Estimation of Transmission Line Parameters Using

PMU Measurements
Markos Asprou and Elias Kyriakides
Department of Electrical and Computer Engineering, KIOS Research Center for Intelligent Systems and Networks
University of Cyprus
Nicosia, Cyprus
{asprou.markos, elias}@ucy.ac.cy

Abstract—Transmission line parameters are affected by several The transmission line parameters that are stored at the
factors, such as ambient temperature, mutual coupling of control center are used in several tools for monitoring and
parallel lines, and erroneous transmission line length. Further, control effectively the power system operation [3]. In this
the several joints, especially at the line ends, cause additional sense, the line parameter values should be as precise as
changes in the calculated line parameters. These factors are possible in order not to degrade the accuracy of the control
rarely taken into consideration resulting to obsolete line center applications. For instance, the state estimator is a vital
parameter databases. The error that is encompassed in the tool in the control center providing in consecutive time
values of the transmission line parameters can be up to 30% of intervals the power system states (i.e., bus voltage magnitudes
their nominal value. In this paper, a methodology for estimating
and angles) which are useful for assessing the power system
the actual parameters of transmission lines represented as a
nominal pi model is proposed. In this methodology the PMU
operating condition. At the same time, it is important to ensure
measurements at the one end of the line and the estimated states the accuracy and reliability of the estimated states for the
from a hybrid state estimator for the other end of the line are advanced system applications. The erroneous line parameters
required. The proposed methodology is applied to the IEEE 14 considerably limit the performance of the power system state
bus system and the results are demonstrated in this paper. estimation, which takes as input the power system network
parameters [4]; it is therefore important to identify and
Index Terms—Parameter estimation, phasor measurement unit, estimate the erroneous line parameters stored in the control
state estimation, transmission lines. center database.

I. INTRODUCTION In some cases the identification of the erroneous

parameters is based on the experience of the power system
The values of the power system network parameters (i.e., operators [5]; however, an automated routine for the
resistance, reactance, and shunt admittance) are usually identification of the erroneous branches is more convenient.
calculated by the electric utilities using ideal models and data The identification procedure of the erroneous line parameters
provided by the manufacturers. In the real field, the is usually based on the normalized residuals of the
parameters of the transmission lines are affected by measurements, calculated using the Weighted Least Squares
environmental factors, human errors, or modeling (WLS) state estimator [6]. In [7], the identification of the
inaccuracies. More specifically, the resistance of the erroneous network parameters is achieved through the
transmission line conductor is affected by the ambient Langragian multipliers, while the methodology is improved by
temperature and the soil resistivity [1], while mutual coupling using multiple measurement scans [8] and measurements from
of parallel lines affects the transmission line reactance. In strategically placed Phasor Measurement Units (PMUs) [9]. It
some cases, it is impossible to know the exact values of is to be noted that in the proposed methodology the
transmission line parameters, given the several joints made identification of the erroneous parameters is not necessary.
throughout the length of the line (e.g., the line may be This is because the proposed methodology is valid only for
overhead, whereas the last few meters going into the particular lines that have one PMU at their ends, and therefore
substation may be underground). The changes in the network the estimation of the line parameters can be performed for all
configuration without updating the control center database and these lines, without needing to identify erroneous transmission
the miscalculation of the length of the transmission line also lines.
cause considerable mismatches between the actual values of
the transmission line parameters and the ones stored in the The methodologies for estimating the erroneous line
control center database. The errors between the actual and the parameters are separated in two main categories: methods
stored values of the line parameters can reach up to 30% of based on the residual sensitivity analysis [10] and methods
their nominal values [2]. based on the augmentation of the existing state vector of the
state estimation with the erroneous parameters either using

This work was supported by the European Regional Development Fund and
the Republic of Cyprus through the Research Promotion Foundation (Project
ǻǿǹȀȇǹȉǿȀǼȈ/ȀȊ-ȇȅȊ/0713/39)

978-1-4673-8040-9/15/$31.00 ©2015 IEEE

WLS state estimation [8] or Kalman filter theory [11]. With ~ ~ ~~
I sVr  Vs I r (1)
the advent of the synchronized measurement technology and ysr ~ ~ 
the deployment of the PMUs in the measurement layer of the Vs2  Vr2
power system, the network parameters can be calculated using ~ ~
the voltage and current phasor measurements provided by the
Vs  Vr (2)
ysh ~ ~
PMUs [12]. Although the estimation of the line parameters Is  Ir
using PMU measurements is accurate and reliable, such ~ ~
parameter estimation scheme requires the existence of PMUs where Vs and Vr are the voltage phasors of buses s and r
~
at both ends of a transmission line. The installation of PMUs respectively while I s is the current phasor that flows from bus
is done incrementally by the electric utilities and power ~
systems are partially observable by PMUs. Therefore, the s and I r is the current phasor that arrives to bus r as shown in
presence of two PMUs at both ends of a line might be rather Fig. 1.
