Charles Kim, "Lecture Notes on Fault Detection and Location

in Distribution Systems," 2010.

Source: www.mwftr.com

3. Fault Location Algorithms

Charles Kim

June 2010

Fault Location Overview

Traditional Methods of determining the location
of a fault on T&D lines
Impedance Approaches (Our Focus)
Traveling Wave Approaches
Problems in Distribution Network
Other Methods
Short-circuit analysis software
Customer calls (distribution case)
Line inspection
Fault Indicators (4)
New Opportunities
Smart Sensors (6)
Smart Meters (6)

2
 Impedance-Based Measurement Technique Overview

Calculation of the fault location from the

apparent impedance seen looking into the line
from one end (or two ends).
Steady-State Approach
Phase-to-ground voltages and current in each
phase must be determined.
Fault Impedance Influence
Loading Influence
Ground Fault Case
Zero-sequence Impedance Information
Ground Compensation Factor

Simple Impedance Equations (with Rf = 0)

k=(ZoL Z1L)/(3 * Z1L) ground compensation factor

ZoL: zero-sequence line impedance
m: per unit distance to fault
IR: Residual Current 4
 Basic Model with System Parameters

Simplifiedtransmissionlinewithtwosources
5

Impedance (Distance) Equation

IG:LineCurrentduringfault
If:Faultcurrentthroughthefault
resistorRf.

6
 Derivation of If/IG

Derivation Continued

8
 Reactive component of fault resistance

2 factors
Current distribution factor, ds
Determined by system impedances
Angle of ds () = 0 if system is homogeneous (Same R/X
ratio of lines)
Circuit loading factor, ns
Determined by the load current (IL) presence in the system
The angle of ns () is not zero if there is a load flow in the
system
If IG is much bigger than IL, the angle will approach zero.
Sum of the angles (+) determines the reactive
component caused by fault resistance, Rf.

Practice with MathCad

10
Mutual Impedance vs. Sequence Impedance

11

k factor from mutual impedance

12
 Practical Approaches:
(1) Simplification & (2) Limiting errors

Various techniques to accommodate the two

factors or to reduce the factors
Fault location without using source impedance
(Takagi et al)
Fault location using source impedance: US Patent
5839093 System for locating faults and estimating
fault resistance in distribution networks with tapped
loads Novosel et al. 1998, ABB
Simplification
Reactance Method
Voltage Sag
Rms V and I
13

Reactance Method
Measures the apparent impedance
Determines the ratio of the measured reactance
and to the reactance of the entire line, which is
proportional to the distance to the fault.
Assumptions:
The current through the fault resistance is in phase
with the current at the measurement point
There is no load prior to the fault
One of the earliest algorithms that compensate
for the fault resistance by measuring only the
imaginary part of the apparent line impedance

14
 Reactance Method

m: per-unit distance to the fault

15

Reactance Method for Ground Fault Case

16
Further Approximation

17

Reactance Approach

18
Further Approximation

19

Performance

20
 Fault Location without Using Source Impedance

Improvement
Elimination of load current by determining the
change in current on occurrence of a fault.
Takagi
Uses the net fault line current (IG )

21

Takagi Method - Patent

22
 Takagi Method

23

Takagi Approach Example

24
 Variation of A2

Modifiedmethod
Useszerosequencecurrent(IR)(insteadofthe
netfaultcurrent)
Usestheangle ofthecurrentdistributionfactor
derivedfromthesourceimpedancedata
Accountsfornonhomogeneoussystem
Reducesthereactanceeffecterror
Problemwithaccuratecorrectionof.

25

Further Improvement (Modified Takagi &SEL)

26
 Differential Equation Approach

L calculationfromsampleddata
ofv andI
TimedomainApproach
ExtractionofRandX
Differentwaysofdealing
Derivative(d/dt)forcomputing
fromsampleddata
NumericalAnalysisofDerivative
Conversionoftheequationinto
integral,thentrapezoidalrule
Discretized elementmodel

27

Discretized Model of Elements EMTP approach

28
 Trapezoidal Approximation

Source:ArumG.Phadke andJamesS.Thorp,Computerrelaying
forpowersystems,RSP.Ltd August1994,pp.118131
29

Final Form and Practice with MathCad

30
Differential Equation Approach on SEL

31

Fault Location with Actual Data

32
 Fault location using Source Impedance

The current distribution factor, ds, is a function of the source

impedance, line impedance, and the unknown fault location.
If the source impedance is known, fault location can be
estimated without the assumption related with the
33
distribution factor.

