0% found this document useful (0 votes)
78 views8 pages

Fast Decoupled Power Flow Unbalanced Radial Distribution Systems

This paper presents a novel power flow formulation and solution method for general unbalanced radial distribution systems. It introduces a new problem formulation that significantly reduces the number of power flow equations by exploiting the radial structure of distribution networks. It also presents a fast decoupled solution algorithm that exploits both the numerical and structural properties of distribution systems. The proposed method is evaluated on two large test systems and shows promising results.

Uploaded by

Abdulrahman
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
78 views8 pages

Fast Decoupled Power Flow Unbalanced Radial Distribution Systems

This paper presents a novel power flow formulation and solution method for general unbalanced radial distribution systems. It introduces a new problem formulation that significantly reduces the number of power flow equations by exploiting the radial structure of distribution networks. It also presents a fast decoupled solution algorithm that exploits both the numerical and structural properties of distribution systems. The proposed method is evaluated on two large test systems and shows promising results.

Uploaded by

Abdulrahman
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 8

IEEE Transactions on Power System, Vol. 10, No.

4, November 1995

2045

Fast Decoupled Power Flow for Unbalanced Radial Distribution


Systems
Ray D.Zimmerman

Hsiao-Dong Chiang
School of Electrical Engineering
Cornel1 University, Ithaca, NY 14853 USA

Abstract This paper presents a novel power flow fofi


mulation and an effective solution method for general
unbalanced radial distribution systems. Comprehensive
models are considered including lines, switches, transformers, shunt capacitors, cogenerators, and several types
of loads. A new problem formulation of three-phase distribution power flow equations taking into account the radial
structure of the distribution network is presented. A di5
tinguishing feature of the new problem formulation is that
it significantly reduces the number of power flow equations, as compared with the conventional formulation. The
numerical properties as well as the structural properties
of distribution systems are exploited resulting in a fast
decoupled solution algorithm. The proposed solution algorithm is evaluated on three-phase unbalanced 292-bus and
394-bustest systems with very promising results.

K-r&
power flow, load flow, radial network, distribution system, fast decoupled
INTRODUCTION

h a d flow is a very important and fundamental tool for


the analysis of any power system and is used in the operational as well as planning stages. Certain applications,
particularly in distribution automation and optimization
of a power system, require repeated load flow solutions. In
these applications it is very important to solve the load
flow problem as efficiently as possible. Since the invention
and widespread use of digital computers, beginning in the
1950s and 1960s, many methods for solving the load flow
problem have been developed [l]. Most of the methods
have grown up around transmission systems and, over
the years, variations of the Newton method such as the
fast decoupled method [2], have become the most widely
used.
Unfortunately, the assumptions necessary for the simplifications used in the standard fast decoupled Newton
method often are not valid in distribution systems. In particular, R/X ratios can be much higher. However, some
work has been done to attempt to overcome these difficulties [31.
On the other hand, some of the methods based on the
general meshed topology of a typical transmission system

95 WM 219-6 PWRS A paper recommended and approved


by the IEEE Power System Engineering Committee of
the IEEE Power Engineering Society f o r presentation
a t the 1995 IEEEJPES Winter Meeting, January 29, t o
February 2, 1995, New York, NY. Manuscript submitted
January 4, 1994; made available f o r printing
November 2 3 , 1994.

are also applicable to distribution systems which typically


have a radial or tree structure. Specifically, we will compare the proposed method to the standard Newton method
141, and the implicit z b , Gauss method [5,6]. These methods do not explicitly exploit the radial structure of the system and therefore require the solution of a set of equations
whose size is of the order of the number of buses.
Our goal was to develop a formulation and solution
algorithm for solving load flow in large three-phase unbalanced systems which exploits the radial topological structure to reduce the number of equations and unknowns and
the numerical structure to further reduce computation as
in the fast decoupled methods for transmission systems.
Some algorithms specific to radial systems have been
presented in [71-[131. These methods are all based on the
concept of doing backward andlor forward sweeps of a ladder network. In [7], the backward sweep involves the computation of a driving point impedance equivalent at each
bus and the forward sweep is used to update voltages and
currents. The method proposed in [8] updates voltages and
currents during the backward sweep and the forward
sweep is a direct voltage correction. Various combinations
of voltage, current, and power flow updates are used in the
backward and forward sweeps of [9]-[14]. A fast decoupled
Newton update, similar to the one proposed in this paper,
was used for the backward sweep in 1131.
In this paper we propose a new problem formulation
and solution algorithm which has a flavor of several of the
above methods. It is probably most closely related to the
methods presented in [9] and [13]. Our numerical results
also include a comparison with the backlforward sweep
approach of [91.