rare case today. Based on (1) and (2) it is necessary to know the voltage
In this paper a methodology for estimating the erroneous and current phasor for the sending and receiving bus in order
line parameters in the case that the transmission line is to calculate accurately the parameters of the transmission line.
observed by only one PMU is proposed. In particular, a linear This can be achieved by the installation of two PMUs at the
parameter estimator is developed for using the phasor sending and receiving ends of the line. However, this is still
measurements that are provided by the PMU at the one end of not a realistic case since PMUs are typically installed sparsely
the transmission line and the estimated states that are provided in order to cover a wide area of the network.
by the hybrid state estimator (state estimator that use both In this paper, this constraint is relaxed by proposing a
conventional and PMU measurements) for the other end of the methodology for estimating the transmission line parameters
line. It is to be noted that the line that its parameters will be in the presence of only one PMU at one of the two ends of the
estimated is virtually eliminated from the network in the line. The algorithm of the proposed methodology is shown in
execution of the hybrid state estimator in order not to bias the Fig. 2 in a flowchart form. As it is aforementioned, the
state estimation in the case that the line parameters are identification of the erroneous line parameters is not necessary
erroneous. since the line parameter estimation can be performed for all
The paper is organized as follows: the proposed the lines that have one PMU at their ends. Initially, some
methodology for estimating the line parameters is presented in measurement snapshots from historical data for a relatively
Section II. The results when the proposed methodology is small time period are chosen. It is preferable that the chosen
tested on the IEEE 14 bus system are shown in Section III and measurement snapshots do not include bad measurements.
the paper concludes in Section IV. This can be achieved by filtering the measurement snapshots
using bad data detection and elimination techniques [5].
II. METHODOLOGY FOR ESTIMATING TRANSMISSION LINE However, even in the presence of bad measurements in the
PARAMETERS measurement snapshots, unless the bad measurement is not
When a PMU is installed at a bus, it can provide the critical, the proposed methodology is not affected. Further, the
voltage phasor of that bus and the phasors of the current that measurement snapshots should belong to the same time
flows through the transmission lines emanating from that bus. period, so that the values of the line parameters will be time
The transmission lines are usually modeled using the nominal invariant.
pi model. As shown in Fig. 1, based on the nominal pi model Based on the current situation, the power systems are
the transmission line is characterized by three parameters: the usually measurement redundant (i.e., there are more than one
series conductance, gsr, and susceptance bsr, and the shunt measurement that make a bus observable) and the elimination
admittance in which the conductance is usually omitted and of some measurements should not cause any unobservability
thus only the shunt susceptances bsh are considered. It is to be issues. Therefore, in this methodology all the measurements
noted that the shunt susceptances are usually considered zero that are related to the line parameters to be estimated are
for short lines. removed from the selected measurement snapshots. These
measurements could be power flow, power injection, or the
~ ~ current phasor provided by the PMU. This step is performed
Vs
~
Is
~ Vr in order to obtain an accurate estimation (without being
Ir affected by possible erroneous parameters) of the states (i.e.,
voltage magnitude and angle) at the end of the transmission
line that no PMU is installed.
