FAULT MANAGEMENT IN ELECTRICAL DISTRIBUTION SYSTEMS

Final report of the CIRED Working Group WG03 Fault Management

FOREWORD
This text comprises the final report of CIRED Working
Group 03 Fault Management. The working group was
established in November 1995. The initiative came from
Mr. Yves Harmand of Electricit de France, who was the
convenor of the predecessor working group Distribution
Automation.
On behalf of CIRED organisation, the work has been
supported and supervised by Dr. Frank Otto, Stadtwerke
Dresden, Germany.

One of the main issues is the techniques for fault location

and fault indication. For these, both different equipment
and functions are considered. Of high importance also, are
the various computer systems used in network operation.
To analyse the state of art of these, also a detailed analysis
of the properties of present SCADA-systems is given.
In the latter part of the report, two special issues are
considered. One is the comparison of different solutions of
fault statistics and fault indices. The second subject is the
various approaches when assessing the economic feasibility
of fault management solutions.

The members of WG 03 have been

LIST OF ABBREVIATIONS
Dr. Matti Lehtonen, VTT Energy, Finland (convenor)
Dr. Damian Cortinas, Electricit de France, (secretary)
Mr. Rino Anelli, ENEL, Italy
Mr. Jean-Paul Krivine, Electricit de France
Mr. Iaki Ojanguren, Iberdrola, Spain
Mr. Philippe Perusset, Electricit Neuchteloise,
Switzerland
Prof. Peter Schegner, TU Dresden, germany
Mr. Philip Tempelaere, Electrabel, Belgium
Dr. Walter Tenschert, OKA, Austria
Mr. Antonio Gomes Varela, Electricidade de Lisboa,
Portugal

DCC

Distribution control centre

DLC

Distribution line carrier

DMS

Distribution data management system

END

Energy not delivered

FM

Facilities management

GIS

Graphic information system

HV

High voltage (> 60 kV)

Low voltage (< 1 kV)

LED

Light emitting diode

MMI

Man machine interface

MV

In Espoo, Finland 22. December 1998

Matti Lehtonen

Automatic mapping

LV

I wish to thank all the working group members and also

Dr. Otto for a very good and pleasant co-operation when
working towards this final report on fault management
issues.

AM

Medium voltage (1 kV . 60 kV)

NCC

Network control centre

RTU

Remote terminal unit

SCADA

Supervisory control and data acquisition

INTRODUCTION
In electrical distribution systems, fault management is one
of the main functions to reduce outage times. For this
purpose, various methods are used in different countries.
This report is an attempt to estimate the current state of
these functions and to give recommendation for
improvement of the present solutions.
The approach of the report is to find the different technical
and organisational solutions for fault management,
depending on network structure, protection philosophy and
other circumstances. Various methods are presented and
compared for advantages and disadvantages, as well as
for costs and benefits.

1 NETWORK PROFILES
1.1 Distribution network structures
This section focus on the structure of MV (1 kV to 60 kV)
and LV (1 999 V) networks. The distribution utilities that
have been studied in this report are very heterogeneous in
size (they operate from 1 to 100 distribution networks
HV/MV substations) and in structure. Nevertheless, the
group has managed to produce a summary of network
structures of all utilities that highlights the differences and
the similarities between them. We will show only those
characteristics that will be useful for the discussion of fault
management.

Usually we find differences in the typical numbers of

feeders of rural and urban substations, but the trends are
different from one utility to another.
The tendencies regarding the total number of primary
substations are not same around Europe:

In Finland the number is decreasing, in order to

diminish the maintenance costs.
In Austria the number is increasing in order to cope
with new demand.
In EDF and ENEL the number is increasing in order to
improve the quality of supply.

1.1.3 MV feeder profile

There is one major issue that differentiate the
characteristics of a network : its urban or rural nature.
Urban networks are usually short and underground, rural
networks are usually long (but not necessarily overhead : in
some parts of Germany and in Electrabel they are mostly
underground, and we see a trend in all utilities to increase
the part of underground in rural networks). Many
characteristics of networks can be explained when
considering the urban or rural feature.
Two remarks should be made about this chapter:

This document reflects the state of the art on network

structures, and some changes might occur in the future.
All the utilities contributing to this survey use MV
networks without distribution of the neutral wire.
Some conclusions dont apply to networks of the
distributed neutral kind (very common in the United
States and Canada).

1.1.1 Structure and operation

The structure of the MV networks is mainly meshable, so
that feeders can be backfeed by adjacent lines. Urban
networks are almost completely meshable, but that is not
the case of rural networks for obvious reasons of cost. The
percentage of rural networks that can be backfeed varies
between 50% and 90% on the utilities studied.
LV networks are much less meshable than MV networks.
Usually only some urban networks can be backfeed. The
percentage varies between 0% and 30%.

Rural MV feeders are much longer than urban MV feeders.

Average values vary from 10 to 35 km for rural feeders, and
from 3 to 10 km for the urban ones.
Regarding the typical load of a MV feeder, there is no
obvious trend to differentiate urban and rural lines. Some
utilities design rural feeders more loaded than urban ones,
other utilities do the contrary, and sometimes there is no
significant difference. Typical loads of rural feeders vary
from 1 to 8 MVA, typical loads of urban ones vary from 1
to 10 MVA.
MV networks are almost always 3-phased in the utilities
that have contributed to this report. Some 2-phase branches
(connected to a 3-phase backbone) are used by EDF to feed
low-consumption rural areas.
1.1.4 Secondary substations
The average numbers of secondary substations (MV/LV) by
MV feeder are different for rural and urban feeders :

Averages from 5 to 15 secondary substations by urban

MV feeder.

Averages from 15 to 50 secondary substations by rural
MV feeder.
The average number of LV customers by secondary
substation varies from 40 to 100, the maximal number
beingalways under 500.
1.1.5 Power generation

MV and LV networks are almost always operated radially.

No utility have plans to loop networks in the future.
1.1.2 Primary substations
Primary substations (HV/MV substations) have generally
from 1 to 4 power transformers. 2 transformers is the most
typical value.
The average number of MV feeders by substation varies
between 4 and 20, most typical values being around 10.

All the utilities reflect a step increase in the number of

power generators that apply for connection in MV and LV
networks.
The range of rated power of the dispersed generators
connected to MV networks varies largely from one utility to
another. The minimum rated power goes from 25 kW
(ENSA) to 1 MW (Iberdrola), the maximum power allowed
for connection at MV level goes from 1 MW (Finland) to
25 MW (Iberdrola).

included. This is the case of some networks in

Austria and Germany.

1.2 PROTECTION PHILOSOPHIES

1.2.1 Short Circuit protections

Types of protections
Short Circuit Faults are usually the easiest to detect, as the
fault current is important when compared to the load
current. All the utilities studied use definite time current
relays to detect this kind of faults: if the measured time is
superior to a fixed threshold, the protection considers that a
fault is present on the network. Time delays are used to coordinate the transformer, the busbar and the feeders
protections.
Some utilities (Austria, Germany, ENSA) install distance
protections that measure the current and the voltage at the
time of a fault, and calculate the impedance of the fault that
is a good indicator of the fault distance to the substation.
This information is very useful to locate the fault.
Sometimes the protection directly calculates the distance.
1.2.2 Earth Fault protections
Neutral grounding
The type of neutral grounding has important consequences
over the earth fault protection scheme and the types of
faults encountered by each company. Throughout the
European utilities studied, we can find three main types of
neutral grounding :

All the utilities use definite time current relays to detect

high-current faults (short circuits and some ground faults
with low-impedance earthing). Concerning higher
impedance faults (up to a few k), several techniques are
used :

Inverse-time current relays

zero-sequence relays (wattmetric,
ampermetric)

Distance protections

Neutral voltage relays

voltmetric

or

Sometimes a combination of these three types is used.

Also, as stated previously, EDF and ENEL use the
protection called shunt to eliminate non permanent earth
faults without opening the feeder.
Sustained operation with fault
The compensated grounding technique permits a very
interesting practice called sustained operation with fault. As
the earth faults dont produce a significant fault current, it is
possible to search for the fault without opening the feeder.

Systems with isolated neutral, in which the neutrals of

transformers and generators are not intentionally
connected to earth, except for high impedance
connections for signalling, measuring or protection
purposes. Mainly used in ENEL and Finland, with
some networks also in ENSA, Austria and Electrabel.

Among the utilities that use compensated grounding,

Austria and Germany use this technique.

System with resonant earthing, in which at least one

neutral of a transformer or earthing transformer is
earthed via an arc suppression coil and the combined
inductance of all arc suppression coils is essentially
tuned to the earth capacitance of the system for the
operating frequency. Mainly used in Austria,
Germany, some Swiss companies, and to some extent
in Finland. EDF and ENEL are studying the
conversion of part of their rural networks to
compensated grounding.

All the utilities contributing to this survey use reclosers at

the head of MV overhead feeders in order to eliminate nonpermanent faults without a long power cut. The reclosing
schemes are basically of three types :

System with low-impedance neutral earthing, in

which at least one neutral of a transformer is earthed
directly or via an impedance designed such, that due to
an earth fault at any location the magnitude of the fault
current leads to a reliable automatic tripping due to the
magnitude of the fault current. Mainly used in
Electrabel, EDF, EDP and Iberdrola.
NOTE: Systems with isolated neutral or resonant
earthing, whose neutral is earthed within a short
time at each occurrence of an earth fault, are

1.2.3 Reclosing schemes

Reclosing philosophies

Substation reclosers alone (Germany, ENSA, Finland,

Iberdrola).

Substation reclosers and on-network devices, located
somewhere in the middle of the MV feeders. They are
mainly automatic sectionalisers (Austria, Electrabel,
EDF, ENEL, EDP), and sometimes automatic reclosers
(Electrabel, EDF).

Substation reclosers, on-network devices substation
shunt - EDF, ENEL.
Substation reclosers
The principle of operation of the substation recloser
automatism is as follows :
If a fault is detected by the feeder protection (short circuit
or earth fault), the circuit breakers of the faulted lines open.

As sometimes faults are automatically cleared after the

disappearance of the phase voltage, the recloser
automatism serves to restore service as quickly as possible.
It usually has two cycles :

First, a fast cycle : some time after the opening

(usually 0.3 to 1 second) the recloser forces the closing
of the circuit breaker.

If the fault has not been cleared, we have the slow
cycle : the recloser waits between 20 seconds and 3
minutes, and orders the closing of the circuit breaker.

If the fault is always there (permanent fault), the protection opens definitively.
FAULT
Close

CIRCUIT
BREAKER

Open
0.3 - 1 seconds

TIME

20 seconds - 3 minutes

Fig. 1.1 The reclosing cycle for a permanent fault.

in comparison with the sectionalisers is that they act
faster, as they dont have to wait for the slow cycle of
the substation recloser. They need some kind of coordination with the other automatism.

Reclosers and on-network devices

Besides the substation reclosers, there are on-network
devices. These can be of two types :

Automatic sectionalisers. They will open only if two

conditions are met : the fault is seen beyond the
sectionaliser, and after two openings of the feeder (they
open during the slow cycle of the substation recloser).
This device reduces the portion of the network that is
de-energised in case of a permanent fault.

On-network reclosers. They act just like the substation
reclosers, with a fast and a slow cycle. Their advantage
Network
recloser

Sometimes if there are automatic sectionalisers the

substation recloser will have a second slow cycle. This
allows a first slow cycle before opening the sectionalisers,
or permits the use of several sectionalisers in series along a
feeder.

Network recloser,
Sectionaliser

Sectionaliser

FAULT
Close

CIRCUIT
BREAKER

Open
Fast cycle

Second slow cycle

First slow cycle

TIME

Fig. 1.2 The automatic sectionaliser reclosing cycle in a permanent faut.

Reclosers, shunt and on-network devices

Some utilities use also an automatism called shunt . This
technique is mainly used in networks with high impedance
or isolated grounding. The shunt is just a circuit breaker
between each phase and the earth, that is normally open but
that closes in case of an earth fault. By closing only the

faulted phase, it allows for the disappearing of temporary

earth faults without cutting the feeder. The co-ordination of
the shunt with the reclosing sequence is achieved by an
order of retard sent by the shunt to the recloser if it detects
an earth fault. The following graphic describes the
functioning of the system for permanent faults :

FAULT
Close

SHUNT

Open
0.2 seconds

TIME
Network
recloser

Network recloser,
Sectionaliser

Fast cycle

First slow cycle

Sectionaliser

Close

CIRCUIT
BREAKER

Open
Second slow cycle

TIME

Fig. 1.3 The sectionaliser reclosing cycle with a shunt.

1.3 DEGREE OF REMOTE CONTROL

1.3.1HV/MV Substation
HV/MV substations are always remotely controlled in the
distribution utilities that have participated to this survey.
We can distinguish between remote signals (binary values
associated to an event) and remote measurements (numeric
values associated to a measurement device at the substation
or at counting devices).

Other kinds of remote measurements found in some utilities

are :

Remote signals
The remote signals associated to switch position, circuit
breaker position, fault detection and relay operation are
found in almost all substations.
Sometimes other types of signals are sent to the
Distribution Control Centres (DCC), as cause of event,
recloser and automatism operations, and type of fault.
The remote signals sent to the DCC form the set of events
to be interpreted by operators or by alarm processing
software. Differences from one utility to another may
explain the need or the difficulty for alarm processing
software.
Remote measurements
Remote measurements can be obtained by several different
ways, such as cyclic polling, polling on demand, or
threshold trespassing. There are some differences about the
type of measures sent to DCCs and the frequency of the
polling.
The two measures that we can find in most of the
substations are the MV feeders current and the MV busbar
voltage. We usually find also a measure of active and
reactive power (sometimes at the MV side, sometimes at
the HV side, sometimes both of them).

HV busbar voltage.
Measures associated to the resonant system.
Counting for billing purposes.
Position of the tap changers.
Transformer temperature.
Fault current (the value of the current during a fault).
This is the most important issue in fault management,
because this measurement may allow fault distance
calculations. Sometimes the electric distance is
calculated locally and directly sent to the DCC.

No utility sends today to the DMS the time recording of the

current pattern after a fault, that would be much richer to
made fault distance calculations. Some utilities are
nevertheless studying the issue: the technology to record
and send this pattern exists, and is becoming standard for
most of the new digital protections. Nevertheless, the
frequencies of those protections range between 600 Hz and
2 kHz; those frequencies are not sufficient for the analysis
of the current pattern studies show that 10 kHz should be
needed for the analysis of ground faults.
1.3.2 MV Network
The situation regarding the degree of remote control for
MV networks is very different from one utility to another.
We can distinguish the following items :

Remotely controlled switches : some utilities have

none, some have very few, and some have between zero
and four by feeder ; usually there are much more in
rural networks than in urban networks. Some utilities
have today an active policy to increase the number of
remotely controlled switches.

Remotely controlled fault detectors : most utilities link
them to the remotely controlled switches in overhead
networks (whenever they install one, they install also
the other to take full profit from the communication
link).

