3252 IEEE Transactions on Power Apparatus and Systems, Vol. PAS-101, No.

9 September 1982

AN ACCURATE COMPUTER METHOD FOR OBTAINING BOUNDARY

CONDITIONS IN FAULTED POWER SYSTEMS
E. B. M4akram G. G. Koerber K. C. Kruempel
Siemans-Allis Inc. Senior Member, IEEE Member, IEEE
Raleigh, NC Electrical Engineering Department
Iowa State University
Ames, Iowa
Abstract. This paper presents an accurate tech- The foregoing comments apply primarily to the
nique for solving-the three-phase power system line. The boundary conditions are of equal importance
equations under fadlted conditions. The technique and we propose two amendments to what we conceive to
makes use of frequency domain simulation to provide a be current standard practice. These are described
means of representing the transmission lines with fre- below.
quency dependent parameters. Faults can be simulated
The usual practice is to treat the sending end
at aanloaionaogteln
any location along the line and n tine
n any al
ieoof fault
boundary condition as a source in series with a fixed
surge impedance. Since the generator inductances vary
The technique enables a more accurate represen- with-rotor angle, this is obviously simplistic.
tation of synchronous machines to be combined with an Accordingly, we have developed a sequence of trans-
untransposed line model having distributed frequency formations that couple the generator equations with
dependent parameters. A typical power system example the line equations. The only ad-hoc assumption in
along with the modeling details illustrate the practi- this process is that the rotor speed will remain sub-
cality and accuracy of the method. stantially constant during the first few cycles after
fault inception [6],
Introduction
Our second proposal is; we replace the usual zero
This paper is concerned with the simulation of fault impedance or the arc resistance with a fault
fault induced transients on a three-phase power trans- impedance that is presumably present as an early fault
mission line.. For this particular application the condition.
critical features of the simulation center around the
proper characterization of the line parameters and the We present a simulation study in which the conse-
development of physically realistic boundary condi- quences of the proposals suggested above are compared
tions. with the results obtained by using standard boundary
conditions. The line per se is untransposed and has
It is now widely recognized that accurate frequency dependent parameters. This-simulation study
transient analysis requires the inclusion of frequency does not include any treatment of the nonlinearities
dependent line parameters and assymetric magnetic of the arc resistance or transformer and generator
coupling in untransposed lines. Many attempts have saturation.
been made in the past to approximate the frequency
dependence of the line parameters by the utilization The Transmission Line Equations
ot exponential weighting functions for direct time
domain solution tl1r31. The computational effort in A general three phase transmission line consists
the weighting process is considerable and the accuracy of a set of three phase conductors and a set of ground
is not fully proven yet £13. wires. The ground and bundle wires can be eliminated
algebraically [7] to reduce the transformed line
We believe that the frequency domain treatment, equations to
as exemplified in reference [4] offers the flexibility
as a basis for simulation studies. The major features dx V(x,s) R I(x,s) + sL I(x,s) = Z(s) I(x,s)
of this approach are: (1) Frequency dependent dis- x
tributed line parameters can be readily incoperated d I (2)
into the analysis, and (2) except for a contingency dx - ' s V Y
noted later, a modal transformation,[5] can be used to
uncouple the line equations. This leads to a signifi- where
cant reduction in programming effort, and reduction in s = a + jw
computational time.
and V represents a set of three phase voltages, I
When the lines are untransposed, the problem of represents a set of three phase currents, and x is the
decoupling the line equations is algebraically distance along the line. The elements of the 3x3
intractable for a general complex frequency. This impedance and admittance matrices can be calculated
makes it possible to write an analytical expression at any frequency by techniques presented in reference
for the impulsive response [3]. For any particular
frequency, however, all of the critical matrices can
E7].
be diagnalized. If the steady state response to a Differentiation of (1,2) makes it possible to
sinusoidal stimulus is calculated for a range of fre- eliminate either the voltage or current vectors. The
quency components consequent to a step-sinusoidal results are
input [5].

dx V(x,s) = A V(x,s) ( 3)

