World Academy of Science, Engineering and Technology 39 2010

A Simple Approach of Three phase Distribution

System Modeling for Power Flow Calculations
J. B. V. Subrahmanyam, C. Radhakrishna

systems need to be analyzed on a three phase basis instead of

Abstract—This paper presents a simple three phase power flow single phase basis.
method for solution of three-phase unbalanced radial distribution There have been a lot of interests in the area of three phase
system (RDN) with voltage dependent loads. It solves a simple distribution load flows. A fast decoupled power flow method
algebraic recursive expression of voltage magnitude, and all the data has been proposed in [4]. This method orders the laterals
are stored in vector form. The algorithm uses basic principles of instead of buses into layers, thus reducing the problem size to
circuit theory and can be easily understood. Mutual coupling between
the number of laterals. Using of lateral variables instead of
the phases has been included in the mathematical model. The
proposed algorithm has been tested with several unbalanced radial
bus variables makes this method more efficient for a given
distribution networks and the results are presented in the article. 8- system topology, but it may add some difficulties if the
bus and IEEE 13 bus unbalanced radial distribution system results network topology is changed regularly, which is common in
are in agreements with the literature and show that the proposed distribution systems because of switching operations. In [5], a
model is valid and reliable. method for solving unbalanced radial distribution systems
based on the Newton-Raphson method has been proposed.
Keywords—radial distribution networks, load flow, circuit Thukaram et al. [6] have proposed a method for solving three-
model, three-phase four-wire, unbalance. phase radial distribution networks. This method uses the
forward and backward propagation to calculate branch
I. INTRODUCTION currents and bus voltages. A new three-phase decoupled
OAD flow technique is very important tool for analysis of power flow method has been proposed in [7]. This decoupled
L power systems and used in operational as well as planning
stages. Certain applications, particularly in distribution
power flow method method uses traditional Newton-Raphson
algorithm in a rectangular coordinate system.
automation and optimization require repeated load flow In recent years the three-phase current injection method
solutions. As the power distribution networks become more (TCIM) has been proposed [8]. TCIM is based on the current
and more complex, there is a higher demand for efficient and injection equations written in rectangular coordinates and is a
reliable system operation. Consequently, the most important full Newton method. As such, it presents quadratic
system analysis tool, load flow studies, must have the convergence properties and convergence is obtained for all
capability to handle various system configurations with but some extremely ill-conditioned cases. Further TCIM
adequate accuracy and speed. In many cases, it is observed developments led to the representation of control devices [9],
that the radial distribution systems are unbalanced because of [10]. Miu et al., [11] have also proposed method for solving
single-phase, two-phase and three-phase loads. Thus, load three-phase radial distribution networks.
flow solution for unbalanced case and, hence special treatment A fast decoupled G-matrix method for power flow, based
is required for solving such networks. on equivalent current injections, has been proposed in [12]
Due to the high R/X ratios and unbalanced operation in .This method uses a constant Jacobian matrix which needs to
distribution systems, the Newton-Raphson and ordinary Fast be inverted only once. However, the Jacobian matrix is
Decoupled Load Flow method may provide inaccurate results formed by omitting the reactance of the distribution lines with
and may not be converged. Therefore, conventional load flow the assumption that R>>X; and fails if X>R. In [14], a method
methods cannot be directly applied to distribution systems. In has been suggested for three phase power flow analysis in
many cases, the radial distribution systems include distribution networks by combining the implicit Z-bus method
untransposed lines which are unbalanced because of single [13] and the Gauss-Seidel method. This method uses
phase, two phase and three phase loads. Thus, load flow fractional factorization of Y-bus matrix. Thus, large
analysis of balanced radial distribution systems [1-3] will be computational time is necessary for this method. The Network
inefficient to solve the unbalanced cases and the distribution Topology method uses two matrices, viz. bus injection to
branch current (BIBC) and branch-current to bus-voltage
(BCBV) matrices, to find out the solution [15]. The Ladder
J. B. V. Subrahmanyam is with the TRR Engineering College, Hyderabad Network theory [16], [17] based on approaches trace the
500 059, Andhra Pradesh, India (corresponding author to provide phone: +91-
network to and fro from its load end to source end. However,
96761-28777; e-mail: [email protected]).
methods proposed by researchers reviewed above are very
C. Radhakrishna is with Global Energy Consulting Engineers,Hyderabad cumbersome and large computation time is required.
500034, Andhra Pradesh, India (e-mail: [email protected]).

