0142 0615 (95) 00050 0 PDF

You might also like

You are on page 1of 12

Electrical Power & Energy Systems, Vol. 17, No. 5, pp.

335-346, 1995
UTTERWORTH
I N E M A N N Copyright 1995 Elsevier Science Ltd
Printed in Great Britain. All rights reserved
0142-0615(95)00050-X 0142-0615/95/$10.00+0.00

Simple and efficient method for


load flow solution of radial
distribution networks
D Das
Department of Electrical and Electronic Engineering,
Birla Institute of Technology and Science, Pilani,
Rajasthan-330 031, India

D P Kothari
Centre for Energy Studies, Indian Instititue of
Technology, Hauz Khas. New Delhi-110 01 6, India

A Kalam
Department of Electrical and Electronic Engineering,
Victoria University of Technology, Footscray,
Melbourne. Australia

fi(m2) voltage angle of node m2


The paper presents a simple and efficient method for solving
LP(j) real power loss o f b r a n c h j
radial distribution networks. The proposed method involves
LQ(j) reactive power loss o f b r a n c h j
only the evaluation of a simple algebraic expression of
voltage magnitudes and no trigonometric functions as IS(j) sending-end node of branch j
IR(j) receiving-end node of branch j
opposed to the stan&~rd load flow case. Thus, com-
PLOSS total real power loss
putationally the proposed method is very effic&nt and it
requires less computer memory. The proposed method can QLOSS total reactive power loss
easily handle different types of load characteristics. Several
Indian rural distribution networks have been successfully I1. I n t r o d u c t i o n
solved by using the proposed method.
Load flow analysis of distribution systems has not
received much attention unlike load flow analysis of
Keywords: distribution loadflow, mathematical techniques,
radial networks transmission systems. However, some work has been
carried out on load flow analysis of a distribution net-
work but the choice of a solution method for a practical
system is often difficult. Generally distribution networks
I. Nomenclature are radial and the R / X ratio is very high.
NB total number of nodes Because of this, distribution networks are ill-conditioned,
LN1 total number of branches (LN 1 = NB- 1) and conventional Newton-Raphson (NR) and fast
PL(i) real power load of ith node decoupled load flow (FDLF) methods 1-4 are inefficient
QL(i) reactive power load of ith node at solving such networks.
Many researchers 5-7 have suggested modified versions
Iv(i)l voltage magnitude of ith node
of the conventional load flow methods for solving
R(j) resistance o f j t h branch
x(j) reactance o f j t h branch ill-conditioned power networks.
I(j) current flowing through branchj Recently researchers have paid much attention to
P(m2) total real power loadTed through node m2 obtaining the solution of distribution networks.
Q(m2) total reactive power load fed through node m2 Kersting and Mendive s and Kersting9 have presented a
load flow technique based on the ladder network theory
Received 17 May 1993; revised 17 November 1993; accepted 4 and it appears to work very well. Shirmohammadi et al.lO
August 1994 have presented a compensation-based power flow

335
336 Method for solution of radial distribution networks. D. Das et al.

method for weakly meshed distribution and transmission


systems. Baran and Wu 11 and Chiang u have obtained the
load flow solution in a distribution system by the iterative
solution of three fundamental equations representing real -~LI4
power, reactive power and voltage magnitude. These
Substation
three equations are very useful as they appear to be
useful in real physical systems. Renato 13 has proposed
one method for obtaining load flow solutions of radial I' ) 5

