Electrical Power Quality and Utilisation, Magazine Vol. II, No.

1, 2006

Low Voltage Fault Detection and Localisation Using

the Topas 1000 Disturbance Recorder
Stephanie L. HORTON, Nico G. van LUIJK
LEM, EDF Energy, England

Summary: During the past 20 years an increasingly competitive power industry has recognised
the importance of addressing the issue of power quality. Many companies now take an active roll
in addressing the problems associated with power quality. They are investing in the research and
development of equipment to overcome power dips, surges and interruptions, and offering the
customer a wide and varied choice of solutions and services to their power quality needs. These
solutions are usually based on overcoming the limitations of individual equipment, which is
being used by the customer, rather than making improvement to the quality of power supplied.
This paper aims to bring to light the incipient problems experienced on LV underground networks, the reasons for the problems and their affect on the end customer and a utilities CML
statistics. (Customer Minutes Lost CML is a method of measuring the performance of the
network in the UK) This paper presents the research work carried out by LEM and EDF Energy
to identify how to detect and locate faults using online techniques.
The finding of this research is being used to reduce the downtime experienced by customers by
decreasing the time involved in resolving these reoccurring problems.
1. BACKGROUND
EDF Energy and LEM recognise the value of
combined R&D when investigating and
developing solutions. LEM has a vested
interest in developing new markets and
products and EDF in the reducing costs and
increasing the efficiency of their network
through investment.
EDF Energy was created in 2003 as re-branding
exercise of the London Electricity Group. The
London Electricity group acquired Eastern
Power Networks (EPN) in January 2002 and
South East Electricity Board (SEEBOARD) in
July 2002, making it one of the largest electricity
suppliers in the UK. EDF Energy is a subsidiary
of EDF.
LEM is a Swiss company serving all markets, in
which electric energy plays an important role.
Its profession includes the specialised
measuring of electric parameters. It develops,
manufactures and sells components,
instruments, highly complex measuring systems
and services.
The aim of this combined research was to work
with a utility to develop a product which would
offer the best solution to assist in the
identification and location of a growing problem.
The TOPAS 1000 Power Quality analyser was
used extensively in the study. It provides an
unprecedented range of functionality, including
mains voltage analysis, disturbance analysis,
frequency analysis, and transient analysis
enabling the characteristic fault fingerprints to
be identified and recorded. Based on the

Key words:
power quality,
fault detection,
location method

findings and research conducted the TOPAS is

being developed further to simplify the process
and integrate the algorithms outlined in the
research.
2. INTRODUCTION
The effect of transient disturbances on the
normal high quality of electricity supplied to
domestic, industrial and commercial customers
in the UK, is more than just a nuisance, in many
case it can have an adverse effect on the
customers business. Voltage dips have become
the most important concern for many industrial
and commercial consumers due to the
susceptibility of sensitive electronic equipment.
As a voltage varies from its normal magnitude
sensitive critical loads can be affected, and can
cause the system to experience a voltage dip
and subsequently cause equipment to
malfunction.
Faults on the system can be caused by many
different phenomenon, however to date most
of the investment made has been to increase
the robustness of the MV and HV networks.
Most major cities around the world employ
underground LV networks to supply their
customers, as cables and constructions remain
static it becomes evident that each of these
networks are susceptible to the same types and
frequencies of faults.
Some of the oil impregnated cables used in
Londons underground network date from the
early nineteen hundreds. Consisting of around
20.000 km of distribution cable and an equal

S.L. Horton and N.G. van Luijk: Low Voltage Fault Detection and Localisation...