770
 World Academy of Science, Engineering and Technology 39 2010

In this article, a simple algorithm is developed which is ⎡V pa ⎤ ⎡V qa ⎤ ⎡ z aa pq z ab ac an

pq z pq z pq
⎤ ⎡ I pq
a
⎤
based on basic systems analysis method and circuit theory. ⎢ ⎥ ⎢ ⎥ ⎢ ⎥⎢ ⎥
The purpose of this paper is to develop a new computation ⎢V p ⎥ ⎢V q ⎥ ⎢ z pq
b b ba
z pq z pq z pq ⎥ ⎢ I pq ⎥
bb bc bn b

model for unbalanced radial distribution network, which ⎢ c ⎥ = ⎢ c ⎥ + ⎢ ca ⎥⎢ c ⎥ (1)

requires lesser computer memory and is computationally fast. ⎢V p ⎥ ⎢V q ⎥ ⎢ z pq z cb cc cn
pq z pq z pq ⎥ ⎢ I pq ⎥
⎢ n ⎥ ⎢ n ⎥ ⎢ na ⎥⎢ n ⎥
The proposed method involves only recursive algebraic ⎢⎣V p ⎥⎦ ⎢⎣V q ⎥⎦ ⎢⎣ z pq z nb nc nn
pq z pq z pq ⎥
⎦ ⎢⎣ I pq ⎥⎦
equations to be solved to get the following information:
• Status of the feeder line, and overloading of the
conductor and feeder line currents. Also representable in matrix form as
• Whether the system can maintain adequate voltage
level for the remote loads. ⎡V pabc ⎤ ⎡Vqabc ⎤ ⎡ Z abc
pq z npq ⎤ ⎡ I abc ⎤
⎢ n ⎥=⎢ n ⎥+⎢ T ⎥⎢
pq
• The line losses in each segment. ⎥ (2)
V V ⎢ n nn
⎥
⎣⎢ p ⎦⎥ ⎣⎢ q ⎦⎥ ⎣ z pq z pq ⎦ ⎣⎢ pq ⎦⎥
I n
• It can also suggest the necessity of re-routing or
network reconfiguration for the existing distribution If the neutral is grounded, the voltage V pn and Vqn can be
network.
The algorithm has been developed considering that all loads considered to be equal. In case, from the lst row of eqn. (2), it
draw constant power. However, the algorithm can easily is possible to obtain
n −1 T
accommodate composite load modeling, if the composition of I pq = − z nn n abc
pq z pq I pq (3)
load is known. The algorithm has good convergence property
and substituting eqn.(3) into eqn. (2), the final form
for practical unbalanced radial distribution networks.
corresponding to the Kron’s reduction becomes
II. SYSTEM MODELING V pabc = Vqabc + Ze abc abc
pq I pq
(4)

For the purposes of power flow studies, we model a radial Where

distribution system as a network of buses connected by ⎡ ze aa ab ac
pq ze pq ze pq
⎤
distribution lines, switches, or transformers to a voltage −1 T ⎢ ba bb bc ⎥
pq = Z pq − z pq z pq z pq = ⎢ ze pq ze pq ze pq ⎥
nn (5)
Ze abc abc n n
specified source bus. Each bus may also have a corresponding
⎢ ca cb cc ⎥
load, shunt capacitor, and or cogenerator connected to it. The ⎢⎣ ze pq ze pq ze pq ⎥⎦
model can be represented by a radial interconnection of copies abc
of the basic building block shown in Figure 2. The dotted I pq is the current vector through line between bus p and q,
lines from the cogenerator, shunt capacitor, and load to can be equal to, the sum of the load currents of all the buses
ground are to indicate that these elements may be connected in beyond line between bus p and q plus the sum of the charging
an ungrounded delta-configuration. Since a given branch may currents of all the buses beyond line between bus p and q, of
be single-phase, two phase, or three-phase, each of the labeled each phase.
quantities is respectively a scalar, 2 x 1, or 3 x 1 complex Therefore the bus q voltage can be computed when we
vector. For the simplicity of presentation we will occasionally know the bus p voltage, mathematically, by rewriting eqn. (4)
assume everything is three-phase, although both single and ⎡Vqa ⎤ ⎡V pa ⎤ ⎡ ze aa ab ac
⎤ ⎡ I pq
a
⎤
pq ze pq ze pq
two phase laterals are handled by our program. ⎢ ⎥ ⎢ ⎥ ⎢ ⎥⎢ ⎥
pq ze pq ze pq ⎥ ⎢ I pq ⎥
⎢Vqb ⎥ = ⎢V pb ⎥ − ⎢ ze ba bb bc b
(6)
A. Distribution system line Model
⎢ c ⎥ ⎢ c ⎥ ⎢ ca cb cc
⎥ ⎢ c
⎥
⎢⎣Vq ⎥⎦ ⎢⎣V p ⎥⎦ ⎢⎣ ze pq ze pq ze pq ⎥⎦ ⎢⎣ I pq ⎥⎦