distribution networks. His technique seems to be quite - -15


promising because it gives a solution for bus voltage
magnitude only. Goswami and Basu 14 have presented a 7 12
direct solution method for solving radial and meshed
distribution networks. However, the main limitation of 13
their method is that no node in the network is the junction
of more than three branches, i.e. one incoming and
Figure 1. Single line diagram of an existing distribution
two outgoing branches. Jasmon and Lee 15'16 have feeder
proposed a new load flow method for obtaining the
solution of radial distribution networks. They have Table 1. Branch number, sending-end and receiving-end
derived the fundamental equations for solving a load nodes in Figure 1
flow problem of a distribution network using a single-line
Branch Sending- Receiving-
equivalent.
number end node end node
In India, all the 11 kV rural distribution feeders are
radial and too long. The voltages at the far end of many
(j) IS(j) IR(j)
such feeders are very low with very high voltage 1 1 2
regulation. 2 2 3
In this paper the main aim of the authors has been to 3 3 4
develop a new load flow technique for radial distribution 4 4 5
networks. The proposed method involves only the 5 2 9
evaluation of a simple algebraic expression of voltage 6 9 l0
magnitude and no trigonometric terms, as opposed to the 7 2 6
standard load flow case. Computationally the proposed 8 6 7
method is very efficient. Another advantage of the 9 6 8
proposed method is that it requires less computer 10 3 ll
memory. Convergence is always guaranteed for any ll 11 12
type of practical radial distribution network with a 12 12 13
realistic R/X ratio while using the proposed method. 13 4 14
Loads, in the present formulation, have been represented 14 4 15
as constant power. However, the proposed method can
easily include composite load modelling if the break up of
the loads is known. The proposed load flow technique has I IVll~l I(I) ,v2,~ 2
_ ......
been implemented on an IBM PC-AT. Several practical
rural radial distribution feeders in India have been
successfully solved using the proposed method. Relative R (I) +j x (I) -~ P (2),Q(2)
speed and memory requirements of the proposed method
have also been compared with the methods proposed by Figure 2. Electrical equivalent of Figure 1
Baran and Wu ~1.
From Figure 2, we have the following equations:
I ( l ) = ]V(1)]16(1) - ] V(Z)]L6(2)
III. A s s u m p t i o n s R(1) +iX(l) (1)
We assume that the three-phase radial distribution net- P ( 2 ) - j Q ( 2 ) = v*(2)I(1) (2)
works are balanced and can be represented by their
equivalent single line diagrams. This assumption is From equations (1) and (2) we have (details in Appendix 1):
quite valid for 11 kV rural distribution feeders in India [V(2)I = {[(P(2)R(1) + Q(Z)X(1) - 0.5] V(1)]2) 2
and elsewhere. Line shunt capacitance (different from
shunt capacitor banks that are considered as loads) is - (R2(1) + xz(1))(p2(2) + Q2(Z))]U2
negligible at the distribution voltage levels as is found in
most practical cases. - (P(Z)R(1) + Q(Z)X(1) - 0.5]v(1)[2)}'/2
(3)
IV. S o l u t i o n m e t h o d o l o g y where P(2) and Q(2) are total real and reactive power
loads fed through node 2.
Figure 1 shows a single-line diagram of an existing
rural distribution feeder. Branch number, sending-end (P2) = sum of the real power loads of all the nodes
and receiving-end nodes of this feeder are given in beyond node 2 plus the real power load of
Table 1. Figure 2 shows the electrical equivalent of node 2 itself plus the sum of the real power
Figure 1. losses of all the branches beyond node 2.
Method for solution of radial distribution networks: D. Das et al. 337