amount of service cable, it results in the low

voltage network being four times larger than
the medium and high voltage network put
together.
Due to the size and age of the LV underground
network, low voltage related incidents in urban
areas are responsible for almost 50% of the total
Customer Minutes Lost, impacting on customer
confidence and the utilities performance
markers. This number has been slowly rising as
the impact of investment in MV and HV networks
has resulted in a decrease in faults and faster
restoration times. On the low voltage network,
faults are numerous and restoration times are
slow. The result is that faults on the low voltage
network are responsible for up to 45% of the
total network operation costs.
Diagnosing the cause of a failure on a circuit is
a difficult task for the utility companies. Faults
can be caused by a variety of reasons and in
most a fuse operation is the only evidence. This
evidence does not help to distinguish if the
fault is caused by an underground cable failure,
where the failure occurred, or if high load
currents caused the fuse to operate. If the cause
cannot be identified the supply is usually
restored by replacing the fuse. If the circuit fails
again in a short time span it is unlikely that the
fuse operation is caused by a high load
demands, and other solutions have to be
sought to deal with the problem. An autorecloser can be installed to replace the fuses on
the circuit. These reclosers will trip on fault
currents and reclose after a set time interval.
However this can cause multiple interruptions
for customers on a circuit but does minimises
the supply down time.
Low voltage faults are becoming more frequent,
as corrosion and ageing start to affect their
performance. Finding low voltage faults can be a
difficult task, as off line fault-locating techniques
cannot be used because of the permanently
connected loads on the circuits. The lack of online
fault finding equipment that complies with the
EN50160 norm, also makes it difficult to locate or
pin point low faults on live circuits.
To drive these operation costs down and at the
same time improve network performance,
improved methods are needed for fault
detection and location.
3. SUSCEPTIBILITY OF UNDERGROUND
LV CABLES
Cables are subject to both internal and external
influences and forces which can eventually lead
to penetration of the protective sheathing of
the cable. This in turn can lead to moisture
ingress and results in various dielectric changes

&

inside the cable, and the degradation process

that is responsible for many of the faults on the
network begins.
In addition to moisture, temperature in oil-paper
cables can have a negative impact on their life
expectation and failure rate. As electricity
demand has increased the cables are operating
at higher temperatures. This increase in
temperature can cause the oil to expand and
results in the aluminium or lead sheath being
permanently stretched, leading to tiny empty
volumes between the paper insulation and earth
sheath when the temperature decreases. These
voids have a lower pressure than the cable
surroundings, which may cause moisture to be
sucked into the cable. Besides deformation of
the cable, increased temperature leads to higher
dielectric losses, especially in cables with
increased moisture content. This may
eventually lead to thermal breakdown of the
insulation and premature aging.
Eventually moisture ingress into the damaged
cable will result in a fault developing as the
paper insulation becoming conductive enough
to allow a low voltage arc between the
conductor(s) or conductor to earth to develop.
The heat from the arc will vaporise most of the
moisture, leaving carbon particles behind, but
providing a short-term solution to the cause of
the fault. However in time the fault will develop
further and become more frequent on the circuit.
Incipient low voltage fault occur sporadically
and reoccurring incipient faults can take weeks
to redevelop.
It is worth mentioning that higher operation
temperatures have a positive effect on repelling
moisture. Furthermore, oil will expand ten times
more than copper when temperature increases,
leading to a pressurised cable. The increased
pressure will cause the water to repel out of the
cable. This may explain why many faults occur
when a cable is depressurised i.e. when the
cable has been taken off-line and then reenergised or when operating under low load
conditions.
4. TOPAS 1000 MONITORING SYSTEM
The TOPAS 1000 is a Power Network Analyser
developed in conjunction with utilities to
identify and record sources of disturbances and
assess mains voltage quality in compliance with
the applicable standards. The equipment
records all current and voltage parameters
including:
Harmonics
Interharmonics
Flicker
Voltage variations, dips, surges