B. Load Model
The loads generally available in the three phase unbalanced
distribution systems are spot and distributed loads. All the
loads are assumed to draw complex power
( )
SL q = PLq + jQL q . It is further assumed that all three-phase
loads are star and delta connected and all double- and single-
phase loads are connected between line and neutral and line to
Fig. 1 Model of the three-phase four-wire distribution line line respectively.
Figs. 2 and 3 show the three phase unbalanced spot load
For instance a four-wire grounded wye overhead model of star and delta connected three-phase loads at bus q,
distribution line shown in fig. 1 results in a 4×4 impedance
matrix. The corresponding equations are
SLaq , SLbq and SLcq can be of different values or even zeroes.
In fact, two-phase and single-phase loads are modeled by
setting the values of the complex power of the non-existing
phases to zero.

771
 World Academy of Science, Engineering and Technology 39 2010

C. Line Shunt charge model

Fig. 2 Star Load Model Fig. 3 Delta Load Model

In the case of three phase loads connected in start or single

phase loads connected line to neutral, the load current
injections at the qth bus can be given by: Fig. 4 Shunt capacitance of line sections
⎡ ⎛ Vqa ⎞
n
⎤ ⎡ ⎛ Vqa ⎞
n
⎤
⎢ SLa
⎢ q0
( )⎜⎜ ⎟
a ⎟
⎥
⎥ ⎢
(
⎢ PLa + jQLa
q0 q0 )⎜⎜ ⎟
a ⎟
⎥
⎥
The previous line section model can be improved by the
⎢ ⎝ V0 ⎠ ⎥ ⎢ ⎝ V0 ⎠ ⎥ inclusion of line charging representation. The shunt
a
(
⎡ SLq ⎤ ⎡ PLq + jQL q ⎤ ⎢
a a
) ⎛ Vqb
n⎥
⎞ ⎥ ⎢ ⎛ Vqb
n⎥
⎞ ⎥