Basic Concept
Our approach is based on the fact that, given the voltage and current at one end of a distribution feeder and the
currents injected into each sub-lateral, it is possible to
compute all voltages and currents in the rest of the feeder.
Since the current at the end of the feeder is zero, the end
voltage can be taken as the unknown. And since the voltage at the source is specified we have a boundary condition
we can use to compute a voltage mismatch.
Exploring the numerical characteristics of a distribution line, we can make decoupling approximations to the
Jacobian used to update the unknown end voltages. This
results in a very fast Newton-like solution to a reduced set
of load flow equations.
For comparimn, this fast decoupled algorithm was
implemented along with the standard Newton method, the
implicit Z b Gauss method, and the backlforward sweep
method of [9]. Each method was evaluated and compared
on a 292-bus and a 394-bus test system.

0885-8950/95/$04.00

Ca 1995 IEEE

2046
BUS & LATERAL INDEXING

In most typical load flow formulations, a set of equations and unknowns is associated with each bus in the network and these equations and unknowns are organized by
a particular bus ordering. Due to the radial structure of
the systems under consideration, the number of equations
and variables can be reduced so that each set of equations
and unknowns corresponds to a n entire lateral instead of
an individual bus. Our formulation therefore calls for a n
appropriate lateral indexing to order these equations and
variables.
Aradial system can be thought of as a main feeder with
laterals. These laterals may also have sub-laterals, which
themselves may have sub-laterals, etc. So first we define
the level of lateral i as the number of laterals which need
to be traversed to go from the end of lateral i to the source.
For example, the main feeder would be level 1,its sub-laterals would be level 2, their sub-laterals level 3, etc.
The laterals within level 1 are indexed according to the
order visited during a depth-first traversal of the network.
Each lateral can be uniquely identified by a n ordered pair
( k , m ) where 1 is the lateral level and m is the lateral
index within level 1.
Buses are also indexed within each lateral starting
with the first bus on the lateral, so that each bus is
uniquely identified by an ordered triple ( I , m, n) where R.
is the bus index. So ( I , m, n) refers to the nth bus on the
mth level 1 lateral. The source is given an index of
(1,1 , O ) . Figure 1 shows an example of this indexing
scheme on a sample 63-bus system. The boxed numbers

show the reverse breadth-first (RBF) ordering of the laterals found by sorting the lateral indices in reverse order,

firstby level, then by lateral index.


We will also use the following shorthand notation when
i is an ordered pair referring to a lateral and k is an
ordered triple referring to a bus. Lateral i - 1 refers to the
parent of lateral i, and bus k - 1 refers to bus ks parent
bus. Bus k + 1 may also be used to refer to the bus follow-

ing bus k on the same lateral. This notation is used in


indexing voltages, currents, impedances, etc.
SYSTEM MODELING

For the purposes of power flow studies, we model a


radial distribution system a s a network of buses connected
by distribution lines, switches, or transformers to a voltage specified source bus. Each bus may also have a corresponding load, shunt capacitor, andlor cogenerator
connected to it. The model can be represented by a radial
interconnection of copies of the basic building block shown
in Figure 2. The dotted lines from the cogenerator, shunt
capacitor, and load to ground are to indicate that these elements may be connected in a n ungrounded delta-configuration. Since a given branch may be single-phase, twophase, or three-phase, each of the labeled quantities is
respectively a scalar, 2 x 1,or 3 x 1complex vector. For the
simplicity of presentation we will occasionally assume
everything is three-phase, although both single and twophase laterals are handled by our program.
One of the key concepts behind our formulation is that
the voltage and current at one bus can be expressed as a
function of the voltage and current at the next bus. If we
let

r,,

(2.1.4

(2.1.3

we can write the branch update function as

(2.1.2

(2.1.1

wk-l

(2)

