1
Including Voltage Measurements in Branch Current State
Estimation for Distribution Systems
Mesut E. Baran Jaesung Jung
Dept. of ECE, NC State University
AbstractThis paper considers the incorporation of voltage
measurements in a branch-current-based state estimation
(BCSE) program. Original BCSE is designed to include only
power and current measurements. The motivation for enhancing
BCSE is that with the adoption of large scale automated meter
infrastructure (AMI) technologies, voltage measurements will be
available at the distribution level. Hence, including these
measurements has the potential to improve the accuracy of the
state estimation. The paper elaborates the technical approach
taken to accomplish this task, and the test results for assessment.
KeywordsPower System State Estimation, Power Distribution
System.
I. INTRODUCTION
U tilities have been improving their means of monitoring
their distribution systems mainly to improve the service
reliability [1-8]. Recently, there has been additional incentive
to improve monitoring of feeders to improve the efficiency
by adopting advanced functions such as voltage control for
demand management. Effective management of distribution
systems require analysis tools that can estimate the state of
the system (the operating condition) and predict the response
of the system to changing load and weather conditions. The
main tool used for system analysis is the power flow analysis.
But this tool is not very suitable for real-time monitoring as it
requires accurate load and system data.
To better monitor the system operating conditions for system
management, recently some utilities have started installing
limited SCADA systems at distribution level. Some utilities
started deploying large scale Advanced Metering Infra-
structures (AMI) also. With the availability of real-time
measurements, new methods are proposed for monitoring the
operating point (state) of distribution system. One of the
approaches is power flow based [5-8] and the others [2,3] are
extensions of the conventional state estimation (SE) method
for three-phase analysis. Although SE is preferred over the
power flow approach, its computational complexity may
prevent its use in practical applications.
This work was supported in part by the U.S. Department of Energy under
Grant DE-FG02-06ER84647.
M. Baran is with Department of ECE, NC State University, Raleigh, NC
USA (e-mail: baran@eos.ncsu.edu).
J. Jung is a graduate student in the Dept. of ECE at NC State University in
Raleigh, NC (jsjung@ncsu.edu)
T. E. McDermott is with EnerNex Corp. Pittsburgh, PA, 15236 (email:
tom@enernex.com)
978-1-4244-4241-6/09/$25.00 2009 IEEE
Thomas E. McDermott
EnerNex Corp. Pittsburgh, PA
In this paper, a branch-current-based three-phase SE (BCSE)
method [4] is considered, as it is computationally more
efficient and more insensitive to line parameters than the
conventional node-voltage-based SE methods. BCSE is very
efficient in handling line-flow and power-injection
measurements for radial networks. In the original algorithm
the voltage measurements have not been included as the
voltage measurements have not been available [4]. But with
AMI, voltage measurements will be readily available. In [12],
a method for handling voltage measurements in BCSE is
introduced. This method converts the voltage measurements
into equivalent measurements. In this paper, we show the
enhancement of the original BCSE to include voltage
measurements more accurately.
II. BRANCH-CURRENT-BASED STATE ESTIMATION
A. Branch-Current-Based State Estimation (BCSE) Method
The branch-current-based SE method, like conventional node-
voltage-based SE methods, is based on the weighted least
square (WLS) approach. Rather than using the node voltages
as the system state x, the method uses the branch currents as
the state, i.e.
xi = [ I ri , I xi ]
where I ri is branch current real part and I xi is branch current
imaginary part. Hence, BCSE solves the following WLS
problem to obtain an estimate of the system operating point
defined by the system state x :
w ( z h ( x))
m
m i n J ( x) = i i i
2
= [ z h( x )]T W [ z h( x)] (1)
x
i =1
where and represent the weight and the
wi hi ( x )
measurements function associated with measurement zi
respectively. For the solution of this problem the conventional
iterative method is adapted by solving following the normal
equations at each iteration to compute the update
x k +1 = xk +Vx k
[G ( x k )]x k = H T ( x k )W [ z h( x k )] (2)
where
G ( x) = H T ( x)WH ( x)
is the gain matrix and H is the Jacobian of the measurement
function h( x) .
Hence the only difference between the node voltage based
 2

