=Paper=
{{Paper
|id=Vol-3195/paper5
|storemode=property
|title=PMU-based Fault Localization in Distribution Networks
|pdfUrl=https://ceur-ws.org/Vol-3195/paper5.pdf
|volume=Vol-3195
|authors=Denis Sodin
|dblpUrl=https://dblp.org/rec/conf/lambda-net/Sodin21
}}
==PMU-based Fault Localization in Distribution Networks==
https://ceur-ws.org/Vol-3195/paper5.pdf
PMU-based Fault Localization in Distribution Networks
Denis Sodin 1,2
1
Institute Jozef Stefan, Jamova 39, 1000 Ljubljana, Slovenia
2
ComSensus d.o.o., Brezje pri Dobu 8a, 1233 Dob, Slovenia
Abstract. In this paper, a fault localization method for distribution networks,
based on PMU measurements and compensation theory, is presented. Voltage
and current phasors of pre-fault and post-fault are used to determine the faulted
bus in the network. The method was verified using the Real Time Digital Simu-
lator (RTDS) with the simulation of real electric power system.
Keywords: Fault Localization, Distribution Network, Real Time Digital Simu-
lator.
1 Introduction
In today’s world, all industrial and economic branches are increasingly dependent on
electric power. However, the continuity of electric power supply is often
compromised due to faults or short circuits caused by either natural conditions
(lightning strikes, extreme wind, snow etc.), poor vegetation management, animal
contact or simply as a consequence of the aging equipment. Therefore, the faults must
be detected and localized quickly and accurately to ensure swift restoration and
minimize the time of customers not being supplied.
2 State-of-the-Art
Fault localization is theoretically well-known and developed field with variety of
applicable methods [1], which can be roughly divided into three groups.
The impedance based methods are characterized by low complexity and offer a fairly
accurate performance in case of the grounded systems. Their main drawback is that
they often rely on the iterative processes to produce results or even yield multiple
possible solutions [2], [3].
The second category of methods relies on the use of the travelling waves. A travel-
ling wave is a high-frequency electro-magnetic pulse due to unexpected change of the
current at the faulted point which propagates away from the fault. Therefore, costly
measurement equipment with high-sampling frequency (order of MHz or higher) and
precise time synchronization is needed to detect wavefronts. Furthermore, most of
those methods were developed for transmission networks that are more homogenous
Copyright © 2021 for this paper by its authors. Use permitted under Creative Com-
mons License Attribution 4.0 International (CC BY 4.0).
in terms of line parameters and far less branched than distribution networks. Some
representative traveling wave techniques and their application are provided in [4]–[6].
The last category of methods is based on pattern recognition and are also known as
knowledge-based methods [7]–[9]. They are based on a large training database that
contains reference fault cases for a given network. Even though those methods gener-
ally do not require complicated formulation, it is worth noting that they need to be
retrained in case of any modification to the existing system.
It is evident that each group of methods have their pros and cons in terms of accu-
racy and complexity. Given that the phasor measurement units (PMUs) are becoming
increasingly adopted in the context of smart grids, a fault localization method relying
on PMU data is used in this paper.
3 Methodology
The method similar to the one proposed in [10] was implemented for the purposes of
this paper and is explained hereinafter. To produce credible results, a real electric
system was simulated with electro-magnetic transient (EMT) based simulation tool
RTDS, which is currently the closest approximation (to the real-life conditions)
achievable with simulations.
The main idea of the algorithm is to estimate fault location with a minimal number
of PMU devices using their respective voltage and current phasors. Optimal observa-
bility of the feeder, when using just two PMUs, is achieved when one PMU is in-
stalled at the beginning of the feeder (substation) and the other is placed at the end of
the feeder. Both PMUs stream information about frequency and Rate-of-Change-of-
Frequency (RoCoF) and time-tagged phasors for phase voltages and currents (all three
phases), that serve as a basis for the operation of the algorithm. A criterion for identi-
fying the fault location is finding a bus with minimal difference in voltage, as seen
from both PMUs. To achieve this, each bus voltage needs to be calculated twice (once
with each PMU data).
