00488055

7/28/2019 00488055

1/8

New method of power swing blocking for digitaldistance protectionP.J .MooreA.T.J ohns

Indexing terms: Power swings, Distance relay, Reactance transients, Fault detection

Abstract: A new method for preventing digitaldistance relay operation during power swings ispresented based on detection of transients in themeasured reactance. The transients are detectedby the difference between two reactancescalculated using two parallel FIR filters based oncoefficients used in the discrete Hartleytransform. The transients are present undergenuine fault conditions but not during powerswings. Fault transient detection therefore enablesthe relay for faults but leaves it inherentlyblocked for power swings. Simulation results arepresented showing that the fault transientdetection method does not lead to increased relayoperating time, but allows the relay todifferentiate between faults and power swings.The significant advantage of this new method ofblocking is that the relay can correctly operate ifa genuine fault occurs during the power swingperiod.

Glossary o f symbolsa, b, c =subscripts denoting phase quantities0, 1 = subscripts denoting zero and positive phasesequence quantitiesR, L =resistance and inductance measured by relayX , X , = reactance and directional reactance measuredby relayvr[n] ,Qn] =discrete time relay voltage and current sig-nals at time index nTi[n]=discrete time current signal time derivativeT =sampling intervalf , =sampling frequency,f, = l /Tp = number of discrete time samples taken per powersystem frequency cycleJb =nominal power system frequencyK =residual compensation factor, K = 1/3(Z10/Z~,-1)Zl =transmission line impedanceD =derived signal being proportional to square of cur-@ IEE, 1996IEE Proceedings online no. 19960055Paper first received 9th February 1995 and in final revised form 29thAugust 1995The authors are with the School of Electronic and Electrical Engineering,Universityof Bath, Bath BA2 7AY, UK

rent magnitudea = angular delay between solutions of fundamentalline equationHh(k)= coefficients of Hartley filter (for k = 0, ...,K,(k ) =time reversed coefficients of Hartley filter (fork =0, ...,N-1)1 Introduct ionUnder certain conditions of power system instabilityunwanted operation of distance relays can occur due tothe apparent impedance sensed by the relay fallingwithin the characteristic. This is due to an abnormaltransmission line voltage profile. For example, if thevoltages at either end of a homogenous transmissionline are in anti-phase, then there exists a position ofvoltage zero at the middle of the line which will beinterpreted as a fault by distance protection at eitherend. Although such conditions are undesirable on apower system, distance relays cannot provide the cor-rective action needed to restore stability since the phe-nomenon is not associated with any defect in theprotected line. Consequently distance relays are usuallyblocked from operating during power swings. This isespecially important where the instability is recoverableprovided critical transmission lines remain in service.The conventional method of blocking relay operationduring power swings is to time the movement of theimpedance through the zones of the relay. Under genu-ine, in-zone, fault conditions the impedance measuredby the relay will move from its prefault position intothe characteristic almost instantaneously. Thus, zones1, 2, 3 and an outer, power swing detection, zone willpick up simultaneously. However, under power swingconditions the impedance trajectory, being governed bythe inertial constants of the generating plant attachedto the line, will move slowly, typically taking hundredsof milliseconds to move into the characteristic. In thissituation, the outer power swing detection zone willpick up before any other zone. This time delay is usedas the basis of the conventional power swing blockingmethod. Two problems can be experienced with thismethod. First, the relay may not respond to genuinefaults occuring during the power swing period since itis blocked from operation. Secondly, the time delay hasto be set with a knowledge of the likely speed of move-ment of the impedance during the power swing. It istherefore possible that the relay may not correctlyblock if fast changes in apparent impedance occur.