Electric Power Quality and Utilization, Magazine Vol. II, No 1, 2006

 Measurement of signaling frequency

Waveform capture
It offers both triggering functions and fast
transient capture (10MHz) as standard (6.4kHz),
making the instrument and ideal tool for this
research project.
5. ONLINE LOW VOLTAGE FAULT
DETCTION AND LOCATION
METHODS
5.1 Detecting an incipient low voltage fault
Detecting an incipient event caused by a low
voltage fault requires the recording equipment
to be able to detect sharp voltage dips caused
by the arcing event, as illustrated in Figure 1. A
trigger event can be achieved by measuring the
transient voltage dip or the sudden increase in
(short circuit) current.
The methods of triggering, on fault events,
used by the TOPAS 1000 are:
Mask trigger: The mask trigger option applies
a mask around the signal of a predefined
magnitude. When the measured signal
exceeds this mask the unit will trigger
capturing voltage and current signal at the
basic 6.4KHz sampling rate;
Over current trigger: The over current trigger
is a simple exceed value trigger, which is
set to a rating exceeding the absolute load
condition. When set to 1.5 times the fuse
rating of the circuit this trigger is ideal to
identify the short current bursts or to detect
overload current;
Transient trigger: The transient trigger is a
fast edge trigger that operates on a sample
frequency between 100 kHz and 10 MHz,
which can be user defined. Using this option
to trigger is extremely effective when setting
the unit to trigger on transient events of more
than 30V. This will minimise the unit from
triggering on small transient events. When
the trigger occurs it will capture the voltage
signals at the predefined sample rate, plus it
will capture the voltage and current signal at
6.4 kHz.
The pre trigger setting of the TOPAS 1000
should be set to one to two cycle(s) and the
post trigger to ten cycles to provide sufficient
data for a clear overview of the fault event.
5.2. Voltage gradient
Voltage gradient [1] is a method based on
measuring the voltage collapse during the arc
event, simultaneously at different locations,
along the circuit. This method does require a
minimum of three voltage measuring devices,

Fig. 1. Example of the

sharp voltage depression caused by an arc
event

of which two must be in between the point of

supply and the fault location. An illustration of
this method can be found in Figure 2.
The theory of operations is as followed: When
a fault event takes place, the voltage will
collapse and be recorded by each instrument
and in a simplistic manner the voltage collapse
can be described as in equation 1.
u (x) = i zc (x) + uarc

" "
! {
a

Fig. 2. Example of a
voltage gradient
measurement

(1)

were x a variable position between the point of

supply and the fault location [m], u is the
voltage [V] as function of the position, i is the
fault current [A], z is the impedance of the
system between the point of supply and fault
location as a function of the position [Ohm]
and uarc is the arc voltage at the fault location
[V]. This equation neglects the load current in
the circuit.

S.L. Horton and N.G. van Luijk: Low Voltage Fault Detection and Localisation...

'

The example in Figure 2 shows that equipment

m1 and m2 will measure the voltage which is
affected by the gradient a as given equation
1. The equipment m3 will measure the constant
b as given in equation 1, which is the arc
voltage. Drawing a line between points m1 and
m2 and a constant line from point m3, will reveal
the fault location. The fault location is the
intersection point of both lines.
In this example it has been assumed that the
cable dimensions are similar throughout the
network, thus there is a linear coefficient in
impedance. In real network conditions, varying
cable sizes make it difficult to achieve accurate
fault locations with only three devices.
The TOPAS 1000 can be used successfully for
voltage gradient analysis, although it is over
specified for this job. In addition the size
restriction in pillar and link boxes and the
requirement of a minimum of three units makes
this method expensive.
5.3. Transient Impedance Method (TIM)
The transient impedance method is based on
measuring the current and voltage during the
low voltage fault [2]. Using the current and
voltage waveforms it is possible to calculate
the system impedance in the time domain using
equation 2. Applying the cable parameters per
length unit, the location of the fault can be
determined.

ratioCable(i ) >Cable (j ) =

Rcable (j )
Rcable(i )

(3)

Taking the 400mm2 cable as a reference cable

the ratio factors can be calculated using
equation 3:
ratio185> 400 = 164 = 2 ,1
78
ratio240 > 400 = 125 = 1, 6
78

Using the ratio factors the cable lengths can be

recalculated to the 400mm2 reference size.
Resulting in:
L185 > 400 = ratio185> 400 475 = 1001 metres

L250 > 400 = ratio240 > 400 150 = 241 metres

2 = (n) =
2

= (Rc i (n) + Lc i (n) + uarc us (n))2

(2)

were ? is the error [V], us is the voltage [V], i the

total current during an arc event [A], Rc the
cable resistance including return path [Ohm],
Lc the inductance including return path [Ohm],
uarc the arc voltage [V] , n is the number of
samples.
Different dimensions of cables in the same
circuit also affect the TIM method. However,
with TIM it is possible to correct for this by
normalising the circuit to one cable size as
explained in the example below.
Table 1. Properties of cable sections

!