= gk(wk)

where wUR
is a 12 x 1vector containing the real and imaginary parts ofthe voltages and currents at bus k. The func-

cogeneratob

shunt1

capacitor

Figure 1 Example of Bus & Lateral Indexing

/////////////-

Figure 2 Basic Building Block

2041

tion g, is determined by the sub-laterals attached at bus

k as well as the models for distribution lines, switches,


transformers, loads, shunt capacitors, and cogenerators.
From v k we can compute the currents injected by the
loads, shunt capacitors, and cogenerators from (4)-(8),
according to the models presented below. Given I , +; and
the currents I j injected into sub-laterals branching off
from bus k, we apply KCL at bus k to give

I,' =

IGk+ICk+ILk-Ik+l-

Ij
je

Table I Update Equations

Element

Update Equation

...................................................................................

(4)

(5)

(6)

(3)

where A, is the set of buses adjacent to bus k on sub-laterals.


So for each bus k we can express the voltage and current at bus k - l as a function of the voltage and current
at bus k in the form of (2) by using (3) and the appropriate
equations from Table I.
Load Model
The load model used is a general model which allows
each load to be either wye-connected or delta-connected
and either constant impedance, constant current, or constant complex power. For the wye-connected case the
injected currents can be computed as shown in (4)-(6).The
bar denotes a constant value and the division in (4) and (6)
is element-wise. Injected currents for deltaconnected
loads are computed by taking the differences of the appropriate elements in (4146). This model could easily be generalized to be a linear combination of all of the above
types.
Shunt Capacitor Model
Shunt capacitors are modeled as wye-connected or
deltaconnected constant admittance. The injected current
as a function of voltage for the grounded wye-connected
case is given in (7) where the multiplication is elementwise.
Cogenerator Model
Cogenerators are modeled as wye-connected or deltaconnected constant complex power devices. The injected
current given by (8)is therefore in the same form as (6).
Line Model
The line model used is the standard pi-model. The
impedance of distribution line k is represented as a series
impedance Z, and the line charging effects are divided
between the two shunt arms, each with an admittance of
Y,/2.The impedance Z, and the admittance Yk are
both n x n complex matrices, where n is the number of
phases in the line. The voltages and currents at the sending end of the line are computed via (9) and (10).
Switch Model
Sectionalizing switches are modeled as branches with
zero impedance. The corresponding voltage and current

line

I 'h-1 =
I I , = -I;

'h

I I, = Y p , - ,

+Y p ,

From (15),we can solve for the voltage and current at the
primary given the voltage and current at the secondary
resulting in (13) and (14).

PROBLEMFORMULATION
Here we present a new load flow formulation with a
reduced number of equations and unknowns. The load flow
problem is typically formulated as a set of non-linear
power mismatch equations as functions of the bus voltages. The number of equations and unknowns in an n-bus
all three-phase system is 6 (n - 1) ,since the source is the
only voltage specified bus in a distribution system. Our
formulation reduces the number of equations and
unknowns to 6 times the number of laterals in the system.
To illustrate our formulation we start with a system consisting of a single main feeder.
Single Feeder
Given the voltage and current at either end of the
feeder we can compute the remaining voltages and currents. Note also that throughout this formulation currents
could be replaced by complex power flows as in [131. We
have two boundary conditions, the current I,,+l at the
end of the feeder is equal to zero, and the voltage V , at
the source is a specified constant Bo. So we can take

relationshipsare given by (11) and (12).

l'kansformer Model
Three-phase transformers are modeled by the admittance matrix equivalent proposed in E151 for grounded-we
to grounded-wye connections.

vo = Po
/o+

"-------t-H-]

source

V"
-In+,
end bus

Figure 3 Single Feeder

=0

2048
either the current I , at the source, or the voltage V, at
the end of the feeder to be the unknown. Unlike I , , an initial guess to V, is readily available without computation
(balanced 1 P.u.), so we choose V, , which we refer to as
the end voltage, as the independent variable.
= 0 we can compute
Using (2) and the fact that I ,
the source voltage (and current) as a function of V,.
Starting with w , we apply the appropriate branch update
function gk consecutively until we reach w ,
+

bus k = ( I , m,, n)