82 WM117-0 A paper recommended and approved by the and

IEEE Power System Engineering Commnittee of the IEEE
Po er3ngineering Society for- presentation at the IEEE d2
PES 1982 Winter Meeting, New York, New York, January 31- ~ 2 I(x,s) =A I(x,s) 4
February 5, 1982. Manuscript submitted August 13, 1981; dx
made available for printing Decemnber 16., 1981.
OO18-95110/82/O9OO-3252$OO.75 © 1982 IEEE
 3253

where r
.rl ---t- r2
A = Z(s) Y(s). R

Att = Y(s) Z(s) I

Following the method that is discussed in reference v
[5], equations (1-4) are transformed to the modal set x
as
vv
v+ S- v1
+ -1f
I= (5) Figure 2. Fault simulation

Where S is the matrix of the eigenvectors of A and Q v (SD S + (Z-

t -Si -c- )Vr +(Z S Qi 2 rY
is the matrix of the eigenvectors of A . This pro-
duces the uncoupled sets -sl Q ri+Q(ZO) D S 1 V
2 +sl -1lQ
+
V (x,s) = S 1A S V(x,s) (6)
Q-osli-9
Similarly, the receiving-end equations are
dx
~~~~~~~~~V
=(S-D QD s2
S Vr + (Z o Q)
and
and -s2 --c2- Ir2
d2I+(x
dx
S) =9-1 At I+(x,s) (7) Is2 -(Q D2Qc+ (Q D O V
(10)

The solution of (6,7) can be used for finding the V I V

three phase voltages and currents at any distance x -sl' -l' Vs2' Is2'.-r -l' -r2
on the line and it is given in reference [5] with the boundary conditions at the sending end, the receiving
distance x measured from the receiving end as end and at the fault location.

-x - cx -
-V V
Y-s2D
D
-o - -sx -
-I
-s2
The load and the fault models are discussed in
reference [5].

x - -cx 2 1s2 + 2 Ysx

2 Z Vs2 (8) Synchronous Machine Equations in the
Frequency Domain
where: Dcx and Dsx are diagonal matrices The machine equations are easy to formulate in
the direct phase components (a-b-c), but the co-
D cx = cosh (y efficients are functions of angular displacement, and
i,ix) and D = sinh (y. .x)
sx 1,1 the equations per se are mathematically refractory.
i,i isi As.a consequence, numerous transformations have been
developed to produce a more tractable set of equations.
V and Ix are 3xl voltage and current vector at dis- The most important of these is the Park transformation
tance x from the receiving end [6] in which the (0-d-q) coordinates are a set of
V and I are 3xl voltage and current vector at equations with constant coefficients. A second trans-
-s2 -s2 formation called the (0-1-2) transformation (symmetri-
x =o (receiving end) cal components) is also available to simplify the
-1 -1 analysis of unbalanced three-phase circuits [73. This
= r transformation does not pertain to the machine per se,
but it can be used in.conjunction with the (0-d-q)
transformation to solve machine problems.
r 2 = s -1 A S =Q -I At Q
The speed of the synchronous machine is assumed
Fault Transient Equations to be constant in the first few cycles. This assump-
tion enables the machine equations to be transformed
If vr irl' and ir2 are voltage anc current vec- into the frequency domain and to be combined with the
tors at the fault location as shown in figure (1,2), transmission line equations.
then the sending end voltage anc current (v s'si
s s) ~~~~Machine-Equations in the'Dir~ct
were obtained in reference [5]as and Quadrature Axis'(d-q)
Vsl Vr s2
'rl Vr Vs2 The modified Park's transformation is used to
rl gr2 transform the machine equations from phase components
- _ / _ ~~~~~~~~~(a-b-c)to (0-d-q) components as shown in figure 3.
r The details of.formulating the.nachine equations inWto
s1 s xl _>< x2 - ?- S2 (O-d-q) are given in reference F6). A method for
1 ~~~~~~~~~~~~~~modelling
the synchronous machine via the elimination
