Phase selection based on superimposed components

D.W. P. Thomas
M.S.Jones
C. Christopoulos

Indexing terms: Phase selection, Superimposed voltuges, Distunce protection, Fault conditions

[Q]= modal current transformation matrix

Abstract: A method is described for faulted phase x = distance along a transmission line
selection for non-unit-distance protection using xf = distance of the fault location from busbar 1
the superimposed voltages and currents initiated
by faults. Even though superimposed voltages [Y,] = fault phase admittance matrix
and currents are comprised of many incident and [ Yf,] = fault modal admittance matrix
reflected waves it is shown for the first time that 2,= transmission line surge impedance of the ground
the criteria used for phase selection remain valid mode
throughout the fault period. The proposed
ZI = transmission line surge impedance of modes 1 and 2
method is hence very powerful. To demonstrate
the potential of the scheme, its response to a [Z,] = transmission line modal surge impedance matrix
number of fault conditions is presented. [z,b] = modal surge impedance of a busbar
I I [Zpb] = phase surge impedance of a busbar
[Zml], [Z,,] = modal surge impedance of busbars 1 and
2, respectively
List of Symbols
[Z,,], [Z,?] = phase surge impedance of busbars 1 and
AI, = superimposed or incremental modal current 2, respectively
AIp = superimposed or incremental phase current yo, yl, y2 == propagation constants of modes 0, 1 and 2,
[k,,] = modal voltage reflection matrix respect ivelly
R,, = phase a fault resistance to ground [p,~], [pm2] = modal reflection matrices for busbars 1
and 2, respectively
R u b = fault resistance between phase a and phase b
R b b = phase b fault resistance to ground 1 Introduction
SI,S2, S,, S4,S,, S,, S7 = seven phase selection param-
eters The speed and accuracy of digital distance relaying
[a = modal voltage transformation matrix schemes can be improved by accurate and fast phase
selection I[ 11 and this also allows single-pole tripping
[Ul = unit matrix schemes to be employed [2].
AV, = superimposed or incremental modal voltage Whein a fault occurs on a transmission line it causes a
AV, = superimposed or incremental phase voltage sudden change in the voltage at the fault location. This
Vo, V I , V, = modal voltages of modes ground (0), 1 change causes travelling waves, in voltage and current,
and 2, respectively to propagate away from the fault location and these
are superimposed on the power frequency signal. Each
V,, Vb, V, = phase voltages of phases a, b and c, particular type of fault produces its own characteristic
respectively
set of travelling waves. These superimposed signals
V', = voltage amplitude of a mode propagating in the provide the information necessary for extremely fast
positive x direction phase selection (within 1Oms). Following fault incep-
V,, = voltage amplitude of a mode propagating in the tion many travelling waves quickly develop due to
negative x direction reflections throughout the power network. Previous
[VA]= modal voltage amplitude of an incident wave work [2, 31 has shown that phase selection based on
superimposed components (obtained by subtracting the
[ V d ]= modal voltage amplitude of a reflected wave
steady-state power frequency from the transient signals
[V,,] = initial voltage amplitude of the modal waves 121) can be implemented during a period of the order of
propagating from a fault the round trip time between the fault location and the
0IEE, 1996 relaying point. It is shown in the present paper that
IEE Proceedings online no. 19960292 these criteria are valid throughout the whole period
Paper first received 17th July 1995 and in revised form 1st December 1995 that superimposed components can be obtained (typi-
Dr. D.W.P. Thomas and Prof.C. Christogoulos are with the Department cally a half-cycle data window). This then provides a
of Electrical & Electronic Engineering, University Park, Nottingham NG7 rapid, secure phase selection which is valid for all
2RD, UK transmission line lengths. The potential of the scheme
M.S. Jones is with GEC Alsthom T&D Protection and Control Ltd, St is demonstrated using an EMTP [4] simulation of an
Leonards Works, Stafford ST17 4LX, UK example power system.
IEE Pvoc.-Gener. Transm Distrib.. Vol 143, No. 3, May 1996 295
2 Fault transients be found by considering the Thevenin equivalent circuit
for superimposed quantities [lo] as given in Fig. 2. The
Faults create superimposed transients which propagate Thevenin source voltage at the fault (x = xf) is the
in the power network. From the theory of modal prop- reverse polarity prefault voltage at the fault location
agation [5, 61, the transient modal voltages AV,, and [vi,,,]. The amplitudes of the incident travelling-wave
currents AI, are derived from the superimposed or transients at the relay locations R1 and R2 can be
incremental phase voltages and currents AV, and AI, found by splitting this transmission system into two