After eliminating the measurements, a hybrid state
estimator is executed for all the measurement snapshots. In the
last step of the methodology the line parameters are estimated
Fig. 1 Transmission line representation in a pi model by executing a linear parameter estimator. More details about
The line parameters of the transmission line can be the linear parameter estimator and a brief description about the
calculated as [3], hybrid state estimator that is used in this paper are provided
next. It is to be noted that the proposed methodology is
executed offline.
 z >P fl P flpse Pinj Q fl Q pse
fl Q inj șVPMU VPMU @T . (8)
B. Linear parameter estimator
The types of the line parameters that can be estimated are
the series conductance and susceptance and the shunt
susceptance as shown in Fig. 1 (assuming that the shunt
conductance is very small and thus it can be neglected). By
assuming that a PMU is installed at bus s, the model that
relates the line parameters with the current phasor
measurements can be formulated as,
ª g sr º
ª I sreal º ª Vsreal  Vrreal  Vsimag  Vrimag  Vsimag º « »
« imag » « imag » bsr (9)
«¬ I s »¼ «¬Vs  Vrimag Vsreal  Vrreal Vsreal »¼ « »
«¬ bsh »¼
Fig. 2. Flowchart of the parameter estimation methodology where,
A. Hybrid state estimator I sreal and I simag are the real and imaginary part respectively
In this paper, the hybrid state estimator proposed in [13] is of the current that flows from bus s to bus r
used. The model of the hybrid state estimator is the same as
the conventional state estimator [14], Vsreal , Vsimag , Vrreal , and Vrimag are the real and imaginary
parts of the voltage of buses s and r respectively.
z h(x)  e, (3)
According to (9), the quantities that are needed for the
where z is the measurement vector, h(x) is the vector
estimation of the line parameters are: the current phasor that
containing the equations that relate the measurements to the
flows from bus s to bus r, the voltage phasor of bus s, (these
system states, x is the state vector containing the power system
quantities are provided as measurements by the PMU installed
states and e is the Gaussian noise in the measurements.
at bus s), and the voltage phasor of bus r which is obtained by
Based on the WLS formulation, the state vector can be the state estimator (since no PMU is installed at bus r).
determined by minimizing the function J(x) as,
The problem in (9) is under-determined and therefore the
min J (x) [z  h(x)]R 1[z  h(x)], (4) line parameters cannot be estimated since the measurements
x  are less than the unknowns. In order to make feasible the
where R is the measurement error covariance matrix. Thus, estimation of the line parameters several snapshots from a
(5) short time period are used in the parameter estimator and thus
xk 1 xk  [G(xk )]1 HT (xk )R 1[z  h(xk )],  (9) becomes,
where H(x) is the Jacobian matrix and is equal to w h ( x ) , ªª I real º º ªª V real
wx s s  Vrreal  Vsimag  Vrimag  Vsimag º º
1 «« imag » » «« imag » »
and G(x) is the gain matrix, given by H ( x ) R H ( x ).
T
««¬ I s »¼1 » ««¬Vs  Vrimag Vsreal  Vrreal Vsreal »¼1 »
« real » « real »
In the proposed hybrid state estimator, the measurement «ª I s º » « ª Vs  Vrreal  Vsimag  Vrimag  Vsimag º » ª g sr º
vector z consists of conventional SCADA measurements «« imag » » «« imag » »« »
««¬ I s »¼ 2 » ««¬Vs  Vrimag Vsreal  Vrreal Vsreal »¼ 2 » « bsr » (10)
(power flows, power injections and bus voltage magnitudes) « » « » ¬« bsh ¼»
and synchronized measurements (bus voltage phasors and line « » « »
current phasors). Since the reporting rate of the PMU «ª I real º » «ª V real  Vrreal  Vsimag  Vrimag  Vsimag º »
s s
«« imag » » «« imag » »
measurements is much faster than the SCADA measurements, «¬¬« I s ¼» n »¼ «¬¬«Vs  Vrimag Vsreal  Vrreal Vsreal ¼» n »¼
in the execution of the hybrid state estimator only the last
PMU measurement snapshot that is closer (in time) to the The above problem can be solved using the least squares
SCADA measurement snapshot is used. The problem is non- formulation and therefore the line parameters can be
linear, and the state vector x is estimated through an iterative calculated as,
procedure [5]. The presence of the current phasor p (M T M ) 1 M T I
measurements may cause convergence problems to the hybrid (11)
state estimator, due to numerical ill-conditioning [15]. Thus, where p is the line parameter vector, M is the Jacobian matrix
the pseudo flow measurements are used instead of the direct that relates the current phasor measurements to the line
current phasor measurements. The flow measurements can be parameters, and I is the vector with the current measurements
calculated as, from several snapshots of a short time period.