______________ £ ______ of the *rotor circuits is discussed in reference ES,9].
The final results of voltage equations at the machine
IT'gure 1. Single line diagram of faulted trans- terminals are
.mission systems
3254

d axis a axis Then,

bi Odq = Z1Sy_m Z12tr Z22

-
21 (Odq
q axisMachine Equations a Components (0-1-2)
Z (s) Z -1 Z s

b axis The transnission line equations could be given in

i t: > (S(0-1-2) components. To combine such transmission line
c W .-dq equations with the machine equations, it is necessary
lb ' ~~~~totransform both (0-1-2) and (0-d-q) components into
c axis antother form which is transformable to either (0-1-2)
or (O-d-q) components. This intermediate transforma-
Figure 3. model of synchronous machine axi
Full0~~~~~~~~~ tion is known as the forward and backward (0-f-b)
anote for whc io) tasomal toe hr 0
transformation matrix [8] and can be explained as 2
e t mv0=-(rip+c
n follows:

ad Vd 0-(r id + O oqq
Lq i +io QQ~ ~tanfomaio Iarx[]adcn eepanda
t KMQ iQ

ji f-i +i (13)
d d a
o q q o Q Q
(when the q-axis lags the d-axis as Figure 4).
-wKN~~~~~~L+Li +KM i)
v =-(r i - w L i d w KMF ~ ~ ~ ~ j - +i
i
(3
f Ib 1

o u%iD+ q iq + Q Q if I

VF rF iF + KMFid+LFiF+ D

0 = -(rD
vi-(r w-LrD +0 R wiFD +OL i(~
iDi + F%
-
D)

By transforming the above equations into the frequency Figure 4. Forward-backward components
or s-domain with z,ero initial stored energy, they can
r matrix form
be writtenv in the In order to make theC KfM b
transformation power invariant
-(r
0 ~ ~ ~ ~ ~ (T~ = [T ~ ] ),0 the transformation matrix that trans-
p vO(80 I7r,+8LO) O O 9 O O1KIl(8 forms voltage or current for (d-q) axis to (f-b)
q0 oLd -(ra+sLq) aMd w F KQ

0 sK ° (FF+sLF) SHR ° 0 f b

01 0 -sK)0 0 D -sMD -(r+sl) d I O

0 0 -sMQI 0 0
°
(rQ+sLQ)J frIQs | O
1//2 /

which can be partitioned to produce i/ th1 i/ur (14)

V Odjq(SYj 11 121 IodqET inverse transformation matrix transforms the volt-

m s ---
-- - (0-f-b) components to (f-d-q)
from
age or current
d_ a d 0 q Z IFDQ ) components as

Elimination of Rotor Variables

v0m = To
-Ofb(-o v(f
-Odq (15)

The modified Park's Transformation P in reference

Shinccan thev
the
FDI si
exctitationedz21vol
superimposed value
IFD (s
isrcnstnt,the
of this voltage is zero. The
2JZ F [6] which is also power invariant is used to trans-
form the voltage or current from (0-d-q) components
superimposed voltage V(s), V(s), and V(s) is con- to phase components. Then equation (15) can be
sequently written as
| 11 OdqTd)
Idq(s) -f12

- Q(s j
 3255

T Pv or
-Ofb -o P -abc

-o P -012) x02 YOfb) = t012

and
-2p -012 = 2(16) (1)!fbA
I'Ofb(s) = Tt 10'12.( )
where
s represents the s-domain and sA represents the
T2p -o P .012
-1 P h12Where
shifted s-domain. This means that there. is no shift
0 1 2 in zero sequence, the positive sequence voltage is
shifted jw and the negative sequence voltage is
by
shifted by -jiw. Tt is a transformation matrix that
- = fe ° transforms the voltage or current from (0-f-b) com-
K[
b Lo Oe |
ponents to (0-1-2) components. The superimposed
voltage or current can be obtained as

0= w 0 t + 6 + /2, w 0 =377 rad/sec, (s) Z -s0dq(s)

(s)
-90dq' s) dq 1

6= rotor angle 0 -Qfb -Odq o gQf(

and
T T *V (-s') (s) * T
F 1
1 11 ~~~~~~~~~~~~~~~~~-o
-g0dq~=
Z

-t

* T IA (s')
-012 h a :21|
~~~012 -90~~~~~~12
t a a22 gO12( ) ~~~ ~~~~~~~~~~~~~-12
-Odq( -12 .O2(
= T * - T I- (s
'h= Vr for the power invariant transformatLon -2 -12
-Od
From equation (16), the superimposed voltages at the
machine terminals are =z (s) * I (s9) (22)
_e
-i 0O12 0O12
vf =vl e
A