,
using the [SI and [Q] voltage and current transforma- regions. Region 1 is to the left of the fault location (x <
tion matrices [6]. For three-phase fully transposed xf) and region 2 is to the right of the fault location (x >
transmission lines there are three modes of propaga- Xf).
tion; the ground mode which is highly dependent on
ground conductivity and two aerial modes which prop-
agate with negligible attenuation below 100kHz [6]. relov RI relav R 2
For fully transposed transmission lines the transforma-
tion matrices [SI and [Q] are identical and frequency @
@
*-
independent and this is assumed in the work using the busbar 81 - busbar 82
definitions given in [7]. fault
At a given frequency the general solution for the Fig. 1 Power .swiem with a,fault on a transmission line
Circuit schematic diagram
modal voltages at a point a distance x along a trans-
mission line is given by [8]
zm zrr
[V:l[e(-rX)l
[ w n ]= + [KI[e(S'x)l (1)
where [ V,'] and [V,-] are the modal wave amplitudes
propagating in the positive and negative x directions,
respectively and [e(yx)] is a diagonal matrix with ele- Zm2
ments exp(y,x), where yl = yo, yl or y2, which are the
propagation constants of the ground mode 0 and the
two aerial modes 1 and 2, respectively.
The modal current is given by
P]I = [ z m i - l ( [ v L , l [ e ( - ~ x ) l - [v;I[e(rx)l) (2)
where [Z,,] is the modal surge impedance matrix and is
a diagonal matrix. region 1 region 2
All faults on a three-phase transmission line can be
represented by a fault-phase conductance matrix [ Yf] x =o X Z X f X =L
[SI. When propagating modal transients of amplitude Fig.2 Poirei system with afault on a transmi,ssion line
Equivalent circuit for superimposed quantities
[V,i] are incident at the fault location they will be
reflected with an amplitude [Vn:] given by [8]
For clarity it is assumed that the relays are connected
[G,l= [ k v m I [ K l (3) close to the terminating busbars of the transmission
where line. This does not affect the generality of the solution.
The travelling-wave reflection coefficient at a busbar i s
[ k u m ]= -@[U3 + [ z m l [ ~ f m ] ) - l [ z m l [ y f , l (4) given by [8]
[Yfm] = [QI [YfI[SI (5) [Pmb] = ( [ z m b ] + [ Z m ] ) - l ( [ z m b ] - [zm]) (7)
and [U] is the unit matrix. where
The initial voltage amplitude of the transient travel- [zmb] = [ZpbI [QI (8)
ling wave [V,,] which propagates away from the fault
can also be given in terms of [k,,] and the modal volt- [Zph] = the busbar phase impedance matrix,
age [vf,] at the fault location under healthy conditions Note that for ehv systems the busbars should present
a balanced load so that [ Z J is a diagonal matrix and
[91 therefore [pmh] is also a diagonal matrix.
[Vim] = [kum1[7JfmI (6) At the busbars we can relate the modal propagating
The travelling wave [ V,,] therefore contains informa- waves in the positive and negative x directions by the
tion about the fault type and forms the basis of travel- busbar modal reflection coefficients. From eqns. 1 and
ling-wave phase selection schemes as proposed in [2, 31. 2 the modal travelling-wave voltages and currents
The problem is that on most ehv transmission lines along region 1 are given by
the period when the travelling-wave transient is given
by eqn. 6 is very short (typically less than a millisec- PK%l().I = [W([e(rx)l+ [Pmll[e(-rx)l) (9)
ond). Within a few milliseconds many travelling-wave
In1ml(x)l = [zmI-'iv;l([Pmll[e(-~x)l - [e(rx)l) (10)
transients are established and the superimposed tran-
sients approach a steady-state condition. For a travel- where [V,'] are the incident modal voltages at the bus-
ling-wave scheme to be reliable and applicable to bar B1 (x = 0) propagating in the negative x direction.
transmission lines of all lengths, it must also remain Similarly the modal transient voltages and currents
valid for the steady-state conditions following the fault. along region 2 are given by
Let us consider a fault on a transmission line of ) 1 [~,l([e(-rx)l
[ . c l ~ m 2 ( ~ := + [LJrnzl[e(r(x- 2L))I) (11)
length L with relays R1 and R2 located at its terminat-