Example:
Consider a circuit consisting of three-cable
sections all connected in series. All sections
have different diameters, giving them different
resistance values. Table 1 summarises the
conductor area, cable length and resistance
value for each section.
To normalise the cables back to one default size,
the following equation can be applied:

Circuit

Conductor area
[mm2]

Length [metres]

Resistance
[/m]

Section 1

400

250

78

Section 2

185

475

164

Section 3

240

150

125

Applying TIM and applying calculations to the

data a resistance of for example 100 milli Ohm
would represent 1285 metres of 400mm2 cable
length. To calculate the fault distance back into
real values, it is necessary to take the fault
distance minus the length of section one. For
this example it will result in 34 metres (1285-2501001).
The above shows that the fault is in section
three. To transfer the calculated value back to
its real value (normalised to 400mm2), the
number of metres must be divided through the
ratio factor of the cable. The result in a cable
length will be 21 metres of the start of section
three, or 746 metres from the start of section
one. The same method can be applied for the
inductance of the cable.
Although TIM allows compensation for
different cable diameters, it cannot discriminate
the faulty section in a circuit and can lead to
multiple fault locations as illustrated in Figure 3
by f1 and f2. Additional equipment can resolve
this issue, by using a tool like the CableSniffer
to pin point the fault position.

Electric Power Quality and Utilization, Magazine Vol. II, No 1, 2006

TIM does require relative higher sample

frequency of at least 10 kHz to acquire an accurate
measurement of the arc igniting and
extinguishing in the voltage waveform. The
transient option on the TOPAS 1000 is more than
sufficient option to measure these fault events.
6. CASE STUDY
The TOPAS equipment is usually installed after
a number of fuse operations, as the number of
TOPAS 1000 units available within EDF Energy
is limited. Generally the network has been
rearranged and sectionalised and the faults
have progressed into an intermittent stage. This
can make it difficult to produce a fault location
as insufficient data can be gathered before the
fault evolves into a permanent fault.
In this case study the circuit targeted consists
of 160mm2 aluminium distribution cable with
various joints and spurs branching off the main
cable. It supplies several shops in a high street
plus a significant amount of domestic
customers.
Over a period of several months a number of
fuse operations were recorded, the TOPAS unit
was installed in the affected sub station to
investigate the root cause of the problem. The
TOPAS recorded multiple events during a tenweek period and the data retrieved indicated an
active fault as presented in Figure 4.
The waveform shows a clear voltage depression
together with a short current surge, which is
typical in low voltage faults. The short duration
of this event (and others measured) in
combination with the relatively low current did
not result in a fuse operation. In addition it can
be concluded that the low current value
indicates that the fault is a comparative distance
from the point of measurement.
Using the recorded data with TIM (equation 2) it
is possible to determine the system impedance.
The resistance value is applied as in example 1
and results in a predicted fault location of 300m
on the 160mm2 distribution cable. This location
coincides with a known transition joint and is a
plausible location for a fault, as joints are typically
the weak points in the system.
Since analysing the data and identifying the
possible location of the fault it has become
dormant, and may not reoccur for some time.
The next time the fault occurs it may cause an
incipient event or a fuse operation, this will
provide the opportunity to investigate further
using a CableSniffer, to pinpoint the exact
location of the fault.
TIM is still being developed and each fault
provides more data to refine the location software.
Currently 95% of the fault events are captured

Fig. 3. Example of
possible multiple fault
locations using TIM

using the TOPAS system. However, lack of

resources to assist in verifying fault locations,
means that no accurate success rate for locating
low voltage faults can be given at present.
However, it is expected that TIM will develop into
a successful online location tool that can be used
in conjunction with the Topas monitor and
CableSniffer to achieve maximum benefit.