Pij pseudo Vi Iij cos(Ti  Tij ) (6) III. APPLICATION OF THE METHODOLOGY FOR LINE
 PARAMETER ESTIMATION
Qij pseudo Vi Iij sin(Ti  Tij ) (7)
A description for the generation of measurements used in
Hence, the measurement vector of the proposed HSE is the proposed methodology and the test results are presented in
formed as, this Section.
A. Measurement generation TABLE I
MEASUREMENT DEVICES MAXIMUM ERRORS
The measurements used in this methodology are obtained Voltage Current
from power flow software [16]. To generate the virtual Real/reactive Real/reactive Phase angle
magnitude magnitude
power injection power flow PMU
measurements, the power flow results are subjected to random (%) (%)
PMU PMU
(degrees)
noise, assuming that all the systematic errors are eliminated (%) (%)
from the measurement process [17]. The Gaussian noise added ±3 ±3 ±0.02 ±0.03 ±0.54
to the PMU measurements is assumed to be introduced by
mainly two components of the measurement chain: the TABLE II
instrument transformer, and the measurement device. Thus the VOLTAGE AND CURRENT TRANSFORMERS OF 0.5 ACCURACY CLASS
MAXIMUM ERRORS [19]
PMU measurements are created as,
Voltage transformers Current transformers
V Percentage
Vmeas Vnetwork  N (0, uVT )  N (0, uVMU )  (12)
of voltage
Phase ± Percentage of current ± Phase displacement at
I I displacement error at percentage of percentage of rated
I meas I network  N (0, uCT )  N (0, u MU ) (13) magnitude
(degrees) rated current current (degrees)
error
V V TV TV
T meas T network  N (0, uVT )  N (0, u MU ) (14) ±0.5 ±0.333
1 5 20 100 120 1 5 20 100 120
- 1.5 0.75 0.5 0.5 - 1.5 0.75 0.5 0.5
I I TI TI
T meas T network  N (0, uCT )  N (0, u MU ), (15) B. Test results-IEEE 14 bus system
V I
where Vnetwork , I network , ,T network T network
are the input The proposed methodology for parameter estimation was
quantities to the instrument transformers and measurement applied to the IEEE 14 bus system, which is shown in Fig. 3.
V TV I TI In order to better illustrate the proposed methodology, a step
devices; uVT , uVT , uCT , u CT are the voltage and current by step estimation of the parameters for the transmission line
transformer standard uncertainties associated with the that connects buses 6 and 12 (hereafter line 6-12) is shown.
magnitude and the phase angle of the respective quantity; and The locations of the conventional measurements in the IEEE
T T 14 bus system are shown in Table III. Moreover, 3 PMUs are
uVMU , uMU
V I
, u MU , uMU
I
are the standard uncertainties of the
assumed to be installed at buses 2, 6, and 9, as shown in Fig.
measurement device (PMU) associated to the magnitude and
3. Based on that PMU configuration, no transmission line has
the phase of the voltage and current.
PMU at both ends and therefore the direct calculation of any
Further, the conventional measurements, such as the power line parameter using phasor measurements is not feasible. It is
flows and the power injections that are used in the hybrid state to be noted that there is not a specific number of PMUs for
estimator are created as, executing the methodology. The more PMUs are installed in
the system, the more line parameters of the power system can
P,Q
Pmeas Ptransf  N (0, uMU ) (16) be estimated.

