World Academy of Science, Engineering and Technology

International Journal of Electrical, Computer, Energetic, Electronic and Communication Engineering Vol:3, No:1, 2009

An Effective Approach for Distribution System

Power Flow Solution
A. Alsaadi, and B. Gholami

algorithms specially designed for distribution systems have

Abstract—An effective approach for unbalanced three-phase been proposed in the literature [5]–[13].
distribution power flow solutions is proposed in this paper. The Some of these methods were developed based on the
special topological characteristics of distribution networks have been general meshed topology like transmission systems [5]–[9].
fully utilized to make the direct solution possible. Two matrices—the From those methods, the Gauss implicit -matrix method [7] is
bus-injection to branch-current matrix and the branch-current to bus- one of the most commonly used methods; however, this
voltage matrix— and a simple matrix multiplication are used to
method does not explicitly exploit the radial network structure
International Science Index, Electrical and Computer Engineering Vol:3, No:1, 2009 waset.org/Publication/14357

obtain power flow solutions. Due to the distinctive solution

techniques of the proposed method, the time-consuming LU of distribution systems and, therefore, requires the solution of
decomposition and forward/backward substitution of the Jacobian a set of equations whose size is proportional to the number of
matrix or admittance matrix required in the traditional power flow buses. Recent research proposed some new ideas on how to
methods are no longer necessary. Therefore, the proposed method is deal with the special topological characteristics of distribution
robust and time-efficient. Test results demonstrate the validity of the systems [10]–[15], but these ideas require new data format or
proposed method. The proposed method shows great potential to be some data manipulations. In [10], the authors proposed a
used in distribution automation applications. compensation-based technique to solve distribution power
flow problems. Branch power flows rather than branch
Keywords—Distribution power flow, distribution automation currents were later used in the improved version and presented
system, radial network, unbalanced networks.
in [11]. Since the forward/backward sweep technique was
adopted in the solution scheme of the compensation-based
I. INTRODUCTION
algorithm, new data format and search procedure are

AS an important tool and the foundation of Distribution

Management System (DMS), power flow calculation
problem has been paid more and more attention. Many
necessary. Extension of the method, which emphasized on
modeling unbalanced loads and dispersed generators, was
proposed in [12]. In [13], the feeder lateral based model was
programs of real-time applications in the area of distribution adopted, which required the “layer-lateral” based data format.
automation , such as network optimization planning, One of the main disadvantages of the compensation-based
switching, state estimation, and so forth, require a robust and methods is that new databases have to be built and maintained.
efficient power flow method [1]–[3]. Such a power flow In addition, no direct mathematical relationship between the
method must be able to model the special features of system status and control variables can be found, which makes
distribution systems in sufficient detail. The well-known the applications of the compensation-based algorithm difficult.
characteristics of an electric distribution system are radial; The algorithm proposed in this paper is a novel technique.
multiphase and unbalanced operation; unbalanced distributed The only input data of this algorithm is the conventional bus-
load; extremely large number of branches and nodes; wide- branch oriented data used by most utilities. The goal of this
ranging resistance and reactance values. Those features cause paper is to develop a formulation, which takes advantages of
the traditional power flow methods used in transmission the topological characteristics of distribution systems, and
systems, such as the Gauss-Seidel and Newton-Raphson solve the distribution power flow directly. It means that the
techniques, to fail to meet the requirements in both time-consuming LU decomposition and forward/backward
performance and robustness aspects in the distribution system substitution of the Jacobian matrix or the Y admittance matrix,
applications. In particular, the assumptions necessary for the required in the traditional Newton Raphson and Gauss implicit
simplifications used in the standard fast-decoupled Newton- Z matrix algorithms, are not necessary in the new
Raphson method [4] often are not valid in distribution development. Two developed matrices, the bus-injection to
systems. Therefore, a novel power flow algorithm for branch-current matrix and the branch-current to bus-voltage
distribution systems is desired. To qualify for a good matrix, and a simple matrix multiplication are utilized to
distribution power flow algorithm, all of the characteristics obtain power flow solutions. The proposed method is robust
mentioned before, need to be considered. Several power flow and very efficient compared to the conventional methods. Test
results demonstrate the feasibility and validity of the proposed
method.
A. Alsaadi and B. Gholami are with the Islamic Azad University,
Kazeroon branch, Iran.