For a detailed explanation of the method some indexes need to be introduced. The
total number of busses on the main feeder is n, so each bus is assigned its respective
position as a subscript. Counting starts at the beginning of the feeder, where the first
PMU is installed and ends with the last, n-th bus, where the second PMU is installed.
Bus subscripts are also used to designate other quantities referring to that particular
bus, for instance voltages and currents measured from the first PMU will be called
and , whereas voltages and currents measured from the second PMU are denoted as
and , respectively. We already mentioned that the voltage of each bus is calcu-
lated twice, therefore superscript needs to be introduced to avoid ambiguities. When
using data from the first PMU and applying the propagation model from 1-st toward
n-th bus the quantaties are denoted by superscript f (forward propagation). In the case
of the second PMU the backward propagation is denoted with b superscript.
Knowledge of current and voltage from the first PMU (at the beginning of the line)
and information about line parameters of first line (marked in Fig. 1) enables us to
calculate the voltage and current of the second bus. If there is a load connected to that
bus, we need to subtract load current from current flowing into a bus to get the value
of current flowing into the next line segment. This procedure can then be repeated
until we get to the end of the feeder and voltage of each network bus, as seen from the
first PMU, is known. A similar process is repeated for the second PMU, with the only
difference being, that this time we start with the voltage and current of the last bus
(obtained by the second PMU) and that load currents need to be added (instead of
subtracted) to bus currents, to get values of currents flowing into next segment.
Fig. 1. Forward and backward calculation of bus voltages
In the pre-fault condition, forward and backward voltages in each bus should be the
same and therefore their difference equal to zero. This condition changes however
with the introduction of a fault. Let’s assume that the fault occurs at bus j. Voltages
and currents from the first PMU will be propagated correctly up to this bus (meaning
that voltages will still be correctly calculated), but since we do not take
fault current into account, the calculated voltages for the next bus, and all subsequent
busses, will be wrong (meaning will be incorrect values). Similar
observation can be done for the second PMU. Again, voltages and currents will be
correctly propagated up to j-th bus ( - correct values), but voltages for
subsequent busses will be wrong ( will be incorrect values).
It is evident, that voltages are correctly calculated from both sides only for the bus,
where the fault occurred. Therefore we can expect the difference between backward
and forward voltage to reach its minimum in a faulted bus, which can be formulated
as:
( )
where
| |
and
Note that would have almost identical value, have we decided to only use
post-fault measurements of PMUs, since the following condition holds true for pre-
fault:
.
The reason for including the pre-fault measurement is, that current and voltage
transformers can introduce some error in measurements which is stable over a short
period of time. With the subtraction of pre-fault and post-fault measurements, we can
achieve to cancel this error out.
It is worth pointing out that with this method only the busses of the feeder can be
identified as a fault location. This means that when a fault happens between two bus-
ses, a bus that is electrically closer to the fault location will be identified as faulted,
however the actual position on the line will not be calculated.
4 Results
Performance of the method was tested on the 14-bus MV system for various fault
locations and fault types. The resistance of the fault was set to for all scenarios.
On the x-axis of the diagrams busses of the main feeder are presented, whereas their
corresponding voltage differences (introduced in section 3 and labelled as ) are
given on the y-axis of the diagram.
The performance of the method for three different types of fault is presented in Fig.
2.
Fig. 2. Results for phase to ground, phase to phase and three phase to ground fault.
In all three cases, bus 8 was correctly identified as the faulted one. However, differ-
ence in magnitude of voltage mismatch can be observed between different fault sce-
narios. As already explained in Section 3, the voltage mismatch (voltage difference as
seen from two PMUs) of a certain bus is a consequence of not involving fault current
in post-fault analysis. Therefore it comes to no surprise, that voltage mismatches are
much higher in cases of 3-phase to ground and phase-phase faults, as the fault cur-
rents of those events are much higher than in case of a single phase-ground fault.