P,Q In the case of line 6-12, it was assumed that its parameters
Qmeas Qtransf  N (0, uMU ), (17)
g6-12 and b6-12 are at a 30% error from their nominal values. As
where Ptransf and Qtransf are the power measurements illustrated in Fig. 3, the one end of the transmission line (bus
including the uncertainties that are introduced by the 6) has a PMU installed, while the other end (bus 12) has no
instrument transformers, and can be calculated as, PMU. Based on the flowchart of Fig. 2, the measurement
snapshots that will be used for the estimation of the line
V I
Ptransf Vtransf Itransf cos(Ttransf Ttransf ) (18) parameters are chosen. In this case study 50 measurement
 snapshots for a 5-hour time interval was used, given that the
V I
Qtransf Vtransf Itransf sin(Ttransf  Ttransf ). (19) line parameters are time invariant for this time interval. Then,
P ,Q all the measurements that are related to the erroneous line
and u MU is the power measurement device standard
parameters are eliminated from the updated measurement
uncertainty. vector. In the case of line 6-12, the real/reactive power flow
Both the instrument transformer and the measurement measurements of line 6-12, the real/reactive power injection
device standard uncertainties (PMUs and power measurement measurement of bus 6, and the real/reactive pseudo flow
devices) are calculated based on the maximum errors measurements that are calculated using the PMU current
introduced to the measurements, as specified by the phasor measurement are eliminated from the measurement
manufacturers [18], [19]. Further, it is assumed that the vector of the hybrid state estimator.
measurement errors lie with a 95% probability in the interval In the next step, the hybrid state estimator is executed for
bounded by the maximum errors [17]. The maximum errors the 50 measurement snapshots. In the last step of the
for both the measurement devices and the instrument methodology, for each snapshot, the estimated voltage phasor
transformers are shown in Tables I and II respectively. of bus 12, as well as the current phasor measurement for line
Without loss of generality it is assumed that the instrument 6-12 and the voltage phasor measurement of bus 6 are used in
transformers belong to the 0.5 accuracy class. The maximum the parameter estimator. The estimated values of the
Total Vector Error (TVE) of the PMUs is assumed to be 1% parameters of all the transmission lines (including line 6-12)
for being compliant with [20]. that have a PMU at only one end are shown in Table IV. It is
to be noted that in all the line parameters a 30% error was
considered and their erroneous values are shown in Table IV
in the column “Values used in the simulation”. Based on the at a time and the computational requirements are those of a
estimated values of the line parameters, the largest estimation state estimator which is anyway performed in any power
error for the series conductance (gsr) corresponds to line 9-10 system. In this work, it is assumed that only one line with
and is about 5.4%. In the case of the series susceptance (bsr) erroneous parameters exists in the system; however, it is likely
the largest estimation error is about 6.6% and corresponds to to have more than one line that has erroneous parameters. This
line 7-9. Further, the largest parameter estimation error for the issue will be considered for future investigation.
shunt susceptance is 8.5% and corresponds to line 2-5. It is to
be mentioned that the methodology is not affected by the error REFERENCES
magnitude that is encompassed in the line parameters, since [1] M Bockarjova and G Andersson, "Transmission line conductor
the erroneous parameters are eliminated from the state temperature impact on state estimator accuracy," IEEE PowerTech,
estimation process (thus the estimated states are not affected). Laussane, Jul. 2007.
[2] K. J Kusic and D. L Garrison, "Measurement of transmission line
parameters from SCADA data," IEEE PES Power System Conference
TABLE III
and Exposition, New York, USA, Oct. 2004.
PLACEMENT OF CONVENTIONAL MEASUREMENTS IN THE
IEEE 14-BUS SYSTEM [3] G. Sivanagaraju, S. Chakrabarti, and S.C. Srivastava, "Uncertainty in
Injection Voltage transmission line parameters: Estimation and impact on line current
Flow measurements location differential protection," IEEE Trans. on Instrumentation and
measurements magnitude
(bus # - bus #) Measurement, vol. 63, no. 6, pp. 1496-1504, Jun. 2014.
location (bus #) (bus #)
1-2, 1-5, 3-4, 4-7, 4-9, 5-6, 6-12, 6- 1, 2, 3, 4, 6, 9, 3, 8 [4] M Asprou, E Kyriakides, and M Albu, "The effect of parameter and
13, 7-8, 7-9, 12-13, 13-14,10-11 10, 11, 13, 14 measurement uncertainties on hybrid state estimation," IEEE Power and
Energy Society General Meeting, San Diego, CA, USA, 2012.
TABLE IV [5] A. Abur and A. G. Exposito, Power system state estimation: Theory and
ESTIMATION OF TRANSMISSION LINE PARAMETERS implementation. New York: Basel, 2004.