International Scholarly and Scientific Research & Innovation 3(1) 2009 1 scholar.waset.org/1999.5/14357
 World Academy of Science, Engineering and Technology
International Journal of Electrical, Computer, Energetic, Electronic and Communication Engineering Vol:3, No:1, 2009

II. UNBALANCED THREE-PHASE LINE MODEL *

⎛ P + jQ ⎞
Fig. 1 shows a three-phase line section between bus i and j. I ik = I ir (V i k ) + jI ii (V i k ) = ⎜⎜ i k i ⎟⎟ (5)
The line parameters can be obtained by the method developed ⎝ Vi ⎠
by Carson and Lewis [2]. A 4×4 matrix, which takes into
Where
Vi k and are the bus voltage and I ik equivalent current
account the self and mutual coupling effects of the unbalanced
threephase line. Ir
injection of bus at the k-th iteration, respectively. i and
I ii are the real and imaginary parts of the equivalent current
injection of bus i at the k-th iteration, respectively.

A. Relationship Matrix Developments

Fig. 1 Three-phase line section model

A 4×4 matrix, which takes into account the self and mutual
International Science Index, Electrical and Computer Engineering Vol:3, No:1, 2009 waset.org/Publication/14357

coupling effects of the unbalanced threephase line section, can

be expressed as:

⎡ Z aa Z ab Z ac Z an ⎤
⎢Z Z bb Z bc Z bn ⎥⎥
[Z abcn ] = ⎢ ba (1)
⎢ Z ca Z cb Z cc Z cn ⎥
⎢ ⎥
⎣ Z na Z nb Z nc Z nn ⎦ Fig. 2 Simple distribution system

After Kron’s reduction is applied, the effects of the neutral or

A simple distribution system shown in Fig. 2 is used as an
ground wire are still included in this model as shown in (2)
example. The power injections can be converted to the
equivalent current injections by (5), and the relationship
⎡ Z aa − n Z ab − n Z ac − n ⎤ between the busmcurrent injections and branch currents can be
[Z abc ] = ⎢⎢ Z ba − n Z bb − n Z bc − n ⎥⎥ (2) obtained by applying Kirchhoff’s Current Law (KCL) to the
distribution network. The branch currents can then be
⎢⎣ Z ca − n Z cb − n Z cc − n ⎥⎦ formulated as functions of equivalent current injections. For
example, the branch currents B1, B3 and B5, can be expressed
The relationship between bus voltages and branch currents in by equivalent current injections as
Fig. 1 can be expressed as
B1 = I 2+ I 3+ I 4+ I 5+ I 6
⎡Va ⎤ ⎡V A ⎤ ⎡Z aa−n Z ab−n Z ac−n ⎤ ⎡I Aa ⎤
B3 = I 4+ I 5 (6)
⎢V ⎥ = ⎢V ⎥ − ⎢Z
⎢ b ⎥ ⎢ B ⎥ ⎢ ba−n Z bb−n Z bc−n ⎥⎥ ⎢⎢ I Bb ⎥⎥ (3)
B5 = I 6
⎢⎣Vc ⎥⎦ ⎢⎣VC ⎥⎦ ⎢⎣ Z ca−n Z cb−n Z cc−n ⎥⎦ ⎢⎣ I Cc ⎥⎦
Therefore, the relationship between the bus current injections
For any phases failed to present, the corresponding row and and branch currents can be expressed as
column in this matrix will contain null-entries.
⎡ B1 ⎤ ⎡1 1 1 1 1⎤ ⎡ I 2 ⎤
III. ALGORITHM DEVELOPMENT
The proposed method is developed based on two derived
⎢ B ⎥ ⎢0 1 1
⎢ 2⎥ ⎢ 1 1 ⎥⎥ ⎢⎢ I 3 ⎥⎥
matrices, the bus-injection to branch-current matrix and the ⎢ B3 ⎥ = ⎢0 0 1 1 0⎥ ⎢ I 4 ⎥ (7a)
branchcurrent to bus-voltage matrix, and equivalent current ⎢ ⎥ ⎢ ⎥⎢ ⎥
injections. In this section, the development procedure will be ⎢ B 4 ⎥ ⎢0 0 0 1 0⎥ ⎢ I 5 ⎥
described in detail. For distribution networks, the equivalent- ⎢⎣ B 5 ⎥⎦ ⎢⎣0 0 0 0 1 ⎥⎦ ⎢⎣ I 6 ⎥⎦
current-injectionbased model is more practical [5]–[13]. For
bus , the complex load Si is expressed by Equation (7a) can be expressed in general form as:
S i = Pi + jQi i = 1... N (4)
And the corresponding equivalent current injection at the -th [B] = [BIBC ][I ] (7b)
iteration of solution is