Recently, new techniques for power swing blockinghave been reported and include measuring the rate of

N-1)

19EE ProccGenev. Tvansm. Distrib., Vol. 143, No. I , January 1996

7/28/2019 00488055

2/8

change of line current [l ], measuring the rate of changeof apparent resistance [2] and calculating the anglebetween busbar voltages [3]. The new method to bedescribed here differs from all these techniques in thatit detects the condition of a f au l t , rather than the con-dition of a power swing. This provides the benefit thatno assumptions regarding the nature of the power sys-tem are needed. This work is a further development ofearlier research in the area of power swing blockingusing fault transient detection [4] and is applicable todigital distance relays.

filtering mixingI I

element relayimpedance fault trip tripII

enable flagFig. 1 Structure o digital distance relay

2 Digital distance relay structure2.1 OverviewA block diagram of the structure of the digital distancerelay used for this work is shown in Fig. 1. The dis-crete time input signals to the relay, i.e. the three phasevoltages and three line currents, are assumed to besampled at a frequency of 4kHz using a 16 bit ana-logue to digital converter. Sampling is assumed to notcause aliasing and to occur on all inputs at the sameinstant of time, i.e. each input channel has a separatesample and hold amplifier. The main filtering functionof the relay is performed digitally, for each input sig-nal, using a finite impulse response (FIR) filter basedon a sine function and having 20 symmetrical coeffi-cients. The filtered signals are then mixed in order toprovide the inputs for three phase elements and threeearth elements of zone 1. Each zone 1 element com-prises of an impedance calculation stage, where theinput discrete time voltage and current signals are con-verted to discrete time resistance and reactance signals,and a fault evaluation stage, where the impedances arecompared against a characteristic in order to set theelement trip flag. Any relay element trip will cause arelay trip if the ENABLE flag is set which, in turn, is afunction of the power swing blocking feature of therelay to be described in the next Section. For simplic-ity, this work has considered only the zone 1 elementssince the power swing blocking feature is not related tothe operation of other zones. Zones 2 and 3 may easilybe implemented by providing extra fault evaluationstages reflecting the larger characteristics of these ele-ments, no extra impedance calculation is required.2.2 Impedance calculationThe calculation of measured resistance, R, and reac-tance, X , in the fault loop is performed using a well-established technique [5] which solves the fundamentalfirst-order line equation, i.e. v =Ri +L dildt, for R andL (fault loop inductance) by using two sets of discretetime relaying voltage v(n) and current i(n) separated intime by six samples, which is equivalent to 1.5ms. The20

equations for calculating R and X re given below:v[n]i[n 61- [n- 6]i[n]i[n]Z[n- 1- [n-6]Z[n]R Z

v[n- 6]i[n]~[n]i[n61i[n]Z[n 1- i[n-6]i[n]x =2xfoL %

wheref o s the system frequency and the time derivativeis approximated by the central difference formula,

( 3 ), di[n] i[n+11- [n- 11 1

d t 2T f sz[n]=- T = -wheref, s the sampling frequency.+

--.--.-.1,2x . -9 .- ..........1 ..... . .-

y - , -1 ._.

. . .__-- ...............................

RO -9 R,Fig.2 elay quadrilateral characteristic and counting strategy2.3 Relay characteristic and countingstrategyThe quadrilateral characteristic of Fig. 2 was used inall of the relay elements. To fulfil the requirements oftransmission protection where fast relay operation isrequired, a target relay operating time of less thanlOms for the majority of faults was chosen. Althoughthe relay digitally filters its inputs, this filtering is notperfect and results in some noise appearing on theimpedance estimates. Consequently the singleoccurence of the impedance falling within the charac-teristic is not a good indication of a fault. To providethe extra filtering needed to ensure that the relay givesgood reach point behaviour, a counting strategy is usedas shown in Fig. 2. Each relay element is providedwith a counter which is initially set to zero. The coun-ter can increase provided the measured impedance fallswithin the characteristic and that the directional checkdescribed in the next Section is positive. For faultsleading to impedance estimates falling within the first80% of the characteristic, the counter increases at therate of nine for each sample in this area. When thecounter reaches 45, i.e. five coiisecutive samples in thisarea, the element trips. If the fault is closer to the relayboundary, the counting value is reduced to 4 for meas-ured impedances between 80 and 90% of the character-istic. The counting value is set to 1 for impedancesbetween 90 and 100% of the characteristic. The count-ing values are reflected negatively for impedances fall-ing beyond the reach point. The counting strategy hasbeen carefully designed in order to provide the extrafiltering required for exacting relay operation under allconditions. The counter cannot exceed 90 or go lessthan zero. After a relay trip occurs, the relay resets20ms after the element counters return to zero.