Power producers : about half the utilities monitor some

of the power producers connected in MV. The usual
measures are active power, reactive power, voltage,
current and switch state. Some other utilities announce
plans to monitor them. Only IBERDROLA sends
tripping orders to power producers.

Power quality monitoring : few utilities monitor power
quality at all, and even in this case the monitoring is not
directly linked to the SCADA system. The only
exception is EDF, with a program aimed to monitor the
quality of the entire MV network (but not linked to
SCADA neither).
This issue is extremely important to fault management in
Distribution Management Systems.
The ratio of remotely controlled MV/LV substations is low
in general, and specially in rural areas. The main exception
is ENEL, who is going towards a global ratio of 12.5% for
all kind of areas.

No utility has today any kind of remote control for LV

networks. Some utilities are working on remote metering
and monitoring systems for LV customers

2 FAULT LOCATION FUNCTIONS AND

PRACTICES
In this section the most important techniques and solutions
for fault location are outlined. The items considered are
fault indicators, computational fault location and the
methods for the detection of high resistance earth faults.
With respect to the last item, also the aspects of sustained
operation with earth faults is considered.
2.1 FAULT INDICATORS
Short circuit indicators
Short circuit indicators are mainly used in Medium Voltage
(MV) networks, either radial or open ring operated, for
finding the fault location. They can be installed on the
current conductors to be monitored, busbar, cable or
overheadline.

1.3.3 LV Network

Open separation path

Transformer station
Short circuit indicators tripped
Short circuit indicators not tripped

Fig. 2.1. A network with short circuit indicators .

The magnetic field of a conductor through which current

flows triggers the short circuit indicator. The exceeding of
the rated threshold current of the indicator results in a
signal, either optical or electronical, thus marking the
direction of the short circuit current from the feed in point
to the fault location by tripped short circuit indicators. (See
Figure 2.1)

The basic layout of a short circuit indicator consists of a

yoke and an attached display system. The display system
can be mechanical, e.g. a pivot bearing rotor for local
indicating and/or a microcontact for remote signaling, or
electronical processed for local indicating or remote
signaling.
Every conductor through which current flows is surrounded
by a magnetic field, see Fig. 2.2. The magnetic field

strength H is used to discriminate between normal load

conditions and overload or short circuit currents.

glass zylinder

H
Yoke

Fig. 2.3. Particles pulled up by magnetic field.

Fig. 2.2. Conductor surrounded by magnetic field strength
H
Short circuit indicators with rotor system.
At currents higher than the threshold current the induced
magnetic field strength H is sufficient to overcome the
retaining force of a catch spring moving a rotor to the
tripped position. The reset of this indication has to be done
manually.
Typical technical data:
Threshold value:
Adjustment pulse:
Clamping range:

200 2000 A, 10%

100 ms
Conductor diameter 8 - 80
mm for circular
conductors 20x4 - 60x30
mm for rectangular
conductors

Electronic short circuit indicators with remote signaling

system
In installations where the short circuit indicators cannot be
easily read you apply indicators with remote signaling at
the supervised conductor. The short circuit indicators
consists of three sensors, a wiring equipment and a display
unit. The sensors are mounted directly on the insulated
cable and transmit their measured values to the display
unit. To monitor bare cables or busbars the wiring harness
has to consist of optocouplers and highly insulated fibre
optical cable. The display unit is attached to an accessible
location, e.g. on the front wall of a switchgear cubicle.
Typical technical data:
Threshold value:
Length of wiring:
Indication:
Clamping range:

Used on outgoing cable feeders or busbars in switchyards

Short circuit indicators with fluid system.
Another indication principle is based on fluids. They get
along without moving parts. In the event of a short circuit a
composite body is pulled up by the magnetic field stirring
up red particles in a glass-clear fluid, see Fig. 2.3. The
coloured particles remain in suspension in the fluid for
approximately 4 - 8 hours indicating a short circuit. Due to
the gravity they will sink and the display turns clear again.
Typical technical data:
Threshold value:
Adjustment pulse:
Clamping range:

400, 600 or 1000 A, 20%

200 ms
Conductor diameter 8 - 80
mm for circular
conductors 20x4 - 60x30
mm for rectangular
conductors
Used on outgoing cable feeders or busbars in switchyards

200 2000 A, 20%

16m
flashable led, rotor or fluid
Conductor diameter 6 100 mm for circular
conductors 20x4 - 60x10
mm for rectangular
conductors

Mechanical overhead line short circuit indicator

These short circuit indicators may be attached to the
overhead line. The indication of a short circuit is done by
turning an all-round visible zylindrical flag indicator from
neutral to red.
The indication has to be reseted manually by means of an
actuation rod. The indicator can be installed and removed
under power with a special actuation rod (max. 30 kV).
Typical technical data:
Threshold value:
Adjustment pulse:
Visibility:

300 / 400 A, 10%

100 ms
up to 200 meter
(depending on daylight
brightness)
Clamping range:
Conductor diameter 8-25 mm
Electronic overhead line short circuit indicators with LED
or strobe light

Here the indication is accomplished by extremely bright

flashing LEDs. The indicator is excited by the short circuit
current. Reset is done automatically after expiration of a
previously set time. The indicator is also turned off when
the line power is switched back on and the current flow is
greater than 3 A. For powersupply a replaceable lithium
battery with shelf life of approx. 15 years. The indicator
can be installed and removed under power is acieved with a
special actuation rod (max. 30 kV).
Typical technical data:
Threshold value:
Actuating time:
Reset time:
Visibility LED:

Visibility strobe light:

Clamping range:

100 2000 A, 10%

20 ms
2, 4 or 8 h
up to 200 meter
(depending on daylight
brightness)
300m (cloudy sky) up to
900m in dark
Conductor diameter 8mm

Combined short circuit and earth fault indicators

Fault indicator for cable networks.
A fault current indicator of medium voltage cable networks is
shown in Figure 2.4. The indicator works both for short
circuit and earth fault cases. The sensor for short circuit
faults is based on a small coil, which energizes a light
emitting diode (LED ). The intensity of light varies with the
magnetic field magnitude, which in turn depends on the
current of the conductor. The light is transmitted to the
detector unit via a fiber optic cable and the measured
magnitude compared to the threshold value.
The sensor for earth faults is a simple sum current
transformer, formed by a metal strip wound around the

phases. Also the earth fault sensor is connected to the

detector unit via a fiber optic cable. The remote reading of
fault indicators speeds up the fault location and cuts the
outage times especially in the urban networks, where the
remote control of disconnecting switches is too expensive. In
most utilities, the fault indicators already exist in the medium
voltage / low voltage substation. The inclusion of these
devices into the automation system only demands the data
transmission system. This can be based on
telecommunication cables, or leased telephone connections,
for instance.
Fault indicator for overhead line networks.
A fault indicator for medium voltage overhead lines is shown
in Figure 2.5. The device is based on the coil, which
measures magnetic field and on a antenna to detect the
electric field. There is no contact to the live parts of the line,
but the indicator is mounted in the pole 3 to 5 meters below
the conductors. The sensitivity of the device is tuned simply
by altering this distance.
The indicator is able to detect both short circuit and earth
faults. The earth fault detection is based on the measurement
of magnetic field produced by the zero sequence current of
the line.
Another restriction of the device is, that it can not be
mounted at the line crossings. This is because of the
interference of magnetic fields of the lines. If used in
association of sectionalizer stations, for instance, the
indicators must be mounted into the next poles, and wired to
the station. This is costly and expensive, however. This kind
of problems are not met, if the fault indication is based on the
direct current measurement. For this purpose, small current
transformers, built on the top of an insulator, are nowadays
available. The corresponding fault indicator is actually a
small relay, which can be set remotely. The currents can be
recorded during the normal operation state too.

DATA TRANSMISSION

SHORT CIRCUIT SENSOR

DETECTOR UNIT

EARTH FAULT SENSOR

Fig. 2.4. An example of a fault indicator for cable networks.

dl/dt

&

FLIPR FLOP

A1

BLINK
UNIT

6h

ELECTRIC
FIELD
MEASUREMENT

1
A2

8s

LIGHT OR RELAY

Fig. 2.5. An example of a fault indicator for overhead line networks.

Restrictions of present fault indicators
The main restriction of fault indicators in the present market
is the lack of directional function. This would be of high
importance in the case of earth faults in compensated
networks. Another application is the networks with dispersed

generation. Here the directional characteristics is needed also

in the case of two or three phase faults.
The bottleneck when developing a directional fault indicator
is the voltage measurement. This should be arranged using a
technology cheaper than the present inductive voltage

transformers typically used in the high voltage / medium

voltage substations. A possible solution is a voltage divider
based on resistive, capacitive or inductive principle. This
kind of devices are not generally available in the market,
however.

2.2 FAULT DISTANCE COMPUTATION

TECHNIQUES
Especially in rural and suburban networks, where distances
are long, a very good supplement to the fault indicators is
the fault distance computation. In the case of short circuit
faults, the distance estimate can be obtained by using
distance relays. An alternative solution is to utilise the
measured fault current only and to obtain the distance
estimate by comparing the measured fault current to the
results of network calculations.
Computational fault location techniques are now available
for the case of short circuit faults. In the high impedance
grounded systems, the single phase earth fault currents are
so small that they do not allow for a reliable fault distance
estimate. For these cases transient based solutions have
been proposed instead.
2.2.1 Fault distance computation using distance relays
The method described here is part of the practical fault
location process in the distribution network at OKA, an
electricity utility of the state Upper Austria of Austria
[13]. It is in use in all HV/MV-substations of OKA and the
practical experiences could be collected for more than five
years.
In addition to distance computation fault detectors and
telecontrol of MV-stations are in operation. In combination
with the local staff a complete system has been established
to speed up the fault location process and to reduce the
duration of supply interruptions. Because the MV-network
of OKA is compensated, the method is first of all designed
for multi-phase faults.
Network characteristics
The MV-network has a nominal voltage of 30 kV with a
portion of 11 % cables. It supplies mainly rural regions.
The feeders are principally radial. Few meshed feeders are
split up automatically in case of a fault. After the
customary autoreclose ( 0,5 s ) the conditions of radial
feeders are fulfilled.
All the feeders are protected by distance relays,
approximately 40 % of them are of digital type, the rest is
electromechanically. In substations with electromechanical
relays the backup protection relays at the infeeding
transformer have been replaced by digital relays. The fault
reactance measured by the relays is telemetered by the
SCADA-system to the control centre. The reactance is

transferred from the relay to the SCADA-system partially

by analogue signals or by digital protocols.
Method
The basis of the method is a MV-network analysis model,
that is kept and maintained for planing purposes by the
network analysis group. Every feeder is split up into
segments, that are physically defined by MV-stations and
branching points. The segments have a length between 50
and 500 m, in few special cases longer.
With this network model an off-line short-circuit
calculation is performed for every segment automatically
by a standard power system analysis software. The shortcircuit calculation is updated once a year. This results in a
static reactance list for every feeder, that contains the fault
reactance at the beginning and the end of every segment as
well as long names for the stations or branching points.
If a fault occurs, the telemetered fault reactance is
transferred automatically from the SCADA-system to a
selection program on a PC or a SCADA-computer, that
selects the suitable segments. There is no on-line analysis
to compute the fault location, just a segment selection.
These segments are presented to the operator as probable
fault locations in form of a small list ( 1 to 5 segments ) on
the screen.
The practical experiences of the last years are excellent.
There is a considerably reduction in time for locating faults
as well as in the requirement for local staff. The accuracy is
better than 5 %. Inaccuracy of the metering is compensated
by the knowledge of the local staff about frequently faulted
sections. This led to the fact, that fault distance metering
became an integrated part of the fault location process and
it was accepted by the staff in a very short time. The
assumption of a standard topology without on-line analysis
is largely acceptable.
Reactance measurements by the backup relays of infeeding transformers are less accurate than direct feeder
measurement, especially if the transformers are highly
loaded. Nevertheless the results are still an important aim
in fault locating.
Permanent single-phase-to-ground faults
Location of permanent single-phase-to-ground faults in
compensated networks is sometimes difficult. If sustained
operation is not wanted, there is no point of keeping the
compensation. Therefore in two substations the Peterson
coil is by-passed by a low resistive impedance in case of a
permanent earth fault after 3 s. This leads to a short-circuit
current that is detected by the distance relay and switched
off by the circuit breaker of the faulted feeder.
The intention is to use from now on the same fault location
devices as for multi-phase faults ( reactance metering, fault

detectors, telecontrol ). Presently two problems are

investigated:

fault detectors for single-phase-to-ground currents less

than load current
adaptation of the fault reactance value given by the
relay according to earth-fault currents

If these problems are solved, short term grounding will

become standard at OKA-substations.
Further aspects
The present method of fault distance computation with
distance relays allows still some improvements:

automatic incorporation of telemetered fault detectors

graphic display in the control centre ( GIS )
consideration of transformer and line loading
consideration of generation and network topology
extension to single-phase-to-ground faults ( short term
grounding )

Because of the good experiences so far, there is not much

pressure to start these improvements
2.2.2 Short circuit fault location based on fault current
measurement
In this section, the use of measured fault currents to the
location of short circuit faults is discussed. A practical
implementation of this automation function is first presented.
The factors affecting the fault location accuracy are then
outlined, and some practical results are given based on the
experience with existing fault location systems.
The practical implementation of a fault location system
A modern short circuit fault location system is based on the
integration of distribution data management system (DMS),
substation telecontrol system (SCADA) and the relay
protection. The main idea is the comparison of the measured
and the computed fault currents. The computation is made
on-line, assuming the same network topology as it was when
the fault occurred. As a result, the estimated distance of the
fault from the substation is obtained. This distance is then, in
turn, compared to the network diagram, and the possible fault
locations are shown on a graphical display of the distribution
network in the DMS system.
When a fault happens, the operation of the fault location
system is as follows [3]:
1)

2)

the protective relays store the fault information

(currents, fault type, phases involved, feeder
involved, information on reclosing steps)
the recorded information is transmitted to the
SCADA central unit

3)

4)
5)

the SCADA system adds some more information to

the received fault data. This information includes
the measured load current of the feeder concerned
and the active and reactive load flows of the
substation main transformer
the information is transferred to the DMS system
the DMS computes the corresponding fault currents
of the feeder concerned and compares the measured
data to the computation results. Before the
comparison, the load currents, superposed on the
measured currents during the fault, are compensated
for