Figure 4 Mismatch Calculation

We will denote the voltagf part of the composite function g , ... g , - g , as Vo . At the solution the mismatch between the specified source voltage V O and the
computed source voltage Vo(Vn) must be zero. The power
flow equations can therefore be written as
f(V,,

V0(V,)-Vo

= 0

(17)

Since the voltage Vis a complex 3 x 1 vector this equation


is equivalent to 6 real equations in 6 real unknowns independent of the number of buses on the feeder.
The Jacobian off can be expressed using the chain rule
and the branch Jacobians. The branch Jacobian Gk for
bus ks incoming branch is the Jacobian of equation (2)

reverse breadth-first (RBF) ordering illustrated in


Figure 2.
To complete the voltage mismatch calculation for each
lateral we take the difference of the two voltages computed
for each branching bus. Suppose lateral (I, ml) has a
level 1 + 1 sub-lateral i branching off at bus k, where i is
the ordered pair ( 1 + 1,m2) and k is the triple ( I , ml, n ) .
The mismatch for lateral i is taken as the difference
between Vk as computed from the end voltage of sub-latera1 i, and V k as computed from the end voltage of the
supplying lateral (I, ml) .
For lateral i starting at bus k we now have an equation
similar to (17), but with V, being replaced by a vector x
containing the end voltages of all laterals.
fi(x)

(20)

= Vk(x)-vk(x) = 0

The function f i does not depend on all elements of x, but


only on the voltages of end buses supplied through bus k
since they affect currents injected into the sub-laterals of
lateral i. This relationship will be seen more clearly in
Figure 5 when we look at the structure of the Jacobian.
Putting these equations together in RBF order we can
express the new load flow equations compactly as

F(x) = 0
We express the system Jacobian for this simple single
feeder case as

where the first term in the product is just the top half of
G , and the last term is the left half of G, ,
General Radial Structure
To generalize this formulation to handle an arbitrary
radial structure we first note that the voltage at the beginning of any lateral can be computed as a function of the
end voltage if the currents injected into each sub-lateral
are given. In a system with L levels, the level L laterals
have no sub-laterals and can therefore be computed first.
After all level L laterals have been computed, the currents
injected into the sub-laterals of each level L - 1 lateral are
known, hence the level L - 1 laterals can be computed.
Next the level L - 2 laterals are computed, and so on,
until the main feeder has been computed. This is the

(21)

Note that V k ( Z ) is no longer constant except when i is the


main feeder and k is the source bus, i.e. the last set of
equations when in RBF order.
Assuming we separate real and imaginary parts, for a
system with m laterals, (21) is a set of 6m non-linear equations in 6m real unknowns. This is also assuming that all
buses are three-phase; single and two-phase laterals
would reduce these numbers accordingly.
SOLUTION METl?HD
The reduced set of load flow equations in (21) could be
solved by any of several iterative methods for finding zeros
of a set of general non-linear equations. The well-known
Newton method proceeds as follows:

Newton Method

1. Choose a n initial guess for the solution, xo


2. Set k = 0 .
3. Evaluate Fk = F ( x k ) .
4. Stop if llFkll 5 some tolerance.

2049

4zk.

5. Evaluate the Jacobian, J , = 6. Solve J k S k = -Fk.


7. Let xk
= xk + sk
8. Let k = k + 1 and go to step 3.
As is typical in the Newton method, the majority of
computation time is spent in step 5 evaluating the Jacobian, and step 6 solving for the update step. Any reasonable approximation which results i n fewer computations
in these two steps will greatly improve the speed of the
algorithm, barring an increase i n the total number of iterations.
First, we think of the Jacobian in terms of block elements corresponding to the laterals. The block element in
block row i and block columnj is the sensitivity of the voltage mismatch at lateral i with respect to variations in the
end voltage of lateral j. Each block element of the system
Jacobian can be expressed, using the chain rule, as a product of branch Jacobians along the path between the start
bus of lateral i and the end bus of lateral j.