capacitances phase to phase and phase to ground, depicted in
⎢ b⎥ ⎢ b ⎥
( )
⎢ b
⎢ SLq ⎥ = ⎢ PLq + jQL q ⎥ = ⎢ SLq 0
b
( )⎜⎜ b
⎟ =
⎟ ⎥
(
⎢ b + b
⎢ PLq 0 jQL q 0 )⎜⎜ b
⎟
⎟ ⎥
Fig. 4, can be taken into account through additional current
⎢ c⎥ ⎢ c
( ⎥ ⎢
) ⎝V 0 ⎠ ⎥ ⎢ ⎝ V0 ⎠ ⎥ injections.
⎢⎣ SLq ⎦⎥ ⎣⎢ PLq + jQL q ⎦⎥ ⎢
c
n⎥ ⎢ n⎥ These current injections for representing line charging,
⎢ c ⎛ Vqc ⎞ ⎥ ⎢ c ⎛ Vqc ⎞ ⎥
⎢ SLq 0 ( )⎜⎜ c
⎟
⎟ ⎥
(
⎢ PLq 0 + jQL q 0
c
)⎜⎜ c
⎟
⎟ ⎥
which should be added to the respective compensation current
⎢⎣ ⎝ V0 ⎠ ⎥⎦ ⎢⎣ ⎝ V0 ⎠ ⎥⎦ injections at buses p and q, are given by
⎡Ishqa ⎤ (
⎡− yaa ab ac
pq + y pq + y pq) yab
pq yac
pq
⎤ ⎡Vqa ⎤
⎡⎛ SL a ⎞ ⎤ ⎡⎛ PL aq + jQL aq
∗
⎞ ⎤
∗
⎢ ⎥ 1⎢ ⎥ ⎢ ⎥ (9)
⎢⎜ q ⎟ ⎥ ⎢⎜ ⎟ ⎥ ⎢Ishq ⎥ = ⎢
b ba
y pq (
− y pq + y pq + ybc
ba bb
pq ) ybc
pq ⎥ ⎢Vqb ⎥
⎢⎜⎝ V qa ⎟ ⎥ ⎢⎜ V qa ⎟ ⎥ ⎢ c⎥ 2⎢ ⎥ ⎢ c⎥
⎡ Il qa ⎤ ⎢
⎠
∗
⎥ ⎢
⎝ ⎠
∗
⎥ (7) ⎣⎢ q ⎦⎥
Ish ⎣⎢ y ca
pq y cb
pq (
− yca cb cc
pq + y pq + y pq ) ⎦⎥ ⎣⎢Vq ⎦⎥
⎢ b ⎥ ⎢⎛ SL bq ⎞ ⎥ ⎢⎛ PL bq + jQL bq ⎞ ⎥
⎢ Il q ⎥ = ⎢⎜⎜ b ⎟ ⎥ = ⎢⎜
⎟ ⎥ ⎢⎜
⎟ ⎥
⎟ ⎥
⎢ c ⎥ ⎢⎝ V q ⎠ ⎝ V qb ⎠
⎢⎣ Il q ⎦⎥ ⎢ ∗
⎥ ⎢ ∗
⎥ D. Line Shunt charge model
⎢⎛ SL cq ⎞ ⎥ ⎢⎛ PL cq + jQL cq ⎞ ⎥
⎢⎜⎜ c ⎟ ⎥ ⎢⎜ ⎟ ⎥ Fig. 5 shows phase a of a three-phase system where lines
⎢⎣⎝ V q ⎟ ⎥ ⎢⎜ V qc ⎟ ⎥
⎠ ⎦ ⎣⎝ ⎠ ⎦ between buses p and q feed the bus q and all the other lines
connecting bus q draw current from line between bus p and q.
The current injections at the qth bus for three phase loads
connected in delta or single phase loads connected line to line
can be expressed by:
⎡ ⎛ Vqab ⎞ ⎤
n
⎡ ⎛ Vqab ⎞ ⎤
n