International Scholarly and Scientific Research & Innovation 3(1) 2009 2 scholar.waset.org/1999.5/14357
 World Academy of Science, Engineering and Technology
International Journal of Electrical, Computer, Energetic, Electronic and Communication Engineering Vol:3, No:1, 2009

where BIBC is the bus-injection to branch-current (BIBC) the j-th bus and fill the line impedance (Zij ) to the position of
matrix. the j-th bus row and the k-th column.
The constant BIBC matrix is an upper triangular matrix and Step 6) Repeat procedure (5) until all line sections are
contains values of 0 and 1 only. included in the BCBV matrix.
The relationship between branch currents and bus voltages The algorithm can easily be expanded to a multiphase line
as shown in Fig. 2 can be obtained by (3). For example, the section or bus. For example, if the line section between bus
voltages of bus 2, 3, and 4 are and bus is a three-phase line section, the corresponding branch
current Bi will be a 3×1 vector and the in the BIBC matrix will
V2 = V1 − B1 Z 12 (8a ) be a 3×3 identity matrix. Similarly, if the line section between
bus i and bus j is a three-phase line section, the Zij in the
V3 = V2 − B2 Z 23 (8b) BCBV matrix is a 3×3 impedance matrix as shown in (2).
V4 = V3 − B3 Z 34 (8c) It can also be seen that the building algorithms of the BIBC
and BCBV matrices are similar. In fact, these two matrices
were built in the same subroutine of our test program.
where Vi is the voltage of bus i, and Zij is the line impedance Therefore, the computation resources needed can be saved. In
between bus i and bus j. addition, the building algorithms are developed based on the
Substituting (8a) and (8b) into (8c), (8c) can be rewritten as traditional bus-branch oriented database; thus, the data
International Science Index, Electrical and Computer Engineering Vol:3, No:1, 2009 waset.org/Publication/14357

preparation time can be reduced and the proposed method can

V4 = V1 − B1 Z 12 − B2 Z 23 − B3 Z 34 (9) be easily integrated into the existent DA.

From (9), it can be seen that the bus voltage can be expressed C. Solution Technique Developments
as a function of branch currents, line parameters, and the The BIBC and BCBV matrices are developed based on the
substation voltage. Similar procedures can be performed on topological structure of distribution systems. The BIBC matrix
other buses; therefore, the relationship between branch represents the relationship between bus current injections and
currents and bus voltages can be expressed as branch currents. The corresponding variations at branch
currents, generated by the variations at bus current injections,
can be calculated directly by the BIBC matrix. The BCBV
⎡V1 ⎤ ⎡V2 ⎤ ⎡Z12 0 0 0 0 ⎤ ⎡ B1 ⎤ matrix represents the relationship between branch currents and
⎢V ⎥ ⎢V ⎥ ⎢Z
⎢ 1 ⎥ ⎢ 3 ⎥ ⎢ 12 Z 23 0 0 0 ⎥⎥ ⎢⎢ B2 ⎥⎥ bus voltages. The corresponding variations at bus voltages,
⎢V1 ⎥ − ⎢V4 ⎥ = ⎢Z12 Z 23 Z 34 0 0 ⎥ ⎢ B3 ⎥ (10a) generated by the variations at branch currents, can be
⎢ ⎥ ⎢ ⎥ ⎢ ⎥⎢ ⎥ calculated directly by the BCBV matrix. Combining (7b) and
⎢V1 ⎥ ⎢V5 ⎥ ⎢Z12 Z 23 Z 34 Z 45 0 ⎥ ⎢ B4 ⎥ (10b), the relationship between bus current injections and bus
⎢⎣V1 ⎥⎦ ⎢⎣V6 ⎥⎦ ⎢⎣Z12 Z 23 0 0 Z 56 ⎥⎦ ⎢⎣ B5 ⎥⎦ voltages can be expressed as

Equation (10a) can be rewritten in general form as: [ΔV ] = [BCBV ][BIBC ][I ] = [DLF ][I ] (11)