All the data needed for fault location computation is stored in

the working memory of the DMS system. The purpose of this
arrangement is to make the processing fast enough to meet
the practical requirements. Each fault case data is stored in a
separate record, which can be picked up by the operators for
processing. Hence, phase 5 of the above procedure is not
initiated automatically, but must be trigged manually. This
arrangement makes it possible, that several fault cases can be
studied simultaneously, and in an organized manner, by a
number of power system operators.
Performance of the fault location system
The main factors affecting the accuracy of short circuit fault
location are the errors of measurement transformers and
other measurement equipment, the variation of network
component impedances, the load current superposed on the
measured fault current and the fault resistance [4].
The errors due to the measurement transformers are usually
small, only a few per cent of the actual fault distance. The
most difficult is the case where the fault is close to the
substation. Because of the high fault levels, the current
transformers may become saturated, which deteriorates their
accuracy. If the voltage falls below 2 % of the nominal value,
the performance of voltage transformers may also collapse.
When the fault location is based on an analysis of the current
measurement solely, the result is also sensitive to a variation
of network impedances. In addition to the variation of the
resistance and reactance of the line concerned, the variation
of short circuit level in the medium voltage busbar may also
cause problems. The last one depends, in addition to the
changes in the grid impedance, mostly on the position of the
on-load tap changer of the substation transformer. For
accurate fault location, the on load tap changer position
should be telemetered.
The problem with the superposed load current is, that its
magnitude is changed dynamically with the voltage change
during the fault. For the load behavior, theoretical models
have been developed, which can be used for load current
compensation [4].
The largest errors are usually caused by the fault resistance,
however. In the case of power arc, this can be calculated as
follows [8]:

8750
l
R =
(ohm)
1.4
0.305
I

Earth fault transients

2.1

where l is the arc length (m) and I the fault current (A).
Resistance is increased when the fault current is decreased.
Hence it is of importance especially in weak systems and for
faults distant from the substation.
The fault current is known by the measurement and the
maximum arc length can be deduced from the line geometry.
Consequently, it is possible to estimate the maximum value
of the fault resistance. In practice however the fault location
computation is made with the assumption of zero fault
resistance. This gives the estimated maximum fault distance.
Because of the fault resistance, the actual fault point must
then be closer to the substation.
The best statement on fault location accuracy is based on
practical experience. An automatic fault location system,
described in this section, is in active use in several
distribution companies in Finland. According to the practical
experience in rural overhead line networks, the average error
in fault distance estimation has been about 1.2 km. For
comparison, the corresponding average fault distance is 13
km. For close faults the absolute errors are smaller, whereas
for distant faults they are larger, respectively. This degree of
performance is good enough for defining the faulty line
section between two isolating disconnectors.
The computational means however give only the fault
distance. If there are several branches or laterals in the line
concerned, several possible fault locations are also obtained.
Among these the actual fault point must be distinguished by
some other means. One possible solution is the remote
monitored fault current indicators in the line branching
nodes.
2.2.3 Earth fault distance computation based on
transient signals
In networks with an isolated or a compensated neutral, the
fundamental frequency earth fault currents usually are so
small that no reasonable fault distance estimates can be based
on them. Under these conditions some other techniques must
be used. A very promising possibility is the utilization of
earth fault initial transients. In this case the best results are
obtained using the methods, which estimate the line
impedance during the transient. These methods can be
classified into differential-equation algorithms [7], Fouriertransform methods [2] and least squares fitting methods [5].
In the following, the theory of earth fault transients is first
briefly discussed. The possible fault location methods are
then presented.

When an earth fault happens, the voltage of the faulty phase

falls and the charge stored in its earth capacitances is
removed. This initiates the discharge transient. Because of
the voltage rise of the two sound phases, another component,
called charge transient, is created. The latter has a lower
frequency and in most cases also a higher amplitude. Hence,
it is the component best suitable for fault location purposes.
The maximum amplitude of charge transient, compared to
the capacitive steady state fault current, approximately varies
with angular frequency. Since it is not unusual for this to be
5000 rad/s, the maximum amplitudes can be even 10-15
times that of the uncompensated fundamental frequency fault
current.
In real systems there is always some damping, which is
mostly due to the fault resistance and resistive loads.
Damping affects both frequencies and amplitudes of the
transients. The critical fault resistance, at which the circuit
becomes overdamped, is in overhead line networks typically
50...200 , depending on the size of the network and also on
the fault distance. If the resistive part of the load is large,
damping is increased, and the critical resistances are shifted
into a lower range.
The earth fault transients have also been examined by field
tests in real systems. According to the results, in the case of
the discharge component the amplitude was typically 5 to 10
% that of the charge component. The frequencies varied
through a range of 500-2500 Hz and 100-800 Hz for
discharge and charge components respectively.
Methods for earth fault distance estimation
Differential-equation algorithms solve the line inductance
directly in time domain. Consider the first order model,
which includes the series connection of the line resistance R
and inductance L. The voltage and current of the faulty phase
have the following relation:

u(t) = Ri(t) + L

di(t)
dt

2.2

which can be solved for inductance L, if three equally spaced

pairs of samples are available. Since differentiation is
sensitive to higher frequency noise, the solution is usually
obtained by integrating the above. Differential-equation
algorithms work in theory for all the voltage and current
components which satisfy equation (2.2). The best result is,
however, obtained if all the other frequencies are first filtered
out, except the charge transient.
Fourier-transform methods solve the line impedance in
frequency domain. In the case of the first order model,
reactance of the faulty line length is obtained directly as
imaginary part of the impedance calculated from

the
the
the
the

corresponding Fourier frequency spectrum components of

voltage and current.
In the prototype system described in reference [2], the fault
distance is calculated as a weighted average of the estimates
made for the n dominating frequencies in the spectrum. Also
a higher order model, which allows for the phase to earth
capacitances, is presented.
The least squares fitting methods solve first the parameters
the voltage and current transients directly in time domain.
This is done by fitting a function, that models the transient
waveform, to the measured samples of phase voltage and
phase current. In the case of earth fault transients, the
function used is a damped sinusoid.
For proper computation, least square fitting methods require
signal filtering and preprocessing similar to differentialequation algorithms. Once the amplitudes and phase angles
of transient phase currents and voltages are known, the fault
distance is solved as reactance of the faulty line length.
Accuracy of the fault distance estimation

a Swedish 22 kV overhead line network was used. The

network was with a 100 % compensated neutral, the rating of
the suppression coil being 17 amperes. Altogether 60 fault
cases were considered.
According to the results the least square methods have
somewhat better accuracy compared to the others. The
differential-equation algorithms and Fourier-transform
methods are essentially impedance relay algorithms. Their
main advantages are the numerical stability and relatively
small computation burden, which makes them suitable for
on-line calculations.
The highest fault resistance, that allows for reliable distance
estimation, is about 50 . With zero resistance, more than 95
% of the faults can be located reliably. According to the
experience in field test, the average errors are 1.6 km for the
first order line model and 1.2 km for the higher order line
model .
The transient based algorithms require a measurement
technique with a relatively high sampling frequency,
preferably in the order of 10 kHz. This has been a limiting
factor for a widespread use of these techniques so far.

The accuracy of the different methods is compared in Table

2.1. For the comparison, data recorded during staged faults in
Table 2.I Comparison of the accuracy of transient based fault location methods. Sixty faults with fault resistance 0
. First
order and higher order algorithms of Differential-equation methods (a and b) [7], Fourier-transform methods (c and d) [2]
and curve fitting methods (e and f) [5] respectively.
___________________________________________________________________________________________________
error class in kilometers:
0.0-0.6
0.6-1.2
1.2-1.8
1.8-2.4
>2.4
___________________________________________________________________________________________________
a
15
11
3
5
26
b
7
11
8
3
31
c
14
11
14
2
19
d
9
14
8
7
22
e
24
19
7
1
9
f
21
20
5
7
7
___________________________________________________________________________________________________

2.3 METHODS FOR THE DETECTION OF EARTH

FAULTS

2.3.1 The most important earth fault detection

methods [11]

This chapter gives a short overview of earth fault

location methods presently in use. Conventional methods
which use the zero sequence voltage or wattmetric
relays have severe problems in detecting high impedance
earth faults. This is the case especially in compensated
neutral systems. To mitigate this problem, several
techniques have been under research. Of these some
examples are given.

Transient residual voltage and current

Transient earth fault identification
During the ignition of an earth fault the faulted phase
unloads itself with a current impulse and a following
transient oscillation ("ignition oscillation"), while the
fault-free phases load themselves at the same time. The
amplitude of the ignition oscillation depends thereby on
the phase angles of the earth fault entrance. During these
high frequency procedures the petersen coil is ineffective
by their high inductance. Therefore direction of the earth

fault is determined outlet from the phase position

between residual-current and -voltage.
Advantage:

large,
however
variable
amplitudes of
the ignition
current
disadvantage:

if the earth fault accurse nearby the voltage zero (e.g.
by foreign influences) the ignition oscillation is very
small

the measurement is only unique (with the ignition
oscillation)

after disconnection of one base point during a double
ground fault the detection of the remaining earth fault
is not possible
important parameters:

ignition amplitude

L/C distribution in the network
remarks:

depends on the phase angel of the earth fault
appearance

determines the ignition frequency and damping

Residual voltage
Earth fault detection relay
Based on the unsymmetrical phase to earth voltage the
residual voltage is not longer zeroed. An overvoltage
relay observes the residual voltage. The threshold is set
to approximately 30% of the nominal voltage. If the
residual voltage exceeds this level a status signal is set
Advantage:

The detection of the earth fault is
independent of the fault location.
Therefore earth faults in the whole
network could be detected.
disadvantage:

no feeder selective earth fault detection is
possible; there are problems with high-resistance
faults
important parameters:

capacitive asymmetry of the network

earth fault resistance

remarks:

huge capacitive asymmetry combined with low
network damping or resonance tuning of the
petersen coil lead to high residual voltage during

normal operation
the residual voltage at a high-resistance earth
fault is low

Residual current
Variation of compensation by switching of capacitors
(pulse method)
The working method of this equipment is described at
the beginning of this chapter
Advantage:

the magnitude of the measured
residual current is free dimensional
disadvantage:

A central auxiliary equipment is necessary.

the dimension of the switching capacitor
depends on the network size

Residual voltage and current

Wattmetric earth-fault detection
The residual current at the fault location consists of a
resistive and a reactive-component. In resonant earth
networks the reactive-component depends on the
compensation level. Therefore in this networks only the
resistive-component is being used for the fault location
with a wattmetric earth-fault relay. The quantity of the
resistive-component depends approximately to two thirds
of the losses of the petersen coil and to one third on the
line losses in the ground fault path. The small
resistive-component has the same magnitude as the
transducer errors. Therefore after the occurrence of an
earth fault a resistance at the auxiliary winding is switched
on. That rises the resistive-component and improves the
operation of the relay.
Advantage:
the resistive-component is constant

disadvantage:

false indication in meshed networks based
on current splitting or at parallel lines based
on phase splitting are possible

huge requirements on the phase accuracy of
the current transducers are needed

cable-type current transformers should be
used instead of the three single phase current
transformers
important parameters:

losses of the petersen coil

maximum current loading of the auxiliary
winding for earth-fault detection of the
petersen coil

remarks:

determines the useful signal level

limits the signal level

important parameters:

level of the harmonics

unbalanced residual current

number of branches

L/C distribution in the network

Reactive-power relay
In isolated networks the reactive-component is used
for the earth-fault detection. The relays work with the
so-called sin  arrangement.
Advantage

independent from central
installations

normal a high signal level

auxiliary

disadvantage:

the signal level of the reactive-component
depends on the part of the network without
earth-fault

important parameters

capacitive earth-fault-current
Harmonic components
In compensated network the harmonics of the residual
current and voltage have the same characteristics than
the fundamental signals in isolated networks. The
petersen coil is ineffective by their high inductance to the
harmonics. An audio-frequency remote control system
(AF remote control) can also be used in state of the
natural 5. Harmonic. This system is switched on during
the earth fault. The frequency of this AF remote control
system should nearby 250 Hz. The evaluation of the
residual current and voltage preferably takes place either
with ampere-metric relays or directional relays, which
are adjusted, to the AF remote control system frequency.
advantage:

Independence from central installations
on use of the 5. Harmonic. Particular in
cable systems are strong currents and
therefor good results with this relays

disadvantage:

Fluctuations in the level of the 5. Harmonic
causes adjusting problems in some networks
with ampere-metric relays (less with ground
fault directional relays).

there are wrong direction decisions possible
by unfavourable distribution of the capacities
and inductivity of the zero system, e.g. in
expanded overhead line nets with central
arranged capacities (cable system parts)

remarks:

the level of the harmonics is e.g. dependent
on the load

the signal level of the harmonics is
proportional to the unbalanced residual current

the sensitivity of the ampermetric relays
increases with the number of branches

determines the ignition frequency and
damping
5. harmonic magnitude of feeder residual magnetic and
electric field
Operational principle similar to 5. Harmonic relays. The
measured signals processed by the equipment are not
directly the residual current and voltage, but the caused
electromagnetic fields. These are the magnetic field of
the residual current and the electrical field of the residual
voltage. There are only clear field conditions at overhead
line, however not in switchgear bays or at underground
cables.
Advantage:
Independence from central installations
on use of the 5. Harmonic
disadvantage:

Fluctuations in the level of the 5. harmonic causes
adjusting problems in some networks

there are wrong direction decisions possible, by
unfavourable distribution of the capacities and
inductivity of the zero system, e.g. in expanded
overhead line nets with central arranged capacities
(cable system parts)
important parameters:

harmonics

unbalanced residual current

L/C distribution in the network
remarks:

the level of the harmonics is e.g. dependent on
the load

the signal level of the harmonics is proportional to
the unbalanced residual current

determines the ignition frequency and damping

Phase current
Current magnitude relay
In impedance earthed networks the magnitude of the
earth fault current is in order of the nominal current of
the line. Therefore it is possible to use the normal short
current selective protection technology. These are for
example time relays, permission- and blocking-systems
or current differential protection systems. Protective
relays, which evaluate only the residual current, can be
attached to the cable-type current transformers. A set of
three pole current transformers is necessary for distance
protection relays. Since the fault current can be lower
then the nominal current of the line, suitable starting
schemas are necessary.
Advantage:

earth fault detection with existing
short current protection systems
disadvantage:

interruption of the electrical power supply


with temporary impedance earthed networks the
interruption could be avoided.
Important parameters:

magnitude of the earthing impedance

small earth-fault-current

big earth-fault-current
remarks:

determines the magnitude of the
earth-fault current

starting problems
problems with EMC and voltage dips
Short circuit fault indicator
The working method of this equipment is described at
the beginning of this chapter
Advantage:

shorten the time of the earth fault
detection

disadvantage:

the registration of the fault indicator decision is
only detectable directly or by remote control

important parameters:

magnitude of the earthing impedance

network structure
remarks:

Starting problems with small short circuit
currents.

only in radial networks the decision of the fault
indicators are clear
Analysis
earth fault distance protection
The numerical distance protection device reports the
measured fault distance. If also the distance protection
device records single-phase earth faults, the faulted
section can be easily and fast determined.
There is some research programs to determine the fault
location in an isolated or resonant earthed network by
the evaluation of the fault distance with the transient
signals of the earth fault
Advantage:

very fast fault location
disadvantage:

a centralised fault location in the power system
control station is possible
important parameters:

zero sequence impedance
remarks:

this impedance is often unknown
The different earth fault detection methods
have been summarized in Table 2.2.