Performance of the method for three different fault locations is presented in Fig-
ure3. Again the method correctly identified the faulted bus for all three cases.
Fig. 3. Results for phase to ground fault in 3 different busses.
References
1. N. Gana, N. F. Ab Aziz, Z. Ali, H. Hashim, and B. Yunus, ‘A Comprehensive Review of
Fault Location Methods for Distribution Power System’, Indonesian Journal of Electrical
Engineering and Computer Science, vol. 6, p. 185, Apr. 2017, doi:
10.11591/ijeecs.v6.i1.pp185-192.
2. E. Personal, A. García, A. Parejo, D. F. Larios, F. Biscarri, and C. León, ‘A Comparison of
Impedance-Based Fault Location Methods for Power Underground Distribution Systems’,
Energies, vol. 9, no. 12, Art. no. 12, Dec. 2016, doi: 10.3390/en9121022.
3. J. Mora-Flòrez, J. Meléndez, and G. Carrillo-Caicedo, ‘Comparison of impedance based
fault location methods for power distribution systems’, Electric Power Systems Research,
vol. 78, no. 4, pp. 657–666, Apr. 2008, doi: 10.1016/j.epsr.2007.05.010.
4. H. Nouri, Chun Wang, and T. Davies, ‘An accurate fault location technique for distribu-
tion lines with tapped loads using wavelet transform’, in 2001 IEEE Porto Power Tech
Proceedings (Cat. No.01EX502), Sep. 2001, vol. 3, p. 4 pp. vol.3-. doi:
10.1109/PTC.2001.964908.
5. R. Jalilzadeh Hamidi, H. Livani, and R. Rezaiesarlak, ‘Traveling-Wave Detection Tech-
nique Using Short-Time Matrix Pencil Method’, IEEE Transactions on Power Delivery,
vol. 32, no. 6, pp. 2565–2574, Dec. 2017, doi: 10.1109/TPWRD.2017.2685360.
6. S. Robson, A. Haddad, and H. Griffiths, ‘Fault Location on Branched Networks Using a
Multiended Approach’, IEEE Transactions on Power Delivery, vol. 29, no. 4, pp. 1955–
1963, Aug. 2014, doi: 10.1109/TPWRD.2014.2302137.
7. D. Thukaram, H. P. Khincha, and H. P. Vijaynarasimha, ‘Artificial neural network and
support vector Machine approach for locating faults in radial distribution systems’, IEEE
Transactions on Power Delivery, vol. 20, no. 2, pp. 710–721, Apr. 2005, doi:
10.1109/TPWRD.2005.844307.
8. J. Mora-Florez, V. Barrera-Nunez, and G. Carrillo-Caicedo, ‘Fault Location in Power Dis-
tribution Systems Using a Learning Algorithm for Multivariable Data Analysis’, IEEE
Transactions on Power Delivery, vol. 22, no. 3, pp. 1715–1721, Jul. 2007, doi:
10.1109/TPWRD.2006.883021.
9. L. Lovisolo, J. A. Moor Neto, K. Figueiredo, L. de Menezes Laporte, and J. C. dos Santos
Roch, ‘Location of faults generating short-duration voltage variations in distribution sys-
tems regions from records captured at one point and decomposed into damped sinusoids’,
Transmission Distribution IET Generation, vol. 6, no. 12, pp. 1225–1234, Dec. 2012, doi:
10.1049/iet-gtd.2011.0553.
10. M. Farajollahi, A. Shahsavari, E. M. Stewart, and H. Mohsenian-Rad, ‘Locating the
Source of Events in Power Distribution Systems Using Micro-PMU Data’, IEEE Transac-
tions on Power Systems, vol. 33, no. 6, pp. 6343–6354, Nov. 2018, doi:
10.1109/TPWRS.2018.2832126.