(
⎢ SLab
⎢ q0
)⎜⎜ ⎟ ⎥
ab ⎟ ⎥ ⎢
(
⎢ PLab + jQLab
q0 q0 )⎜⎜ ⎟ ⎥
ab ⎟ ⎥

⎢ ⎝ V0 ⎠ ⎥ ⎢ ⎝ V0 ⎠ ⎥
⎡ SLab
q
⎤ ⎡ PLab(
q + jQLq
ab
) ⎤ ⎢
⎛ Vqbc ⎞ ⎥⎥
n ⎢ ⎛ Vqbc ⎞ ⎥⎥
n
⎢ bc ⎥ ⎢ bc ⎥ ⎢ bc
(
⎢ SLq ⎥ = ⎢ PLq + jQLq
ba
) (
⎥ = ⎢ SLq 0 ) ⎜ ⎟ =
⎜ V0bc ⎟ ⎥
(
⎢ bc
⎢ PLq 0 + jQLq 0
ba
) ⎜ ⎟
⎜ V0bc ⎟ ⎥
⎢ ca ⎥ ⎢ ca ⎥ ⎢
(
⎣⎢ SLq ⎦⎥ ⎣⎢ PLq + jQLq
ca
) ⎦⎥ ⎢
⎝ ⎠ ⎥
n⎥
⎢
⎢
⎝ ⎠ ⎥
n⎥
⎢ ca ⎛ Vqca ⎞ ⎥ ⎢ ca ⎛ Vqca ⎞ ⎥
(
⎢ SLq 0 ) ⎜ ⎟
⎜ V0ca ⎟ ⎥
(
⎢ PLq 0 + jQLq 0
ca
) ⎜ ⎟
⎜ V0ca ⎟ ⎥ Fig. 5 Single phase line section with load connected at bus q between
⎣⎢ ⎝ ⎠ ⎦⎥ ⎣⎢ ⎝ ⎠ ⎦⎥
to phase a and neutral n
⎡⎛ SL ab ⎞ ⎤ ⎡⎛ PL ab ⎞ ⎛ PLcaq + jQL caq ⎞ ⎤
∗ ∗ * *
⎞ ⎛ SLcaq q + jQL q
ab
⎢⎜ q ⎟ −⎜ ⎟ ⎥ ⎢⎜ ⎟ −⎜ ⎟ ⎥
⎢⎜⎝ V qab ⎟ ⎜ V ca
⎠ ⎝ q
⎟ ⎥ ⎢⎜
⎠ ⎝ Vq ab ⎟ ⎜
⎠ ⎝ V qca ⎟ ⎥
⎠
⎡ Il qa ⎤ ⎢ ∗
⎥ ⎢ ⎥
ab ∗ ⎥ ⎢⎛ PL q + jQL q ⎞ ⎛ PL q + jQL q ⎞ ⎥
⎢ b ⎥ ⎢⎛ SLbcq
* *
⎞ ⎛ SL q ⎞ bc bc ab ab

⎢ Il q ⎥ = ⎢⎜⎜ bc ⎟ −⎜ ⎟ ⎥ = ⎢⎜
⎟ ⎜ V ab ⎟ ⎥ ⎢⎜
⎟ −⎜
⎟ ⎜
⎟ ⎥
⎟ ⎥
⎢ c ⎥ ⎢⎝ V q ⎠ ⎝ q ⎠ ⎝ V qbc ⎠ ⎝ V qab ⎠
⎢⎣ Il q ⎦⎥ ⎢ ∗ ∗
⎥ ⎢ *
⎥
bc *
⎢⎛ SLcaq ⎞ ⎛ SL ⎞ ⎥ ⎢⎛ PL q + jQL q ⎞ ⎛ PL q + jQL q ⎞ ⎥
bc ca ca bc

⎢⎜⎜ ca ⎟ − ⎜ q ⎟ ⎥ ⎢⎜
⎟ ⎜ V bc ⎟ ⎥ ⎢⎜
⎟ −⎜
⎟ ⎜
⎟ ⎥
⎟ ⎥
⎢⎣⎝ V q ⎠ ⎝ q ⎠ ⎦ ⎣⎝ V qca ⎠ ⎝ V qbc ⎠ ⎦

(8)
Eqns. (7) and (8) represents a generalized model for star and
delta load models respectively. Where n is defined as follows: Fig. 6 An eight bus system
n=0, constant power
n=1, constant current Consider the eight bus three-phase radial distribution
n=2, constant impedance system [14] shown in Fig. 6. The total line current supplied
through the phase a of the line connected between buses 1 and
2 or effective current of phase a at bus 2 is

772
 World Academy of Science, Engineering and Technology 39 2010

I 12a = Il 2a + Il3a + Ish2a + Ish3a (10)

Thus, in general, the line current at any phase of line between
buses p and q may be expressed as
⎡ N ( pq ) N ( pq )
⎤
⎢ ∑ Il{IE ( pq ,i )} + ∑ Ish{IE ( pq ,i )} ⎥
a a

⎡ I pq ⎤ ⎢ i =1
a i =1
⎥
⎢ ⎥ ⎢ N ( pq ) N ( pq )
⎥ (11)
I pq = ⎢ I pq ⎥ = ⎢ ∑ Il{IE ( pq ,i )} + ∑ Ish{IE ( pq ,i )} ⎥
abc b b b

⎢ c ⎥ ⎢ i =1 i =1 ⎥
⎣⎢ I pq ⎦⎥ ⎢ N ( pq ) c N ( pq ) ⎥
⎢ ∑ Il{IE ( pq ,i )} + ∑ Ish{IE ( pq ,i )} ⎥
c