where
vY e--jw 0 t * e j6 (17)
v =v ewot
1 6 (18) ~~~~~~~~~~~~~~~~~~~Z
-gO12 =T-12 z_Odq (s) -12

where 6= 6 ,T/2
+ 0VO(s) ] [16(s)

Equations (17) and (18) are in the time-domain. They Vg0(s) = (s+jw0) , I (s)
9012 g012 ijwj0
can be transformed into s-domain by means f(t) -V(sj
L2 o
)
2
-

exp(jw t) < = > F(s - jw0), then

andT a12 = To -t
Vf(s) = V'(s + jiw) exp(-j6 )

[1 0 0 1 0 0 1
Vs(s) = V5(s - jlw) exp(j6 ) (19)
-0o 0 exp(-j 6) 0
Similarly, V jiv<1 - I 0 exp(j6')j
I'(s) = I(s + jw ) exp(-j6 )
2- 0 0 1
I'(S) = I'(s - exp(j6')
w)jw Ic(s) I~~~~~~~~~~~~~~ - (20) - ~~~~~~~exp
1 exp(-j&-)
(-j
exp(j6)
e ( )A
Equation (19) can be written in a matrix form as
Y
O j exp(-j6 ) -j exp(j) J
Equation (22) gives the terminal machine voltage in
[Vs) [1 0 sequencee cmponents and in the shifted s-domain (s ).
combine the machine equations with
1LV s-ioO)
l -z 'aO _d O V(+j In order to
Vf(S) _ e . ,^-
e~~~~~~~hangedintoto theget shifted
the line equations,

s-domain and into sequencee

the line equations have to be

V9(s) °O J V2(s- j@O) ~components the solutions of the voltage

at the fault location and at t.he'sending
andend.
bV.(es-j (21) current
This can be done by starting from the decoupled trans-
mission line equations ref,erence [51 by using the
 3256

transformation matrix A to get the sequence com- Send. End. Cur. CAS1O
ponents in the s-domain, then the positive sequence is Send. End. Curr. CBS A
shifted by jw and the negative sequence by -jw to Send. End. Curr. CCS +
obtain the line equation in the s'-domain. The same o
steps can be followed to transform the load equations
and the transformer equations into sequence components o
and s'-domain. o_
The total solution for voltages and currents can CD
be obtained by the following steps:
C4J
The shift in the frequency domain must be H o
changed back. The positive sequence component
is shifted by -ju0 and the negative sequence | 4
0 (L~~~~~~~~~~~~~~~~~~~~~~~)
component is shifted by jw for both voltage CfI
and current components.
The voltages and currents at the fault location ___________
and at the sending-end are transformed from '0.00 0.50 1.00 1.50 2.00 2.50 3.00
(0-1-2) components to (a-b-c) components by time in sec. (x10-2)
using the transformation matrix A-012' current for single-line-to
Fig. 5b. Sending-end
ground fault on phase a.
The above voltages and currents are transformed
into time domain by using the fast-Fourier Simple machine mor1l
transform (FFT algorithmr IT ).Therefore, the Accurate machine model
total solution for voltages and currents can
be obtained by adding the steady-state solution
to the transient solution at the point of