[ΔV ] = [BCBV ][B ] (10b) And the solution for distribution power flow can be obtained
by solving (12) iteratively
*
where BCBV is the branch-current to bus-voltage (BCBV) ⎛ P + jQ ⎞
I k
= I (V i ) + jI (V i ) = ⎜⎜ i k i
r k i k
⎟⎟ (12a)
matrix. i i i
⎝ Vi ⎠
[ΔV ] = [DLF ][I ]
k +1 k
(12b)
[V ] = [V ]+ [ΔV ]
B. Building Formulation Development
k +1 0 k +1
Observing (7), a building algorithm for BIBC matrix can be (12c)
developed as follows:
Step 1) For a distribution system with m-branch section and n- According to the research, the arithmetic operation number
bus, the dimension of the BIBC matrix is m×(n-1). of LU factorization is approximately proportional to N3. For a
Step 2) If a line section (Bk ) is located between bus i and bus large value of N, the LU factorization will occupy a large
j, copy the column of the i-th bus of the BIBC matrix to the portion of the computational time. Therefore, if the LU
column of the j-th bus and fill a 1 to the position of the k-th factorization can be avoided, the power flow method can save
row and the j-th bus column. tremendous computational resource. From the solution
Step 3) Repeat procedure (2) until all line sections are techniques described before, the LU decomposition and
included in the BIBC matrix. From (10), a building forward/backward substitution of the Jacobian matrix or the Y
algorithm for BCBV matrix can be developed as follows. admittance matrix are no longer necessary for the proposed
Step 4) For a distribution system with m-branch section and n- method. Only the DLF matrix is necessary in solving power
bus, the dimension of the BCBV matrix is (n-1) ×m. flow problem. Therefore, the proposed method can save
Step 5) If a line section is located between bus i and bus j , considerable computation resources and this feature makes the
copy the row of the i-th bus of the BCBV matrix to the row of proposed method suitable for online operation.

International Scholarly and Scientific Research & Innovation 3(1) 2009 3 scholar.waset.org/1999.5/14357
 World Academy of Science, Engineering and Technology
International Journal of Electrical, Computer, Energetic, Electronic and Communication Engineering Vol:3, No:1, 2009

V. TEST RESULTS (equivalent 13-node system), including the three-phase,

The proposed three-phase power flow algorithm was double-phase, and single-phase line sections and buses as
implemented using MATLAB. Two methods are used for tests shown in Fig. 4 is used for comparisons. The final voltage
and the convergence tolerance is set at 0.001 p.u. solutions of method 1 and method 2 are shown in Table I.
Method 1: The forward/backward method [10]. From Table I, the final converged voltage solutions of method
Method 2: The proposed algorithm. 2 are very close to the solution of method 1. It means that the
accuracy of the proposed method is almost the same as the
A. Accuracy Comparison commonly used forward/backward method.
For any new method, it is important to make sure that the
final solution of the new method is the same as the existent B. Performance Test
method. An eight-bus system A main feeder trunk with 3x90-phase buses, is used for this
test. The single and double-phase laterals have been lumped to
form the unbalanced loads for testing purposes. This trunk is
then chopped into various sizes for tests as shown in Table II.
The substation is modeled as the slack bus.
Table III lists the number of iterations and the normalized
execution time for both methods. It can be seen that method 2
International Science Index, Electrical and Computer Engineering Vol:3, No:1, 2009 waset.org/Publication/14357

is more efficient, especially when the network size increases,

since the time-consuming processes such as LU factorization
and forward/backward substitution of –admittance matrix are
not necessary for method 2. For a 270-node system, method 2
Fig. 4 Eight-bus distribution system is almost 24 times faster than method 1.

C. Robustness Test
TABLE I
FINAL CONVERGED VOLTAGE SOLUTIONS One of the major reasons, which make the power flow
program diverge, is the ill-condition problem of the Jacobian
matrix or admittance matrix. It usually occurs when the
system contains some very short lines or very long lines. In
order to prove that the proposed method can be utilized in
severe conditions, IEEE 37-bus test feeder is used [14]. The
test feeder is adjusted by changing the length of eight line
sections. Four of them are multiplied by ten, and the other four
are divided by ten. The test result shows the number of
iterations for this case is 4 and the execution time is 0.0181 s.
It means that the proposed method is robust and very suitable
for online use.