⎣ i =1 i =1 ⎦
E. Line Shunt charge model
Eqn. (11) provides a method to compute the currents
through the three phase of the branch between buses p and q.
Power fed into the phase a of line between buses p and q at
( ) . Power fed into the phase a of line
bus p is V pa ⋅ I pq
a ∗

between buses p and q at bus q is V ⋅ (I ) . Therefore real

q
a a ∗
qp
and reactive power losses in the line between buses p and q
may be written as:
⎡ LS pq
a
⎤ ⎡ LPpq
a a
+ jLQ pq
a
( ) a ∗ a
( )
⎤ ⎡⎢V p ⋅ I pq − Vq ⋅ I qp ⎤⎥
a ∗

⎢ ⎥ ⎢ ⎥
⎢ LS bpq ⎥ = ⎢ LPpq
b
( ) ∗
( )
+ jLQ bpq ⎥ = ⎢⎢V pb ⋅ I bpq − Vqb ⋅ I qp
b ∗⎥
⎥
(12)
⎢ c ⎥ ⎢ c ⎥
c

⎣
( ) ∗
( )
⎢⎣ LS pq ⎥⎦ ⎢⎣ LPpq + jLQ pq ⎥⎦ ⎢⎢V pb ⋅ I bpq − Vqb ⋅ I qpb ∗⎥
⎥⎦ Fig. 7 Flow chart to identify the buses and branches beyond a
particular bus
III. FLOW CHART FOR THREE PHASE LOAD FLOW
The complete algorithm is presented in the flow charts are
shown in Figs. 7 and 8. Fig. 7 shows the algorithm to identify
the buses and branches beyond any one particular bus. Fig.8
shows the algorithm for load flow solution. In every iteration,
the following steps are followed. In Fig. 5, only one line
connecting the bus q to the substation bus p feeds the bus q.
The total line current supplied through this line to bus q is
determined using eqn. (11).
With the knowledge of current flowing between buses p
and q or at the qth bus, from eqn. (11), the proposed algorithm
computes the voltage at receiving end bus q by using eqn. (4).
In this method, the algorithm computes the voltages at all the
buses of the system starting from the substation to all the
buses downstream. The algorithm stops if the change in the
computed bus voltage magnitudes in successive iterations is
within tolerance limit (i.e. IT ≥ ITMAX).

Fig. 8 Flow chart for Load Flow solution

773
 World Academy of Science, Engineering and Technology 39 2010

From the above discussion, it is observed that the number

IV. RESULTS AND ANALYSIS of iterations for low tolerance and time of execution by the
The effectiveness of the proposed method has been proposed method is superior when compared with the existing
explained with two unbalanced radial distribution systems. method.