IEE Proc.-Gener. Transm. Distrib., Vol. 143, No. 1, January 1996

7/28/2019 00488055

3/8

2.4 DirectionalityIn order to directionalise the relay for busbar faultswhere the voltage collapses to zero, a separate imped-ance is calculated using the post fault current and theprefault voltage which is stored in memory. This reac-tance is referred to as the directional reactance, X,,and is given as

(4 )~ [ n - 6]i[n]~ [ n ] i [n- 61

Z[n]i[n 1' - [ n- 6]i[n]'x, M 2 T f (1p = S = 8 0f o

X, will be positive for forward faults and negative forreverse faults and is used to directionalise the relay forclose-up faults. Hence, in Fig. 2, a directional thresh-old is established which X m must exceed in order forthe element counter to increase. In eqn. 4, a one-cycledelay for the voltage is achieved by taking valuesstored in memory 80 samples previously since, with asampling frequency of 4kHz, one cycle of a 50Hz sig-nal is made in exactly 80 samples. Under dynamic con-ditions where the power system frequency varies, 80samples will not correspond exactly to one cycle, how-ever, since X, is used to form only an approximatemeasure of direction, this will not lead to relay malop-eration. In fact, errors of up to 25" in the memory volt-age do not affect directionality. Delays of greater thanone cycle can be used by taking an integer multiple of pin eqn. 4.2.5 Relay elementsThe six zone-1 relay elements each have separateimpedance calculation and fault evaluation stages. Theearth elements need residually compensated line cur-rents and so the signal mixing stage of Fig. l performsthe following calculation:

where is,, is the residual current, and

and similarly for ib, and i,, where K is the residual com-pensation factor defined by K = 1/3 (Zlo/Zl~-l).incethe modulus of K is taken, residual compensation isachieved using a scalar multiplication in eqn. 6. This issuitable for overhead line applications where the argu-ment of K rarely exceeds 5". For underground cableapplications this would not be acceptable and a com-plex residual compensation technique must be used [6].Hence, each earth element is provided with the respec-tive phase voltage and residually compensated line cur-rent inputs. Since earth faults are always accompaniedby zero sequence components in the relay input voltageand current signals, earth elements can be preventedfrom operation for clear-of-earth faults by testing forthe presence of residual current. The magnitude ofresidual current can be found using the following equa-tion:

ires[n]=i +ib[n]+ic[n]i a [ n ]= a [ n ]+lKlires[n]

(5)

(6)

D =-2~folI,,,I~ sincu(7)=ires[n]ires[n- 61 - res[n - ]ires[n]'6a =2 T f 0 ,

J sProof of this equation is shown in the Appendix. ThusD is compared against a suitable threshold with theeffect that earth element operation is prevented if insuf-IEE Proc-Genes. Transm.Distrih., Vol. 143, No . 1, January 1996

ficient residual current is detected. Note that residualcurrent can be present under normal conditions due toeffects such as assymetric loading and nonideally trans-posed lines, thus the threshold level needs to considerthese effects.

The presence of phase faults and three phase faultswill be detected by the phase elements. The a-b elementis provided with the following inputs:U[n]=U a [ n ]- b[n] Z[n]=Za[n] Zb[n] (8)

and similarly for the b-c and c-a elements. Biasing thephase elements against the presence of negativesequence current is not carried out since it would pre-vent relay operation for three phase faults.+ault inceptionIposition B y I> 1 /position cposition AvA .............. ......... .

.t

. ..

