41st Southeastern Symposium on System Theory M2C.

1
University of Tennessee Space Institute
Tullahoma, TN, USA, March 15-17, 2009

Some Algorithms for Transmission Line

Parameter Estimation
Yuan Liao1, Senior Member, IEEE

Moreover, for special non-compensated parallel line

Abstract— Transmission line parameters are inputs to various configurations, this paper shows it is possible to derive
power system analysis algorithms and precision of these algorithms for estimating the line parameters by employing
parameters are thus important in ensuring the analysis accuracy. voltage and current measurements from the local bus only.
This paper proposes methods for estimating positive-sequence Section II illustrates the proposed method for series
line parameters of series compensated lines. Voltages and compensated line utilizing data from both ends. Section III
currents from two ends of the line are utilized. In addition,
presents the methods for non-compensated parallel lines
algorithms for estimating parameters of non-compensated
parallel lines utilizing one end data for special scenarios are utilizing data from the local bus. The preliminary studies are
presented. summarized in Section IV, followed by the conclusion.

Index Terms— Transmission line parameter estimation, II. PROPOSED METHODS FOR ESTIMATING PARAMETERS OF
Distributed parameter line model, Series compensated line. SERIES COMPENSATED LINE

I. INTRODUCTION
P R Q

T ransmission line parameters are important inputs to

diverse power system analysis programs and algorithms
EG

l1 l2
EH

such as protective relaying algorithms and power flow

analysis [1-2]. Precision of line parameters is thus essential in
ensuring the accuracy of the obtained analysis results. This
paper is concerned with estimating line parameters including
series resistance, series reactance and shunt susceptance at the Figure 1. Series compensated line considered for analysis
fundamental frequency, which are usually required by power
system applications. Consider the line, assumed to be a transposed line, between
An optimal line parameter estimation method is proposed in terminals P and Q as shown in Figure 1, where EG and E H
[3], which utilizes voltages and currents from two ends of a represent the Thevenin equivalent sources. The series
line to determine the line parameters. The characteristics of compensation device is located at location R. The
the method are that bad measurement data may be identified compensation device can be a simple capacitor bank, or more
and thus only the reliable measurements are utilized to complicated thyristor-controlled power flow controller. The
calculate the line parameters. distance of line segment PR and QR are l1 and l 2 ,
Built upon [3], this paper further proposes an online
respectively.
algorithm for estimating the positive sequence line parameters
The proposed algorithm draws on the steady state voltage
for series compensated transmission lines. The proposed
and current phasors at terminal P and Q that are measured at
solution is based on the fundamental frequency voltage and
different moments such as every hour during normal
current phasors from two ends of the line, which can be
operations. The aim is to estimate the positive sequence series
calculated from recorded waveforms or directly obtained from
resistance and reactance and shunt susceptance of the line per
measuring devices such as phasor measurement units. We
unit length.
employ a two terminal transmission line model for illustrating
Note that zero sequence components in the circuit during
the solution.
normal operations are usually negligible, thus the proposed
Similar to [3], the equivalent π circuit based on distributed
algorithm mainly aims at estimating positive sequence line
parameter line model is harnessed to automatically and fully parameters instead of zero sequence parameters. Nonetheless,
consider shunt capacitances and distributed parameter effects if significant zero sequence components arise due to various
of long lines. reasons such as unbalanced load conditions, existence of faults
external to the studied line, etc., the proposed algorithm will
1
Yuan Liao ([email protected]) is with the Department of electrical and be equally applicable for estimating zero sequence parameters
computer engineering, University of Kentucky, Lexington, KY 40506, USA. of the line.

978-1-4244-3325-4/09/$25.00 ©2009 IEEE 127

Authorized licensed use limited to: HIGHER EDUCATION COMMISSION. Downloaded on June 10,2024 at 10:09:49 UTC from IEEE Xplore. Restrictions apply.
 M2C.1

The proposed approach for estimating positive sequence Z c = z1 / y1 (5)

parameters is delineated as follows.
The following discussion assumes that synchronized voltage γ = z1 y1 (6)
and current measurements at P and Q during normal Where,
operations are available. Figure 2 depicts the positive δ the synchronization angle between measurements at
sequence network of the system during normal operations. P and Q, representing any possible synchronization
error.
P Vsi Q z1 , y1 positive-sequence series impedance and shunt
I pi I ri I ri I qi admittance of the line per mile or km, respectively.