Table 2.2. Overview on earth fault location methods (based on Verband der Elektrizittswerke sterreichs :
Sternpunktbehandlung in MittelspannungsnetzenWorking Group Report, 1996, ISBN3-901411-19-4 Authors: Fickert L.,
Tenschert W., and others ) [11].

Method
Residual voltage
Permanent earth fault identification
Automatic feeder interruption (OFF and ON)
Manual feeder interruption (OFF and ON)
Two-Transformer-Method (no interruption)
Earth fault indicators (residual voltage field)
Transient residual voltage
Transient earth fault identification
Residual current
Magnitude
Active component
Variation of compensation by Peterson coil
Variation of compens. by capacitor switching
Phase current
Magnitude
Reactance computation (feeder I and U)
Short circuit indicator
Harmonic components
5. harmonic magn. of feeder residual U and I
5. harmonic magn. of magn. and electric field
5. harmonic phase comparison of residual I
Analysis
Numerical analysis of feeder residual I and U

faulty
feeder
select.

Treatment of neutral point

temp.
Isoreson.
imped.
Imped.
lated Earthed earthed
Earthed

No
Yes
Yes
Yes
Yes

No
No
No
No
No

Yes
Yes
Yes
Yes
Yes

Yes
Yes
Yes
Yes
Yes

No
No
No
No
No

Yes
No
No
No
Yes

Yes

No

Yes

Yes

No

Yes

Yes
Yes
Yes
Yes

No
No
No
No

No
Yes
No
No

No
Yes
Yes
Yes

Yes
Yes
No
No

Yes
No
No
No

Yes
Yes
Yes

Yes
Yes
Yes

No
No
No

No
No
No

Yes
Yes
Yes

Yes
Yes
Yes

Yes
Yes
Yes

No
No
No

Yes
Yes
Yes

Yes
Yes
Yes

No
No
No

No
No
No

Yes

Yes

Yes

Yes

No

No

2.3.2 New Methods

To improve the detection of high ohmic earth faults new
methods have been developed. The main features are:

Faulty
section
locat.

relative measurement of signals

centralised signal processing

Admittance Method
The admittance method developed in Austria determines
the unbalance of the zero sequence admittances of every
feeder. The unbalance of the zero sequence admittance
can be caused by the unbalanced network impedances or
by high ohmic earth faults. The unbalance for feeders
with symmetrical impedances under no fault conditions
is zero [10].
The calculation of the unbalance takes place every 20
ms. If the unbalances of all the feeders are below a preset value, these unbalance values are stored as reference
values for every feeder.
The behaviour of the admittance unbalance of every
feeder is continuously monitored. If the difference of the

magnitude of the unbalance compared to its reference

value exceeds a threshold value, an earth fault is
detected. Because of the relative measurement this
threshold value can be set very sensitive.
Current Method (DESIR-Method by EdF)
If the zero sequence voltage is not available, the
vectorial position of the zero sequence currents relative
to a computed reference voltage can be monitored. The
direction of the in phase component of an earth fault
current is opposite to the direction of unfaulted feeders.
This method can be applied for steady state signals
(static DESIR-method) as well as on steady state relative
measurements (dynamic DESIR-method).
Practical Experiences
Field tests and practical experiences show excellent
results. With DESIR-method and admittance method
earth faults up to 50 kohms can be detected. The
sensitivity of the admittance method can be set for earth
fault detection up to 100 kohms (proved by field tests).
The practical experiences with the admittance method at
OKA confirm the reliability and sensitivity of the device.

2.4 SUSTAINED OPERATION WITH EARTH

FAULTS
Because of the increasing demand of decreasing the supply
interruptions experienced by the customers, there is a
rising interest in sustained operation with earth faults also
in public medium voltage distribution networks. Sustained
operation has so far been used in Austria and in some
utilities in Germany. In this section, the sustained
operation is discussed according to the Austrian
experience.
2.4.1 Austrian MV-networks
The neutral treatment of the MV-networks of the public
supply in Austria is quite unique:

6,0
93,3
0,03

%
%
%

0,7

isolated
compensated
low impedance
grounded
short term low
impedance grounded

Except some smaller networks that are operated isolated,

almost all the MV-networks are compensated with
Peterson coils. In isolated and compensated networks
sustained operation with permanent earth faults is usual.
Few rural compensated MV-networks are equipped with
short term low impedance grounding in case of an earth
fault. In these networks no sustained operation with earth
faults is performed.

Table 2.3. Neutral treatment of public MV network in Austria.

neutral treatment
isolated
compensated
low impedance grounded
short term low impedance grounded

2.4.2

Voltage
10 kV
2.824 km
7.871 km
15 km
-

Range of application, restrictions, limits

A sustained operation with a permanent earth fault is

allowed only under some conditions that are further
explained. The following chapters are valid for
compensated MV-networks only. The values given here
correspond to the present Austrian standards. Values
defined in the new European Standard prEN50179,
presently in final discussion, are given in parenthesis
.
Touch and step voltages
All earthing systems must be designed so, that the earth
potential rise is less than 125 V (150 V). In this case no
special proof of the touch voltages is required.
If 125 V (150 V) cannot be kept, explicit measurements are
allowed, to prove that the touch voltages are below 65 V
(75 V). Alternatively equivalent means ( building, fences )
are required to prevent too high touch voltages. The
requirement for step voltages is not defined explicitly.
Common earthing system for high voltage and low
voltage systems
For TN-type LV-systems the earth potential rise must be
less than 65 V (75 V, 150 V if the PEN conductor of the
LV-system is grounded at several points outside the
common earthing system). For TT-type LV-systems the
earth potential rise must be less than 125 V (250 V).

20 kV
249 km
27.733 km
151 km

30 kV
98 km
14.111 km
237 km

Influence on communication networks

Communication networks may be affected by capacitive,
inductive or resistive influence. Capacitive influence has
not to be investigated. Inductive influence has not to be
investigated, if the resulting earth current is less than 60 A
in MV-networks up to 20 kV or less than 67 A in 30-kVnetworks. If these currents are exceeded, the inductive
influence at double-earth faults must be investigated. Up to
these currents temporary earth faults are considered as selfextinguishing. Resistive influence has only to be
investigated, if the above current limits are exceeded.
Influence on pipe lines
Investigations on inductive or resistive influence on pipe
lines have to be performed for networks with nominal
voltages of 110 kV and higher.
Immediate start of fault locating
A sustained operation with permanent single-phase-toground faults is permitted, if the process of fault locating is
started immediately. After locating the fault a sustained
operation may be continued, if it is ensured, that no danger
for persons can arise.

Duration of sustained operation

There are technical restrictions for the duration of a
sustained operation with earth faults:

Phase-to-ground voltage transformers may be

destroyed by the increased voltage of the nonfaulted phases after 8 hours.

Peterson coils are usually designed to sustain the
full earth fault voltage for 2 hours.

Multiple faults. The increased voltage of the nonfaulted phases is a stress on every phase-to-ground
device. The probability of multiple faults is
increasing with time and with the extension of the
network.
Single phase interruption
Single phase interruption can cause a residual voltage
similar to a single-phase-to-ground fault. Locating this type
of fault by switching operations may become rather
difficult, therefore the time of sustained operation is long.
Single phase interruption in the MV-network unfortunately
can cause unbalanced conditions in the LV-network with
damage of consumer equipment.
2.4.3 Fault location
The process of locating a permanent earth fault in
compensated networks is a four step sequence:

detection of the fault

detection of the faulted feeder

locating the faulted section

locating the fault
Detection of the fault
The detection of earth faults is usually done by monitoring
the residual voltage by a relay. If the residual voltage
exceeds e.g. 25 % of nominal phase-to-ground voltage an
alarm signal is generated. The alarm signal is usually 5 s
delayed to prohibit a signal for temporary selfextinguishing faults. High resistive earth faults cannot be
detected by this simple relaying. Such faults can be
recognised by analytical methods (e.g. controller of the
Peterson coil).
Detection of the faulted feeder
The detection of the faulted feeder is performed by
protection devices, by the controller of the Peterson coil or
by switching operations in the HV/MV-substation.
Locating the faulted section
The faulted section of a long feeder is located usually by
switching operations in the network outside the HV/MVsubstation. If there are normally open rings to other
HV/MV-substations or to sound feeders of the same

substation, the location can be performed without

interruption of supply.
Locating the fault
If further sectionalising of a feeder is no more suitable, the
location of the direct fault point is done by patrolling the
line. Present R&D tries to develop reliable methods for the
detection of the fault and of the faulted feeder. Reliable and
accurate methods in compensated networks for locating the
faulted section are presently not known.

3 AUTOMATIC SWITCHING SYSTEMS

In this chapter, different solutions for automatic fault
location, fault isolation and supply restoration are outlined.
The alternatives considered are pause-switching, pauseswitching complemeted by signalling and full automatic
systems based on the integration of relays, SCADA and
AM/FM/GIS systems.
3.1 Pause-switching
General
Pause-switch are commonly used in radial operated
medium voltage (MV) networks. The circuit-breaker
located at the feed-in of the spurline switchs off the
overhead line or cable at overload or short circuit
conditions. It is frequently also responsible for the
protection of further transformer substations on this
spurline, which are insufficiently or totally unprotected
(Fig. 3.1).
Fitting out these substations with circuit-breakers and
appropriate protection equipment to enable more selective
disconnection fails often due to the lack of room or due to
the high purchase costs. Therefore a fault leads always to
the disconnection of the total spurline. Fault locating,
disconneting the faulty line section and connecting the
healthy parts is complicated and time intensive.
Function principle
The principle of pause-switching helps to achieve a
selective disconnection of faulty lines in corresponding
designed nets in an economical manner. Economical
because the investment costs are low and no additional
room, which may not be available at all, is required.
The occurrence of a fault in the net will lead to
disconnection of the spurline by the circuit-breaker, located
at the feed-in. Reclosure however can take place after a
relatively short time period, there is no need to wait till the
faulty part is located and disconnected by the staff. The
reclosing of the circuit-breaker will be done automatically
by a timing relay.

Within this settable intervall between tripping and

reclosing all isolating points between the circuit-breaker
and the fault location will open automatically. If the fault
is located somewhere between the circuit-breaker, at the
beginning of the spur line, and the paus-switch the
reclosure will lead to another tripping of the circuitbreaker, comparable to an unsuccessfull autoreclosing.
The operation of the pause-switch take place under
deenergized and off- circuit conditions. Nevertheless the
pause-switch is build from an upgraded switch
disconnector and therefore also capable of switching
operations under normal load conditions.
The pause-switch is a loadbreaking capable device with
direct overcurrent release. The tripping linkage of the
direct release starts a settable time delay. Expiration of the
setted time results in a tripping of the switch disconnector.

The faultless working depends on the correct setting of the

dead time tp of the circuit-breaker and the time delay tv of
the pause-switch, which both depend on the structure of the
net.
Figure 3.1 shows an example: The tripping time ta2 of the
overcurrent relay of the circuit-breaker  at the feed-in of
the spurline is usually fixed by the time grading schedule.
As the starting condition of the direct overcurrent release
of the paus-switch  cease to exist after the tripping of the
circuit-breaker  the tripping time ta3 has to be shorter than
ta2 . The time delay tv has to guarantee that the paus-switch
trips not before the line is disconnected. As a failure of the
circuit-breaker  cannot be excluded it is wise to set the
time delay tv greater than the tripping time ta1 of the
superior circuit- breaker .

ta1

ta2

tp

1
ta3

tv

circuit-breaker
feeder

circuit-breaker
feed-in spurline

3 switch disconnector
t (sec)
Fig. 3.1. Time grading scheme of a pause-switching system.
Conclusion
Pause-switching is an economical solution to improve the
power supply reliability, where certain circumstances stay
against the use of circuit-breakers instead.
Field of application
The pause-switch was put in action mostly during the 60s
in Germany. Despite the economical advantages only a few
electricity companies made use of them. Today new
solutions are considered, e.g. signalling between the two
stations in combination with modern compact relays.

3.2 PAUSE-SWITCHING COMPLEMETED BY

SIGNALLING
In this section, the fault management practices in the medium
voltage distribution networks of ENEL are presented. The
items covered are the extension of a pause-switching system
with signaling and an application where also fault indicators
are integrated in the automatic fault management system.
In the case of overhead lines, the MV network fault
management is based on secondary substations with SF6
insulated motor driven switchgear and voltage detectors. The
control of these is based on a combination of local logic
similar to that described in the previous section and data

transmission based on the distribution line carrier (DLC)

technology.
The restriction of a DLC system is that the communication
path can be blocked by the fault. To mitigate this problem,
when there is a fault, all the switches of the faulty line are
opened by the local logic after a certain time delay. The
existence of a fault is noticed by the voltage detectors as a
disappearance of the voltage. Next, the circuit breaker at the
HV/MV substation is reclosed, and if the first line section is
sound, communication up to the first line switch is possible.
This switch is then closed by the DLC-system. If there is no
fault in the corresponding line section, the procedure is
repeated for the next line switch etc. When the faulty section
is finally found, the voltage disappears again. The loss of
voltage, immediately (less than 1 s) after the reclosing, is
detected by a timer circuit and the last switch is opened
again. Next, to complete the procedure, the circuit breaker is
reclosed with a time delay longer than the previous one, but
short enough to prevent the reopening of the other line
switches.
In the case of underground cables, the MV network fault
management is based on the secondary substation equipment
similar to the above, but complemented with fault detectors.
In this case, the procedure is somewhat different in order to
avoid the reclosing of the faulty section. For the operation of
the automation system, consider Figure 3.2:
When a fault happens, the circuit breaker of the supplying
HV/MV substation is first opened. During the dead time of
the line, the remote terminal unit at the substation 1 (RTU 1),
sends a message downstream to check the status of the fault
indicators at the second substation. Having received the
message, RTU 2 responds to RTU 1 and sends a

corresponding message to RTU 3 (Fig. 3.2 b). From RTU 3

there is no response, which means that either the fault
indicators at substation 3 are not 'on', or that the
communication path has been blocked by the fault. Now
RTU 2 makes a conclusion, that the fault is in the following
section, opens the corresponding switch and sets it in the
'blocked' mode (Fig. 3.2 c). Next, after a certain time delay,
the circuit breaker is reclosed, and the substations are reenergized. The return of voltage for a sufficiently long period
resets the fault indicators connected to the sound sections
(Fig. 3.2 d).
The goal is to provide 10% of the secondary (MV/LV)
substations by an automatism described above. For this a
new generation of secondary substation equipment has been
designed. In particular, ENEL has decided to use SF6
insulated motor driven switchgear, because the power of the
motor and the capacity of the battery are reduced, due to the
lighter masses of the moving parts.
As regards the injection of the transmission (DLC) signals, a
new generation of MV/LV transformers have been
developed, to allow the installation of the coupling device
through a plu-in operation on the transformer case.
The automatic switching systems described here is a part of a
more extensive automation system, the aim of which is:

remote contol of distribution power network

automation of metering service for MV and LV
customers

The system is called SITRED.

a)

b)

c)

d)

B RE AK ER CLO S E D

BREAKER OPEN

S W IT C H C L O S E D

S W IT C H O P E N

FA U LT D E T E C TO R O N

FA U LT D E T E C TO R O FF

Fig. 3.2. The MV network fault management procedure of ENEL in the case of underground systems.