A. Case Study 1: 8-bus URDS B. Case Study I1: 37-bus IEEE URDS
A sample 8 bus unbalanced radial distribution network is Short and relatively highly loaded for a 4.16 kV feeder is
shown in Fig. 7 has been taken from the Taiwan Power very small and yet displays some very interesting
Corporation [14]. The base values of the system are 14.4 kV characteristics.. One substation voltage regulator consisting
and 100 kVA. The convergence tolerance specified is 0.001 of three single-phase units connected in wye, overhead and
p.u. The converged solutions (voltage magnitudes and phase underground lines with variety of, phasing Shunt capacitor
angles) are given in Table 1. banks, in-line transformer and unbalanced spot and distributed
loads. For a small feeder this will provide a good test for the
TABLE I
VOLTAGES AND ANGLE OF THE 8-BUS UNBALANCED RDS most common features of distribution analysis software. The
Forward backward line data and load data of the system are given in [20-21]. It is
Proposed Method
Bus sweep method [6] assumed that the transformer at the substation is balanced,
No |V| Angle |V| Angle voltage regulators and capacitors at various buses is neglected.
(p.u.) (deg.) (p.u.) (deg.)
1a 1.0000 0.00 1.0000 0.00
For the load flow, base voltage and base MVA are chosen as
1b 1.0000 -120.00 1.0000 -120.00 4.16 kV and 100 MVA respectively. The load flow results are
1c 1.0000 120.00 1.0000 120.00 presented in Table 3.
2a 0.9830 0.18 0.9830 0.18
2b 0.9714 -119.76 0.9714 -119.76 TABLE III
2c 0.9745 119.97 0.9745 119.97 TEST RESULTS OF THE IEEE 13-BUS UNBALANCED RDS
3a 0.9822 0.18 0.9823 0.19 |Va| ∠Va |Vb| ∠Vb |Va| ∠Va
Bus
4b 0.9655 -119.73 0.9655 -119.73 p.u. deg. p.u. deg. p.u. deg.
4c 0.9716 119.93 0.9717 119.94 1 1.00000 0.00 1.00000 -120.00 1.00000 120.00
5b 0.9643 -119.74 0.9644 -119.74 2 0.95376 -2.12 0.97153 -122.63 0.94217 117.46
6c 0.9697 119.92 0.9697 119.92 3 0.92698 -5.24 0.97167 -122.77 0.87823 115.04
7c 0.9731 119.96 0.9731 119.96 4 0.92698 -5.24 0.97167 -122.77 0.87823 115.04
8c 0.9719 119.95 0.9719 119.95 5 0.95064 -2.20 0.96953 -122.68 0.93937 117.45
6 0.95064 -2.20 0.96953 -122.68 0.93937 117.45
Table 1 shows comparison of the results obtained by 7 - - 0.95528 -123.27 0.94730 117.42
Forward backward sweep method [6] and proposed method. 8 - - 0.94965 -123.62 0.94944 117.43
9 0.92698 -5.24 0.97167 -122.77 0.87823 115.04
For proposed method, the maximum deviation of voltage and 10 0.91870 -5.41 0.97289 -122.86 0.87424 115.15
its phase angle from the Forward backward sweep method is 11 0.92527 -5.29 - - 0.87492 115.01
0.0001 p.u and 0.01 deg.. Thus, the two discussed methods are 12 - - - - 0.87163 114.93
quite accurate. The minimum voltages are obtained by both 13 0.92005 -5.22 - - - -
the methods are highlighted in Table 1. The total system losses were found to be the following in
For both the methods, load flow converged in 2 iterations each phase of the radial system:
for the tolerance of 0.001 p.u.. When the tolerance limit is set • Phase A: 34.70 kW, 150.49 kVAr
as 0.0001, the number of iterations required for the • Phase B: 18.67 kW, 87.26 kVAr
convergence is 3 for Forward backward sweep method and 2 • Phase C: 95.90 kW, 197.25 kVAr
for proposed method. The summary of test results is given in
Table 2. V. CONCLUSIONS
The execution time is 0.048 seconds for the Forward
In this paper, a simple and efficient computer algorithm has
backward sweep method and 0.016 seconds for the proposed
been presented to solve unbalanced radial distribution
method on P-IV computer with 1.6 GHz frequency and 128
networks. The proposed method has good convergence
MB RAM.
TABLE II
property for any practical distribution networks with practical
SUMMARY OF TEST RESULT OF 8 BUS UNBALANCED RADIAL DISTRIBUTION R/X ratio. Computationally, this method is extremely efficient,
NETWORK as it solves simple algebraic recursive equations for voltage
Tolerance Tolerance phasors. Another advantage of the proposed method is all the
Load Flow Method
0.001 0.0001
data is stored in vector form, thus saving enormous amount of
Forward backward sweep Method
2 3 computer memory when tested for large systems. The
[6]
Proposed Method 2 2 proposed algorithm can be used effectively with Supervisory
Control and Data Acquisition (SCADA) and Distribution
Automation and Control (DAC) as the algorithm quickly gets
the voltage solution and efficient operation of the system.

774
 World Academy of Science, Engineering and Technology 39 2010

REFERENCES [14] J. H. Teng, 2002. A Modified gauss-seidel algorithm of three–phase