Z pr V ri Z qr The real part and imaginary part of z1 and the imaginary

V pi R Vqi
part of y1 are the three line parameters to be determined.
Y pr Yqr Yqr Referring to Figure 2, based on the line segment to the left
Y pr
of the compensation device, we can derive the following
2 2 2 2
equations:

V pi = Vri cosh(γl1 ) + I ri Z c sinh(γl1 ) (7)

Figure 2. Positive sequence network of the system during normal operation Vri
I pi = I ri cosh(γl1 ) + sinh(γl1 ) (8)
Zc
In the figure, the following notations are adopted.
V pi , I pi i th phasor measurement of positive sequence Based on the line segment to the right of the compensation
voltage and current at P at ith moment; device, we can write
Vqi , I qi i th phasor measurement of positive sequence voltage
Vri − V si = Vqi cosh(γl 2 ) + I qi Z c sinh(γl 2 ) (9)
and current at Q; i = 1,2,..., N , N being the total number of
measurement sets, with each set consisting of V pi , I pi , Vqi Vqi
I ri = I qi cosh(γl 2 ) + sinh(γl 2 ) (10)
and I qi ; Zc
Vri : the voltage at the left side of location R; One needs to note that the adopted current direction at both
Vsi : the voltage drop across the series compensation device at terminals of the line is flowing from terminal P to terminal Q.
location R;
Eliminating Vri and I ri , we get
I ri : the current flowing through the series compensation
device at location R;
V pi = Vsi cosh(γl1 ) + Vqi cosh(γl ) + I qi Z c sinh(γl ) (11)
Zc characteristic impedance of the line;
γ propagation constant of the line;
sinh(γl1 ) Vqi
Z pr , Z qr equivalent series impedance of the line segment PR I pi = Vsi + I qi cosh(γl ) + sinh(γl ) (12)
Zc Zc
and QR;
Y pr , Yqr equivalent shunt admittance of the line segment PR
Where l = l1 + l 2 is the total length of the line.
and QR; For one set of measurements, we can obtain the above two
l1 , l 2 length of the line segment PR and QR in mile or km. complex equations, which can be arranged into four real
equations. We have five real unknowns, i.e. three parameters
The equivalent line parameters are expressed as for the line, real and imaginary part of Vsi . If we obtain
another set of measurements, we have four more real
Z pr = Z c sinh(γl1 ) (1)
equations and two more real unknowns for the new Vsi . Then
with eight equations, the seven real unknowns can be solved.
Z qr = Z c sinh(γl 2 ) (2) It follows that, with more equations obtained according to
different operation conditions, the classical least squares based
2 γl method can be applied to obtain a more robust estimate.
Y pr = tanh( 1 ) (3) Several algorithms based on synchronized and
Zc 2 unsynchronized measurements are presented below.
2 γl
Yqr = tanh( 2 ) (4) A. Least squares method utilizing synchronized measurements
Zc 2

128
Authorized licensed use limited to: HIGHER EDUCATION COMMISSION. Downloaded on June 10,2024 at 10:09:49 UTC from IEEE Xplore. Restrictions apply.
 M2C.1

We define the unknown variables as

X = [ x1 , x 2 ,..., x 2 N , x 2 N +1 , x 2 N + 2 , x 2 N +3 ]T (13) H is composed of the derivatives of the functions with
respect to the unknown variables.
Where,
N total number of measurement sets; B. Least squares method utilizing unsynchronized
x2i −1 , x2i i = 1,..., N , variables used to represent measurements
voltage across the compensation device, i.e.
We define the unknown variables as
Vsi = x 2i −1e jx2 i ;