Values used in the Estimated value True value [6] M. R. M. Castillo et al., "Offline detection, identification, and correction
Line of branch parameter errors based on several measurement snapshots,"
simulation (p.u) (p.u) (p.u)
gsr bsr gsh gsr bsr gsh gsr bsr gsh IEEE Trans. on Power Systems, vol. 26, no. 2, pp. 870-877, May 2011.
1-2 6.5 19.8 0.069 5.0 15.1 0.056 4.9 15.2 0.052 [7] J. Zhu and A. Abur, "Identification of network parameter errors," IEEE
2-3 23.4 9.9 0.056 17.4 7.4 0.044 18 7.6 0.043 Transactions on Power Systems, vol. 21, no. 2, pp. 586-592, May 2006.
2-5 2.2 6.7 0.045 1.7 5.2 0.038 1.7 5.2 0.035 [8] L. Zhang and A. Abur, "Identifying parameter errors via multiple
2-4 2.2 6.7 0.042 1.7 5 0.032 1.7 5.1 0.034 measurement scans," IEEE Trans. on Power Systems, vol. 28, no. 4, pp.
6-11 2.5 5.3 0 1.9 3.9 0 1.9 4.1 0 3916-3923, Nov. 2013.
6-12 2 4.1 0 1.5 2.9 0 1.5 3.1 0
[9] L. Zhang and A. Abur, "Strategic placement of phasor measurements
6-13 4 7.9 0 3.1 5.8 0 3.1 6.1 0 for parameter error identification," IEEE Trans. on Power Systems, vol.
7-9 0 11.8 0 0 9.7 0 0 9.1 0 28, no. 1, pp. 393-400, Feb. 2013.
9-14 1.8 3.9 0 1.4 3 0 1.4 3 0
[10] T. Van Cutsem and V. Quintana, "Network parameter estimation using
9-10 5.1 13.4 0 3.7 10.8 0 3.9 10.3 0
online data with application to transformer tap estimation," IEE
Proceedings, vol. 135, no. 1, pp. 31-40, Jan. 1988.
[11] I. W. Slutsker, S. Mokhtari, and K. A. Clements, "Real time recursive
parameter estimation in energy management systems," IEEE Trans. on
Power Systems, vol. 11, no. 3, pp. 1393-1399, Aug. 1996.
[12] K. Dasgupta and S. A. Soman, "Line parameter estimation using phasor
measurements by the total least squares approach," 2013 IEEE Power
and Energy Society General Meeting , Vancouver, BC, Jul. 2013.
[13] M. Asprou and E. Kyriakides, "Enhancement of hybrid state estimation
using pseudo flow measurements," IEEE Power and Energy Society
General Meeting, Detroit, MI, USA, Jul. 2011, pp. 1-8.
[14] F. C. Schweppe and J. Wildes, "Power system static state
estimation,Part I: Exact model," IEEE Transactions on Power
Apparatus, vol. 89, pp. 120-125, Jan. 1970.
[15] S. Chakrabarti, E. Kyriakides, G. Ledwich, and A. Ghosh, "Inclusion of
PMU current phasor measurements in a power system state estimator,"
IET on Generation, Transmission, and Distribution, vol. 4, no. 10, pp.
1104-1115, Oct. 2010.
Fig. 3. IEEE 14 bus system. [16] PowerWorld. PowerWorld Corporation,[Online] Available
http://powerworld.com.
IV. CONCLUSIONS
[17] "JGCM:Evaluation of the measurement data-Guide to the expression of
In this paper, a methodology for estimating the uncertainty in measurements, JGCM100," 2008.
transmission line parameters in the case that PMU [18] "Model 1133A GPS-Synchronized Power Quality/Revenue standard,
measurements are available from only one end of the line is Operation manual, Arbiter systems Inc.,CA, USA," Sep. 2006.
proposed. Based on the simulation results, the proposed [19] "IEC standard for instrument transformers, IEC 60044," 1996.
methodology estimates accurately the true values of the [20] Power System Relaying Committee, "IEEE Standards for
parameters. Although the methodology is tested in a small Synchrophasor Measurements for Power Systems- IEEE Std C37.118.1-
scale system, the methodology can be applied successfully in 2011," New York, USA, 2011.
larger systems given that the estimation is performed one line