Quadratic Equation Derivation

34
 Continued

35

Continued-

Problem
Source Impedance in program must be the same as
the actual source impedance of a network
36
 Factors affecting the accuracy of the fault estimation
The combined effect of the load current and fault
resistance (reactance effect)
Influence of zero-sequence mutual effect on the
components
Uncertainty about the zero-sequence impedance
Difficult to obtain an accurate zero-sequence
impedance (ZoL) for line
The value is affected by soil resistivity, which can be
difficult to measure, and may be changeable.
A 20% error in the ZoL can introduce a 15% error in
fault calculation
The impedance is not uniformly distributed along the
line length (100 to 1 variation in earth resistivity
produces about 2 to 1 change in Zo)

37

Other Factors of Error

Insufficient accuracy of the line model (un-

transposed lines are represented as being
transposed and charging capacitance is
not considered)
Presence of shunt reactors and capacitors
Load flow unbalance
Measurement errors, CT/PT errors
Low resolution or sampling rate at
measurement station
38
 Two-Terminal Data Methods
Advantages
More accurate than one-terminal methods
Able to minimize or eliminate the effects of fault resistance,
loading, and charging current.
Positive Sequence components are used instead of zero-
sequence, eliminating the adverse effect of zero-sequence
components
Drawback
The data from both ends must be collected at one location
RequiredEquipment
Measuring devices for 3-phase voltages and currents at each
end with time stamping.
Communication equipment
Central computer for collection of data and calculation of fault
location
39

Some other Fault Location Methods

40
More

41

More

42
 Traveling Wave Approach

TransientWaveArrivalTime
TransientWaveFrequencyAnalysis
Features
Highsamplingofdata
TransientFrequencycorrelationtofaultdistance
Moresuitableforlongdistancefaults

43

Overview of High Frequency/Traveling Wave Methods

Reflectionandtransmissionoffaultgenerated
travelingwaveonthefaultedpoint
Accuratebutmorecomplexandexpensivein
implementation
Addedequipment
GPSsystem
Faulttransientdetectorsanddiagnosissoftware
Difficultiesintheconfigurationanlocationof
faulttransientdetectorsduetocomplex
distributionnetwork
44
 Traveling Wave Method
CorrelationofIncidentandreflectedwaveform.
Singleendedanddoubleendedapproaches
Bigprobleminmultiplediscontinuity(reflection
points)innetworks
Variations
Highfrequencysignalsmeasuredatthesubstation(with
Waveletanalysis)F.H,Magnago andA.Abur (1999)Anew
faultlocationtechniqueforradialdistributionsystems
basedonhighfrequencysignals.ProcofIEEEPESSummer
Meeting,1:426431

45

Summated Voltage and Current Wave

Inatotaln lines,withthesamevalueZo,connected
toacommonbusbar,thesummatedwavesonthe
linecarryingtheincidentwaveis:

Asthenumberoflinesconnectedtoabusbarincreases
Thesummatedvoltagewilltendtozero
Thesummatedcurrentwavewilldouble
Observationofcurrentwaves(viaCT)maybepreferable
Butbothhavebeenapplied.

46
 Accuracy Limitation
Assumption
Thelightspeed:3*108 m/s
Discontinuitiesinelectricalsystemproduceswave
reflections
Twoterminalmethodallowtimingfromtheinitiationof
thefault,hencereflectedwavesarenotused.
Accuracy
300metersevenforlonglines
Wavedetectionerrorduetointerpretationofthe
transientisamajorsourceoferror.Manytransients
and/orreflectedtransientsappearatthesametime.
Oneterminalmethodneedstobemoresophisticated
signatureanalysisrequired.

47

Traveling Wave Method Modes

TypeA(singleended)mode
Flashoveratthefaultpointlaunchestwowavesthattravelinopposite
directionsawayfromthefault
Theeffectiveimpedancesatthelineterminalsareassumedtobe lowerthan
thelinesurgeimpedancesothatsignificantreflectionsareproducedwhich
thentravelbackalongthefaultylinetothefaultpoint.
Ifthefaultarcstillexists,andalsopresentsaneffectiveresistancelowerthan
thesurgeimpedanceoftheline,thenanywavesarrivingatthefaultwillbe
almosttotallyreflectedbacktothelineterminals.
TypeD(twoended)mode
Differenceinthetimesoffirstarrivalofthetwofaultgeneratedwavesatboth
lineterminalsaredetermined.
Reflectionsfromotherdiscontinuities,branches,tappedloads,cablesections
becomeunimportant.
TypeE(SingleEndedCircuitBreakerTransient)mode
Usesthetransientscreatedwhenalineisreenergizedbyclosingacircuit
breaker(closetotheImpulseCurrentMethodoffaultlocationwidelyusedon
undergroundcables)

48
 Loss Free Overhead Line
Detection
DeviceatS

Detection
DeviceatR

The distinction between the reflected

wave from the fault point and that
from the remote bus bar is vital. 49

Example Fault Locator

50
 Test 1

51

Test 2 (with load)

52
 Can we apply the algorithms to Distribution
Systems Faults?
Numerous factors affecting the algorithms in
distribution networks
Conductor size change
Multiple feeder taps and laterals
Inaccurate models and system data and dynamic
configuration
Effects of fault impedance
Different Grounding Methods
Solid grounding
Ungrounded Network
Petersons coil
Resistance Grounded