ing busbars B1 and B2 as shown in Fig. 1. The solution [ ' u n Z ( ~ ) l= [zlml-l[v,l([e(-rx)l - [Prnzl[e(r(x - 2L))I)
for the steady-state of the superimposed transients can (12)
296 IEE Proc.-Ceneu. Tramn7. Distrib., Vol. 143, No 3, Muy 1996
where [ V i ] are the incident modal voltages at busbar steady-state ratio of the incident mode voltages given
B2 (x = L) propagating in the positive x direction by eqn. 13 is
Eqns. 9 - 12 can be solved to give the incident voltages
[ V,'] and [ V i ]in terms of the fault voltage [vfm].Enforc-
ing voltage continuity and Kirchhoffs current law and
observing that [ rf,] is the only nondiagonal matrix, we
obtain aftcr some manipulation
[VS= - ( k ( 7 X f ) l + [ P m l l [ e ( - ? X f ) l ) - l (13)
x I%[( -t[Zm][Yfm]) '[zm][Yfm]!uUfm] Therefore, the final steady-state ratio of the incident
aerial mode voltages I and 2 will also be three to one
and this can be used to phase select confidently. Similar
relationships can also be found for other faults.
Faults which involve two phases and the ground can
have a variety of configurations. If we consider a phase
a-to-phase 6-to-ground fault then, it has been found
-~../Ori - p.."p-/"("i ~ 2 L ) p Y 0 Z f - ploe-r"zi that if the fault impedance to ground is negligible, from
(I, =e ___- (15)
Y(1ri + o,Oe-/u(r;i -ah) + p,-fozi + plOe-Yozf eqn. 6
<: -Y l s ' i eY12:t -. p11 - Y1"f
U; - v i +U% - -
21
- p21eYl(rf-21J) e
li= c-71 U 1 + p21cy1(zf-2L) - + e-? l z t + plle-*/lzt (16) 1); 20
and the steady-state solution of eqn. 13 gives
yo, yi and y2 are the mode 0, I and 2 propagation con-
stants, respectively, plo, p I 1 and p12 are the busbar B1 U6 - U ! +vi a -
21
_-
-
reflection coefficients for modes 0, 1 and 2, respectively b 2 0
and p20, p21 and pZ2 are the busbar B2 reflection coeffi-
cients for modes 0, 1 and 2, respectively. The factor alb in eqn. 22 represents a phase shift in the
In this analysis, it is assumed that the transmission frequency domain which translates as a time delay in
lines are fully transposed and the loads balanced so the time domain. This apparent delay in the ground
that the aerial modes 1 and 2 will have the same mode can be found and corrected for by crosscorrelat-
parameters (propagation constant reflection coefficient ing the ground mode waveform with the aerial mode
and h). A similar expression to eqn. 13 can be derived aerial waveforms. A phase a-to-phase h-to-ground fault
for the incident voltage at R2 [ V i ]in region 2. can then be identified using the following criteria:
Eqns. 6 and 13 give the modal travelling-wave ampli-
_-210 - 211 + 212
5 2-
21
(23)
tudes which are incident at the relaying points. These t'O ZO
incident travelling-wave transients can be resolved at
the relay location from the superimposed or incremen- The criterion given by eqn. 23 will identify faults with
tal voltages and currents [AV,] and [ATm]by [2] resistances to ground of up to 150Cl. For greater resist-
ances to ground the faults look like phase-to-phase
[V"] = -[AV,]
[Zrnl[urLl
- - faults.
m
2 (17) Symnietiric faults can be identified by an absence of
To summarise, when a fault occurs the wave travelling ground mode voltage and the fact that at least two
towards the relay will have the form given by eqn. 6. phase-to-phase faults will be simultaneously indicated.
This will generate further waves due to reflections.
Within a few milliseconds steady-state will be described 4 Practical considerations
by eqn. 13. For travelling-wave phase selection to be
both reliable and applicable to transmission lines of For clarity, the phase selection procedure has been
any length, the criteria used must satisfy both eqn. 6 described in terms of modal components. However, for
and the form given by eqn. 13. computational simplicity, the above criteria can be
translated in terms of the phase voltage amplitudes.
3 Phase selection criteria The voltage amplitude of the incident travelling waves
can be given in phase quantities by transforming
As an example we will consider the phase-a-to-ground eqn. 17 into the phase domain.
fault. The fault conductance matrix of a phase a-to- The aerial phase voltages are given by
ground fault is =v; v, vo
- (24)