SE and BCSE is the measurement functions associated with measurement introduces coupling terms between the phases
the type of measurements to be processed. of branch currents and the real and imaginary parts of branch
Note that we need a three phase model for representing un- currents. Hence, this voltage measurement Vk contributes the
balanced distribution feeders [2], as both the feeder circuit
following terms to the measurement Jacobian matrix H :
can have single phase laterals and also the loads are
unbalanced. The main model for a feeder branch (line section) hvk
= X i sin k Ri cos k
shown in Fig. 1 is : I ri
Vr ,1 VS ,1 z11 z12 z13 I l ,1 hvk

V
= V
l z 21 z22 z 23 I l ,2 ( 3) = Ri sin k X i cos k (9)
r ,2 S ,2 I xi
Vr ,3
VS ,3 z31 z32 z33 I l ,3
where Z i = Ri + jX i is the line impedance of branch i and
Or in compact form:
the k is expressed as :
Vr = VS Z l I l (4) k k
where Z l is the branch impedance matrix and l is the line k = tan (V0 r real ( Zi Ii )) /(V0 x imag ( Zi Ii ))
1
i =1 i =1
(10)
length.
where substation voltage is V0 = V0 r + jV0 x . Hence, both the
Jacobian matrix H and the gain matrix G need to be revised to
include voltage measurements in BCSE.
The Jacobian matrix H will have the following structure:
H sr H sx
H = H cr H cx (11)

H vr H vr
Fig. 1.A three-phase feeder line model. Where H sr , H cr , H vr corresponds to elements for the power,
current, and voltage measurements, respectively. The Gain
matrix G can be constructed as follows:
The main advantage of the BCSE is that the measurement
functions are simplified for power and current measurements
m
taken from an unbalanced radial distribution feeder. For G = wi H i


T

[ Hi ] (12)
example, power flow measurements on the feeder can be i =1
converted to equivalent complex current measurement by
using the current estimate of the node voltage: where H i indicates the row of H.
Pi Vri + Qi Vxi Pi Vxi Qi Vri
I ri = 2 2
I xi = 2 2 ( 5) Note that, since H is now coupled, the real and imaginary
Vri + Vxi Vri + Vxi
parts have to be solved together in (1). Also, a good estimate
where I i = I ri + jI xi at branch i . Hence, the resulting functions of initial angle is needed for both voltage and current
of power measurement are linear as the state variables are the measurements in the construction of gain matrix. So, voltage
complex branch currents, and current measurements are excluded in the first iteration
hIri ( I r ) + jhIxi ( I x ) = I ri + jI xi (6) for the good estimate of initial angle.
Thus, the Jacobian matrix elements are :
III. TEST RESULT
hIri hIxi
I ri
=1
I xi
=1 (7) For testing the revised BCSE with voltage measurements, a
test feeder is used. The test feeder is an IEEE 34 node radial
B. Including Voltage Measurements in BCSE test feeder with 23 kV three-phase primary [13]. A reduced
version of this test feeder is used to facilitate debugging and
A voltage at the node of a radial feeder Vk is the voltage at
assessment. A one-line diagram of the reduced feeder is given
the substation minus the voltage drop on the line sections in Fig. 2. For test purposes, distributed line section loads are
between the substation and this node, and hence, the lumped equally at terminal nodes of the line sections. The
measurement function for the voltage measurement Vk can be nominal load data is used as actual load which is determined
written in terms of the branch currents as : by power flow results as the correct measurements. The
feeder is modeled to have a heavy loading condition on feeder
(8 )
k
Vk = V0 Zi Ii and the line data is used given in [14]. The total load is
i =1
Sload , a = 490 + j 315 kVA
where V0 is voltage at the substation. This voltage magnitude
 3

Sload ,b = 415 + j 275 kVA Sload ,c = 430 + j 280 kVA

Small number of iteration indicates fast convergence of the
BCSE method. Figure 3 shows the rate of convergence, i.e.,
k
the reduction in maximum b after each iteration, for four
cases in logarithmic scale. These results indicate that the
convergence is almost quadratic, and is not sensitive to the
type of measurement. Hence, the voltage measurements do
not degrade the performance of the BCSE method.
Fig. 2.One-line diagram of reduced test feeder.