X = [ x1 , x 2 ,..., x 2 N , x 2 N +1 , x 2 N + 2 , x 2 N +3 , x 2 N + 4 ]T (23)
x2 N +1 , x2 N + 2 and x2 N +3 positive-sequence transmission line
series resistance, series reactance and shunt
Where,
susceptance per unit length, respectively.
x1 ,..., x2 N +3 same as those in (13)
By employing the defined variables, (11-12) can be written x2 N +4 synchronization angle;
as f 2i −1 ( X ) = 0 and f 2i ( X ) = 0, respectively as follows

f 2i −1 ( X ) = x 2i −1e jx2 i cosh(γl1 ) + Vqi cosh(γl ) When unsynchronized measurements are utilized, we will
(14) have
+ I qi Z c sinh(γl ) − V pi = 0
f 2i −1 ( X ) = x 2i −1e jx2 i cosh(γl1 ) + Vqi cosh(γl )
sinh(γl1 ) (24)
f 2i ( X ) = x 2i −1e jx2 i + I qi cosh(γl ) + I qi Z c sinh(γl ) − V pi e jx2 N + 4 = 0
Zc
(15)
Vqi
+ sinh(γl ) − I pi = 0 sinh(γl1 )
Zc f 2i ( X ) = x 2i −1e jx2 i + I qi cosh(γl )
Zc
(25)
Vqi
Where, + sinh(γl ) − I pi e jx2 N + 4 = 0
i = 1,2,..., N , representing the index of the measurement set; Zc

Z c = ( x 2 N +1 + jx 2 N + 2 ) /( jx 2 N +3 ) (16) The unknown variables can be estimated similarly to case

A.

γ = ( x 2 N +1 + jx 2 N + 2 )( jx 2 N +3 ) (17) C. Optimal estimator

Since voltage and current measurements may contain errors,

Define function vector F ( X ) as an optimal estimator may be designed by exploiting multiple
sets of measurements to minimize the impacts of measurement
F2i −1 ( X ) = Re( f i ( X )) , i = 1,...,2 N (18) errors and improve the estimation accuracy.
The voltage and current phasors at line terminals and the
F2i ( X ) = Im( f i ( X )) , i = 1,...,2 N (19) synchronization angle are considered as known measurements.
Define
Then the unknown variable vector X can be derived as
follows M = [V p1 , I p1 , Vq1 , I q1 ,..., V pN , I pN , VqN , I qN , δ ] (26)

X k +1 = X k + ΔX (20) δ will be assigned a value of zero since synchronized

measurements are utilized. Modeling δ in the system
ΔX = −( H T H ) −1[ H T F ( X k )] (21) equations could detect potential synchronization error due to
synchronizing device failures. M i , i = 1,..., (4 N + 1) ,
∂F ( X k ) designates the i th element of M .
H = (22) We define the measurement functions for each measurement
∂X
as
Where,
X k , X k +1 variable vector before and after k th iteration; Yi ( X ) = x 2i −1e jx2 i , i = 5n + 1, 5n + 2, 5n + 3, 5n + 4,
n = 0,..., N − 1 (27)
k : iteration number starting from 1;
ΔX variable update. Y4 N +1 ( X ) = x10 N + 4 (28)

129
Authorized licensed use limited to: HIGHER EDUCATION COMMISSION. Downloaded on June 10,2024 at 10:09:49 UTC from IEEE Xplore. Restrictions apply.
 M2C.1

Where X denotes the unknown variable vector, defined as Elements of function vector F ( X ) are:

X = [ x1 , x 2 ,..., x10 N , x10 N +1 , x10 N + 2 , x10 N +3 , x10 N + 4 ]T (29) F2i −1 ( X ) = Re( f i ( X )) , i = 1,...,2 N (38)

Where, F2i ( X ) = Im( f i ( X )) , i = 1,...,2 N (39)

T vector or matrix transpose operator;
x10i −9 , x10i −8 , x10i−7 , x10i −6 , x10i−5 , x10i −4 , x10i−3 , x10i−2 F2i + 4 N −1 ( X ) = abs (Yi ( X )) = x 2i −1 ,
i = 1,..., N , variables required to represent i = 5n + 1, 5n + 2, 5n + 3, 5n + 4, n = 0,..., N − 1 (40)
the 4 N complex measurements;
x10i −1 , x10i i = 1,..., N , variables required to represent F2i + 4 N ( X ) = angle(Yi ( X )) = x 2i ,
Vs , the voltage across the compensation i = 5n + 1, 5n + 2, 5n + 3, 5n + 4, n = 0,..., N − 1 (41)
device;
x10 N +1 , x10 N +2 and x10 N +3 positive-sequence F12 N +1 ( X ) = Y4 N +1 ( X ) = x10 N + 4 (42)
transmission line series resistance, series
reactance and shunt susceptance per unit
length, respectively; The measurement vector and function vector are related by:
x10 N +4 synchronization angle.
S = F(X ) + μ (43)
By employing the defined variables, (11-12) can be written
as f 2i −1 ( X ) = 0 and f 2i ( X ) = 0, respectively as shown below. Where, μ is determined according to meter characteristics.
The optimal estimate of X is obtained by minimizing the
f 2i −1 ( X ) = x10i −1e jx10i cosh(γl1 ) + x10i −5 e jx10i −4 cosh(γl ) cost function defined as:
(30)
+ Z c sinh(γl ) x10i −3 e jx10i − 2 − x10i −9 e jx10i −8 e jx10 N + 4 = 0
J = [ S − F ( X )]T R −1[ S − F ( X )] (44)