Fig. 3 Effect of fault on relaying voltage

3detectionPower swing blocking using faul t disturbance

3. I The effect of disturbances on measuredimpedanceThe occurrence of a fault in the relay protected zonewill result in instantaneous changes in the magnitudeand phase of the relay input signals. This is illustratedin Fig. 3 where the relaying voltage is subject to areduction in amplitude and displacement in phase.Since the relaying algorithm processes a fixed numberof input samples at any point in time, the concept of atime window, equal to the length of the relaying algo-rithm, can be applied to Fig. 3 . The window length,T,s, corresponds to the sum of the group delays of fil-tering, impedance calculation and fault evaluation usedin the relay. The window moves towards the right atevery sampling instant and so the right hand edge ofthe window passes through the most recent voltagesample. In position A, the window is seen to containonly prefault information. More than T,s after thefault, in position C, the window is seen to contain onlypostfault information. However, in position B the win-dow is seen to traverse the fault point and hence therelay processes both prefault and postfault informa-tion. This leads to errors in the impedance measure-ment process since the waveform present in window Bdoes not represent a single frequency sinusoid, uponwhich the relay processing is based.

This effect is more conveniently described by examin-ing reactances measured by the relay for a typical fault.Fig. 4 shows reactances measured for three faults, ineach case the prefault and postfault conditions areidentical except that the point on wave of the voltageat the fault inception is different. It is seen that prior tothe fault, i.e. equivalent to position A of the window,all three reactances are the same, similarly after the21

7/28/2019 00488055

4/8

fault, i.e. position C, again the reactances are identical.However, during the fault, i.e. position B, the reactancebehaviour is erratic and dependent on the fault incep-tion angle. It is also discovered that changing the FIRfilter coefficients used in the relay also affects thebehaviour of reactance during the fault. A result simi-lar to Fig. 4 can also be produced by using just onefault, but processing it through the relay three timeseach using a different FIR filter. This effect- hefuulttransient effect - hich is also present on the meas-ured resistance, has hitherto been regarded as a nui-sance. For example, without careful design of thecounting strategy it is possible for the relay to over-reach for certain out of zone fault conditions due to thefault transient effect causing the measured impedanceto enter the characteristic long enough for the relay totrip. The basis of the work described in this paper isrecognition of the fact that the fault transient effect canbe used to provide the relay with important informa-tion since the fault transient effect is present, to agreater or lesser extent, at the inception of every faultfor which the relay must operate. Furthermore, thefault transient effect will not be present under powerswing conditions which are caused by changes in gener-ator rotor positions resulting in gradual, rather thaninstantaneous, changes in relaying voltage and currentsignals. Hence the potential exists for preventing relayoperation under power swings, not, as is the conven-tion, by blocking operation when swings occur, but byenabling operation when faults occur, thus leaving therelay inherently blocked to all power swing effects.

t ime , msFig. 4 Effect o fault inceptton angIe on measured reactance3.2 Detection of fault disturbancesIn order to detect the disturbances described earlier it isnecessary to detect only the transient region of thereactance measurement caused by the fault. This wasinitially attempted by a form of numerical differentia-tion of the measured reactance [4] which, with referenceto Fig. 4, aimed to provide an output during the tran-period, yet entirely remove the constant reactancepresent during the prefault and postfault periods.Despite some encouraging results using this approach,its high sensitivity to noise can cause spurious outputsoutside of the fault instant. Superimposed techniqueswere also considered yet these would require the relayto use synchronous sampling to afford correct fault dis-turbance detection when changes occur in the powersystem frequency. The faregoing difficulties were over-come by using an approach based on the calculation oftwo reactances using identical input signals filteredusing two different FIR filters. As discussed earlier,dissimilar FIR filters will react differently to the faulttransient period thus leading to differences in the calcu-

lated reactance. The two reactance outputs can then besubtracted to yield an output which is nonzero onlyduring a fault disturbance. The operation of this tech-nique relies upon the use of two filters which haveidentical magnitude responses, giving immunity tochanges in system frequency or harmonics, but differ-ing phase responses, thus promoting individual behav-iour to fault discontinuities. These constraints rule outthe use of conventional Fourier type filters based onsine and cosine waveshapes since the magnituderesponses are similar, but not identical, which leads toproblems with noise susceptibility and system fre-quency changes. Instead, filters based on coefficientsused in the discrete Hartley transform are used.3.3 The discrete Hartley transformThe discrete Hartley transform was proposed byBracewell [7] based upon an integral transform pre-sented by Hartley [8]. The Hartley transform can beused to evaluate the Fourier transform and bears theadvantage that only real computation is required, asopposed to the Fourier transform which uses complexcomputation. The N coefficients used in the fundamen-tal component of the discrete Hartley transform, H&),are evaluated from:

N. \ for k =0 , 1, . ,N - 1 (9)and can be considered as coefficients of a FIR filter inthe same way that Fourier coefficients constitute 'Fou-rier filters'. Hartley filter coefficients have the property,not found in Fourier coefficients, that a filter imple-mented using coefficents Hh(k), and a filter imple-mented using time reversed coefficients, K h ( k ) , .e.have identical magnitude responses but differing phaseresponses. This effect is a consequence of the fact thatthe direct and inverse forms of the Hartley transformare identical and has recently lead to the use of theHartley transform in systolic array parallel processingPI.The frequency magnitude response of the Hartley fil-ters is similar to the general form of the Fourier filtersfrequency magnitude response and is of the formsin(x)lx. For N = 8, the magnitude and argument ofthe frequency response of the Hartley filters is shown inTable 1 for several spot frequencies.Table 1: Frequency responses for Hartley filters (N=8,sampling frequency =4000Hz)

H - * ( k ) =H*(N - C - 1) (10)

~Frequency 50Hz 100Hz 1200HzHh( l 0.754176.5" 1.481L63" 0.662154'& ( f l 0.7541108" 1.4811-126" 0.662118"3.4 Implemen tation of fault disturbancedetection processThe detection of fault disturbances using the fault tran-sient effect was implemented using the above tech-niques. Fig. 5 shows the processing required. Inputsamples are filtered, in parallel, using the Hartley filterswith coefficients described in eqns. 9 and 10 . Two reac-tances are calculated from the filtered signals and sub-tracted to form a difference, X@@ which is zero exceptunder fault disturbance conditions. Fig. 6 shows thebehaviour of X d ~o a single phase-to-earth fault at the

IEE Pioc -Gener Transm Distiib Vol 143 No 1 January 1996L

7/28/2019 00488055

5/8

relay boundary. To facilitate fault detection using Xdirf tthis signal is, in turn, squared and then averaged usingan eight point moving average filter to yield the signalX2d,v, shown in Fig. 7. It is clear that X2d,v veryquickly responds to a fault disturbance. Finally X 2 d a v iscompared to a threshold.

10050

mE 0 -

enable

fauI occurrence----

prefault , , posttaultl

- UFig. 5 Fault detection process

o 58 59 60

defined using the above mixing equations. A thresholdof 10Q2 was used for detecting Pdavs shown inFig. 7.40 I I/

61 1 62 63 64 65 55

5 0 0

time, msFig. 8 Reactance of a to earth element

ul 80E

-. threshold

-150 1 IFig. 6 Re,Yponse of XdirJ o an earth fault

0 -

:::q fi2 5 0 0

: : : i ; : i : 48 52 56 60 64 68 72 76time, ms

58 59 60 61 62 63 6 4 65 66 67time, msFig. 7 Resp0n.w of X2davo un earth fault

To ensure fast fault transient effect detection, thenumber of coefficients used in each Hartley filter, N ,was chosen to be 8, which is less than the 20 coeffi-cients used in the relay digital filter. The process shownin Fig. 5 is able to detect fault disturbances arisingfrom all types of fault for which the relay must operate.For this reason, the discrete time input signals are pro-duced by mixing the relay input signals v,[n], v ~ [ u ] ,e [ ~ ] ,& [ U ] , 4 7 2 1 and i e [n] ,with the effect that only one faultdisturbance detection process is needed for all faulttypes. Ideally the signals would be mixed to produce apositive phase sequence quantities, however, althoughpossible [ l o ] , his would be computational intensive,and so an alternate, simpler form of mixing wassought. Initially a modal form of mixing, using ele-ments of Karrenbauers modal transformation matrixwas tried, however, this was not successful for allfaults. The mixing was consequently adjusted toandExtensive testing of the arrangement for a wide varietyof faults, short circuit levels, loading and fault resist-ances up to 15052 showed that X2davwas always well

v,i,[n] =v,[n] ub[n]- v,[n]imza[n] ,[n] - 2ib[n] 3ic[n]