The minimum voltage corresponding to this load is Iteration k

Vmin = V21, a = 0.9402 3.057

which indicates a heavy loading condition on the feeder.

The revised algorithm was implemented using the C language

bk/b0
on Microsoft Visual Studio .NET 2003. To assess the impact
of voltage measurements on the state estimation, four cases
have been considered:
Case 1: forecasted load data.
Case 2: The same measurements as in Case 1 plus three
voltage measurements m2-m4 from the feeder nodes.
Case 3: Power measurement at the substation (both real and
reactive), indicated as mo in Fig. 2, plus a current Fig.3. Convergence characteristics of BCSE
measurement on the feeder, m1, and forecasted load data.
Case 4: The same measurements as in Case 3 plus three Figures 4-5 compares the estimated branch currents for the
voltage measurements m2-m4 from the feeder nodes. four cases considered. The figures show the error in the
estimated state ( x = [ I r , I x ] ) of branch currents (phase a
To generate measurement data for testing purpose, first the
actual measurements have been obtained by running a power only) of the feeder. Fig. 4 compares the results for Case 1 and
flow for the given load. Then measurement error was added to Case 2. These results indicate that adding voltage
the actual measurements. measurements decrease the error in branch current estimates.
Note that the improvement in the estimation is more on the
Z = Z eZ
a
(19) real part of the current than the imaginary part. Since, in this
a
where Z is actual data and eZ is the measurement error. The case the currents have small imaginary components, the
overall reduction of the error in the current magnitude is
forecasted load data is created by perturbing the actual load
considerable especially towards the substation end of the
data by adding error of 30%. The power and current
feeder. Hence, these results indicate that having voltage
magnitude measurement errors are selected from Normal
measurements helps improve the estimation over the
distribution with a standard deviation of 0.0233 (accuracy is
conventional one that is based on the forecasted load only
7% of their measured values). The voltage measurements data
(case 1).
are generated by adding measurement error with a standard
deviation of 0.0067 ( 2% measurement error). Figure 5 compares the results for Case 3 and Case 4. Note
that these two cases illustrate the effect of adding voltage
The enhanced BCSE has run for these four cases. The measurements to a system which has some limited power and
convergence criteria used is the standard one, i.e., current measurement from the feeder (Case 3). The results
b
= H T W ( z h ( x ))

< 103 ( 20) indicate that in this case adding voltage measurements does
not improve the estimation as much as it did in the previous
Table I shows the performance of the method under the four
case. The differences between the two cases are not
test cases.
statistically significant.
TABLE I: CONVERGENCE PERFORMANCE OF BCSE UNDER FOUR CASES

Run Max. Residual Obj.

Test itr
(sec) rp(kw) Rq(kVar) J(x)
Case 1 6 0.0179 8.35E-07 1.09E-06 1.64E-09
Case 2 7 0.0201 0.02767 0.05044 14.63
Case 3 6 0.0201 0.1876 0.04298 6.092
Case 4 6 0.0223 0.1856 0.04595 10.82

Error (p.u.)
node
(a) error in branch current real part, Ir in p.u.
Error (p.u.)
node
(b) error in branch current imaginary part, Ix in p.u.
Error (p.u.)
node
(c) error in branch current magnitude in p.u.
Fig. 4. Estimated Branch Currents for Case1 and Case 2
IV. CONCLUSIONS
This paper shows that the basic BCSE can be extended to
include voltage measurements in the estimation. The initial
test results indicate that the voltage measurements on the
estimated values are helpful in improving the accuracy of
BCSE results. The impact of voltage measurements on the
estimation depends on the existence of the other
measurements on the feeder. These results are based on the
simulated measurements with assumed accuracy especially on
the forecasted load values.
4
Error (p.u.)
node
(a) error in branch current real part, Ir in p.u.
Error (p.u.)
node
(b) error in branch current imaginary part, Ix in p.u.
Error (p.u.)
node
(c) error in branch current magnitude in p.u.
Fig.5. Estimated Branch Currents for Case3 and Case 4
V. REFERENCES
[1] Mesut E. Baran, "Challenges in State Estimation on Distribution
Systems," in Proc. 2001 IEEE PES Summer Meeting., vol. 1, pp.
429-433.
[2] Mesut E. Baran and Arthur W. Kelley, "State Estimation for
Real-Time Monitoring of Distribution Systems," IEEE Trans.
Power Systems., vol. 9, pp. 1601-1609, Aug. 1994.
[3] C.N. Lu, J.H. Teng, and W-H.E. Liu, "Distribution system state
estimation," IEEE Trans. Power Systems., vol. 10, pp. 229-240,
Feb. 1995.
[4] Mesut E. Baran and Arthur W. Kelley, "A branch-current-based
state estimation method for distribution systems," IEEE Trans.
Power Systems., vol. 10, pp. 483-491, Feb. 1995.
[5] Royetelman I. and S.M. Shahidehpour, "State estimation for
electric power distribution systems in quasi real-time conditions,"
in Proc. 1993 IEEE PES Winter Meeting., vol. 8, pp. 2009-
2015.
 5