The method of computation of a transient solution in

the time-domain is described in appendix C. CO

0 C:
The example in appendix B is used and the solution >
for the three phase voltages and currents are obtained. v , iN
Methods of finding the boundary condition at the fault D_
location are described in appendix A. The waveforms I.iI
for the three phase sending-end voltages and currents
are obtained for all types of faults. Some of these
waveforms are shown in figures (5a, 5b).
Comparisons are also made between the simple _ _. _,_,_._ _
machine model and the accurate machine model at 0.0 0.5 1.0 1.5 2. 2.5 3.0
different fault locatiQns on the transmission line as time in sec (xlO-5
shown in figures 6 and 7.
Fig. 6a. Sending-end voltage of phase a for three-
phase fault at the middle of the line in case
Send. End. Vge VAS 10@ of the simple and accurate machine model.
Send. End. Vge VBS1 A I_
Send. End. Vge VCS1 +
! ~~~~~Simple
S machine model
time Accurate machin. (xlO-motimedin

00
4-' CD

Oo
4JJ
0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
~
O ~ ~ ~ ~ ~ ~ ~ ~ ~ H

r-4 ~ ~ ron C),l onpaeaC) " _ , _ ___tt _

V)~ ~ ~ i (x0)a.o hesmleadacuaeCahnemdl
 3257
Acknowledgements
The first author would like to thank the Power
Affiliates of Iowa State University for their financial
Simple machine model support. The authors wish to acknowledge the Electri-
*-** Accurate machine model cal Engineering Department of Iowa State University
for the support of this project.
.~ . >References
1. IEEE Tutorial course, Digital Simulation of
Electrical Transient Pheonomena, IEEE Publication
no. 81 EH0173-5-PWR.
a s' 2. D. P. Carrol and F. Nozari, "An Efficient Computer
X -It Method for Simulating Transients on Transmission
lo Line with Frequency Dependent Parameters"', IEEE
/ Trans.on PAS, Vol. PAS 94, No. 4 (July/August
,, o I>, - % , 1 1975),pp.1167-1176.
I o - - 3. R.G. Bergmann and J. M. Pensioen, "Calculation
- W \ / \ v of Electrical Transients in Power Systems." IEE,
126 No. 8 (August 1979), pp. 764-770.
o - 4. M. Wedepohl, "Application of Matrix Methods to
the Solution of Traveling-Wave Phenomena in
X__ Polyphase Systems." IEE, 110, No. 12 (December
0.0 0.5 1.0 1.5 25
2.0 2. 3.0 | 1963), pp. 1200-121 Kruempel, and G. G. Koerber,
time in sec. (x ) 5. DigitalSimulation of Faulted E.H.V.Kor
Lines," ICEE, April 13, 1981.
Fig. 7a. Sending-end voltage of phase a for-three- 6. P. M. Anderson and A. A. Fouad, Power'System
phase fault at the sending end in case of Control and Stability. Ames, Iowa: Iowa State
the simple and accurate machine model. University Presst,' 1-977.
7. P. M. Anderson, Analysis-of'Faulted'Power Systems.
o Simple machine model Ames, Iowa: Iowa State UniversityPress,76 7.
CN ..... Accurate tachine model 8. Y. H. Ku, Electric Energy Conversion. New York:
The Ronald Press, 1959.
o 9. M. M. T. El-Kateb and A. T. Johns, "Frequency Do-
co main Digital Simulation of Synchronous Generators
Operating Under Faulted Conditions?" proceeding
/EE Vol. 124, No. 3 March 1977, pp. 223-229.
10. Westinghouse Electric Corporation. 'Electrical
S ; \s /#J4>v\ /fI Transmission and Distribution Book. East
Eittsburg, Penn.: Westinghouse, 1950.
4 c; \\x ( \ f 11. A. V. Openheim and R. W. Schafer, Digital Signal
Processing. Englewood Cliffs, New
Jersey:
C)
'\ ,.1 Prentice-Hall, Inc., 1975.
12. Sylvia-J. Day, N. Mullineux and J. R. Reed,
0 \ - 8/ \ / "Developments in Obtaining Transient Response
c; \ y \L Using Fourier Transforms, Part II: Use the
Modified Fourier Transform." J. EE'Educ.
o _________________________________ 13. 4(1966): 31-40.
. _._ -i 13. J. P. Bickford, N. Mullineux, and J. R. Reed.
0.0 0.5 1.0 1.5 2.0 2.5 3.0 "Computation of Power System Transients."' IEE
time in sec. (x102) Institution' of Electrical 'Engineers. London,
England, 1976. TEE Monograph series 18.
Fig. 7b. Sending-end current of phase a for three- 14. D. J. Wilcox, "Numerical Laplace Transformation
phase fault at the sending end in case of and Inversion." J. EE Educ.15 (1978 ): 247-265.
the simple and accurate machine model. aP LLUJLa A