(11)(12)

IEE Proc.-Gener. Transm.Distr ib. , Vol. 143, No . 1, January 1996

-I ; i i i i : : I : i I i i : I48 52 56 60 54 68 72 76time, msFig. 9 ReJistance of a to earth element

-1 048 52 56 60 64 68 72 76time, msFig. 10 Directionul reactance oj u to earth element

10080::20 I

t ltrip It Ilenobie

Fig. 11 Counter and trip flag or U to earth element, and ENABLEflugThe threshold prevents unfaulted steady-state valuesof X2d,y triggering the fault disturbance detection proc-ess and could be significantly increased if necessary. Inparticular, the threshold can be increased to affordgreater noise immunity, although the fault detectiontechnique was observed to perform correctly (using the1052* setting) under likely noise conditions experiencedby the relay, e.g. travelling wave noise and harmonics.

23

7/28/2019 00488055

6/8

However, in applications where high levels of noise areanticipated, it is likely that the number of coefficientsin the relay's main FIR filter will need to be increasedto ensure correct relay operation. It would be prudent,in this situation, to increase the Hartley filter coeffi-cients by a similar ratio.When X2duy xceeds the threshold, an ENABLE flagis set for a period of 20ms. This period is designed toallow correct operation of the fault evaluation processof the relay elements described in Section 2. The 20msperiod is reset every time any element counter exceedszero. The ENABLE flag is then used to allow the relayelement trips to cause a relay trip as shown in Fig. 1. 4 Results

260km transmissionline

1=080s

Lthree phase faultFig. 12 Model system used o r power swing study

The performance of the relay under power swingconditions was examined for the model system shownin Fig. 12which depicts a generating station with two600MVA sets connected, via transformers, to a 260km,400kV overhead line which is, in turn, connected to alarge power system. The simulation allows a threephase fault to be applied to busbar R which conse-quently prevents power being exported from the stationto the rest of the system. Parameters used in the simu-lation are given in Appendix 8.2.Fig. 13shows the measured R and X of the a-b ele-ment when only one generating set is connected to bus-bar S . Initially, the generator exports power at ratedload. At time 0.05seconds a fault is applied at busbarR which is cleared after 150ms. During this period Rand X measure the total line impedance, although nooperation occurs since the relay is set to 80 % of thisvalue. Note that the measured R and Xvalues suddenlychange at the point of clearance. After the fault iscleared the system becomes unstable resulting in powerswing and pole slips. Pole slips result in near assymp-totic behaviour in R and X . The ENABLE flag is setby both the fault inception and the fault clearance,since these events result in instantaneous changes in24

relay voltage and current. Furthermore the ENABLEflag responds to the pole slips, however, due to the ass-ymptotic behaviour of X and R at these positions, thereis n o risk of relay operation. At times 0.4s, 0.78s, 1.0sand 1.18s, the measured R and X values are in therelay characteristic, this is illustrated for one case inFig. 14 which shows the passage of impedance throughthe relay quadrilateral characteristic. The a-b elementtrip flag is not set during the power swing due to thebehaviour of the directional reactance, X,, shown inFig. 15. It is seen tha t X m largely remains negative dur-ing the period when R and X re in zone and thus thea-b element counter is inhibited from increasing. Thisvariation in X m is caused by a combination of thechange in system frequency and the change in apparentreactance due to the power swing. Changing the delayperiod over which the voltage used to calculate X, isevaluated has no effect as shown in Fig. 15 for delaysof 1, 2 and 4 cycles. Thus, even without the fault detec-tion enabling of the relay trip flag, the relay remainsstable during this particular power swing.