f 2i ( X ) = x10i −1e jx10i sinh(γl1 ) / Z c + x10i −5 e jx10i − 4 sinh(γl ) / Z c The solution to (32) can be derived following the Newton-
jx10i − 2 jx10i − 6 jx10 N + 4 Raphson method [3]. After X is obtained, (27) can be applied
+ cosh(γl ) x10i −3 e − x10i −7 e e =0
to compute the estimated values of the measurement phasors.
(31) For estimating positive-sequence line parameters, the
proposed method is also applicable to complex topology such
Where, as parallel lines as long as the voltage and current
measurements at two ends of the line are available, since there
i = 1,2,..., N , representing the index of the measurement set; is no mutual coupling between positive-sequence circuits.
To detect the presence of bad measurement data, the method
Z c = ( x10 N +1 + jx10 N + 2 ) /( jx10 N +3 ) (32) based on chi-square test may be utilized [3].

γ = ( x10 N +1 + jx10 N + 2 )( jx10 N +3 ) (33) III. ONE END ALGORITHM FOR ESTIMATING PARAMETERS OF
NON-COMPENSATED PARALLEL LINE
Introduce S and F ( X ) as the measurement vector and
function vector, respectively, with their elements shown Q
below. P Circuit 1
EG EH
Elements of vector S are:
Circuit 2 l
S i = 0 , i = 1,...,4 N (34)

S 2i + 4 N −1 = abs ( M i ) , i = 1,2,...,4 N (35) Figure 3. A parallel line considered for analysis

S 2i + 4 N = angle( M i ) , i = 1,2,...,4 N (36) Figure 3 shows a parallel line for analysis. If voltage and
current measurements from both ends are available, the
S12 N +1 = M 4 N +1 (37) methods presented in Section II can be directly applied to
estimate the positive-sequence parameters of each line, by
Where abs (.) and angle(.) yield the magnitude and angle of setting V si to zero.
the input argument, respectively.

130
Authorized licensed use limited to: HIGHER EDUCATION COMMISSION. Downloaded on June 10,2024 at 10:09:49 UTC from IEEE Xplore. Restrictions apply.
 M2C.1