[Yf]=
0 0
0 o]
0 0
[ (18)
The phase selection criteria described so far assumes a
fully balanced power system. It has been found that a
threshold level of about 0.1 of the per unit voltage is
From eqn. 6 the initial incident modal transient will sufficient to compensate for nontransposed transmis-
have amplitudes given by sion lines and an expected signal noise level of about
5%. The complete phase selection procedure, in phase

[ ]
2zo 220 22"
-1 quantities, is given in Table 1 with the additional
3% 321 321 [us-]
'v,'nl = (12L + 421 + 220) zI zI zI
phase-to-phase-to-ground fault identification given in
Table 2. This procedure is identical to that described in
(19) [3] except for the identification of symmetric and
This implies that initially the incident mode 1 voltage phase-to-phase-to-ground faults.
amplitude will be three times that of the incident mode The current relaying signals must be lowpass filtered
2 amplitude for a phase a-to-ground fault. The final to approximate the bandwidth of the transient voltage
IEE Proc.-Genrr. Tran.sm. Distrih., Vol. 143, No. 3, May 1996 291
signals available from the voltage transformer. It has location which is near the 35GVA source. For the
been found that digitising signals to a 16-bit resolution results of Fig. 8 the fault was located 12km from the
does not degrade performance. relay location which was near the 10GVA source. The
threshold level for a significant wave amplitude was set
Table 1: Phase selection scheme (test must be true for at 0.1 of the per unit phase voltage (33kV).
half cycle 10 milliseconds)
200r 100 km 100 km
Parameter Definition of Selection Indicated fault
test for this
label the parameter type if test is true
parameter

SI y - v,. S,(6 phase c-to-ground

s2 v,.- va' S2( 6 phase b-to-ground
s
3 v,. - v,. Sa( 6 phase a-to-ground
s4 v,. S4( 6 phase a-to-phase b
s
5 S,( 6 phase a-to-phase c
S6 P S,( 6 phase b-to-phase c
s, 113 (V,+ V,+ V,) S,( 6 only symmetric three- time,s
phase Fig.4 S purarneters deduced by the relay using the incremental signals
created by a phase a-to-ground juult of resistance 150Q occurring at phase
d is the threshold level for the parameters and is normally set a voltage niarimunz
at 0.1 p.u. voltage Phase selection threshold level = 33kV
--- s,. - - - s2, - - - S?, - - - - sq, ~ s g , - - sg, - - -- s7

Table 2: Phase-to-phase-to-ground fault identification

lO0Ol
SdS, < 2ZqIZo phase a-to-phase b-to-ground
s,/s,< 22,lZo phase a-to-phase c-to-ground
SdS, < 2ZqIZo phase b-to-phase c-to-ground