(Q2) = sum of the reactive power loads of all the We will now explain IB(j, ip + 1) and IE(j, ip + 1).
nodes beyond node 2 plus the reactive power Consider the first branch in Figure 1, i.e. j = 1, the
load of node 2 itself plus the sum of the receiving-end node of branch 1 is 2 (see also Table 1),
reactive power losses of all the branches therefore, IB(1,ip+ 1) and IE(1,ip+l) will help to
beyond node 2. identify all the branches and nodes beyond node 2 and
node 2 itself.
Equation (3) can be written in generalized form:
This will help to find the exact load feeding through
IV(m2)l = [B(j) - A(j)] 1/2 node 2. Similarly, consider branch 2, i.e. j = 2, the
(4)
receiving-end node of branch 2 is 3. Therefore,
where IB(2, ip + 1) and IE(2, ip + 1) will identify nodes and
branches beyond node 3 and node 3 itself. This will help
A(j) = P(m2)*R(j) + Q(m2)*X(j) - 0.5"1V(ml)l 2 compute the exact load feeding through node 3. For each
node and branch identification ip will be incremented by 1.
(5) Note here that before identification of nodes and branches
B(j) = {A2(j) - [R2(j) + X2(j)] * beyond a particular node, ip has to be reset to zero.
For j = 1, (first branch in Figure 1, Table 1),
[PZ(m2) + QZ(m2)]}l/2 (6) IR(j) = 2, check whether IR(j)= IS(i) or not for
i = 2 , 3 , . . . ,LNI-1. It is seen that IR(j) = IS(2) = 2;
j is the branch number, rn 1 and m2 are sending-end and IR(j) = IS(5) = 2; IR(j) = IS(7) -- 2; corresponding
receiving-end nodes respectively (ml =IS(j) and receiving-end nodes are I R ( 2 ) = 3; I R ( 5 ) - - 9 and
m2 = IR(j)). IR(7) = 6.
Real and reactive power losses in branch 1 can be given Therefore IB(j, 1)= I, IE(j, 1)=2; IB(j, 2)=2,
by: IE(j, 2)=3; IB(j, 3)=2; IE(j, 3)=9; IB(j,4)=2,
IE(j, 4) = 6.
LP(1) = R(1)*[p2(2) + Q2(2)] We will miss out IB(j, 1), because we want to identify
IV(2)] 2 the nodes and branches which are beyond node IR(j) and
(7) we store the receiving-end node in the name of a variable,
LQ(1) = X(1)*[P2(2) + Q2(2)] say KK(ip), i.e. KK(1) = 2, KK(2) -- 3, KK(3) = 9 and
Iv(2)l 2 KK(4) = 6.
Note that there should be no repetition of any branch
Equation (7) can also be written in generalized form as: or node while identifying nodes and branches and this
logic has been incorporated in the proposed algorithm
L P ( j ) = R(J)*[p2(m2) + Q2(m2)] (Algorithm 1). This is further explained in the flowchart
]V(rn2)I2 given in Figure 3.
(8) From the above discussion it is seen that node 2 is
L Q ( j ) = X(J)*[PZ(m2) + Q2(m2)] connected to nodes 3, 9 and 6 and the corresponding
branches are branch 2, (2 ~ 3), branch 5 (2 ~ 9) and
IV(rnZ)l 2 branch 7 (2---+ 6). Similarly, the proposed logic will
Initially, if LP(j) and LQ(j) are set to zero for all j, then identify the nodes and branches which are connected to
initial estimates of P(m2) and Q(m2) will be the sum of nodes 3, 9 and 6. First, it will check whether node 3
the loads of all the nodes beyond node m2 plus the load of appears in the left-hand column of Table 1. It is seen
the node m2 itself. that node 3 is connected to nodes 4 and 11 (branches 3
and 10 in Figure 1, Table 1). Therefore IB(j, 5 ) = 3,
IE(j, 5) = 4, IB(j, 6) = 3, IE(j, 6) = 11 and KK(5) = 4,
V. Explanation of the proposed technique KK(6) = 11. Then it will check whether node 9 appears in
Before giving the detailed algorithm, we will discuss the the left-hand column of Table 1. It is seen that node 9 is
methodology of identifying the nodes and branches connected to node 10 (branch 6 in Figure 1, Table 1).
beyond a particular node which will help in finding the Therefore, IB(j, 7) = 9, IE(j, 7) = 10, K K ( 7 ) = 10.
exact load feeding through that particular node. Similarly, node 6 is connected to nodes 7 and 8.
First we will define the variables: Therefore, IB(j, 8) = 6, IE(j, 8) = 7, IB(j, 9) = 6,
IE(j, 9) = 8, KK(8) = 7, and KK(9) = 8.
J j = 1 , 2 . . . , L N 1 (j indicates branch of From the above discussion, again it is seen that node 3
Figure 1; see also Table 1) is connected to nodes 4 and 11, node 9 is connected to
node count (identifies number of nodes node 10 and node 6 is connected to nodes 7 and 8.
beyond a particular node) Similarly, the proposed logic will check whether nodes
IK(~) node identifier (helps to identify the send- 4, 11, 10, 7 and 8 are connected to any other nodes. This
ing-end and receiving-end nodes which are process will continue unless all the nodes and branches
given in the ith branch of Table 1 (i > j ) are identified beyond node 2.
LL(ip) stores sending-end node of ith row of Table 2 gives the nodes and branches beyond node 2.
Table 1 (i > j ) Computer logic (Algorithm 1) will automatically skip
KK(ip) stores receiving-end node of ith row of the node and branch behind node 2, i.e. node 1 and
Table 1 (i > j ) branch 1.
N(j) total number of nodes beyond node IR(j) Total load fed through node 2 is the sum of the loads of
plus 1 (node IR(j) itself) all the nodes beyond node 2 plus the load of the node 2
IB(j, ip + 1) sending-end node itself (right-hand column of Table 2) plus the sum of the
IE(j, ip + 1) receiving-end node losses of all the branches beyond node 2 (see Table 2).
338 Method for solution of radial distribution networks. D. Das et al.

Read
Receiving-end & sending-end
nodes and total number o f
nodes

FromQ

I ,,_-o I ip=ip+l
From@ _1 /K (ip)=i
LL (ip)=lS (i)
KK (ip)=/R (i)
From~ /E q, ip+l)=lR (i)
IB (j, ip+l)=/S (i)
N (j)=ip+ I

From

tE (j,,p+l)=1Rq) Yes
In clip+ 1)=/s q)
No

Yos Yes
m q, it,+l)=m q)]
ta q, ip+~)=mq)l
Y ~ = INo U (j)=ip+l [
IR(j)=KK(iq)
K=lK(iq+ 1)
I ,n--,." I
Yes ~- Yes

No I Yes ~ No
[IE (LN1, I)=IR (LN1)
(~) Im (t~vl, l)=m (t.m)
I N(LNI)=-!