Conclusions Finding the voltage at fault location (V ) for all

r
Assymetric coupiing between untransposed lines, types of fault:
frequency dependent parameters, fault impedance, and-
boundary conditions at the generator all have impor- Three-Phase-to-Ground Fault
tant effect upon the information that emerges from a
simulation study. In particular, the effect of the In this type of fault, the voltage at the fault
way the source is included in the analysis is shown location for the three-phase in the s-domain can be
in figures (6, 7) where simulations -based upon the written as
conventional sending end boundary conditions are
coupled wlth those in which the generator and line V (Rf+sLf) I + Vf
equations are coupled. When the fault is at the (A.1)
center of the line, the results differ by 10 - 25%._
This differences increases to 30 - 35% when the fault - Zf (Irl±Ir2) + f
is at the sending end bus. By substituting I 1and I 2from Eqs. (10) and (11) in
Extension of the method to accouJnt directly for rernctJ ri 2
the effect of the nonlinearities is not obviously Vr = Zf -2TV-4TV]+V
simplistic. However, an approximate treatment may be -1 -1 -1
possible. A final stage in this study would be a = [U+Zf(T2 T1+T4 T3)] Vf (A.2)
quanititative evaluation of the solution time and a -
comparsion with field test resultp.. = [U+ZfTI Vf
3258

aria - ra *-c
V i r2 Assume FH H3 H3
a~~~~~~~

Tf|r =-|H2 H22 H23

Fig . A. 1. Fault-transient3model for three-phase-tVL

> -b S -f + if (A.3)

ifbaif =0 b c

Single-Line-to-bround Faul

boundargrounditfault.
Rf (A.2+
-1
if= irla +12
Rfa From
ra Eq. + ) If A

v ~ vr Line-to-Lne Fault
ba fb f\ 2 22 2
If = r2 V V fa H+( + Lf Ifa
By~
wriin the equation in.s-omi acorin to. the
otinaa
rla ra -[T + r2a (A-2) to Vfe Zfltage
+ at tAhe) fa
ra~~~

c~~~f ifb 0 f
-r 21
fac
i +i H
(A.
4)~~~~ =-
Vraf31=~-i
a- fara
-~ -r
r2
By s-Folving. (A.4) LHf 3112
and + H +Z 3f 3aO
(A.5),toanb L a
obtained.
fb=o fc cn ls b otibe
r~
ra~~~r
fr2bFg ()a
Fig. A.F. Lf for
Fault-traas6Vrt
model thngeobaee-to-d a f
Vb~~~~~~~~~~r f, f
inl12invb Line-to- ieFault m
fa~0
1- f~f2b
bf the fault occurs on e If te faltt occur beines b
v1 i2
[ =00;1 H12 H130 0 ° 1