53

Distribution Network Topology

HeterogeneousFeeders
Differentsizeandlengthofcables
Presenceofoverheadandundergroundlines
Presenceofsingle,double,andthreephaseloads
Presenceoflateralsalongthemainfeeder
Presenceofloadtapsalongthemainfeederandlaterals.
Causeofestimationerrorinfaultlocations
Model
Lumpedparametermodel
Symmetricalcomponentsonphasorbasedalgorithms
Singlelinetogroundfaultismostcommon
Differentvaluesoffaultresistance

54
 Fault Location at SDG&E
FaultLocationEfforts
DataMeasurement(PQNode)at36Substations
DataAnalysisusingPQView
Algorithm(reactanceapproach)Programmed
byEPRI
OfflineEvaluationforafewSubstationCircuits

55

Fault Location EPRI Example

North Bus Creelman - 12/17/2006 22:52:39.8470
Reactance to Fault
North Bus Creelman - 12/17/2006 22:52:39.8470 Va Vb Vc Vab Vbc Vca Ia
Va Vb Vc Ia Ib Ic Ires
Ib Ic Iab Ibc Ica Ires XTF
10
12
Voltage (kV)

5
Voltage (kV)

10
0 8

-5
2
Current (kA)

-10
2 1

1 0
Current (kA)

Reactance (ohms)

4.50
0
4.25
-1 4.00
3.792 (k1=3.500)
1B 3
3.75
-2
0.00 0.02 0.04 0.06 0.08 0.10 0.12
0.02 0.04 0.06 0.08 0.10 0.12
Time (s) Time
EPRI/Electrotek PQView

56
 Strength and Weakness of the Current Approach

Current
approach
Simple
and
Effective
Load
dependen
cy
Overreac
hing&
Under
reaching
Problem

57

Reproduction of EPRI Approach

3 4
210 210
3
1.94510 1.04910
4

3 4
Ia 110 110
Va
Ib
Vb 0
Ic 0
Vc
Ir 4
3 110
110
4
3
1.05410
1.95910 4
210
3
210 0 0.05 0.1 0.15
0 0.05 0.1 0.15
4 T 0.133
4 T 0.133 2.60310
2.60310
4
2 10

North Bus Creelman - 12/17/2006 22:52:39.8470

Reactance to Fault

Va Vb Vc Vab Vbc Vca Ia

Ib Ic Iab Ibc Ica Ires XTF

12
Voltage (kV)

10

2
Current (kA)

0
Reactance (ohms)

4.50

4.25

4.00
3.792 (k1=3.500)
1B 3
3.75
0.02 0.04 0.06 0.08 0.10 0.12
Time

58
 Strength and Weakness of the Current Approach

Currentapproach
Simpleand
Effective
Loaddependency
Overreaching
&Under
reaching
Problem
MinimumData
Length
Requirement
atleast2cycles
offaulteddata
areneeded.

59

Simpler Approach
RMScurrentOnly
FaultCurrentCalculationateach
everynode
LookupTable

Fault Current Vs Fault Distance

Progress Energy Example
2500

2000
Fault Current
Current [A]

1500 Ambient Load

Eqn
1000

500

0
0 5 10 15 20
60
Fault Distance [mi]
 Characterization of Specific Fault
VoltageDipEnergyIndex(Edip)
Characterizationofspecificfault
Integrationofthedropinsignalenergyoverthe
durationofanevent.
V(t):RMSvoltageovertime
Vnom:Ratedvoltage

61

Fault Location by RMS current main tool

DeterminetheaverageofRMScurrentduringthefault(initial
andsteadystateportions)duration
Determinethecurrentindex:Iindex
p:predictedvalue
Exp:experimentalvalue
Comparethecurrentindexatseveralnodesdeterminedby
DSFL(byfaultcurrent&recloser,etc?)
Pickthelocationwherethecurrentindexisminimum(i.e.,the
leasterrorlocationbetweenmodelvs actual)

62
 Example

63

Other Methods
DistributedDevices
VoltageSensormatrix
Voltagemagnitudeandphaseangletableofallsections
andnodesinthenetwork
Measureddatavs.historicalfaultdata.
HybridMethods
Faultdistancecalculation&DistributedDeviceMethod
FaultIndicatorMethods
UseofSmartMetersandSmartGridCommunication
Infrastructure

64
 References
L.Nastac andA.Thatte,DistributionSystemsFault
Locator ElectricalInfrastructureTechnology,
TrainingandAssessmentProgram,DOETechnical
ReportunderCooperativeAgreementDEFC02
04CH11241,September30,2006
L.Nastac,AdvancedFaultAnalysisSoftware(or
AFAS)forDistributionPowerSystems, Centerfor
GridModernizationProgram,DOETechnicalReport
underCooperativeAgreementDEFC0205CH11298,
July31,2007.
NumerousUSpatentsandpatentpublications

65