Figure 3. Flowchart for identification of nodes and branches beyond a particular node

Similarly, we have to consider the receiving-end nodes The concept of identifying the nodes and branches
of branch 2, branch 3, branch 4 , . . . , branch L N I - 1 in beyond a particular node which helps in computing the
Figure 1, and in a similar way to that discussed above, the exact load feeding through the particular node has been
nodes and branches have to be identified beyond these realized using an algorithm (Algorithm 1) given below.
receiving-end nodes Table 3 and 4 also give the nodes
and branches beyond nodes 3 and 4.
Note that if the receiving-end node of any branch in VI. A l g o r i t h m 1" Identification of nodes
Figure 1 (see also Table 1) is an end node of a particular and branches beyond a particular node
lateral, then the total load fed through this node is the Step 1 read system data
load of this node itself For example, consider node 5 in Step 2 j = 1
Figure 1 (branch 4, Table 1). This is an end node, Step3 k=j+l
therefore the total load fed through node 5 is the load Step4 ip=O;iq=O
of node 5 only. Similarly nodes 7, 8, 10, 13, 14 and 15 are Step 5 i= k
end nodes in Figure 1. The proposed computer logic will Step 6 nc = 0
automatically identify all the end nodes if { I R ( j )
= I S ( i ) } go to step 7
Tables 2, 3 and 4 give the pattern of computer output otherwise go to step 15
For any end node, the computer output will show only Step 7 if {ip = 0} go to step 13
this end node. otherwise go to step 8
Method for solution of radial distribution networks: D. Das et al. 339

Table 2. Nodes and branches beyond node 2 in Figure 1 Step 14 ip = ip + 1


IK(ip) = i
Sending- Receiving- LL(ip) = IS(i)
end nodes end nodes KK(ip) = IR(i)
Node behind IE(j, ip + 1) = IR(i)
node 2 has IB(j, ip + 1) = IS(i)
been skipped U ( j ) = ip + 1
(node 1) 2 Step 15 i=i+l
if{i<~LN1} go to step 6
Branches 2 3 otherwise go to step 16
beyond 3 4 Step 16 if{ip = 0} go to step 17
node 2 4 5 otherwise go to step 18
2 9 Step 17 IE(j, ip + 1) = I R ( j )
9 10 IB(j, ip + 1) = I S ( j )
2 6 N ( j ) = ip + 1
6 7 go to step 20
6 8 Step 18 iq=iq+l
3 11 if {iq > ip} go to step 20
11 12 otherwise go to step 19
12 13 Step 19 I R ( j ) = KK(iq)
4 14 k = IK(iq) + 1
4 15 if {iq<~ip} go to step 5
Step 20 j=j+l
if {j~<LN1 - 1} go to step 3
Table 3. Nodes and branches beyond node 3 in Figure 1 otherwise go to step 21
Sending- Receiving- Step 21 IE(LN1, 1) = IR(LN1)
end nodes end nodes IB(LN1, l) = IS(LN1)
N(LN1) = 1
Nodes behind Step 22 stop
node 3 have
been skipped 3
VII. Load f l o w calculation
Branches 3 --+ 4 Once all the nodes and branches are identified, it is very
beyond 3 ~ 11 easy to calculate voltage magnitudes of all the nodes by
node 3 4 ~ 14 using equations (4), (5) and (6). In this algorithm, we have
4 ~ 15 to obtain the exact load feeding through all the receiving-
4 ~ 5 end nodes (Table 1) and the voltage magnitudes of all
11 ~ 12 these nodes by using equations (4), (5) and (6), as the
12 ~ 13 voltage magnitude of substation is known (V(1)).
Then we can compute the feeder current, branch
loSses etc.
The convergence criteria of the proposed method is
Table 4. Nodes and branches beyond node 4 in Figure 1 that if, in successive iterations the difference between the
Sending- Receiving- real and reactive power delivered from the substation is
end nodes end nodes less than 0.1kW and 0.1kVAr, the solution has
converged.
Nodes behind Before a detailed load flow algorithm is given, note that:
node 4 have
been skipped 4 I S ( j ) = IB(j, 1) ~ for
allj (9)
I R ( j ) = IE(j, 1) J
Branches 4 ~ 14
beyond 4 ~ 15 The proposed load flow algorithm is given in the form
node 4 4 ~ 5 of a flowchart in Figure 4.