Numerical Structure of Branch Jacobian


The branch Jacobian G, of (18)relates the sensitivity
of voltages and currents at bus k - 1 to small variations i n
the voltages and currents at busk. For a three-phase
branch it is a 12 x 12 matrix. Due to the fact that for practical distribution lines the elements of the line impedance
matrix 2, and the h e charging admittance Yk are small
with respect to voltage magnitudes, the branch Jacobian
can be approximated by the identity matrix. To derive an
analytical expression for each of the 144 terms of the
matrix and analyze them to determine which terms can be
neglected is obviously too tedious. So here we just give a
motivation for the approximation.
A small change i n voltage at bus k produces a correspondingly small change in the load, shunt, and cogenerator currents at bus k, as well as the series current in
branch k. However, these currents are multiplied by the
lines small impedance 2, ,making their effect on the voltage change at bus k - 1 negligible. So from (9)we see that,
neglecting the effects of the second term, a change in voltage at bus k produces nearly the same change in voltage at
bus k - 1. This accounts for the upper left block of G, in
(18).Similarly, a small change in the current I,
affects
only I, and will have a negligible effect on the voltage
V, - giving us approximately zero for the upper right
block of G,. Due to the small Y, we see that small
have very little effect on the first
changes in V, or 1,
term in (lo), accounting for the lower two blocks of G, in

tion, which is still based on the detailed models described


in the section SYSTEM MODELING.
Numerical Structure of System Jacobian
Taking into account the numerical structure of the
branch Jacobians, we build up the system Jacobian via the
chain rule. his leads to the special structure, dependent
on the RBF ordering of the laterals, which is illustrated in
Figure 5 for the sample system from Figure 1.
Decoupled Algorithm
Since the nonzero blocks below the block diagonal are
nearly zero we can neglect them. By doing this we are
essentially saying that the mismatch for lateral i is only
affected by the end voltage of lateral i and the end voltage
of lateral is parent.
This approximation greatly speeds up the solution of
the update step in step 6 of the Newton algorithm. The
matrix is now more sparse, but more importantly it is
block-upper triangular and the update step can therefore
be solved by block back-substitution.
Fast Decoupled Algorithm
In the decoupled algorithm, the evaluation of the Jacobian in step 5 is still quite complex and expensive t o the
extent of being impractical. Further approximations can
be made by replacing the near identity blocks by an exact
identity matrix and the near negative identity blocks by
exactZy negative identity. The exact identity blocks are
actually divided by the appropriate tap ratios for blocks
corresponding t o laterals which include transformers.
The Jacobian is now a constant upper triangular
matrix corresponding t o the approximations made t o the
branch Jacobians as mentioned above. This means that

11 01 block element is approximatelyzero


11 I block element is approximately identity

(18).
Approximation of the branch Jacobian by the identity is
the equivalent of replacing loads, shunt capacitors, and
cogenerators by constant injected currents and neglecting
line charging and line impedance. The same approximation is used for sectionalizing switches. For transformer
sections, the upper left block of G, is approximated by the
identity divided by the tap ratio, and the lower right block

by the identity times the tap ratio. In other words the


transformer is approximated by its ideal equivalent. It is
important to note that these approximations are only for
simplifying the Jacobian used to compute tho update step
in Newtons method. They have no effect on the final solu-

Figure 6 Structure of System Jacobian

2050
step 5 has been essentially eliminated and step 6 is now a
simple back-substitution. In fact, it is not even necessary
to explicitly form and store the Jacobian since each row
has a 1on the diagonal and a -1in the column mmesponding to the appropriate phase of the parent lateral.
Here again, a 1 is replaced by the reciprocal of a corresponding transformer tap ratio where appropriate.

t.80

Newton
Implicit Z

2.04

Gauss

BacklFoward
sweep

1.04:

69

:
:

Fast Decoslpled
I

RESULTS

c-

The fast decoupled method described above was implemented in Matlab 4 along with the traditional fornulation
of the Newton [4] and implicit Zb,, Gauss methods [5] and
a backKorward sweep method [9]. The implicit z b , Gauss
method actually uses a n optimally ordered, factored Y b a
instead of forming z b , explicitly.
The test systems are 292-bus and 394-bus unbalanced
systems, each with 6 distribution transformers. According
to our formulation, the first system has a total of 85 laterals and the second has 108 laterals, each including some
single and 2-phase laterals. Three of the algorithms,
excluding the backlforward sweep, require the solution of
a large sparse system of linear equations. In the Newton
and fast decoupled methods the resulting matrix is a Jacobian and in the implicit Zba Gauss method it is Yb,.
Table I1 shows the sizes and types of these matrices for our
two test systems. If all buses were three-phase we would
expect the dimension to be approximately 6n for Newton,
3n for implicit z b , Gauss, and 61 for fast decoupled, where
n is the number of buses and I is the number of laterals.
Since there are some single and 2-phase buses the numbers are slightly smaller than these. Notice that in the fast
decoupled method the matrix is triangular, that is, already
in factored form. Unlike the fist two methods, the proposed algorithm does not require the formation of Y b , or
the factoring of a large matrix.
In the case of the 394-bus system, the number of real
equations and unknowns, and therefore the size of the
Jacobian, was reduced from 1996 for the traditional Newton formulation to 560 for the fast decoupled method. This
decrease in size was accompanied by a correspondingly
large decrease in the computation required to solve the
load flow from a flat start. Figure 6 shows the computational complexity for each of the four methods in megaflops. These figures are based on the total number of
floating point operations as reported by Matlab and should
be indicative of run-time for the algorithms in compiled
languages such as C or Fortran.
Our results show that the fast decoupled method is also
more efficient than the bacWforward sweep method based

TypedtStructure

Figure 6 Computational Complexity


on [9]. The difference is in the amount of computation necessary to update the end voltages. For the backlforward
sweep approach this requires a full forward sweep calculating the voltage drops in each branch. In the fast decoupled approach it is a direct computation from the
triangularJacobian. Since the function evaluation in the
proposed algorithm is equivalent to a backward sweep, we
expect each iteration to require less computation.
In our tests, the Newton method demonstrated quadratic convergence as expected, while the other three
methods converged linearly as shown in Figure 7.

CONCLUSIONS
& FrrmTRE WORK
In this paper we have exploited the radial structure
(physicalproperty) and the decoupling numerical property
of a distribution system to develop a fast decoupled Newton method for solving unbalanced distribution load flow.
It involves a reduced set of equations and unknowns proportional to the number of laterals in the network as
opposed to the number of buses. Due to the reduced number of equations and the fact that the Jacobian is approximated by a constant triangular matrix, it is significantly
faster than the implicit Zb,, Gauss method or the traditional Newton method based on U,,,. Since each function
evaluation involves updating each bus voltage and current

IB Newton

real, general

Newton

Dimension
292bus 394bus

OfMntrix

1472

I Implicit Zbus Gauss I complex, symmetric I


I Bacworward SweeD I
Fast-Decoupled

factoring necessary

736
no matrix necessaw

rea1,triangulaF

426

1996
998

560

megaflops

Table I1 Sparse Matrix to be Factored


Algorithm

I
I
Figure 7 Convergence for 394-bus System

205 1

and the Jacobian is triangular the computation in each


iteration is proportional to n, making it suitablb for very
large radihl systems. It is also shown to be more efficient
than the k@k/forward sweep method of [9], due to the savings during the update of the end voltages.
The load how method proposed in this paper could possibly be improved further in several ways. As described
here, it is limited to radial systems with one voltage-reg&
lated bus treated as the source. The author$ believe that it
could be generalized to handle weakly meshed systems
using a com&nsation method similar to those presented
in [lo, 12, 141. These approaches are based on choosing
breakppints cp convert the system to a radial structure.
The network is then solved using a radial load flow algorithm along with corrections to the brbakpoint currents or
powers, These approaches also allow for PV buses, treated
as artificial breakpoints, making them suitable for weaklymeshed triinsmission systems. Since the methods .in [lo,
12, 141 are based on bacwforward sweep, improvgments
might be possible by basing the radial network solver on
the fast decoupled method proposed in this paper.
It may also be possible to extend this formulation to
include distkibution transformers of other connection
types such as the remainder of those presented in [15].

ACKNOWLZDGMENTS
The authors would like to thank Gary Darling and
acknowledge the financial support of New York State Gas
& Electric. The valuable input of Dr.Jianzhong Tong was
also much appreciated.
REFERENCES
C13

B. Stott, Review of Load-Flow Calculation


Methods, Proceedings of the IEEE, Vol. 2, No. 7,
July 1974, pp. 916-929.