[6] Roy Hoffman, "Practical State Estimation for Electric

Distribution Networks," in Proc. 2006 IEEE PSCE Power Syst.
Conf. and Expo., pp. 510-517.
[7] Celik M.K. and W-H.E. Liu, "A practical distribution state
calculation algorithm," in Proc.1999 IEEE PES Winter Meeting.,
vol. 1, pp. 442-447.
[8] Lubkeman, D.L., Jianzhoung Zhang, Ghosh, A.K. and Jones,
R.H., "Field results for distribution circuit state estimator
implementation," IEEE Trans. Power Delivery, vol. 15, pp. 399-
406, Jan. 2000.
[9] Youman Deng, Ying He and Boming Zhang, "A branch-
estimation-based state estimation method for radial distribution
systems," IEEE Trans. Power Delivery, vol. 17, pp. 1057-1062,
Oct. 2002.
[10] Haibin Wang and Schulz, N.N., "A revised branch current-based
distribution system state estimation algorithm and meter
placement impact," IEEE Trans. Power Systems, vol. 19, pp.
207-213, Feb. 2004.
[11] Whei-Min Lin, Jen-Hao Teng and Shi-Jaw Chen, "A highly
efficient algorithm in treating current measurements for the
branch-current-based distribution state estimation," IEEE Trans.
Power Delivery, vol. 16, pp. 433-439, Jul. 2001.
[12] J.H. Teng, "Using voltage measurements to improve the results
of branch-current-based state estimators for distribution
systems," IEE Proc. Gener., Transm. and Distrib., vol. 149, pp.
667-672, Nov. 2002.
[13] Kersting, W.H., "Radial distribution test feeders," in Proc. 2001
IEEE PES Winter Meeting., vol. 2, pp. 908-912.
[14] Goswami, S.K. and Basu, S.K., "Direct solution of distribution
systems," IEE Proc. Gener., Transm. and Distrib., vol. 138, pp.
78-88, Jan. 1991.

VI. BIOGRAPHIES

Mesut E Baran is currently a professor at North

Carolina State University in Raleigh, NC. He received
his Ph.D. from the University of California, Berkeley
in 1988. His research interests include distribution and
transmission system analysis and control.

Jaesung Jung is currently working towards his

masters in Electrical Engineering at North Carolina
State University in Raleigh, NC. He received his B.S.
in Electrical and Computer Engineering form
ChungNam National University, South Korea in 2006.
His research interests are in the area of power system
dynamics and computer simulations in power systems
and power system wide area monitoring and control.

Tom McDermott (SM 1990) is a Senior Consulting Engineer with EnerNex,

currently working in wind generation, distribution systems, lightning protection,
custom software development, and electromagnetic transient studies. He is
currently Vice Chairman of the Distribution System Analysis Subcommittee, a
U.S. delegate to IEC TC 57 Working Group 14, and has previously chaired the
Pittsburgh Section IEEE and the Working Group on Estimating Lightning
Performance of Transmission Lines. Tom is a registered professional engineer in
Pennsylvania. He has a B. S. and M. Eng. in Electric Power from Rensselaer,
and a Ph.D. in Electrical Engineering from Virginia Tech.