NCC

NCC

3.3 METHODS BASED ON THE INTEGRATION

OF RELAYS, SCADA AND AM/FM/GIS
3.3.1 Introduction
An automatic computer model, called the FI/FL model, for
fault location, fault isolation and supply restoration is
presented. The model works as an integrated part of the
substation SCADA and medium voltage distribution
network automation systems.
In the model, three different techniques are used for fault
location. First, by comparing the measured fault current to
the computed one, an estimate for the fault distance is
obtained. In the second phase, this information is
combined, in order to find the actual fault point, to the
data obtained from the fault indicators in the line branching
points. As a third technique, in the absence of better fault
location data, statistical information of line section fault
frequencies can also be used.
Once the faulty section is identified, it is automatically
isolated by remote control of line switches. Then the
supply is restored to the remaining parts of the network. If
needed, reserve connections from other adjacent feeders
can also be used. During the restoration process, the
technical constraints of the network are checked. Among
these are the load carrying capacity of line sections, voltage
drop and the settings of relay protection. If there are
several possible network topologies, the model selects the
technically best alternative.
3.3.2 System integration
The automatic switching system is based on a close
integration of substation telecontrol (SCADA), network
automation, protection relays and AM/FM/GIS systems. In
the solution presented in this section, the integration of
substation telecontrol and protection relays is based on the
Inter Bay Bus (IBB), which enables spontaneous data
transfer of events from bay level to station level (Figure 3.3).
The protective modules transmit the measured current and
voltage values, and the additional fault information, via the
serial bus to the station level unit (SC) of the SCADA
system. The same data is transmitted through the
communication gateway to the main SCADA computer in
the network control center.
The main task of a SCADA system is to provide a real time
interface to high voltage / medium voltage substations and to
the remote controlled points in the medium voltage
distribution system. The role of the AM/FM/GIS system is to
form a data bank of the detailed network data. It is also used
for real-time presentation of the network connections.

MMI / SC

GW1

S t a r
Bay
Controller

GW2

C o u p l e r
Bay
Controller

Inter Bay Bus

Bay
Controller

Fig. 3.3 The principle of the integrated station automation

and relay protection. NCC is a network control center, SC is
the station controller, MMI is the man-machine interface,
GW is the communication gateway and the Bay Controller is
the integrated protection and control unit.
The FI/FL module works as an integrated part of the
substation telecontrol system. After the FI/FL module has
been started, the communication is between SCADA
system and the network controlled points only. In this
phase the commands sent by the SCADA include the line
switch close/open operations. At every step of the fault
management process, the status of feeding circuit breaker
at the substation is also checked, in order to detect possible
closings against a faulty line section.
3.3.3 Fault location techniques
In the fault location part of the FI/FL model, three different
sources of information are used:
1) the computed fault distance
2) the data from the fault indicators in the line crossings
3) the statistical fault frequencies of different line sections
For obtaining the computed fault distance, the measured and
the computed fault currents are compared. As a result, the
estimated distance of the fault from the substation is
obtained. This distance is then, in turn, compared to the
network diagram, and the possible fault locations are
obtained.
When analyzing the readings of the fault indicators one has
to take into account their possible misoperations. It is also
often the case that only a part of the line sections concerned
are equipped with fault indicators. These problems are
mitigated by statistically combining the indicator readings.
The third source of information are the statistical fault
frequencies of different line sections. This data is useful
especially if the network is a mixture of overhead and
underground line sections. In the latter ones the fault
frequencies are substantially lower than in the former ones.
The final phase of the fault location procedure is to combine
the information from different sources by fuzzy logic. The
final result of the fault location module is the probability

weights for each line section. These weights express the

likelihood of the fault being located in the line section
concerned. In the next step of the fault management
procedure, this data is submitted to the fault isolation and
supply restoration model.
3.3.4 Fault isolation and supply restoration
Once the fault location has been analyzed by the FL
submodule, the automatic function for fault isolation and
supply restoration (FI submodule) is executed. The control of
the network switches is done in such a way, that the line is
processed zone by zone in the smallest possible remote
controlled sections. After every reclosing of a line switch, the
status of the feeding circuit breaker is checked, and if there
has been tripping, the line section concerned is regarded as a
faulty one.
The operation of the FI model is optimized so that the
number of outages experienced by the customers connected
to the faulty line is minimized. For this, there are two
alternative approaches, depending on whether the faulty line
section will be verified by trial switching or not. In the trial
switching mode, the FI model first tries the line sections with
the highest fault probability. When the faulty line section is
encountered, it is isolated, and the remaining sections are
energized. The trial switching mode actually performs a
delayed autoreclosing. This mode is recommended in the
cases, where there is some degree of uncertainty in the fault
location, or where it is desired that the traditional delayed
reclosing by protection relays be replaced by a zone-by-zone
reclosing.
In the second mode, the aim is to avoid the trial switchings.
Here the line sections are energized starting from the smallest
fault probability and when only one section is left, the
procedure is stopped. This mode is usually used only if a full
reclosing procedure is first performed by the relays. To be
effective, this mode also requires a good knowledge of fault
location. In practice this means, that there must be fault
indicators mounted at the line crossing points where the
remote controlled switches are.
The FI model first tries to energize the line section by section
from the initial feeding direction. For those sections which
can not be energized in this way, a reserve connection is
used. Before the coupling, the capacity of the reserve
connection is checked for thermal load carrying ability and
for the maximum voltage drop. Also the settings of the
protection relays of the feeding circuit breaker are checked in
order to ensure relay coordination.
3.3.5 Other automatic models for fault isolation and
supply restoration
SARA is a software package that aims at performing
automatic switching experimented for several years on a
EDF control center. When a fault occurs on a feeder,
predefined restoration actions are recorded. After checking

that the actual topology match the recorded one, switching

orders are sent on circuit breakers. SARA is a package that
is installed on the telecontrol system. The experiment was
successful on the experiment done in several primary
substation. Total amount for fault restoration was strongly
reduced for the faulty situation where SARA was running.
However, the system cannot easily be generalized since it
supposes that the network is in a normal topology. So, EDF
decided to put effort on a new tool, named AUSTRAL, that
provide assistance rather than automatic switching
applications (see chapter 4).

4 SOFTWARE FUNCTIONS FOR DISTRIBUTION

NETWORK OPERATION
4.1 Introduction
4.1.1 The role of SCADA functions for fault
management
This chapter discusses how the remote control system and
associated software functions can provide assistance to
operators for fault management.
Valuable assistance to network operation staff for fault
management will be provided in the following main tasks:

Incident management itself : minimum power cut

duration (diagnosis, fault location, suggestion and
implementation support for power resumption plans) ;
Improved customer information ;
Preparation of network installation and repair
operations (new structure set up, maintenance
shutdowns): operational network modelling and
preparation for manoeuvre sequences enable to
anticipate possible fault situation.

4.1.2 Major function categories

Software functions for fault management can be broken
down into three main time-scale categories:

Real-time functions address the current network

situation, providing assistance in real-time network
control, possibly including direct action on network
structures. The main real-time functions are: event
analysis and diagnosis; fault location; power
resumption; field crew management; and trouble call
analysis and customer information. Real-time function
need a connection in real-time to up to date data from
the SCADA system.

Preparatory functions are provided to help operators

study simulated network situations representing
configurations which are liable to arise or be adopted
in the future (e.g. incident or withdrawal
configurations). With the kind of function, the operator
is primarily seeking to determine baseline

configurations which will be capable of handling

future network situations. This kind of functions
require from the SCADA system a copy of actual or
forecasted network.

Hindsight analysis functions are implemented to

investigate actual-fact operating conditions over
elapsed periods of a day or longer, usually by
analysing records of network operation data. They
generally output results in the form of statistics, quality
indicators, or quantified reports on customer supply

disturbances. This can be very valuable for fault

management by enabling an adapted predictive
maintenance.
Because of the variability in function location we do not
attempt to specify a "standard" arrangement, and we do not
use existing nomenclature (e.g. DMS, or AM/FM/GIS),
which is often poorly defined and sometimes inconsistent.
The figure shows an global view of the telecontrol systems
and the above three main set of functions.
analysis
predictive maintenance

Network

hindsight
analysis

Export
- Customer information
- Economic Design
Remote signals
Remote measurement

Remote orders

SCADA

Action

Supervision
and diagnosis

Elaboration

real-time
study mode
Diagnosis, alarm
processing
Fault location
Service restoration

Load transfer
Outage management
Field crew
management

Study mode
training

Figure 4.1 : Overview of functions

4.1.3 Open discussions

Centralized or decentralized
Should the new functions be installed on the SCADA
system, in a centralized position or should be they
decentralized inside every substation. We note that a
compromise must be struck between improved functions
for the network operating system and improved functions
integrated in new generations of source substations.
Specifically, steps should be taken toward greater
complementarity.
The observed trend in primary substation is towards more
and more intelligence, digital and computer-based
technologies. Some functions that has to be previously

implemented in the central control system cam be now

partly implemented at the substation level. Coupled with
improvement
in
information
technologies
and
telecommunications, data exchange between the SCADA
system and the primary substation will become in a near
future a quite solved problem.
However, rather decided to implement a given function
into the telecontrol system or into the primary substation, it
seems to be more interesting to find complementarily. For
instance, a diagnosis function based on event analysis is
better implemented closer to the field (primary substation).
But as soon as the topology is required, an implementation
in the telecontrol system is best suited. A shared
implementation based on data exchange is certainly the
target for future developments.

Close the loop : decision support function or also active

function?
Another point should be addressed. Until now, functions in
the telecontrol system has been almost always specified as
off-line functions. More precisely, they has been seen as
function that inputs data from real time but don't perform
any action directly on the network. The operator has always
been included in the loop. Automatism that directly acts on
real-time operation was implemented at the field level
(protection, substation). Current developments observed
now let predict that the loop will be closed "one level
higher" in a near future. Some functions implemented at the
telecontrol system level are specified to directly send order
on the network (for instance, a fault restoration function).
So, the strong distinction between automatism (necessarily
on field) and analysis and decision support (necessarily on
SCADA) is will certainly become less true in a near future.
Both will be performed at any level.

4.2 SCADA SYSTEM, DATABASES AND BASIC

SCADA FUNCTIONS

course is a serious advantage if we consider using database

information in fault management programs who are not
developed by the SCADA constructor.
The upgrading of the systems can be divided in 2 types :
upgrading of the energy control software and, upgrading
with new data on the controlled network. Modern systems
allow this upgrading actions during real-time operation.
This is the result of both the client-server architecture and
the fact that most systems work in a master/slave
configuration.
All recent systems provide tools for self examination and
self-control of the SCADA system.
Network monitoring, network control
The basic real-time functions can be divided in two major
items : network monitoring and network remote control.
All modern systems allow monitoring of status's (switches,
breakers, short-circuit indicators, etc.), analogue (current,
voltage, temperature, etc.) and digital values and
accumulators (pulse counting).

4.2.1 Definitions
The kernel of the telecontrol system is basically composed
of the following modules :

the SCADA system for data acquisition. The SCADA

system is connected to RTU and manages a real-time
data base.
the graphical display that presents to the operator the
interface for the complete telecontrol system (network
representation, data visualisation, remote order, etc.).
an archiving system.

Plugged on this kernel, we may find the various functions

illustrated on Figure 4.1.
4.2.2 SCADA and platform
Hardware, operating systems and architecture
The majority of the current SCADA platforms are client
server systems in a master/slave or master/hot-stand-by
configuration. Most used operating system is UNIX on 64bit machines but platforms running under Windows NT on
powerful PC's are also used. This of course is a
consequence of the evolution on the PC market (less and
less limitations and more and more processor capacity).
The use of an NT environment gives advantages on behalf
of the user interface and on behalf of the connectivity with
other software tools running in the different companies. A
combination of both operating systems also exists : main
server under UNIX and workstations with Windows NT.
The database used is mostly of an open standard type who
can be easily accessed from other applications. This of

Alarm handling and alarm processing is available

everywhere. Remote control is principally possible on
every object (switch, set point, etc...) in the network (as far
it is equipped to be controlled).
Front-end (FE)
Database
Two philosophies are used : FE's with particular database
and FE's as client on the SCADA main server. Both
systems have pro's and contra's.
FE's with particular database give the opportunity to
connect to different central control systems. Normally
generic data formats are used in this cases. Final formattranslation is then done on the receiving SCADA platform.
Biggest advantage of this method is the fact that
transmission lines to the different controlled stations can be
reduced and that data are shared after the FE. In case of
failure of the central control devices, in most cases reduced- control operations can be done from the FE
consoles.
A disadvantage is that the data have to be placed and
upgraded in at least two different databases (redundancy
problem).
FE's as client on the main SCADA server get their
network-configuration-data from the main SCADA server.
Redundancy is not a problem in this case and the result is a
more simple architecture. However if the main server fails
no control actions are possible. In this case a hot-standby
configuration is a must.

Exchanging network control with other SCADA platforms

can be done with additional servers where the desired data
is presented. Links can be made with point to point
connections or over the Internet.
The FE as client on the SCADA server is the most common
solution in the examined countries.
Protocols
Most recent FE-systems allow the standard exchange
protocols. These protocols run mostly in a multi protocol
mode.
Polling / Transmission to the control centre
The combination of cyclic polling and polling on demand
is normally used. For the transmission both point to point
lines and switched networks (private and public) are used.
4.2.3 Data and database
Most of the functions outlined above are based on the use
of a fixed database storing data from the substations plus
data on the power system structure. Configuration of this
database will often prove long and fastidious. To ensure
that the functions receive valid input data, it is nevertheless
primordial to ensure that the data in the database is up to
date and coherent. This task is not always easy, given the
data models and technologies offered by most suppliers.
Moreover, some functions may need to access data stored
in several databases, which might not even be operated by
the same company (e.g. trouble call functions will need to
merge data from network operation and customer
information databases).
4.2.4 SCADA hook up and Open architecture
Most of the functions outlined above require fairly tight
coupling to the control system (and to the SCADA in
particular). This can be a very complex matter, because
control systems often use outdated technologies, and there
are no standardised data exchange methods. We should,
however, note that international working groups are
addressing the issue of standardised data exchanges
between control centres. An EPRI working group has put
forward a data exchange model, and there are two IEC
working groups assigned to this subject.
Control application (control software)
Practically all new systems work with open utility standard
control software. Tools as visual basic are commonly
provided to give the customer the opportunity to tune the
system. This is a very important feature for automatic fault
detection and load restoration.