VI. DISCUSSION AND CONCLUSION

In this paper, a direct approach for distribution power flow
solution was proposed. Two matrices, which are developed
from the topological characteristics of distribution systems,
TABLE II
are used to solve power flow problem. The BIBC matrix
TEST FEEDER represents the relationship between bus current injections and
branch currents, and the BCBV matrix represents the
relationship between branch currents and bus voltages. These
two matrices are combined to form a direct approach for
solving power flow problems. The time-consuming
procedures, such as forward/backward substitution of the
Jacobian matrix or admittance matrix, are not necessary in the
proposed method. The ill-conditioned problem that usually
TABLE III occurs during the other traditional methods will not occur in
NUMBER OF ITERATION AND NORMALIZED EXECUTION TIME the proposed solution techniques. Therefore, the proposed
method is both robust and efficient. Test results show that the
proposed method is suitable for large-scale distribution
systems.

International Scholarly and Scientific Research & Innovation 3(1) 2009 4 scholar.waset.org/1999.5/14357
 World Academy of Science, Engineering and Technology
International Journal of Electrical, Computer, Energetic, Electronic and Communication Engineering Vol:3, No:1, 2009

REFERENCES
[1] IEEE Tutorial Course on Distribution Automation.
[2] IEEE Tutorial Course on Power Distribution Planning.
[3] W. M. Lin and M. S. Chen, “An overall distribution automation
structure,” Elect. Power Syst. Res., vol. 10, pp. 7–19, 1986.
[4] B. Stott and O. Alsac, “Fast decoupled load flow,” IEEE Trans. Power
Apparat. Syst., vol. 93, pp. 859–869, May/June 1974.
[5] J. H. Teng andW. M. Lin, “Current-based power flowsolutions for
distribution systems,” in Proc. IEEE Int. Conf. Power Syst. Technol.,
Beijing, China, 1994, pp. 414–418.
[6] T. S. Chen, M. S. Chen, T. Inoue, and E. A. Chebli, “Three-phase
cogenerator and transformer models for distribution system analysis,”
IEEE Trans. Power Delivery, vol. 6, pp. 1671–1681.2, Oct. 1991.
[7] T.-H. Chen, M.-S. Chen, K.-J. Hwang, P. Kotas, and E. A. Chebli,
“Distribution system power flow analysis—A rigid approach,” IEEE
Trans. Power Delivery, vol. 6, pp. 1146–1152, July 1991.
[8] T. H. Chen and J. D. Chang, “Open wye-open delta and open delta-open
delta transformer models for rigorous distribution system analysis,” in
Proc. Inst. Elect. Eng., vol. 139, 1992, pp. 227–234.
[9] K. A. Birt, J. J. Graffy, J. D. McDonald, and A. H. El-Abiad, “Three
phase load flow program,” IEEE Trans. Power Apparat. Syst., vol. PAS-
International Science Index, Electrical and Computer Engineering Vol:3, No:1, 2009 waset.org/Publication/14357

95, pp. 59–65, Jan./Feb. 1976.

[10] D. Shirmohammadi, H. W. Hong, A. Semlyen, and G. X. Luo, “A
compensation- based power flow method for weakly meshed distribution
and transmission networks,” IEEE Trans. Power Syst., vol. 3, pp. 753–
762, May 1988.
[11] G. X. Luo and A. Semlyen, “Efficient load flow for large weakly
meshed networks,” IEEE Trans. Power Syst., vol. 5, pp. 1309–1316,
Nov. 1990.
[12] C. S. Cheng and D. Shirmohammadi, “A three-phase power flowmethod
for real-time distribution system analysis,” IEEE Trans. Power Syst.,
vol. 10, pp. 671–679, May 1995.
[13] R. D. Zimmerman and H. D. Chiang, “Fast decoupled power flow for
unbalanced radial distribution systems,” IEEE Trans. Power Syst., vol.
10, pp. 2045–2052, Nov. 1995.
[14] W. M. Kersting and L. Willis, “Radial Distribution Test Systems, IEEE
Trans. Power Syst.”, vol. 6, IEEE Distribution PlanningWorking Group
Rep., Aug. 1991.

International Scholarly and Scientific Research & Innovation 3(1) 2009 5 scholar.waset.org/1999.5/14357