[2]

[3]

B. Stott and 0. Alsaq, Fast DecouplBd Load Flow,


IEEE Dansactions on Powkr Apparatus and
Systems, Vol. PAS-93, MaylJune 1974, pp. 859-869.

D. Rajieik and A. Bose, A Modifichtion to the Fast


Decoupled Power Flow for Netwbks with High R/X
Ratios, ZEEE hnsactions on Power Systems,
Vol. 3, No. 2, May 1988, pp. 743-746.

[4]

W.F. Tinney and C. E. Hart, Power Flow Solution


by Newtons Method, IEEE Dansactions on Power
Apparatus and Systems, Vol. PAS-86, No. 11,
November 1967, pp. 1449-1460.

[5]

D. I. H. Sun, S.Abe,R. R. Shoults, M. S. Chen, P.


Eichenberger, and D. Farris, Calculation of Energy
Losses in a Distribution System, IEEE
Bansactions on Power Apparatus and Systems,
Vol. PAS-99, NO. 4, JulyIAugust 1980, pp. 13471356.

[6]

T. H. Chen, M. S.Chen, K. J. Hwang, P. Kotas, and


E. Chebli, Distribution System Power Flow
Analysis - A Rigid Approach, IEEE Bansactwns on
Power Delivery? Vol. 6, No. 3, July 1991, pp. 11461152.

R. Berg, Jr,, E. S. Hawkins, and W. W. Pleines,


Mechanized Calculation of Unbalanced Load Flow
on Radial Distribution Circuits, ZEEE Dansactions
on Power Apparatus and Systems, Vol. PAS-86,
No. 4, April 1967, pp. 415-421.

W.H. Kersting and D. L. Mendive, An Application


of Ladder Network Theory to the Solution of ThreePhase Radial Load-Flow Problems, IEEEIPES
1976 Winter Meeting, New York, NY, January 1976,
A 76 044-8.

W. H. Kersting and W. H. Phillips, A Radial Threephase Power Flow Program for the PC, Conference
paper, presented at 1987 Frontiers Power
Conference, Stillwater, OK, October 1987.

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 Dansactions on
Power Systems, Vol. 3, No. 2, May 1988, pp. 753762.

M. E. Baran and F. F. Wu, Optimal Sizing of


Capacitors Placed on a Radial Distribution System,
IEEE nansactwns on Power Delivery, Vol. 4, No. 1,
January 1989, pp. 735-742.

G.X. Luo and A. Semlyen, Efficient Load Flow for


Large Weakly Meshed Networks,
IEEE
Dansactions on Power Systems, Vol. 5, No. 4,
November 1990, pp. 1309-1316.
H. D. Chiang, A Decoupled Load Flow Method for
Distribution Power Networks: Algorithms, Analysis
and Convergence Study, Electrical Power & Energy
Systems, Vol. 13, No. 3, June 1991, pp. 130-138.

D. Rajieic, R. AZkovski, R. Meski, Voltage


Correction Power Flow, IEEEIPES 1993 Summer
Meeting, Vancouver, B.C., Canada, July 1993,
93 SM 570-2.

M. S. Chen and W. E. Dillon, Power System


Modeling, Proceedings of the IEEE, Vol. 62, No. 7,
July 1974, pp, 901-915.
Ray D. Zimmerman received his B.S.in electrical engineering
in 1989 from Drexel University,Philadelphia, PA, and his M.S.in
electrical engineering in 1992 from Cornell University, Ithaca,
NY. He is currently working toward a Ph.D. in electrical engineering at Cornell.
Hsiao-DongChiang received the Ph.D. degree in electrical engineering and computer sciences fmm the University of California
at Berkeley. He is currently an Associate Professor of electrical
engineering at Cornell University. He was a recipient of the Engineering Research Initiation Award (1988) and of the Presidential
Young Investigator Award (1989) both fmm the National Science
Foundation. In 1990 he was selected by a Cornell Merrill Presidential Scholar as the faculty member who had the most positive
influence on that students education at Cornell. He was an Associate Editor of the IEEE Transactions on Circuits and Systems
(1990-1991).He is currently Editor for express letters of the IEEE
Transactionson Circuits and Systems I Fundamental Theory and
Applications. His research interests include power systems, nonlinear systems, optimization theory and neural networks.