4.2.5 Basic SCADA functions

Graphic Interface
More and more functions display results on user interfaces
featuring a graphical representation of the power system.
There are many types of representation, ranging from
schematics to geographical maps. Semi-geographical
representations are highly appreciated by control centre
operators, because they give clear analytical views of the
power system, providing a good perception of the relative
locations of different devices. Geographical representations
are needed when the control centre is required to manage
field crews (i.e. assess travelling distances and probable
repair times). For this reason, control centres are
experiencing an increasing need for hook-ups to the utility
company's GIS; a single centralised point for geographical
data entry avoids the risk of inconsistency between
databases, and eliminates the need for duplicate data
capture.
Strong visualisation ameliorates the knowledge of the
status of the network. Schematic flow displays with
dynamic coloration give an immediate view on the impact
of a fault and the result of fault restoration efforts. This
feature increases drastically the control capacities because
it gives a global overview of the network structure.
Graphical trending and digital trending are features that can
be used in the development of a load forecast application.
In the margin of the graphical tools the different operation
modes are also to be mentioned. Study mode and
dispatcher training mode are mostly graphically orientated.
They give the opportunity to simulate faults and events on
the network. The results of these simulations can be of
great value for mathematical and logical models for fault
restoration.
Archiving / Reports / Overviews / Study mode
All modern systems have these possibilities. The fact of the
open database structure makes it possible for the user to
customise his view on the events and alarms. Exporting
these data from the SCADA database to other
environments gives even more opportunities to examine
these files.
Extra features as "net freeze" and "snapshot" give helpful
hints and historic data.
State estimation, topology analysis
State estimation and topology analysis are basic features
that can be provided to many functions.
Topology analysis involves inputting the static network
model plus information on dynamic switching states in
order to perform connectivity analysis and construct a realtime network model. This model may include load flow
calculation (see Electrical System Computation below). As

well as being a basic function of considerable value to

operators in itself, topology analysis is also a prerequisite
for most advanced functions. Results are presented to the
operator as dynamic colouring on the user interface. Each
feeder has its own colour, which is dynamically updated as
switches change position. The function detects abnormal or
erroneous switching states causing closed loop situations or
de-energisation of line sections and substations. Deenergised line sections and meshed feeders are displayed
on screen to help the operator locate the switching state
errors.
Topology analysis is based on graph analysis. Dynamic
colouring suppose a efficient algorithm in order to provide
real-time visualisation.
State estimation and topology analysis is becoming a
standard function, offered by most remote control systems.
A quite efficient algorithms are necessary in order to
support real-time visualisation and real-time functions.
We must note, in addition, that the actual topology may
differ from that recorded in the database because some
devices on the network are not remote-controlled. If these
are manual devices, the database can be updated with
information received by the control centre, but if they are
automatic devices it is much more difficult to ensure
database accuracy.
Electrical system computation
Electrical system computation affords improved
understanding of network behaviour, to help operators
check compliance with technical constraints, supply quality
requirements and protection schemes. The main
computations likely to be of value to distribution network
operators are as follows :

simulations for forecasted situations. So, electrical system

computation appear to become the basis of many new
advanced functions. Many software packages are available
from SCADA vendors and specialised software companies,
but they are not yet widely used for real-time
operation. However, most of the European utilities have
plan to integrate a load-flow computation for real-time
operation.
Load estimation and load control
Although not commonly introduced, these features are on
the list of nearly every company. This is quit obvious
because load forecast and load control can have a direct
impact on the price of the energy.
Depending on the automation degree of the SCADA
platform, load monitoring, load forecast, adaptive load
forecast, automatic load switch-off and tariff applications
are running on some SCADA platforms. In some cases
these applications are running on a different platform.
Area of responsibility
All new systems give the opportunity to virtually divide the
database so that the network can be monitored in isolated
parts. This multi-control feature gives the customer the
ability to split the controlled area in mini control centres.
This can also be used to control several energy forms
without interference (gas/electricity/data).

4.3 REAL TIME FUNCTIONS FOR FAULT

MANAGEMENT
4.3.1 Diagnosis, alarm processing and event analysis
Technical description

power flow (to check device current limits)

voltage drop (to check quality of supply of customers)

The main difficulty concerns data availability and

accuracy. Load profiles are the weakest point, because
loads can only be estimated from measurements on feeders.
The actual load may differ considerably from the estimated
value used for computation purposes.
In addition, electrical computation may require detailed
technical data that might not always be available during
real-time operation (e.g. : electrical features of lines and
cables).
Most of the electrical system computation packages
available for distribution networks assume a radial network
configuration, but it may be useful to compute special
meshed network situations (e.g. for load transfer).
Electrical system computation are needed to estimate the
actual electrical features on the network and to perform

Diagnosis, alarm processing and event analysis functions

input a stream of events from the network (remote
signalling data, mainly) in order to output diagnostic
information and event-analysis messages, for real-time
transmission to the operator (via the remote control system
log or in a dedicated window).
The precise term for the function ("diagnostic", "event
analysis" or "alarm processing") will depend on whether
the system accommodates automatic device management
and supports the notion of alarms.
Basically, this kind of function works by matching
incoming signalling and alarm data against predetermined
patterns that characterise the messages and diagnostic data
to be output for the operator.

Difficulties

Difficulties

The main difficulties affecting this kind of function are as

follows:

real-time connection

availability and validity of data (especially data
describing the protection system)

update of expert knowledge

The first method requires a specially fitted network, which

can represent substantial investment. The cost of the
second solution is not exactly negligible either.
Another difficulty concerns the reliability of fault detectors
and transmission systems.
Benefits

Benefits
This type of function brings the following benefits:

improved understanding of actual network conditions

and incident situations, for faster and more appropriate
remedial action.
diagnosis of failings in protection system.

Existing and operational functions

The most common situation is that such functions are not
yet operational in the utilities from our study. However,
some advanced project can be mentioned. Several EDF
control centres are about to be equipped with a real-time
diagnosis function. Experiments with the diagnosis
function from the EDF AUSTRAL package has started in
two sites (1998).

Precise fault location means faster remedial action by field

crews and by operators.
If the network is fitted with remote-controlled switches,
power can be resumed for customers on certain network
sections without having to wait for action by field crews.
Existing and operational functions

4.3.2 Fault location function

Technical description
Networks use various devices for locating feeder faults
(network fault detectors, fault distance estimation devices).
Some of the information output from these devices is
integrated in the remote control system, to facilitate the
implementation of fault location functions of varying
degrees of sophistication. Fault location will be used :

by operator for fault isolation, using remote control

devices,
and by field crew management for fault repairing.

For the latest purpose, fault distance calculation is very

useful to ensure a fast repairing. One method involves the
use of detectors in the network. Location is simple, since
faults are located between two adjacent detectors. The
accuracy therefore depends on the number of detectors in
the network..
A second method involves measuring the fault current
(using digital protection systems or oscillographs) in order
to calculate the distance to the fault. However, in a treestructured network there will always ambiguity in
determining the defective section. Greater accuracy and
reliability will therefore be obtained by correlating
calculated values with data from the fault detectors.

Several Finnish utilities are using since 1993 a fault

location software that combine fault distance
calculation and fault detectors information.
An experiment has started (1998) in one EDF control
center using the fault location function that is included
in the AUSTRAL package. This function is based on
fault detector on networks.
Fault distance calculation is under experiment at EDF
for several years in several distribution centres.
Remote access to the fault distance calculation is
performed. But connection to the DCC and integration
into the DMS package is not yet realised
In OKA (Austria) a "Fault Location Function" is used
in a semi real time modus. After a fault in the
distribution system the measured reactance of the
tripping distance relay is sent to the DCC. A program
is started and the probable fault locations (generally
more than one) are presented to the operator on the
screen in alphanumerical form. This procedure if
performed automatically. Together with remote
monitored fault detectors and fault detectors read by
the field crew, the operator decides what tree of the
feeder is most probable the faulted one. "
In ENEL the STM system provides a function which is
able , in case of fault, to detect the section affected by
the fault between two automated substations: The
faulted section is then isolated by opening the switches
in the substations and the rest of the line can be resupplied.

4.3.3 Power restoration

Technical description
Network reconfiguration may be required in the following
situations:

Permanent fault on a portion of the network: the

defective portion will have to be isolated while power

is resumed on other fault-free portions that were

blacked out when the protective device tripped.
Preliminaries for work on a network section: the
section will have to be isolated from the network, and
load transferred to ensure uninterrupted customer
supply.
Overload on a MV/HV transformer: MV/HV
transformers are designed to withstand only temporary
overload, so when an overload situation arises, the
controller will have to transfer load to another
transformers.

Reconfiguration is possible because of the meshed

arrangement of the distribution network. Load transfer
plans seek solutions capable of resuming power or
providing back-up to the largest number of customers,
while complying with electrical engineering constraints.

Existing and operational functions

In ENEL, in case of fault the system gives

automatically the list of the personnel to be called for
intervention.
A real-time positioning system on road map is used in
Electrabel.
The
system
works
with
beacons/transmitters placed on the Belgian border
allowing to give the position and two-way data
transmission. The crews give the status of the task they
are working on (started, finished, etc.). This
information is transferred to the positioning system, to
give the control centre a real-time overview of where
the crews are and what is the job situation.

4.3.5 Trouble call analysis and customer information

Technical description

Power resumption and load transfer functions involve

optimisation problems that can be solved using a number of
different techniques (classic optimisation methods,
artificial intelligence, etc.).
Difficulties
These functions involve highly combinatory problems
which require complex algorithms. To suggest valid plans,
the software requires accurate estimations of loads and
transits on the network, and this is always difficult to
achieve on distribution networks.
Benefits
shorter service interruptions, if power resumption plans are
developed promptly
implementation of optimum (or optimised) plans
network operation closer to permissible limits
Existing and operational functions

In Finland, systems from various vendors are used in

several utilities.
In France, experiment has started in 1998. The fault
restoration function of the AUSTRAL package (called
RTP) is experimented on several EDF control centre.

4.3.4 Field crew management functions

Technical description
Field crews must be able to respond to incident situations
as fast as possible, so fault identification and location are
important factors in determining field crew performance. A
basic version of the field crew management function would
offer real-time location of field crew teams (by means of a
GPS system, for example), so that repair operations could
be assigned to the closest team. A more sophisticated
version would seek to optimise team assignments

Customer calls are usually managed using one of two

methods.
The first method involves a voice server connected to the
control system. When the control system detects an
incident, it sets the voice server to issue a pre-recorded
message covering the affected area. Incoming calls to the
voice server can be recognised automatically (using the
caller ID function), or callers can be asked to identify their
locations by pressing keys on the telephone sets. The
system can also be used to store customer voice messages
for rapid examination by an operator. These messages may
provide notification on incidents that have not yet been
detected (for example, incidents can be detected by
identifying calls originating from the same area), or
additional information on incidents that have been detected
but not yet located. Customers may have experienced
abnormal events on the power system, and their reports
may well accelerate the location process, thus accelerating
service resumption as well.
The voice server system can also be used for calling
specific clients automatically, to inform them on the extent
or probable duration of an incident, if the operator has the
necessary information to predict this. The main advantage
of the voice server system is that it can handle large
numbers of incoming calls following an incident. Servers
are today offered by several suppliers, and are becoming
fairly widespread among electric utility companies.
The second method consists in providing the operator in
charge of handling customer calls with a tool for locating
the caller on a map representing the power system and the
electrical state thereof. Location might be performed by
analysing the customer name, telephone number, etc. The
map will give a clear indication of those areas that receive
power and those that don't. If the customer is located in an
incident-affected area, the operator can explain the extent
of the incident and forecast how long the customer will
have to wait for supply resumption. If the customer is

located in an area not affected by a known incident

condition, full customer details are stored. If there are
several calls concerning the same area, a simple algorithm
is applied to identify the branch most likely to contain the
incident. This information is automatically supplied to the
controllers. By providing automatic real-time information
to the staff responsible for customer calls, the customer call
management function ensures that customers are given
relevant up-to-date information, and that information is
conveyed efficiently and promptly among all concerned.
Customer call management is offered by many suppliers of
DMS functions, and is used by many utility companies,
especially in the United States. The greatest difficulty (and
the most significant cost factor) in setting up this type of
function concerns hook up between the control system and
the customer databases.

Analysis also helps the operator determine exactly what

power cuts are experienced by what users. This can be
useful for revealing, for example, whether priority
customers are receiving the stipulated supply quality.

Difficulties

Most system suppliers offer functions for archiving data in

databases, accessible via market-standard spreadsheet
applications like Excel, which operators can use to run
analysis operations as and when required. Unfortunately,
most data storage software does not store control system
configuration data, which means it is not always possible to
recover stored data following power system upgrades.

The main difficulty concerns hook up between the

operations database and the customer database, for
identifying and locating a given customer.
Benefits

Service quality (customer information)

Fast power resumption

Existing and operational functions

EDF has started (1998) the experiment with its

application called Prevenance Incident . Based on a
service provided by France Telecom, this application
provides to customers a phone number where
information on fault that have occurred are available.
This application relies the fault location and the
customer identification to provide appropriate
information to the customer. Direct connection to the
real-time application is scheduled for 1999. The actual
state of each load point, as known by the real-time
operation, will be exported to the Audiotel system.
ENEL has experimented both the approaches in
different areas. The choice of the system that will be
implemented is on the study.

The operational diagram can be modified to improve the

situation pending a more thorough examination of the
problem. Customer-based determination of power cuts also
enables the operator to identify weak areas within the
power system. This can be followed up by field tests to
determine sharply-focused improvement solutions.
Statistical analysis is also useful for detecting defective
equipment. Data acquired at the substations can be
processed to schedule sharply-focused preventive
maintenance operations.

Difficulties
Implementation of this kind of function relies to data
archived from real-time operation. Some analysis require
topology information that have to be recalculate from the
archived data. This can be more or less complex to design,
according to the type of data archived.
Benefits
Analysis are very profitable for predictive maintenance and
for identifying weakness in the network and then decide
further investment.
Existing and operational functions
Every utility has a more or less sophisticated analytical set
of functions that can operate on data archived from realtime. The level of sophistication can range from simple
data request (SQL request) to more complex computation
based on request, sort and filters.

4.4 ANALYSIS FUNCTIONS

4.4.1 Analytical functions

4.5 STUDY MODE AND TRAINING

Technical description

Overview

In a distribution control centre, all available data can be

stored for subsequent statistical analysis, in order to detect
weak spots on the grid (as revealed, for example, by
repeated incidents on the same substation feeder).
Frequent statistical analysis thus provides operators with
early warning of possible permanent fault conditions,
allowing them to take prompt preventive action.

Training is an important feature for fault management. It

may enable a prompter reaction from operator when a fault
occurs. Training software is usually based on a network
simulator. Fault situation are simulated and the operator
has to deal with this situation as he will do in real time
operation.