Step 8 in = 1 Villi. A l g o r i t h m 2: Load f l o w calculation


Step 9 i f { I S ( i ) = LL(ip) and IR(i) = KK(ip)} or Step 1 iteration count = I T = l
if {IR(i) = KK(ip)} go to step 10 S L P = 0.0; SLQ = 0.0
otherwise go to step 11 Step 2 P L O S S = 0.0; Q L O S S = 0.0
Stepl0 nc= l Step 3 j = 1
Stepll in = in + l Step4 LK = N(j)
if{in<.~ip} go to step 9 Step 5 i = 1
otherwise go to step 12 Step 6 if {IB(j, i) ~ I S ( j ) } go to step 8
Step 12 if{nc = 1} go to step 15 otherwise go to step 7
otherwise go to step 14 Step7 L I = I B ( j , i )
Step 13 IE(j, ip + 1) "-- I R ( j ) L2 = IE(j, i)
IB(j, ip + l) = I S ( j ) ml = I S ( j )
340 Method for solution of radial distribution networks. D. Das et al.

m2 = I R ( j ) Step 14 (a) compute voltage magnitude by using


P(m2) = PL(L2) equation (4), i.e. compute A ( j ) and B ( j )
Q(m2) = QL(L2) by using equations (5) and (6) then compute
go to step 13 voltage magnitude by using equation (4)
Step 8 L1 : IB(j, i) (b) compute branch real and reactive power
L2 = IE(j, i) losses by using equation (8)
Step 9 in= l (c) P L O S S = P L O S S + L P ( j )
Step 10 if {L1 = IS(in) and L2 = IR(in)} go to step 11 QLOSS : QLOSS + L Q ( j )
otherwise go to step 12 Step 15 j = j + l
Step 11 P(m2) = P(m2) + PL(L2) + LP(in) if {j<~LN1} go to step 4
Q(m2) = Q(m2) + QL(L2) + LQ(in) Step 16 DP = I(PLOSS-SLP)[
Step 12 in = in + l DQ = I(QLOSS-SLQ)I
if {in<<.LN1} go to step 10 if {(DP and DQ) < EPS} go to step 18
otherwise go to step 13 otherwise go to step 17
Step 13 i=i+1 Step 17 I T = I T + 1
if {i<~LK} go to step 6 SLP = P L O S S

I Read
line parameters & load data
t
T_
~- No From E
Iteration count 1 T=I

I
t
Set I ,-.,l I
From D SLP=0.0; SLQ=O.0

Set
PLOSS--O.O
QLOSS--O.O
Compute voltage magnitude by
using eqn. (4) i.e. Compute
A(]) & B(]) by using eqns.
From C [ ~ (5) & (6). Then compute voltage
magnitude by eqn. (4). Also
compute branch real and
I LK='U' [ reactive power losses by
using eqn. (8)
t
From B ,~ PLOSS=PLOSS+LP (j) I
Q~.OSS-QrOSSLQO)

No ] j=j+l ]

L l = m O, i)
L2=IE (j, i)
Ll=/B q, i)
L2=iE (j, i)
m l =/S (j) DP=ABS (PLOSS-SLP)
m2=lR (j) DQ=ABS(QLOSS-SLQ)
P (m2)=PL (L2)
Q (m2)=QL (L2)
Solution has converged
1 Write voltage magnitudes,
power losses, etc.

H'= IT+ 1
SLP=PLOSS
SLQ=QLOSS
P (m2)=P (m2)+PL (L2)+LP(in)
Q (m2)=Q (m2)+QL (L2)+LQ (in)
I
]
i
@
I,=,.ll