T s eindasbfor n tisvre
-r
ih rqec.
I
By solving (A.4)
n facneotie
and~~~ -A5voobanIa
fro th prealtVconition
~~~~~~~~~~~~~~~~~fa
T isdefied
beoreand
s t i vared ith reqen Fig.n Va3 Faulte modelne fror line-toliefault.cnito

fT21r T=22 (~~~~~a.3)

bequwrittn
can insdmagiuebtwinhas
prfutvlabovtheboaundar
Acorin
tothe
oitolrt
lcndtion, Eqx.
a From Eqs. (A.4) and (A.6V
H H cn (Atan7)
]

V H H I~~
vrb Line-to-Line'Faul3t 32 f
 3259

Then diagram of the system to be studied is given in figure

Vr -V rc = (H12 - H12 -H1333 + H13) If
fb (A.8)
B.l. The system data are taken from references [6]
rb 23 22 32 and 1J0).

tained as g~ ~ e~nerbator s2 SI
(V rb-Vrc) can also be obtained from Fig. A.3 as S.E. R.E.
Vrb Vrc (Vfb
fb (A.9) fc) f
By solving Eqs. (A.8) and (A.9), then Ifb can be ob- transmission line

tained as I fb
fc
I fb = (H23 -H22fb-H33+H32-
0f tasone
f) (A.10) transformer H load

By knowing Vfb and Vfc from the prefault condition,

then from Eqs. (A.7) and (A.10), Vr can be obtained.
Double-Line-to-Ground Fault Figure B.1. An illustrative transmission system
If the fault occurs between lines b and c, then Transmission Line:
i = In this example, a three-phase transmission line
fa 0 with flat configuration and ground wire s is used as
According to the above boundary condition, Eq. (A.3) shown in Figure B.2. The distances between conductors
can be written as are shown in that figure in ft.
V ra H H 12 H - 0 Conductors are AL, ACSR 26/7, KCmi/Al 636

H H _I]GMR = .03 5 ft.

r11 13 (A1)Line
r 11 2 1(A.ll) length
Voltage
= 200
= 220 KV
miles

Vrb
rI
= H21122 H123 Ifb Conductor diameter = .99" = .0825 ft.
Conductor resistance at 50°c = .1618 ohm

Lvrci L131 32 33 | fcJ Ground wire: steel, diameter = .001 ft., resis-
tance =4 ohm/mi., and GMR = .001 ft.
Vra GROUND WIRES

a ¶ ifa=0 Vfb

R~~J7fb 'f4 f/2R 20' 20'~.~ 5

b W,
Vrc 10'
~~
c ~ ~ ~ ~ ~ ~~r

~~~R'f f4
fb

c R /=Rf +-
55

Ll
Lf Lf*
L2 fx L z/////////////
/ =

S
I Figure B.2. Transmission line configuration
Vfb \-Jfc Synchronous generator:
The following data are from reference 761 which
is based on a 15 KV, 160 MVA system
Rated MVA = 160 MVA, .85 p.f.
- - Rated KV = 15 KV
r = .000742 pu T = .023 sec x = 1.64 pu
Fig. A.4. Fault model for double-line-to-ground f d q
fault. Xf = 1.651 pu Tdd = .98 sec r = .0011 pu
a
V =Z I +V xdd = .185 pu Tddo = 033 sec x k =Qd =q .15 pu
rb f fb fb
xd = .245 pu Td = 5.9 sec r = .016pu
Vrrc Zff Iffc + Vffc (A.12) ddo 2
Xd = 1.7 pu T' = .023 sec x2 = .115 pu
By solving Eqs. (A.11) and (A.12), then Ifb and Ifc x = .185 pu T' = .51 sec x = .1 pu
can be obtained as q q 0
-(Z tV b+Vf ) xq = .38 pu T' =.076 sec T.' .54 sec
ft fb fc q qo qo
1fb Zft (Zf+22) (A13) r= .054 pu x = 1.526 pu r 0131 pu
xD = 1.605 pu
and
-1 xAD + K
MI = K
MF MR
=
=1.7 1.5 = 1.55 pu

~ ~ ~ 20
-

23
b f(1
fc H23 f 22 fb fb xAQ = K MQ = 1.64 = .15 = 1.49 pu

MVA Load bus voltage = 20 KV

where Transformer: Load:
ZfZft =-1 (ZZf+33)
+H Three-phase Load Power = 160 MW, unity
H 15/220 KV power factor
By substituting tf b and Tfc from Eqs. (A.13) and (A.14) xl = x2 = x =.1p
into Eq. (A.ll), Vr can be obtained. 1 - 2 = r 01 pu
Appendix B: Numerical Example 1° 05p
A fault study of the general transmission system Fal meac: r=1 h Lf=.1nH
on the digital computer has been carried out using Appendix C: Method of Computation
both the simple and the full model of the synchronous The time-domain solution is obtained from the
generator. The system parameters have been carefully frequency-domain solution by the inverse Laplace
selected to be of most practical value. The one-line integral
 3260

f(t) - f F(a + jw) exp(at) exp(jwt) dw Numerical integration:

The numerical evaluation of the integral of equa-
(C.1)
where t > 0 tion (C.9) is based on discrete samples of F(a + jt)
taken at points lying along the path of integration. 3Aw
and f(t) =0 for t < 0. .