Technical description
The kernel of the software is the simulator. This simulator
can be more or less complex, according to the target of the
training. It can range from simple state simulation
(switches status) to a detailed simulation of the protection
scheme
Most training simulator are disconnected from real-time
operation and real-time SCADA. Training is often based on
a specific network description designed for training
purposes. Sometime, the training room is not located in the
control centre. However, the trend is to provide integrated
solutions where training is done on a copy of the actual
network.
Difficulties
As for real time operation, load descriptions is an important
and often complex problem.
Benefits
Benefits are obvious. Since fault situations become to be
less frequent, it is important to train operator before these
critical situations.
Existing and operational functions
Training simulators disconnected from SCADA are
commonly available on utilities. Training simulator based
on real-time SCADA are less frequent.

The first class deals with the quality of the service as it is

perceived by the customer. The second one deals with the
faults on the network, so they are useful to detect where to
improve the service. The third investigates on the quality of
the components. The last one can show us how efficient we
are in case of trouble in finding the origin of the fault and
in restoring. These indices can be useful computed also for
operating areas in order to find how and where to improve
the performance of the area itself.
Indices of availability of supply
These indices concern the quality of supply as percieved by
LV and MV customers. A Distribution Study Committee
(50,05 DISQUAL - Unipede) has produced the
prescriptions to compute such indices. In particular the
interruptions of supply are considered in respect of the
effects on a customers.
First distinction is done between PLANNED interruptions
and ACCIDENTAL interruptions. The planned interruption
allow to inform in the most cases the customers in advance.
The accidental interruptions are generally caused by
PERMANENT or TRANSIENT FAULTS and usually are
classified as LONG (if the duration is 3 min or over) or
SHORT (if the duration is up 3 min). So far the following
indices should be provided separately for planned and
accidental interruptions:

5 FAULT STATISTICS
In this chapter the statistics, associated with fault
management functions, are discussed. First a general
overview is given on the data collected in different
countries. This overview is based on an inquiry made in
different countries, and it is considered to give a general
description how this data typically is arranged.
As special cases, two examples are given in more detail.
These are the description of fault statistics in ENEL of Italy
and the use of fault statistics in Electrabel of Belgium.

INTERRUPTION
FREQUENCY:
number
of
interruptions on average per year per customer
(number of interruptions/year)
SUPPLY UNAVALABLITY: number of minutes
wthout supply on average per year per customer
(minute/year)
INTERRUPTION DURATION: average duration of
customer interruptions (minutes/interruption)

Indices of the performance of the network

These indices concern the status of the network and the
components; as they are related mainly to the faults
(transient or permanent) only the accidental interruptions
are considered. They should be provided for MV
CIRCUITS, MV/LV SUBSTATIONS, LV CIRCUITS,
namely

5.1. Main fault statistics indices

Fault statistics indices are computed in different ways in
the company according to the different uses. In order to
emphasize the influence on FAULT MANAGEMENT in
distribution system, it seem suitable to suggest the
following classification.

Indices of availability of supply

Indices of status of network
Indices of availabilty of network components
Indices of effectiveness of emergency teams

INTERRUPTION
FREQUENCY:
number
of
interruptions per consumer per circuit or per substation
FAULT RATE: number of interruptions per consumer
per 100 km of circuit

Indices of availability of components

These indices concern the components of the network; and
usually the fault rate for permanent faults is considered.

Effectiveness of emergency teams

These indices are related to the organization of company in
case fault. In order to monitor the effectiveness o fthe
emergency team commonly the follwing indices are
considered.

alert duration of emergency team

travel duration of e.t.
selection of the faul t part of the line ad resupply time
restore time

5.2 FAULT STATISTICS IN ENEL

From the SCADA systems, which monitor the HV/MV
substations, the following information is derived, taking
into account that normally a short reclosure at 0.3 seconds
and a long reclosure at 30 second are adopted on each MV
feeder:

data and time of each circuit breaker manoeuvre:

opening, reclosing
cause of manoeuvre: overload protection, short circuit
protection, ground fault protection, manual
duration of interruption: transient ( 1 second), short
(1 second < and  3 minutes), long (> 3 minutes)
value of the current immeditely before the opening
number of manoeuvres on the circuit breaker for each
interruption: typically 1 for transient, 2 for short,  3
for long interruption
faulted section of the circuit in case of load switches
on MV feeders locally automated or remote controlled

From the forms filled by

information is obtained:

the agents the following

The threshold and target values are established, according

to the kind of index,taking into account:

the subdivision of areas introduced by the Charts of

Electrical Service for supply quality indices
physical characteristics of the network (i.e. cable or
overhead line, length of the circuit, etc:) for network
performance indices
physical characteristics of the environment (i.e.
keraunic level, pollution, population density, etc.) for
network perfomance indices

These target and threshold values allow to select the

critical circuits (those with indices beyond the target or the
threshold) to be considered for improvement interventions.

MAIN INDICES CONCERNING QUALITY OF

SUPPLY
The main indices concerned with quality of supply to the
customers are the following:
accidental interruptions > 3 minutes

Total duration of accidental interruptions > 3 minutes

Number of accidental interruptions> 3 minutes
Duration of a single accidental interruption > 3
minutes

prearranged interruptions

Total duration of prearranged interruptions

Number of prearranged interruptions
Duration of a single prearranged interruption

accidental interruptions < 3 minutes and voltage dips

stage interrupted: HV or HV/MV, MV, MV customer

or MV/LV, LV, LV customer
type of interruption: only accidental interruptions > 3
minutes, prearranged interruptions
node, branch of the node, date and time of each
manoeuvre, performed automatically or manually by
telecontrol or by emergency teams on site

In real time in each operational area can be desplayed all

the events that have affected the network and equipements
with all the information recorded by SCADA system
The information system provides monthly and yearly per
each MV feeder and each organisational subdivision,
statistical information in terms of:

average values of main indices

characteristic values (threshold and target values
representative of the allowable range of the indices)
number of customers beyond the target and beyond the
threshold

Number of accidental interruptions > 1 second and  3

minutes
Number of accidental interruptions  1 second
Number of voltage dips

5.3 FAULT STATISTICS : METHODS USED IN

ELECTRABEL
Electrabel (Belgium) has created a standard way of making
Fault Statistics which is to be used in the basic form by all
the branches over the country. The branches themselves
can add different items to this standard in order to give it
more significance on behalf of the local situation and the
used materials.
Keeping this kind of statistics makes it easier to have a
view on both the state of the network and the equipment
used. It also allows us to make better investments (a cable
who has failed several times during a certain period may be
used and better replaced).

Because of the fact that the basic form is standard for the
whole company, it is very easy to compare the different
regions on behalf of the number of power failures and the
difference in intervention and restore times. This
comparison on intervention and restoring times can for
example be used as the base for investment in remote
control and command (see example at the end) or as a base
for refining the intervention methods.
Data collected in the basic form is as follows:

Date of fault
Time when fault occurred
Total time in minutes until the last client had current
again (T)
Total number of switchings (remote/manual) necessary
to restore power (S)
Number of MV/LV stations involved (C )
Cause of the fault
On what material did the fault occur
On what main item of the network was it (Cable,
overhead, ...)
Statistical information on the type of material
A calculated value in order to evaluate the weight of
a fault because an interruption time of 10 minutes for
10.000 clients is much more important than an
interruption time of 1 hour for only 25 clients.

Formula : weight = C x T x (1+S)/(2*S) (MV/LV

station minutes)
In the case of a fault which was restored immediately (for
example with recloser) the weight is set to 0.
For example, comparison of the statistics for 96 of a
specific region with the neighbour region, was the start of
investments in extra remote control in the second region.
Evaluation of the figures for 96 pointed out that the
average total restoring time of region 1 (with 1400 remote
controlled switches on MV) was exactly the half of the
average time to restore in region 2 (with very few remote
controlled stations on MV).
These figures made it easy to calculate the volume of the
energy not sold, which resulted on its turn in a cost due
to power failure. Every lost kWh was calculated at a price
of 2.5
   
      



production due to power breakdown varies between 1.25
and 10
        
  

loss of production is the base of claims from the customer.

In this specific example it was shown that an investment for
connecting some 500 crucial MV-feeders on the control
network was profitable over max. 1 year. The cost/feeder to
control it was around 840
    

 

were already motorised and an RTU was almost

everywhere available for the HV/MV side. The profit on
saved crew cost was not taken into account. Faults with
weight = 0 were also not taken into account because f. e.
a successful automatic reclosing gives practically no
interruption.

Most significant facts of the evaluation :

Region 1
faults on cable
1/ 100 km network
faults on overhead
1/100 km network
faults in MV/LV stations
idem/1000 stations
Average time of intervention
Fault frequency/MV-LV station
power shut-off/MV-LV station
restoring time

Although the restoring time of region 1 is the half, the

area is much more spread (larger distances between
clients and towns and the network is much older
(average number of faults is higher). Region 2 is an area
with denser population (larger towns).

Region 2

319
6,4
330
19
108
10,5
1hO5
1,52
2320
31

203
3 (factor 1/2)
28
21
45
3
1h58 (factor 2)
0,37
48
1h03 (factor 2)

Average number of stations/fault : 11,8

Total of faults : 276
Average unsold energy/station : 105 kVA
Price/lost kWh : 2.5

Time gain : 30 min
Profit : 11,8 x 105 x 276 x 0,5 x 2.5 = 427 000
 
Cost : 500 feeders x 840
    

6 COST/BENEFIT ANALYSIS OF FAULT

MANAGEMENT
This paper is an attempt to outline the most important
factors, which will affect the future fault management
applications. The benefits of various functions are first
discussed and, as an example, results of a cost/benefit
analysis are given. Finally two different approaches of
dealing with investment selection and supply quality
appreciation are given.

in overhead lines. The fault often begins with a very high

resistance, and evolves only gradually into a fully
established earth leakage requiring protective actions.
The difficulty with outage cost reduction is, that the benefits
mostly come to the customers, and only a minor share,
corresponding the lost revenues, to the utility. Because of
this reason, several alternative approaches exist for the
appreciation of these costs:
1.

6.1 Benefits of fault management functions

2.
The benefits of fault management applications can broadly
be classified into two categories: 1) direct benefits to the
distribution utility itself or to the customers, and 2) better
adjustment of the utility to the changes in and to the new
demands of the operation environment.

3.

The customers outage costs are fully taken into account

as if they were hard costs
A certain price per kWh is used, less than the customers
costs, but higher than the revenues.
They customers outage costs are ignored

The savings in network investments is the other major

benefit class. Due to fault management, investments can be
reduced mainly for three reasons:

In the first benefit category we have:

1.
- reduction of outage costs
- savings in network investments
- savings in labor costs
and in the second category:
- optimized voltage quality
- better competitiveness of the utility in the free power
market
The basic difference between the two groups is, that the
benefits in the second category are much more difficult to
assess in monetary terms than those in the first one. It is
clear, however, that they will in the longer run be of
extremely high value. In what follows, the benefits in the
first category only are discussed in more detail.
The reduction in outage costs has so far been the major
benefit of network automation. With the remote control of
line switches of overhead line networks solely, for instance,
the distance between the controlled points being about 15 to
20 km, the outage costs have been cut by about a half. The
outage costs can further be reduced by developing fault
location techniques, i.e. fault indicators and computational
fault location.
In the future, also part of the temporary faults and the faults
with high fault resistances can be detected and located. In
this case, the critical fault type is a single phase to earth fault

2.

better knowledge on load behavior, which allows

smaller operation margins
more flexible use of the distribution capacity

Investment savings are made possible by line switch remote

control and effective computerized tools for switching
planning. This is of value especially in abnormal situations.
Suppose, for instance, that the main transformer of a high
voltage / medium voltage substation is temporarily out of
operation. The automation system enables quick switching
of outaged lines to the neighboring stations. Furthermore,
the load of these stations can in turn be relieved by other
adjacent substations. This reduces the investments needed
for reserve connections in the medium voltage system.
A necessary tool here is also the distribution load estimation.
Since this functions gives a better knowledge on the load
currents in different power system parts, it also allows for a
more efficient use of distribution system capacity both in
normal and in fault situations.
Savings in labor costs are expected in operation and
planning of the power system. Examples are routine
switching and trouble shooting in the case of network faults.
In some cases the labor cost savings are the largest benefit
factor of the function. Among this kind of functions is the
processing of telephone calls of the customers experiencing
an outage.

Table 6.1.

The benefits of different fault management functions. The abbreviations used are: OC for outage costs, L for
labor costs, INV for investments, M for competitiveness in power market and Q for supply quality.
------------------------------------------------------------------------------------------------------------------------------Functions
Benefits
____________________________________________________________________________________
Medium voltage switch control
and fault clearing
- line switch remote control
- fault indicator remote reading
- fault distance computation

OC,L,Q

High resistance fault detection

Voltage and var control
Distribution load estimation
Switching planning
trouble call handling
customer information
fault statistics production

OC,L,Q
INV,M,Q
INV,M,Q
INV, M,Q
L, M,Q
L,M,Q
L,M,Q

6.2 RESULTS OF A COST/BENEFIT ANALYSIS

In this section, the results of a costs benefit analysis made for
six Finnish distribution systems is summarized. The
functions covered are all related to fault detection and fault
clearing. The outage costs of the customers are fully taken
into account as hard costs. Both the costs and the benefits of
various fault management functions are partly overlapping.
This makes their economic analysis complicated and tedious.
The feasibility analysis can most conveniently be made using
a computer program for assistance. In this section a software
package for this purpose is briefly presented. The input data
of the software package is divided into four classes:
-

the utility specific data

general calculation parameters
the configuration of automation functions
general cost parameters

The utility specific data includes the information which is

applicable only for the distribution company concerned. In
this file we have, for instance, statistical information of the
utility and some average technical data of its network.
The general calculation parameters include such information
that is common to all the utilities, like the parameters of
outage costs to different customer classes. Another example
is the parameters which control the efficiency of the fault
management functions, like the per cent reduction of outage
times due to remotely controlled switching stations. This data
is often based on experience and must be collected from
several sources.
The configuration of automation functions is simply a list of
the functions that are taken into account in the analysis. The
software package allows for any combination of the functions
mentioned in the latter part of this section.