Figure 4. Flowchart for load flow calculation


Method for solution of radial distribution networks: D. Das et al. 341

SLQ = QLOSS convergence. The memory requirement of the proposed


go to step 2 method is approximately 65% of Baran and Wu's
Step 18 solution has converged, write voltage magnitudes, method. The proposed method is also 1.93 times faster
power losses, feeder current etc. than that of Baran and Wu.
Step 19 stop It is worth mentioning here that the proposed method
is much faster than Baran and Wu's ll' 12method when the
system is very large.
The solution of this system is given in Table 6. The total
IX. Examples real and reactive power losses of this system are
286.52 kW and 180.52 kVAr respectively.
IX.1 Example 1 Further, it is found that the proposed method is 1.39
In this example, we consider Figure 1, which is an existing times faster than the method in Reference 13 and 4.2
rural distribution feeder. Line data and load data of this times faster than the method in Reference 8. Again, the
system are given in Appendix 2. memory requirement is the same as in References 8 and
It took three iterations to converge by the proposed 13.
method. The proposed method has also been compared
with that of Baran and Wu II. Their method also took
three iterations to converge. The memory requirement of
the proposed method is 63% of Baran and Wu's method. Table 6. Load flow solution of Example 3
The proposed method is 1.37 times faster than that of
Baran and Wu. It is worth mentioning here that to Node Voltage Node Voltage
compute the Jacobian matrix, Baran and Wu's method no. magnitude no. magnitude
needs a series of matrix multiplications and the number of
1 (substation) 1.00000 44 0.88316
these multiplications increases with the increase in size of
2 0.99587 45 0.00213
the radial distribution networks. In addition, Baran and
3 0.98977 46 0.88153
Wu's method needs at least one matrix inversion.
4 0.98197 47 0.88143
However, the proposed method needs no matrix
5 0.97816 48 0.88050
multiplication and inversions.
6 0.96443 49 0.88024
The solution of this system has been given in Table 5.
7 0.95599 50 0.87979
The total real and reactive power losses of this system are
8 0.91950 51 0.87945
61.79 kW and 57.30 kVAr respectively.
9 0.91778 52 0.87818
The proposed method has also been compared with
10 0.91394 53 0.87775
the methods proposed by Renato 13 and Kersting8,
11 0.91122 54 0.87749
respectively. The proposed method is 1.1 times faster
12 0.09897 55 0.87777
than the method proposed in Reference 13 and 2.3
13 0.90799 56 0.88014
times faster than the method proposed in Reference 8.
14 0.90769 57 0.91562
However, the memory requirement is the same as that of
15 0.90751 58 0.90975
the methods given in References 8 and 13.
16 0.99557 59 0.90962
IX.2 Example 2 17 0.98916 60 0.90624
18 0.97580 61 0.90517
In this example, we consider a rural distribution feeder of
19 0.97439 62 0.90444
85 nodes. The line and load data of this system are given
20 0.97370 63 0.90533
in Appendix 3.
21 0.97300 64 0.90184
The proposed method took four iterations to converge.
Baran and W U ' S 11' 12 method also took four iterations for 22 0.97234 65 0.90167
23 0.97427 66 0.90154
24 0.95559 67 0.90008
Table 5. Solution of Example 1 25 0.91410 68 0.89722
26 0.90996 69 0.89510
Node no. IVI 27 0.90453 70 0.89455
28 0.90202 71 0.89430
1 (substation) 1.00000 29 0.89741 72 0.89994
2 0.97128 30 0.89319 73 0.89580
3 0.95667 31 0.89122 74 0.89560
4 0.95090 32 0.88998 75 0.89533
5 0.94992 33 0.88900 76 0.89415
6 0.95823 34 0.88476 77 0.90166
7 0.95601 35 0.88248 78 0.91319
8 0.95695 36 0.88239 79 0.91294
9 0.96797 37 0.90969 80 0.90744
10 0.96690 38 0.90380 81 0.90694
11 0.94995 39 0.89700 82 0.90688
12 0.94583 40 0.88934 83 0.90625
13 0.94452 41 0.88841 84 0.90606
14 0.94861 42 0.88828 85 0.90761
15 0.94844 43 0.88819
342 Method for solution of radial distribution networks. D. Das et al.

X. Additional application of the proposed Newton like method' IEEE Trans. gol PAS-101 (1982)
method 3648 3657
8 Kersting, W H and Mendive, D L 'An application of ladder
X.1 Load modelling network theory to the solution of three phase radial load flow
All loads, including shunt capacitators for reactive power problem" IEEE PES Winter Meeting, New York, January
compensation are represented by their active (P0) and 1976, paper no A 76 044-8
reactive (Q0) components at 1.0 p.u. The effect of voltage 9 Kersting, W H 'A method to teach the design and operation
variation is represented as follows: of a distribution system' IEEE Trans. Vol PAS-103 (1984)
1945 1952
P = Pol vI ~ (10) I0 Shirmohammadi, D, Hong, H W, Semlyen, A and Lno, G X 'A
Q = Q01Vl k (11) compensation-based power flow method for weakly meshed
distribution and transmission networks' IEEE Trans. Vol
where PWRS-3 (1988) 753-762