Fig. 4. Fault event

capture using the
TOPAS 1000;
Current scaling:
1V=1A

7. CONCLUSIONS
A large majority of the Utilities around the world
use underground LV systems in urban areas,
and many of the installed systems are still using
cables dating from the early 19th century. In
addition as these systems are generally
deployed in populated urban areas, faults affect
a greater number of customers and restoration
times are slower. In comparison to overhead
systems, faults on the low voltage network are
responsible for up to 50% of the interruption
time experienced by customers connected to
the networks around the world. However where
the HV and MV networks have received
continual improvements to reduce the number
of faults experienced by customers. At low
voltage, little research has been carried out into
ways of detecting and locating faults.
The research outlined in this paper is a first
step towards a more efficient and integrated
way of cable fault management at low voltage

S.L. Horton and N.G. van Luijk: Low Voltage Fault Detection and Localisation...

!

levels. By using the characteristic phenomenon

which occur prior to the fault, and through the
use of off the shelf technology and analytical
calculations, a method of accurately identifying
and locating repetitive underground faults has
been developed. These results show typical
characteristic faults and present the accuracies
of calculations when determining the fault
location. Using the TIM in conjunction with
newly developed pin pointing techniques such
as the CableSniffer will greatly increase the
efficiency of managing faults on the low voltage
system.
Further developments and implementation in
field equipment will result in a better
understanding of incipient low voltage fault and
will give a better insight in the prediction of
incipient faults and their location.
REFERENCES
1. C l e g g , B . Underground CABLE FAULT
Location, BCC Electrical Engineering and Training
Consultancy. 1998, Hertfordshire, England.
2. N . G . v a n L u i j k . Online Incipient Detection
and Localisation for Low Voltage Faults in Oil
Impregnates Cables. 2002. Eindhoven, the
Netherlands, Eindhoven University of Technology.
3. N . G . v a n L u i j k . The impacts of undeground
transients and incipient fault in the quality and
reliability of supply in low voltage networks. 2005.
Conference Proceedings CIRED 2005.

Stephanie Horton

has been involved with Power Quality for

many years and is a well known face in
the engineering fraternity in the UK and
abroad. Her degree in Electromechanical
Power Engineering provided her with the
skills necessary to take in to account all
aspects of PQ, and truly understand the
impacts and causes. After leaving
university she worked for PowerGen's
Power Technology Centre and became involved with the PQ
debate in the UK and later helped develop a PQ based
monitoring project into a business. Later moving to ALSTOM
T&D in Stafford to work as a applications engineer for
protection and measurement, she was an integral part of the
development of the PQ measurement team, and the services
offered, from measurement equipment through to training
programs and software development. In 2002 she accepted a
position within LEM HEME, a PQ equipment manufacturer as
their Engineering Sales Manager, where she drove new PQ
initiatives and built strong partnerships to drive forward the
PQ debate and respond to customers needs. This paper was
the result on one such partnership with EDF.
Stephanie has since accepted a position with ERA Technology
in their Engineering Service Centre (ECS). ERA is one of the
highest regarded consultancies in the UK; the ECS department
provides a range of engineering, fault and failure analysis
services. She continues to drive forward new opportunities
and build strong relationships and partnerships for the future,
and is committed to promotion of the LPQI partnership. Her
unique experiences and abilities are now being used to offer
customers valuable insights into both conventional and
intricate PQ problems.
Address:
Engineering Consultancy Services
ERA Technology Ltd.
Cleeve Road, Leatherhead, Surrey KT22 7SA, UK
Telephone: +44 (0) 1372 367360
Mobile: +44 (0) 7768 876011
Fax No: +44 (0) 1372 367070
Website: http://www.era.co.uk

Nico van Luijk

received his M.Sc. degree from Eindhoven

University of Technology, the Netherlands
in 2001 and started his career with EDF
Energy as a research and development
engineer. During his 4 years at EDF Energy
he studied the behaviour and impact of
cable faults on the reliability and power
quality of distribution networks. He
successfully developed new methods for
detecting and locating LV and MV cable faults and started
experimenting with various recording devices to implement
these methods on the network, with several papers being
published on these subjects. This paper was the result of the
experiences with the LEM Power Quality equipment using it
as a fault detection and location device.
Nico has since accepted a position with Bechtel and is
working as an electrical engineer for the petrochemical
division on various projects around the world and is still
actively involved on issues surrounding distribution networks
and power quality.
e-mail: nico@vanluijk.net

Electric Power Quality and Utilization, Magazine Vol. II, No 1, 2006