-600 n n enable n0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1time, sFig. 13 Relay signals during an instability resulting rom 150ms ault atbusbar R (relay at S)

608oil-!---7

20015010050

.c o- 50-100-1 50 { U-200 1

720 76 0 aoo a4 0 WO 920 960time, msFig. 15 Directional reactance using various voltage delays o r an instabil-ity resulting from I5Oms ault busbar R (relay at S)

IEE Proc -Gener. Transm. Distrib., ol. 143, No . 1, Junuary 1996

7/28/2019 00488055

7/8

If the relay position is now changed to the lv side ofthe generator transformer, Fig. 16, then R, and X ,are consequently higher and the a-b element trip flag isset when R and X enter the characteristic. However,since a relay trip requires both the element trip flag andthe ENABLE flag to be set simultaneously, it is appar-ent from Fig. 17 , that no relay trip will occur since nofault disturbance is detected during the power swingperiod.

6 0 0 ~

-600.-In n enable

Fig. 16 Relay signals during an instability resultingfrom a 150m ault atbus bar R (rela y on lv side o generator trans former, meusured impedancereferred to hv side of generator transformer)

60c:$ 40cg 20?

0-20

20 10 0 10 20 30 40 50 60 70 80 90resistance, LI.Fig. 17 Locus of impedunce for t = 0.7 to 0.8s during an instabilityresulting from 150m fault at busbar R (relay on lv side of generatortransform er, measured imped ance referred to hv side o f generator trans-former)Figs. 18 and 19 show the effect of a 200ms fault atbusbar R with two machines connected and the relaypositioned at R. Due to the higher inertia of the twomachines, the power swing and pole slip period is farslower. This results in a period of over 0.2s when thea-b element trip flag is set. However, again due to thebehaviour of the ENABLE flag, no relay trip willoccur.

5 ConclusionsA new approach to blocking digital distance relay oper-ation during power swings has been proposed based ondetection of fault disturbances which are present duringgenuine faults but not during power swings. The relaydetects fault disturbances by the use of FIR filtersbased on the Hartley transform and calculated reac-tance values. As a result of the disturbance detection,the relay trip flag is enabled during the period whenrelay operation is likely. This leaves the relay inherentlyblocked for all events not associated with fault distur-bances such as power swings. The extra processingneeded for the fault disturbance detection is relativelymodest and is less than the processing needed for a sin-gle relay element. Results have shown that the faultIEE Proc.-Gener. Transm. Distrib., Vol. 143, No . I , January 1996

disturbance detection process does not lead toincreased relay operating times.

-600 n enable fl n0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1

Fi .18 Relay signals during an instability resulting from a 2 3 0 m a ul tat %usbarR with two m uchines connected at S (relay at S)time, s

-2 0 20 10 0 10 20 30 40 50 60 70 80resistance,aFig. 19 L o w of im edance for t = 0.7 to 0.8s during an instabilityresulting from 2 3 0 m s ~ u l t t busbar R with two machines connected at S(relay at S )

On simulated dynamic system tests, the relay isobserved to correctly block under power swing condi-tions which cause the measured impedance to passthrough the characteristic. For some power swing con-ditions the relay does not require power swing blockingsince the directional reactance, calculated by the relayto ensure correct operation during close-up faults,affords a degree of immunity. However, in the generalcase, power swing blocking is needed and the newapproach proved to be reliable. The relay performedcorrectly even under fast pole-slipping conditions wherethe generator was slipping every 200ms. A significantadvantage of this approach compared to conventionalpower swing blocking techniques is that the relay willcorrectly operate if a genuine fault occurs within theprotected zone during the power swing period. Further-more, the only setting associated with the fault distur-bance detection process is a relatively insensitivethreshold used to determine the presence of fault tran-sient effect reactance. Once this threshold has beenestablished, the power swing blocking feature of therelay is likely to be application independent and notrequire user settings.6 AcknowledgmentsThe authors would like to thank the University ofBath, Power and Energy Systems Group, for the provi-sion of facilities to conduct this research. The authorswould also like to extend their gratitude to ThomasHass, University of Aachen, Germany, who helpedwith the simulation results while spending an ERAS-MUS funded study visit to Bath in 1994.