I vl is the voltage magnitude 11 Baran, M E and Wu, F F 'Optimal sizing of capacitors placed
on a radial distribution system' IEEE Trans. Vol PWRD-2
k = 0 for constant power loads
(1989) 735-743
k = 1 for constant current loads
k = 2 for constant impedance loads 12 Chiang, H D 'A decoupled load flow method for distribution
power network: algorithms, analysis and convergence study'
The value of k differ according to the load Int. J. Electr. Power Energy Syst. Vol 13 No 3 (1991) 130-
characteristics. The load flow solution depends on the 138
type of real and reactive loads. It is extremely easy to 13 Renato, C G 'New method for the analysis of distribution
include a real and reactive power loads representation in networks' IEEE Trans. P W R D Vol 5 No 1 (1990) 391-396
the proposed algorithm. For constant current and
constant impedance loads, real and reactive power 14 Goswami, S K and Basu, S K 'Direct solution of distribution
systems' IEEProc. C. Vol 188 No 1 (1991)78-88
loads have to be computed after every iteration.
15 Jasmon, G B and Lee, L H C C 'Stability of loadflow
techniques for distribution system voltage stability analysis'
XI. C o n c l u s i o n s lEE Proc. C. Vol 138 No 6 (1991)
A simple and efficient load flow technique has been 16 Jasmon, G B and Lee, L H C C 'Distribution network reduction
proposed for solving radial distribution networks. It for voltage stability analysis and load flow calculations' Int. J.
completely exploits the radial feature of the distribution Electr. Power Energy Syst. Vol 13 No 1 (1991) 9-13
network. The proposed method always guarantees
convergence of any type of practical radial distribution
network with a realistic R / X ratio. Computationally,
the proposed method is extremely efficient compared
Appendix 1
with Baran and Wu's method as it solves a simple F r o m Figure 2, we have the following equations:
algebraic expression of voltage magnitude only.
Another advantage of the proposed method is that it I(1) = I V(1)llr(1)- I v(2)1l~(2) (12)
requires less computer memory. The proposed method R(1) + j X ( 1 )
can easily handle the composite loads if the break up of
the loads is known. The proposed method has been and
implemented on an IBM PC-AT. Several Indian rural
distribution networks have been successfully solved by
I(1) - P(2) - jQ(2) (13)
using the proposed load flow technique. v*(2)

F r o m equations (12) and (13) we obtain:


XlI. References
I V(1)ILr(1) - I V(2)ILr(2) _ P(2) - j Q ( 2 )
1 Tinney, W F and Hart, C E 'Power flow solution by Newton's
method' IEEE Trans Vol PAS-86 (1967) 1449-1456 R(1) + j X ( 1 ) V*(2)
2 Stott, B and Alsae, O 'Fast decoupled load flow' IEEE Trans.
Vol PAS-93 (1974) 859-869 therefore
3 Stagg, G W and EI-Abiad, A H Computer methods in power
system analysis McGraw Hill (1968) IV(1)I IV(Z)ILr(1) - t5(2) - I V ( 2 ) I 2
4 Nagrath, I J and Kothari, D P Modern power system analysis = [P(2) - jQ(2)][R(1) +jX(1)]
2nd edn, Tata McGraw Hill, New Delhi (1989)
5 Rajicie, D and Bose, A A modification to the fast decoupled therefore
power flow for networks with high R / X ratios IEEE Trans.
Vol PWRS-3 (1988) 743-746 IV(1)I IV(2)I cos[6(1) - 6(2)] - I V(2)I 2
6 Iwamoto, S and Tamura, Y 'A load flow calculation method
for ill conditioned power systems' IEEE Trans. Vol. PAS- +Jl V(1)I I V(2)I sin[6(1) - 6(2)]
100(1981) 1736 1713 = [P(2)R(1) + Q(2)X(1)]
7 Tripathy, S C, Durgaparasad, G, Malik, O P and Hope, G S
'Load flow solutions for ill conditioned power systems by a + j[P(Z)X(1) - Q(Z)R(1)] (14)
Method for solution of radial distribution networks: D. Das et al. 343

Table A2.1 Line data of Example 1


Branch Sending- Receiving- R X
number end node end node (ohm) (ohm)
1 1 2 1.35309 1.32349
2 2 3 1.17024 1.14464
3 3 4 0.84111 0.82271
4 4 5 1.52348 1.02760
5 2 9 2.01317 1.35790
6 9 10 1.68671 1.13770
7 2 6 2.55727 1.72490
8 6 7 1.08820 0.73400
9 6 8 1.25143 0.84410
10 3 11 1.79553 1.21110
11 11 12 2.44845 1.65150
12 12 13 2.01317 1.35790
13 4 14 2.23081 1.50470
14 4 15 1.19702 0.80740

Separating real and imaginary parts of equation (14) we or


obtain
] V(2)I 4 + 2.0[P(2)R(1) + Q(2)X(1)
I v(1)l I v(2)l cos[6(1) - 6(2)] - I v(2)l 2 - 0.50] V(1)[2]I V(2)[ 2 + (R2(1)
= P(2)R(1) + Q(2)X(1)
+ xz(1))(p2(2) + Q2(2)) = 0 (17)
therefore
Equation (17) has a straightforward solution and does
not depend on the phase angle, which simplifies the
I v(1)l I v(2)l cos[6(1) - 6(2)] problem formulation. In a distribution system, the
= ]v(2)] 2 + P(Z)R(1) + Q(z)x(1) (15) voltage angle is not so important because the variation
of voltage angle from the substation to the tail-end of a
distribution feeder is only few degrees. Note that from the
and two solutions of IV(2)] 2 only the one considering the
positive sign of the square root of the solution of the
I V(1)] IV(Z)] sin(6(1) - 6(2)) = P(2)X(1) - Q(2)R(1) quadratic equation gives a realistic value. The same is
(16) applicable when solving for IV(2)].
Therefore from equation (17), the solution of IV(2)I
Squaring and adding equations (15) and (16) we can be written as:
obtain:
] V(2)I = {[(P(2)R(1) + Q(2)X(1) - 0.51V(1)]2) a
] V(1)[2[ V(2)[ 2 = [[ V(2)[ 2 -~- P(Z)R(1) + Q(Z)X(1)] 2 - (R2(1) + X2(1))(p2(2) + Q2(2))]1/2