When the remote ends of the two circuits share a common estimation technique can be drawn on to find a more robust
bus, and thus have a common voltage, this section will show solution.
that it is possible to estimate the positive-sequence line
parameters utilizing one terminal data during certain special
B. Special case two
situations. In this discussion, we assume that data from
terminal P are available.
In this case, the two circuits of the line have identical
parameters, and the sending ends of the two circuits have
A. Special case one different voltages and currents. Then based on positive-
sequence networks during normal operations, we get
In this case, the two circuits of the line have different
parameters. The sending ends of the two circuits can share a V1 pi = Vqi cosh(γl ) + I 1qi Z c sinh(γl ) (49)
common bus, or possess different buses.
Based on the positive-sequence network during normal
Vqi
operations, we obtain I 1 pi = I 1qi cosh(γl ) + sinh(γl ) (50)
Zc
V1 pi = Vqi cosh(γ 1l ) + I1qi Z1c sinh(γ 1l ) (45)
V2 pi = Vqi cosh(γl ) + I 2 qi Z c sinh(γl ) (51)
Vqi
I 1 pi = I 1qi cosh(γ 1l ) + sinh(γ 1l ) (46)
Z 1c Vqi
I 2 pi = I 2 qi cosh(γl ) + sinh(γl ) (52)
Zc
V2 pi = Vqi cosh(γ 2 l ) + I 2 qi Z 2c sinh(γ 2 l ) (47)
For one set of measurements at terminal P, we got 8 real
Vqi equations. The number of unknown variables is 9. If we
I 2 pi = I 2 qi cosh(γ 2 l ) + sinh(γ 2 l ) (48) acquire a 2nd set of measurements, we will get 8 additional real
Z 2c equations and 6 more real unknowns. In total, we will have 16
real equations and 15 unknowns. Thus, it is possible to solve
Where, the line parameters. Similarly to the preceding case, if more
V1 pi , I 1 pi i th phasor measurement of positive sequence sets of measurements are obtained, the least squares based
method, or optimal estimation technique can be harnessed to
voltage and current at P for circuit 1; i = 1,2,..., N , N being estimate the line parameters.
the total number of measurement sets, with each set consisting
of V1 pi , I1 pi , V2 pi and I 2 pi ;
IV. PRELIMINARY RESULTS
V2 pi , I 2 pi i th phasor measurement of positive sequence
voltage and current at P for circuit 2; Preliminary results based on simulation studies have shown
Vqi i th phasor measurement of positive sequence voltage at that the proposed methods are theoretically sound and can
Q; yield quite accurate results. Impacts of possible errors of
voltage and current measurements on the accuracy of the
I1qi , I 2 qi i th phasor quantity of positive sequence current at Q
proposed algorithms will be further studied.
for circuit 1 and 2;
Z c1 , Z c 2 characteristic impedance of circuit 1 and 2;
γ 1 , γ 2 propagation constant of circuit 1 and 2. V. CONCLUSION

For each set of measurement, we can establish the above 4 This paper presents algorithms for estimating the positive
complex equations, which can be separated into 8 real sequence parameters of a series compensated transmission line
equations. The unknowns are: 6 line parameters (6 real by utilizing online voltage and current phasors measured at
unknowns), 1 complex voltage at remote bus (2 real different moments from two terminals of the line during
unknowns), and 2 complex currents at remote bus (4 real normal operations. This paper also proposes methods for non-
unknowns), and thus we have a total of 12 real unknown compensated parallel lines based on local data for special
variables. If we have a 2nd set of measurements, we will have configurations. The developed algorithms are based on
another 8 real equations, and 6 more unknowns. If we have a distributed parameter line model and thus fully consider the
3rd set of measurements, we will have another 8 real equations, effects of shunt capacitance and distributed parameter effects
and 6 more unknowns. Together, we will have 24 real of long lines. The common least squares based technique and
equations and 24 real unknown variables. Therefore, the more advanced non-linear optimization theory may be adopted
unknowns can be solved. If more sets of measurements are to solve for the parameters. More detailed studies will be
obtained, the least squares based method, or optimal reported in the future.

131
Authorized licensed use limited to: HIGHER EDUCATION COMMISSION. Downloaded on June 10,2024 at 10:09:49 UTC from IEEE Xplore. Restrictions apply.
 M2C.1

VI. REFERENCES

[1] S.H. Horowitz and A. G. Phadke, Power System Relaying, Research

Studies Press Ltd., Taunton, Somerset, England, 1995.
[2] Y. Liao, “Fault location utilizing unsynchronized voltage measurements
during fault”, Electric Power Components & Systems, vol. 34, no. 12,
December 2006, pp. 1283 – 1293.
[3] Yuan Liao and Mladen Kezunovic, “Online optimal transmission line
parameter estimation for relaying applications”, vol. 24, no. 1, January
2009, pp. 96 - 102.

VII. BIOGRAPHY

Yuan Liao (S’98-M’00-SM’05) is an Assistant Professor with the

Department of electrical and computer engineering at the University of
Kentucky, Lexington, KY, USA. He was a R&D Consulting Engineer and
then Principal R&D Consulting Engineer with the ABB Corporate Research
Center, Raleigh, NC, USA. His research interests include protection, power
quality analysis, large-scale resource scheduling optimization and
EMS/SCADA design.

132
Authorized licensed use limited to: HIGHER EDUCATION COMMISSION. Downloaded on June 10,2024 at 10:09:49 UTC from IEEE Xplore. Restrictions apply.