25

7/28/2019 00488055

8/8

~

71

2

3

4

56

78

9

eferencesZHANG, Z.Z., and CHEN, D.: An adaptive approach in digitaldistance protection, IEEE Trans . , 1991, PWRD-6, (l), pp. 135-142GAO, Z. D., and WANG, G.B.: A new power swing block in dis-tance protection based on a microcomputer - principle and per-formance analysis. IEE Conf. Publ. 348, 1991, pp. 843-847MECHRAOU I, A., a nd THOMAS, D.W.P.: New blocking prin-ciple to overcome the deficiencies of the conventional powerswing blocking schemes for distance protection. Proceedings of29th Universities Power Engineering Conference, September 1994,(Galway), pp. 216-219MOORE, P.J ., and JOHNS, A.T.: A new method of powerswing blocking for digital distance protection. Proceedings ofInternational Power Engineering Conference, March 1993, (Sin-gapore), pp. 633-637MOORE, P.J., and JOHNS, A.T.: Adaptive digital distance pro-tection. IE E Conf. Publ. 302, 1989, pp. 187-191ZHOU, K. M. , MOORE, P.J., JOHNS, A.T., and XU, B.Y.:Accurate residual compensation for a digital distance relay. Pro-ceedings of 24th Universities Power Engineering Conference,1989, (Belfast), pp. 61-64BRACEWELL, R.N .: Discrete Hartley transform, J . Opt. Soc.Amer., 1983, 7 3 , (12), pp. 1832-1835HARTLEY, R.V.L.: A more symmetrical Fourier analysisapplied to transmission problems, Proc. IRE, 1942, 30, pp. 14 4150LONG-WEN, C . , and SHEN-WEN , L.: Systolic arrays for thediscrete Hartley transform, IEEE Trans. Signal Process., 1991,39, (11), pp. 2411-2418time, IE E Pvoc C, 1992, 139, (1)10 LOBOS, T Fast estimation of symmetrical components in real

8. I Proof of the general form of eqn. 7The form of this equation is the same as the denomina-tors of the right-hand sides of eqns. 1 and 2. It can eas-ily be proved by reference to a sinusoidal signal:12[n]=I , sin(2nfonT) where T =- (13)

f sand thus the time derivative is given byd( i n ) =i [n]=1,2~fo c o s ( 2 n f o n T ) (14)dnT

Hence, forming the expression for D,

D =z[n]i[n ]- z[n- k]i[n] (15)where k is an arbitrary integer number (k is 6 ineqn. 7), and substituting eqns. 13 and 14 into eqn. 15,givesD =I , sin(2~fonT)1,2nf0 os(2nfo(n- ) T ) (16)- , sin(2.irfo(n- )T)lm2nf0 os(2nfo(n - k)T)which rearranges to

(17)

D =2nfO1; s i n ( 2 ~ f o k ~ ) (18)

D =2nfo1k sin2(2nfon~)i n ( 2 n f o k ~ )+2nf& cos2(2xf012~)in(2nfokir)which is equivalent to

8.2 Parameters used in power swing studyLine:Voltage =400kVZ,, =0.023 +JO.3165;2 per kmZ, =0.388 +jl.Ol9 Q per kmLength =2 6 0 hEach unit 221400kV, 600MVALeakage reactance =0 . 1 2 3 2 ~ ~Single machine rating 600MVA, 22kV, 50Hz, 2 poleArmature resistance =0.0031 puDirect axis reactance =2.8puDirect axis transient reactance =0 . 3 6 2 5 ~ ~Direct axis sub-transient reactance =0 . 2 2 8 0 ~ ~Quadrature axis reactance =2 . 7 2 ~ ~Quadrature axis transient reactance = quadratureaxis sub-transient reactance =0 . 2 2 0 2 ~ ~Inertia constant =4.44s

Transformers:

Generating plant:

26 IE E ProcGener. Transm. Distsih., Vol. 143, No. I , January 1996