+ [P(Z)X(1) - Q(Z)R(1)] z - (P(Z)R(1) + Q(Z)X(1) - 0.5] V(1)[2)} U2


(18)

Equation (18) can be written in generalized form:


Table A2.2 Load data of Example 1 V(m2) = [B(j) - A(j)] U2 (19)
Nodes KVA Nodes KVA where
1 0.0 9 100.0
2 63.0 10 63.0 A ( j ) = P(m2)*R(j) + Q ( m 2 ) * X ( j ) - 0.5*]V(ml)l 2
3 100.0 11 200.0
(20)
4 200.0 12 100.0
5 63.0 13 63.0 B ( j ) = [AZ(j) - (RZ(j) + X3(j))*(PZ(m2) + QZ(m2))]l/2
6 200.0 14 100.0
7 200.0 15 200.0 (21)
8 100.0
P o w e r f a c t o r o f the l o a d is t a k e n as c o s q~ = 0.70.
j is the branch number, ml and m2 are sending-end and
R e a l p o w e r l o a d = PL = K V A * c o s ~b. receiving-end nodes respectively ( m l = I S ( j ) and
R e a c t i v e p o w e r l o a d = QL = K V A * sin ~b. m2 = I R ( j ) ) .
~ X
E
q~
C~

c~

C~

c~

~J
Method for solution of radial distribution networks: D. Das et al. 345

Table A3.1. Continued


Branch Sending- Receiving- R X
no. end end (ohm) (ohm)
41 41 42 0.273 0.113
42 41 43 0.455 0.189
43 34 44 1.002 0.416
44 44 45 0.911 0.378
45 45 46 0.911 0.378
46 46 47 0.546 0.226
47 35 48 0.637 0.264
48 48 49 0.182 0.075
49 49 50 0.364 0.151
50 50 51 0.455 0.189
51 48 52 1.366 0.567
52 52 53 0.455 0.189
53 53 54 0.546 0.226
54 52 55 0.546 0.226
55 49 56 0.546 0.226
56 9 57 0.273 0.113
57 57 58 0.819 0.340
58 58 59 0.182 0.075
59 58 60 0.546 0.226
60 60 61 0.728 0.302
61 61 62 1.002 0.415
62 60 63 0.182 0.075
63 63 64 0.728 0.302
64 64 65 0.182 0.075
65 65 66 0.182 0.075
66 64 67 0.455 0.189
67 67 68 0.910 0.378
68 68 69 1.092 0.453
69 69 70 0.455 0.189
70 70 71 0.546 0.226
71 67 72 0.182 0.075
72 68 73 1.184 0.491
73 73 74 0.273 0.113
74 73 75 1.002 0.416
75 70 76 0.546 0.226
76 65 77 0.091 0.037
77 10 78 0.637 0.264
78 67 79 0.546 0.226
79 12 80 0.728 0.302
80 80 81 0.364 0.151
81 81 82 0.091 0.037
82 81 83 1.092 0.453
83 83 84 1.002 0.416
84 13 85 0.819 0.340
346 Method for solution of radial distribution networks: D. Das et a

Table A3.2. Load data of Example 2


Node PL Node PL
no. (kW) no. (kW)

4 56.0 46 35.28
6 35.28 47 14.0
8 35.28 50 36.28
11 56.0 51 56.0
14 35.28 53 35.28
15 35.28 54 56.0
16 35.28 55 56.0
17 112.0 56 14.0
18 56.0 57 56.0
19 56.0 59 56.0
20 35.28 61 56.0
21 35.28 61 56.0
22 35.28 62 56.0
23 56.0 63 14.0
24 35.28 66 56.0
25 35.28 69 56.0
26 56.0 71 35.28
28 56.0 72 56.0
30 35.28 74 56.0
31 35.28 75 35.28
33 14.0 76 56.0
36 35.28 77 14.0
37 56.0 78 56.0
38 56.0 79 35.28
39 56.0 80 56.0
40 35.28 82 56.0
42 35.28 83 35.28
43 35.28 84 14.0
44 35.28 85 35.28
45 35.28
Power factor of the load cos = 0.70.
Reactive power load = QL = PL* tan 8.
Those nodes with no power are not shown.

You might also like