power flow analysis in distribution network, Electrical Power and
[1] D. Das, D. P. Kothari, and A. Kalam, 1995. Simple and efficient Energy Systems, 24(2):97-102.
method for load flow solution of radial distribution networks, [15] J. H. Teng, 2000. A Network-Topology-based Three-Phase Load
Electrical Power and Energy Systems, 17(5):335-346. Power Flow for Distribution Systems, Proceedings of National
[2] S. Ghosh, D. Das, 1999. Method for load flow solution of radial Science Council ROC (A), 24(4):259-264.
distribution networks, IEE Proc. Generation, Transmission and [16] L. L. Grgsby: Editor-in-Chief, The Electric Power Engineering
Distribution, 146(6):641-648. Handbook, CRC Press 1999, pages 6-53 to 6-62.
[3] P. Aravindhababu, 2003. A new fast decoupled power flow method [17] W. H. Kersting, Distribution System Modeling and Analysis, CRC
for distribution systems, Electric Power Components and Systems, Press 2002, chapter 9.
31(9):869-878. [18] J. Nagarath and D. P. Kothari, Power system Engineering, New
[4] R. D. Zimmerman and H. D. Chiang, 1995. Fast decoupled power Delhi: Tata McGraw Hill Publications, 2nd edition, 1998.
flow for unbalanced radial distribution systems, IEEE-PES Winter [19] D. Shirmohammadi and Carol S. Cheng, 1995. A Three phase power
Meeting, paper no. 95, New York. flow method for real time distribution system analysis, IEEE Trans.
[5] S. K. Goswami and S. K. Basu, 1991. Direct solution of distribution on Power System, 10(2):671-679.
systems, IEE Proc. C, 188(1):78-88. [20] IEEE Distribution System Analysis Subcommittee, 1991. Radial
[6] D. Thukaram, H. M. Wijekoon Banda, and J. Jerome, 1999. A Robust distribution test feeders, IEEE Trans. on Power Systems, 6(3):975-
three phase power flow algorithm for radial distribution systems, 985.
Electric Power System Research, 50(3):227-236. [21] Radial Distribution test feeders –
[7] W. M. Lin, Y. S. Su, H. C. Chin, and J. H. Teng, 1999. Three-Phase web:http://www.ewh.ieee.org/soc/pes/dsacom/testfeeders.html.
unbalanced distribution power flow solutions with minimum data
preparation, IEEE Trans. on Power Systems, 14(3):1178-1183. J. B. V. Subrahmanyam received the B.Tech degree from JNTU Kakinada,
[8] P. A. N. Garcia, J. L. R. Pereira, S. Carnerio, V. M. da Costa, and N. and M.E degree from Jadavpur University Kolkata and presently pursuing
Martins, 2000. Three-Phase power flow calculations using the current Ph.D from JNTU-Hyderabad, India. He has very rich experience in application
injection method, IEEE Trans. on Power Systems , 15( 2):508-514. of latest condition monitoring techniques to reduce industry equipment
[9] P. A. N. Garcia, J. L. R. Pereira, and S. Carneiro, Jr., 2001. Voltage breakdowns & application of modern GPS & GIS technologies to power
control devices models for distribution power flow analysis, IEEE utilities to reduce power losses & manage the power distribution utility
Trans. Power Syst., 16(4):586–594. business effectively. At present he is a professor in the Electrical &
[10] P. A. N. Garcia, J. L. R. Pereira, and S. Carneiro, Jr., 2004. Electronics Engineering Department, TRREC, AP, India. He is actively
Improvements in the representation of PV buses on three-phase involved in the research of planning and optimization of unbalanced power
distribution power flow, IEEE Trans. Power Del., 19(2):894–896. distribution systems. His research interests are computer applications in power
[11] K. N. Mui and H. D. Chiang, 2000. Existence, uniqueness and systems planning, analysis and control.
monotonic properties of the feasible power flow solution for radial
three phase distribution networks, IEEE Trans. Circuit and Syst., C. Radhakrishna has more than 35 years of experience in teaching &
47(10):1502-1514. research in Electrical engineering and published more than 85 papers in
[12] W. M. Lin and J. H. Teng, 2000. Three-Phase distribution networks international, national journals & conferences. He is the recipient of Best
fast decoupled power flow solutions, Electric Power and Energy Teacher award from Govt. of Andhra Pradesh, India and Jawaharlal Birth
Systems, 22(5):375-380. Centenary award from IE(I), India. He was associated with Jawaharlal Nehru
[13] T. H. Chen, M. S. Chen, K. J. Hwang, P. Kotas, and E. A. Chebli, Technological University (JNTU)-Hyderabad for several years in various
1991. Distribution system power flow analysis-a rigid approach, positions and was the founder Director of Academic Staff College-JNTU-
IEEE Trans. on Power Delivery, 6(3):1146-1153. Hyderabad, India.

775