The general cost parameters include the data of the costs of

the computer equipment, computer software packages, data
transmission systems, automation related power equipment
etc. The costs are further divided into purchase, installation,
operation and maintenance costs.
The output data of the software system is correspondingly
divided into two classes:
- intermediate results of technical computations
- results of cost/benefit analysis
The intermediate results of the technical computations are
used for checking the consistency of the input data and for
assessing the quality of the analysis in general. The final
computation results are stored in a separate file. The results
include the total costs and benefits for the given function
configuration and their decomposition into various classes.
Benefits modeled in the software package include the savings
in labor costs and the reduction in outage costs. Savings in
labor costs are expected in the operation of the power
system, like in trouble shooting during network faults. In
some cases the labor cost savings are the largest benefit
factor of the function. An example is the processing of
telephone calls of the customers experiencing an outage.
According to the cost benefit analysis, the reduction in
outage costs is the major benefit of fault management
functions. With the remote control of line switches of
overhead line networks solely, the distance between the
controlled points being about 15 to 20 km, the outage costs
can be cut by about a half. The outage costs can further be
reduced by developing fault location techniques, i.e. fault
indicators and computational fault location.
In the future, it will also be possible to detect and locate part
of the temporary faults and the faults with high fault
resistances. In this case, the critical fault type is a single
phase to earth fault in overhead lines. The fault often begins

with a very high resistance, and evolves only gradually into a

fully established earth leakage requiring protective actions.
This kind of fault can be detected well in advance by a
careful analysis of neutral voltage, and in many cases the
permanent fault and the corresponding outage can
completely be avoided.
Typical costs of supply outages for different customer
categories are given in Table 6.2. The values are taken from
a recent Nordic study. The values for residential customers
are based on the willingness to pay, whereas those for
agricultural customers are defined according to the
willingness to accept. The figures of the three other major
Table 6.2

classes are based on the direct evaluation of the costs and

losses by the customers themselves. According to the study,
the costs of short outages are relatively high. It should also
be stressed that the short repetitive outages (3 outages of
about 1 minute within a period of 15 minutes) usually cause
high monetary losses. Thus, such solutions of network
automation should be favored, which not only shorten the
outage time, but also reduce the number of outages. An
example is the use of fault current indicators in the switching
stations in line locations. By the use of the indicators, it is
possible to reduce the number of trial switching needed
during the course of fault location.

Typical values for average outage costs in different customer classes in Finland (
/kW) .
____________________________________________________________________
Outage
1s
3x1min
1h
4h
8h
duration
____________________________________________________________________
Residential
Agricultural
Industrial
Commercial
Public

2.2
1.8
0.5

4.3
4.3
1.2

The above discussion applies to the rural networks with

overhead construction. In the underground systems of urban
areas the situation is somewhat different. According to the
studies in Finland, the outage rates in underground cable
networks usually are so low, that the remote control of line
switches hardly is profitable. Some other solution must be
found for medium voltage network automation in these cases.
One of the most promising alternatives is the remote reading
of fault current indicators, located in the MV/LV distribution
transformer stations. This is a fairly inexpensive way to cut
that part of the outage time, which is due to the location of
the faulty line section.
The costs and benefits of different fault management
functions are presented in more detail in Table 6.3. The
results are given as present values per feeder of a HV/MV
substation. The networks studied were two urban and four
rural ones. The results can be summarized as follows:
-

the remote control of line switches was justified for

all the rural networks but not for the urban ones.
The major part of the benefits, about 90 to 97 %,
comes through the reduced outage costs of
customers. In urban underground systems, the fault
frequencies are substantially lower than in rural
networks. Hence the outage cost reduction also is
smaller.

0.7
5.3
11
12
4

2.8
32
33
11

10
58
58
68
23

Fault indicators at remotely controlled line switches

mainly reduce the costs of repetitive short outages
due to trial switching. The function is economical in
most of the rural systems.

independent fault indicator stations, at locations in

addition and separate from line switch remote
control stations, are cost effective in all the four
rural systems. The remote control of fault indicators
at the secondary substations, or ring main units, was
found to be very cost effective for both the urban
systems. Here it was assumed, that 20% of the
indicators are equipped with communication.

computational fault location, based on the

integration of protective relays, substation SCADA
and the distribution data management system, is
justified in all the rural systems and also for one of
the urban utilities.

For the high resistance earth faults (in high

impedance grounded systems), it was assumed, that
10% of the faults can be detected by neutral voltage
analysis and by partial discharge based techniques.
This function was found to be very cost effective
for all the utilities.

Table 6.3 Benefits and costs (B/C) of various fault management functions in two urban and four rural Finnish utilities.
Sums as present values in 1000
per feeder.

Function
line switch remote control
fault indicators at line switches
fault indicator stations
computational fault location
high resistance fault indication

URB1
-/-/9/2
-/2/1

URB2 RUR1 RUR2 RUR3

-/34/20 57/23 46/17
-/-/4/4
7/3
19/3
7/5
12/5
11/4
6/3
8/4
15/4
9/3
5/1
4/2
7/2
6/1

6.3 WEIGHTING OF QUALITY IMPROVEMENT

INVESTMENTS IN MEDIUM VOLTAGE
DISTRIBUTION NETWORKS
Contrary to the cost/benefit analysis in the previous chapter,
the outage costs of the customers usually are not fully
accepted an argument when deciding on network automation
or fault management investments. Some weighting
procedures are often used instead.
As a first example of weighting the quality improvements
when deciding on network investments, the procedure
adopted
by
an
Austrian
power
company,
Obersterreichische Kraftwerke AG (OKA), is presented.

RUR4
94/20
13/3
23/4
18/3
12/1

reliable investment decision based on technical and

economical aspects.
It is very important to see a target-value always in relation
to the determination method of the value. The target-values
may differ quite a lot, but if the determination methods of
the values are different as well, the different target-values
mean the same target. It is further important to set up a
weighting method that is easy to use in practice. It should
not require data, that are not available, and it should not
require specialists, that the company does not have.

Quality improvements
The MV-distribution network defines many parameters,
that have direct influence on the quality of supply to the
customer (number of short and long term interruptions,
time of interruptions). While the situation in urban
networks is usually acceptable, in rural regions the quality
of supply is often bad. To raise the quality several
improvements to network operation may be introduced:

ring connections in the MV-network

restructuring of historically grown networks
mounting of additional circuit breakers with protection
remotely controlled MV-stations

The aim of such improvements can be quite different:

reduction of time of interruptions after disturbances or

for maintenance
reduction of staff costs
reduction of not delivered energy
improvement of customer satisfaction
increase of technical attraction of rural business areas

Purpose of weighting
The weighting of necessity and gain of a quality
improvement is subjective, depending on experiences and
task of the person. To get an unique quality understanding
in the company a unique weighting of improvement is
required. The target of the weighting method is to get a

Fig. 6.1. A new ring connection in a distribution system.

Technical-Economical-Gain
As example for a weighting method the TechnicalEconomical-Gain for ring-connections is explained, that is
used at OKA. A ring-connection is considered, if several
MV-stations are radially operated (see left side picture). In
case of disturbances or for maintenance purposes in the
first MV-station, all radial MV-stations must be switched
off. An additional ring-connection could solve many
problems (dotted line).
The gain of such an ring-connection depends on the length
and type of the radial feeder as well as on the size of
energy, that is supplied by the radial stations.

TEG

NVist
 DBist  DB%  ND
NVsoll

A simple formula to calculate a target value is shown on

the left hand side:

report quality / cost and of customer services. This offer

has been built on two main principles :

TEG Technical-Economical-Gain
Nvist Not-availability without ring-connection [hours/year]
Nvsoll Target-value for not-availability [hours/year]
Dbist Proceeds of energy sale minus energy costs/yr of
MV-stations [per year]
DB% Target-value, % of DBist, that should be taken [e.g.
0,5 %/year]
ND expected service life [e.g. 30 years]
This rather simple procedure gave good practical results
for the decision, if this investment should be done. If the
costs of the ring-connection are less than TEG, the ringconnection may be constructed.

A solidarity principle : to provide to all customers a

quality superior to a base standard level.

An economic optimum principle : to guarantee the best
use of the financial resources of the company.
This Reference Offer is at the heart of the standard
Emeraude Contract for MV and HV customers, that is
adapted to local constraints depending on the geographical
area and on the goals in quality levels. As for LV
customers, this Offer is contained in the concession
contracts signed with the French local authorities.
Three main action courses have been initiated in order to
ameliorate this reference level at short and mid term :

6.4 DISTRIBUTION POWER QUALITY POLICY AT

EDF
This section describes the policy and methods used by EDF
regarding power quality for distribution customers. It also
describes the method used to calculate the cost of the nonquality.
Customers quality needs
Today EDF observes two strong tendencies regarding their
customers quality needs : a evolution of the exigency levels
and an increasing differentiation of the wanted quality. In
order to respond to this new context, EDF is offering a new
relationship founded on :

A Reference Level adequate for the great majority of

the customers, with a contractual engagement
concerning the quality performance.

A case-by-case upgrade of this level to respond to the
specific needs of certain customers.

Improvement of the everyday quality (reduction of

frequency and duration of customer cuts).

Fight against the major incidents.

Reduction of rms voltage variation margins and
improvement of voltage quality.
The complementary offers
The complementary quality offer aims at responding to
very specific needs from one part of the customers. It is
based on the two following targeted offers :

The personalised offer : for one or several customer

precisely identified. The design of the offer must lead to
an optimal global economic balance. EDF takes in
charge the part of the cost justified by its contribution
to the general interest.

The high performance zone offer : the same as the
personalised one, but offered to a whole zone.

Prospective goals

A Reference Quality Agreement

The characteristics of EDF reference level have been
designed to respond to the expectations of the great
majority of customers, while keeping EDF amongst the
most performing European distribution utilities in terms of

From 1995 to 1998

Rural areas

Maximum number of long cuts

Maximum number of short cuts

6
30

2005

Rural areas

Maximum number of long cuts

Maximum number of short cuts

2
5

The quality goals that serve to define the threshold of the

compensations paid to customers are the following:

Small and middle- Big cities

sized towns
3
3
10
3
Small and middle- Big cities
sized towns
1
1
2
1

This values take into account all the faults perceived by the customers.

City Centres
2
2
City Centres
<1
<1

Putting this policy into practice

A common work of the commercial, customer and
technical teams is necessary. The typical action plan is the
following :

First phase : meeting the customer.

Second phase : analysis and agreement.

To put into practice the general technical orientations, a

decision-help technical-economic approach is used ; this
approach includes :

A standard valorisation of the failure, calculated from

the power cut and from the non-distributed energy
(END), with a cost A (5F  0.76
!   "  
 #
(60F  9.1
!  $
     
 %
(130F  19.7
!    
   

concerning more than 50 MWh of END.

Taking into account the real prejudices of sensitive
customers, when designing personalised solutions.
Distribution Centres conduct an internal feedback system
and several indicators that allow them to evaluate the
customers perception of the personalised offers, and to act
in the designing of these offers.
A great importance is given to clarity. The reference offer
includes simple indicators measuring the global
performance and its dispersion. This indicators are the
same that are written in the standard Emeraude contract.

CONCLUSIONS
The fault management functions can be classified to fault
identification, location and supply restoration, and to
assisting functions, like statistics production and supply
quality reporting and monitoring.

In the case of supply restoration there are two development

lines. One is the fully automatic switching systems. For
these several solutions exist, spanning from simple
predefined sequences to fully integrated systems, where the
network topology is analysed and technical constraints are
checked in real time. However, the latter ones are still in an
experimental phase. The other development line is the
assisting software, used for checking the technical
constraints of the network, but leaving the decision on the
switching actions to the operator. This development line
seems to be more popular nowadays.
In the case of computer tools of fault management, there is
strong development going on. New functions are developed
both in the SCADA and in the DMS systems. Examples of
these functions are tools for distribution system state
estimation, alarm processing and event analysis, fault
location and power restoration, field crew management and
various types of
analysis tools. A very promising
possibility also is the automatic production of fault
statistics.
When assessing the economic feasibility of fault
management functions, the central issue is the outage
costs. When they are reduced, the benefits mostly come to
the customers, and only a minor share to the utility. Because
of this reason, several alternative approaches exist for their
appreciation: 1) The customers outage costs are fully taken
into account as hard costs, 2) A certain price per kWh is
used, less than the customers costs, but higher than the
revenues, and 3) The customers outage costs are ignored. It
is clear that the selection between these alternatives greatly
affects the investments made, and consequently the
techniques used for fault management applications.
References
1.
2.

For the location of permanent faults, two different

approaches exist, namely fault indicators and fault distance
computation. Fault indicators are used both for the
detection of short circuit and earth faults. The main
restriction of the present fault indicators is the lack of
directional function. This would be of high importance
especially in compensated neutral systems.
For fault distance computation, two different techniques are
being used. By comparing the measured fault current to the
computed one, a rough estimate for the faulty line section
is obtained. The other approach is to use impedance relays.
However, these techniques work in the case of short circuit
faults only. In unearthed or compensated neutral systems,
the rated frequency earth fault currents are too small to
allow for a reliable fault distance computation. A possible
solution is the transient based techniques, but the
bottleneck has so far been the lack of measuring technique
with high enough sampling frequency.

3.

4.

5.

6.

Hubensteiner, H: Schutztechnik in elektrischen

Netzen. vde verlag, Berlin, 1989, 282 pp.
Igel,
M.:
Neuartige
Verfahren
fr
den
Erdschludistanzschutz in isoliert und kompensiert
betriebenen Netzen. Signale und Algoritmen im
Frequenzbereich. Dissertation, Universitt des
Saarlandes, Saarbrcken, 1990, 181 pp.
Lehtonen, M. & Antila, E., An integrated system for
distribution automation. 11th CEPSI Conference of the
Electric Power Supply Industry, Kuala Lumpur,
Malaysia, October 21-25, 1996.
Lehtonen, M.: Distribution automation in Finland.
Final report of the research programme 1990-1992.
Ministry of Trade and Industry. Energy Department.
Reviews B:148, June 1993, Helsinki, 58 pp.
Lehtonen, M.: Transient analysis for ground fault
distance estimation in electrical distribution networks.
Espoo 1992, Technical Research Centre of Finland,
Publications No 115. 181 pp + app.
Pundt, H.: Untersuchungen der Ausgleichsvorgnge
bei Erdschlu in Hochspannungsnetzen mit isoliertem
Sternpunkt und induktiver Sternpunkterdung als

7.

8.
9.

10.

11.

12.
13.

Grundlage zur selektiven Erdschluerfassung.

Dissertation, TU Dresden, 1963, 167 pp. + app.
Schegner, P.: Digitaler Erdschluuniversalschutz.
Konzept und erste Realisierung. Dissertation,
Universitt des Saarlandes, Saarbrcken, 1989, 186
pp.
Warrington, C., Protective Relays, Vol. 1, Chapman &
Hall Ltd, 1968, 380 pp.
Willheim, R., Waters, M.: Neutral earthing in highvoltage transmission. Elsevier Publishing Co., New
York, 1956, 669 pp.
Druml G., Detecting High-Ohmic Earth Faults in
Compensated Networks Symposium on Neutral Point
Treatment in Distribution Systems NMT 95, 1995.
Fickert L., Tenschert W. et. al., Treatment of neutral
in MV-networks Verband der Elektrizittswerke
sterreichs, Working Group Report (German) 1996,
ISBN 3-901411-19-4
Tenschert W. ,Planning strategies for MV- and LVnetworks VE-Journal (German), October 1997.
Tenschert W. , Fault location using fault distance
measurement of digital relays CIRED 1993, 4.20.