEIC), 2013 12th Inter-
national Conference on, pp. 346–351, IEEE, 2013.

[30] W. F. Marion, B. Kroposki, K. Emery, J. Del Cueto, D. Myers, and C. Osterwald,

Validation of photovoltaic module energy ratings procedure at NREL. National Renewable
Energy Laboratory, 1999.

[31] A. Menti, T. Zacharias, and J. Milias-Argitis, “Harmonic distortion assessment for a

single-phase grid-connected photovoltaic system,” Renewable Energy, vol. 36, no. 1,
pp. 360–368, 2011.

[32] B. McNeill and M. Mirza, “Estimated power quality for line commutated photovoltaic res-
idential system,” Power Apparatus and Systems, IEEE Transactions on, no. 10, pp. 3288–
3295, 1983.

[33] A. Sarwar and M. J. Asghar, “Multilevel converter topology for solar pv based grid-tie
inverters,” in Energy Conference and Exhibition (EnergyCon), 2010 IEEE International,
pp. 501–506, IEEE, 2010.

[34] S. Pyakuryal and M. Matin, “Implementation of ac to dc converter using thyristor in

atp,” IOSR Journal of Engineering, vol. 2, no. 11, pp. 6–11, 2012.

[35] S. Tunyasrirut, B. Wangsilabatra, and T. Suksri, “Phase control thyristor based soft-
starter for a grid connected induction generator for wind turbine system,” in Control Au-
tomation and Systems (ICCAS), 2010 International Conference on, pp. 529–534, IEEE,
2010.

[36] R. García-Valverde, C. Miguel, R. Martínez-Béjar, and A. Urbina, “Optimized pho-

tovoltaic generator–water electrolyser coupling through a controlled dc–dc converter,”
International Journal of Hydrogen Energy, vol. 33, no. 20, pp. 5352–5362, 2008.

[37] J. B. Zhang, P. Wang, Z. Zhou, and M. M. Cai, “An improved multistage variable-step
mppt algorithm for photovoltaic system,” Applied Mechanics and Materials, vol. 347,
pp. 1833–1838, 2013.

[38] D. Sera, M. Valentini, and A. Raducu, “Real time photovoltaic array simulator for testing
grid-connected pv inverters,”

[39] F. Ding, P. Li, B. Huang, F. Gao, C. Ding, and C. Wang, “Modeling and simulation of
grid-connected hybrid photovoltaic/battery distributed generation system,” in Electricity
Distribution (CICED), 2010 China International Conference on, pp. 1–10, IEEE, 2010.

[40] J. M. M. A. V. Alberto, “Integration of solar and wind power with battery storage sys-
tems,” Master’s thesis, Faculdade de CiÃłncias e Tecnologia da Universidade de Coimbra,
Departamento de Engenharia ElectrotÃľcnica e de Computadores.

[41] S.-K. Kim, J.-H. Jeon, C.-H. Cho, J.-B. Ahn, and S.-H. Kwon, “Dynamic modeling
and control of a grid-connected hybrid generation system with versatile power transfer,”
Industrial Electronics, IEEE Transactions on, vol. 55, no. 4, pp. 1677–1688, 2008.

Master of Science Thesis Paraskevi Breza

106 Bibliography

[42] Solarex, “Msx-60 and msx-64 photovoltaic modules,” 1998.

[43] C. Keles, B. B. Alagoz, M. Akcin, A. Kaygusuz, and A. Karabiber, “A photovoltaic

system model for matlab/simulink simulations,”

[44] S. A. Rahman and R. K. Varma, “Pscad/emtdc model of a 3-phase grid connected

photovoltaic solar system,” in North American Power Symposium (NAPS), 2011, pp. 1–
7, IEEE, 2011.

[45] “Newton Raphson Algorithm.” http://web.as.uky.edu/statistics/users/kviele/

sta601s08/nummax.pdf, 2006. [Online; accessed 30-October-2013].

Paraskevi Breza Master of Science Thesis

 Glossary

List of Acronyms

PV Photovoltaic

EPIA European Photovoltaic Industry Association

EU European Union

MPPT Maximum Power Point Tracking

EMTP Electromagnetic Transients Program

ATP Alternative Transients Program

Si Silicon

ASTM American Society for Testing and Materials

AM Air Mass

STC Standard Test Conditions

BOS Balance of System

IEEE Institute of Electrical and Electronics Engineers

PWM Pulse Width Modulation

ENRC Equivalent Newton-Raphson Circuit

TACS Transient Analysis of Control Systems

MER Module Energy Ratings

SCR Silicon-Controlled Rectifier

THD Total Harmonic Distortion

IEEE Institute of Electrical and Electronics Engineers

Master of Science Thesis Paraskevi Breza

108 Glossary

SVPWM Space Vector Pulse Width Modulation

SOC State of Charge

Paraskevi Breza Master of Science Thesis