The integral in equation (C.1) cannot be evaluated Let these samples be taken at a i 22- -i 2 , etc.
analytically and it is necessary to evaluate it by Application of the midpoint rule of numerical integra-
numerical methods. In order to carry out the numeri- tion then given the following expression for the value
cal integration, it is necessary to truncate the finite of f (t) at a selected time instant t
range of the integral to some finite value, say (-Q,Q). k.
N 2i-l
This introduces a truncation error which, being multi-
plied by exp(at), increases rapidly with (at). f (tk) = exp(at) Real z F(a + i 2i A) a
To examine the nature of the truncation error, i=l
the value of f(t) with the integration range truncated 2i-l 2i-l
is given as - Aw x exp[(j A@) At 0k].
f(t) - ex2(at) f F(a + jw) exp(jwt) dtw where

or
or (0.1) a= N ha ((~~~~~~~~~~~~~~~~~~~~~~2i-1
F ( 1 6'NW
sin k2
f(t) = exp(at)
exp(jwt)
J F(a + jto) 4) (jto)
d2 (0.2)J Ow)aj y2i1-
(C.2)
Aw - 2i-
2 2 IN
where 1 , Ito c Q Proceeding with equation C.10

0 , t> Si f a(tk) exp(at) RealfJ( E F.(w) e( iAw /ttk))

The complete Fourier transform 4(jw) is given in refer-
ence [11-14] as AW *At k

¢O(t) = 2 fQ exp(jtt) dw Tart (C.3) e

One interpretation of this truncation is that the (C..11)

function f(t) is scanned by the passage of the and
Dirichlet Kernel sin(QT/7TT) over it [21. The periodic 2i-1 2i-l
Fi(w) = F(a + j Aw) a 2 A Au
nature of this function gives rise to Gibb's oscilla-
tions -4 ].
Since the period of oscillations of ¢(t) is 27r/Q Since N discrete samples in the frequency domain
as shown in figure C.1, a better representation of f(t), result in N discrete values in the time domain,
say f (t), is achieved by averaging over this period 2 2
in the following way. At ~ A= N
a(t) = f
TQ f(T) d-
] (04) and equation C.12 becomes

exp(at) 2 F(a+j) f (t) = Real

exp(j wT) dt (C.5)

Evaluation of the inner integral (as in reference [12]
gives exp (-i- At k
(t) = exp(at) fQ F(a+j2) (t) exp(jut) dt
ff7 fi ~~~~~~~~~(C.6)Cl
(2 kk)
where u(S) =
CT(w)
sin(TX/0)
snTt/) (0.7)
e Real F(k) exp
'=(.2
The function a(t) is called the sigma factor 112-143
Since the function f (t) is a real function, it implies where F(k) is obtained by FFT algorithm.
that: a The special form of equations (C.12) and (C.13)
allows the use of FFT algorithm which is well-known
F(a + jto) F (a'- jwo) (0.8) to be computationally efficient. This is because the
Therefore FFT algorithm requires only N log2N units of computa-
tion [1p compared with N2 units in the case of the
I direct method. The parameters a, Aw, Q in equation
/1T 2\\aQX (C.13) are not interrelated in a complex way but their
t / F
r t+- choice depends on the time over which the solution is
0 \ required.
/
/!\ Finally, the following parameters were tested on
~t-tt '- sinusoidal and consinsoidal waveforms as a result of
£2+ extensive mathematical experimentation Au = 7.5 v,
: = Ato, Q£2
~~~~~~a 1 x 7.5 X = 24,127.43 rad/sec.
The parameters that have been chosen gave eminently
Figure 0.1. The Dirichlet Kernel sinQ(£2t) satisfactory results for theO isever sstatd werel
rae
f (t) = exp(at) Real { f£Q F(a + jto) a: (to)
exp (jtot) dto } (0 .9)