5 Results

This phase selection routine has been tested on a vari-

ety of system configurations, different fault locations.
for a comprehensive range of initial fault voltage phase relay location
angles and for nontransposed transmission lines. -2000
Results are presented for the simulation of faults on t1me.s
the system which consists of a 200km, 400kV overhead Fig. 5 S parameters deduced by the relay usin the incremental signals
creuted hi, a phusr a-to-ground juult of negligibl resistance occurring at
nontransposed transmission line terminated by sources phase a voltage zero
of 35GVA and 10GVA short-circuit capacity. The sys- Phase selection threshold level = 33 kV
___ s,. - - - S?. - - - sg, - - - - sq, - sg, -- sg, - s7
- - -~
tem studied is shown as an insert in the results Figures.
The geometry and parameters of the overhead non-
transposed line is typical of those found in the UK and 2500 .-
are given in [9]. It was felt that the asymmetry pre-
sented by the nontransposed vertical construction of
the transmission lines would be a realistic test. The ,ooo/ relay location
sources were also typical of those found in the UK. j
' ,/-
The system was simulated using the EMTP power sys-
tems transients simulation program [4].

t1me.s
Fig. 6 S parameters deduced by the relay using the incremental signals
created hy a phase a-to-phase b fault of negligible resistance occurring at
phase a voltage maximum
Phase selection threshold level = 33kV
___ s,, - - - s7. --- s3,-----sq,-s5 --ss6 SI

Fig. 3 shows the amplitudes of the S parameters

(defined in Table 1) during a symmetric fault. Notice
time,s that there is a slight indication of a ground mode volt-
Fig.3 S paranzeters deduced by the relay using the incfemental signals age (&). This is due to the asymmetry of the nontrans-
created by a symmetric three-phase fault occurring at phase a voltage max-
imum posed transmission line which means that the ground
Phase selection threshold level = 33kV. The circuit configuration and fault
location is shown in insert mode as given by S, is not independent of the aerial
---s, ---s2 - - -s3 ----s 4. - sj, - - sg. - - -- s, modes of propagation. However, this distortion is not
significant and the symmetric fault can still be clearly
For Figs. 3-7 all the faults are located at the mid- identified as only S, remains below the threshold
point of the transmission line lOOkm from the relay (33kV).
298 IEE Puor -Gene?-.Transm. Distrib., Vol. 143,No. 3,May 1996
 .. quickly die away so the main component is the 50Hz
lOOkm lOOkm /’
supply frequency. The waveforms show a clear indica-
, ,‘, _ ~ tion of a phase a-to-ground fault (S3below the thresh-
,,,,‘ old level). These results clearly show the generality and
,,,
,‘
,
relay location r’ /- strength 0-f the phase selection criteria described here.

6 Conclusions
3
A method of phase selection has been described which
is based upon the superimposed voltages and currents

-1500
1 .................. initiated by a fault. It is shown that a phase selection
criterion can be derived which is consistent throughout
time,s
the period in which incremental signals are reliably
Fig.7 S parameters deduced by the relay using the incremental signals
created bv a phase a-to-phase h-to-ground fault with a fault resislance a-to- obtained. ‘The method is also a non-unit scheme requir-
ground of 150Q occurring at phase a voltage maxinmm ing information from just the relay location and is
Phase selection threshold level = 33kV
-_- s,,- -S2.---S3,----S4, sg,
-- sg, - - --
~- s7 therefore ideally suited to distance protection applica-
tions. It is assumed that the transmission lines are fully
transposed and that the loads are balanced. From sim-
ulations it is shown that phase selection criteria based
on these assumptions are also valid for ehv systems
with non transposed transmission lines. This phase
selection method is faster than the more traditional
phase current level method as it is valid over a broad
frequency band and not just the power frequency. The
proposed technique does not require a large transducer
bandwidth as it remains valid for all frequencies not
just the high frequency transients normally associated
...
,..:. ,,,..’
with travelling waves.
._.__. -.-..:..:_,,~
..;...............
-250
t1me.s
Fi .8 S parameters deduced by the relay .for U phase a-to-ground juult
7 Acknowledgments
l 2 f m from the relay location occurring at voltage zero and of resistance
ISOQ The authors wish to thank G. Weller and B. Caunce of
Phase selection threshold level = 33kV
---s, ---s2..--- s3,-. s4, s5 -- s6, -----s7 GEC Alsthom T&D Protection and Control Ltd. for
their help and guidance. They are also grateful for the
Fig. 4 shows the amplitudes of the S parameters for financial support of the Engineering and Physical Sci-
a phase a-to-ground fault of resistance 150Q which was ences Research Council and GEC Alsthom T&D Pro-
initiated at phase a voltage maximum. Fig. 5 shows the tection and Control Ltd.
amplitudes of the S parameters for a phase a-to-ground
fault of negligible resistance which is initiated at phase 8 References
a voltage zero. Figs. 4 and 5 demonstrate that the
phase a-to-ground faults can be identified within about 1 BERNARD, P., and BASTIDE, J.C.: ‘A prototype of multiproc-
a quarter of a power frequency cycle for all initial volt- essor based distance relay’, IEEE Trans., 1982, PAS-101, (2), pp.
491498
age amplitudes as within this time there is a significant 2 JOHNS, A.T., and EL-DIN MAHMOUD, M.M.: ‘New
ground mode S7 and the relevant S parameter stays approach to ultra-high-speed phase selection in single-pole
below the threshold level (S, in this example). Notice autoreclosure schemes’, IEE Proc. C, 1986, 133, (4), pp. 188-200
3 BOLLEN, M.H.J.: ‘Extensive testings for algorithms for travel-
that for the resistive fault, given in Fig. 4, the S param- ling-wave-based protection’. Proceedings of fourth international
eters are approaching the threshold level (33 kV) so that conference on Developments in power system protection, Edin-
for higher fault resistances a definite decision is diffi- burgh, 1989, pp. 135-139
cult to obtain. 4 SCOTT MEYER, W.: ‘EMTP rule book’, Bonneville Power
Admin., Portland, Oregan, USA
Fig. 6 shows the S parameters during a phase a - t o 4 5 WEDEPOHL, L.M.: ‘Application of matrix methods to the solu-
fault. This shows a clear indication of a phase a - t o 4 tion of travelling wave phenomena in polyphase systems’, Proc.
fault as only the S, parameter remains below the ZEE, 1963, 110, pp. 2200-2212
6 GALLOWAY, R.H., SHARROCKS, W.B., and WEDEPOHL,
threshold level. Fig. 7 shows the S parameters during a L.M.: ‘Calculation of electrical parameters for short and long
phase a-to-phase b-to-ground fault with a fault resist- polyphase transmission lines’, Proc. IEE, 1964, 111, (12), pp,
ance to ground of 150R. In Fig. 7 the ratio S4/S7is 2051--2059
7 JOHNS, A.T., and WALKER, E.P.: ‘Co-operative research into
smaller than the threshold 22,/Z0 (0.9 for this trans- the engineering and design of a digital directional comparison
mission line) thus indicating a ground return in a phase scheme’, IEE Proc. C, 1988, 135,(4), pp. 335-368
a-to-phase b fault. 8 ANCELL, G.B., and PAHALAWATHTHA, N.C.: ‘Effects of
frequency dependence and line parameters on single ended travel-
Fig. 8 shows the deduced S parameters for a phase a- ling wave based fault location schemes’, I E E Proc. C, 1992, 139,
to-ground fault, of fault resistance 1 5 0 Q occurring at (4), pp. 332-342
voltage zero, which is 12km from the relaying point. 9 THOMAS, D.W.P.: ‘Protection of major transmission lines using
The relay location has also been moved to the line end travelling-waves’. PhD thesis, University of Nottingham, 1990
10 DANIEL, J.S.: AGGRAWAL, R.K., and JOHNS, A.T.: ‘Three
next to the 1OGVA source. For such a close fault to terminal line protection based on a superimposed component
the relay the travelling-wave effects are small and impedance relay’, IEE Proc. C, 1993. 140, (6), pp. 447-454

IEE Proc.-Gene?. Transm. Di.strih., Vol. 143, No. 3, Muy 1596 299