2052

DISCUSSION

W. H. Kersting, New Mexico State University, Las Cruces, NM


88003: The title of this paper is somewhat misleading. The paper
develops a Jacobian matrix to be used in computing bus voltages
and then proceeds to make several assumptions that reduce the
Jacobian to an upper triangular matrix consisting of positive and
negative identity blocks. When all of this is done, the resulting
Jacobian turns out to be only an interesting method of ordering the
radial feeder. This ordering is then used to compute bus voltages
starting at end buses and working back to the source. This
appears to be essentially the same technique that is used in the
BacWorward Sweep.
The technique of this paper is very sensitive to correctly indexing
the laterals. Does the user have to create this indexing prior to
program execution? If not, can the l i e segments be inp@ in a
random order with no concern of up stream and down stream
buses?
While it is interesting to compare number of iterations and number
of computations for various iterative techniques, a far more
important criteria is to compare the results (bus voltages, currents,
etc.). Have the authors run cases where they compare the results of
their technique to others? In particular, it would be interesting to
see the results for one or more of the IEEE Radial Distribution
Test Feeders [l].
Since many distribution feeders use tap changing voltage
regulators or load tap changing transformers, can these be
modeled in the technique described in the paper?
If the authors have been successful in modeling delta-wye
transformer banks correctly for unbalanced voltages and currents, I
urge them to publish the results.
REFERENCE
[l] Distribution Planning Task Force Report, Radial Distribution
Test Feeders, IEEE Transactions on Power System, August
1991, Volume 6, Number 3, pp 975-985.
Manuscript received February 15, 1995.

Ray D. Zimmerman and H. D. Chiang: The authors


would like to thank Dr. Kersting for his interest in and
comments regarding the paper. Each of the issues raised
will be considered in order.
he formulation and presentation of the proposed algorithm highlight the fact that the method can be viewed as
a n inexact Newton method. As with most radial power
flow methods, including the backwardforward sweep
methods, it is based on updating voltages and currents (or
power flows) along the paths from the end buses to the
source. In fact, the function evaluation for the proposed
method is identical to the backward sweep used in [Cl].
While the update of the end voltages corresponds to a for-

ward sweep, it is a direct update which does not require a


forward traversal of the network to compute the voltage
drops in each branch.
The method proposed in the paper is, without question,
similar in nature to backwaraforward sweep methods. It
is not, however, equivalent in the sense of generating the
same sequence of voltage solutions.
The algorithm presented does require a certain ordering of the laterals. In the proposed method, the ordering
ensures that the downstream laterals are processed first
during the function evaluation and that upstream laterals are processed first during the end voltage update. Similarly, a backwardlforward sweep method is sensitive to
the ordering of the buses. A correct bus ordering is
required to ensure that downstream currents are
summed first during the backward sweep and upstream
voltage drops are computed first during the forward
sweep. Neither case requires any special treatment of the
data by the user. In other words, line segments can be
input in a random order. In the implementation used in
the paper, the ordering is done by the program during a
pre-processing traversal of the network used for checking
data integrity.
The authors agree that the correctness of the solution
found is far more important than the efficiency of the algorithm used. Efficiency, however, is a very important criterion to be evaluated in comparing correct implementations
of valid algorithms. Assuming the proposed algorithm is
valid, this becomes an issue of thoroughness in implementation and testing of the software. Our implementations of
each of the algorithms compared in the paper (as well as
several others) yield solutions which match to within the
This
program tolerance, even when set smaller than
is to be expected since all of our implementations use identical modeling. Even a comparison with the results of a
commercial package, which uses slightly different modeling, yields very similar solutions. Unfortunately, our
implementation does not currently read the input data format of the test feeders presented i n [C2].
In its current implementation, the program does not
include the automatic tap changing of voltage regulators
or load tap changing transformers. However, since transformers with off-nominal taps are modeled, it seems reasonable that the automatic tap changing could be handled
as described i n [Cl].
Delta-wye and other transformer connections have subsequently been modeled successfully and the results will
be published in the near future.

[Cll W.H. Kersting and W. H. Phillips, A Radial hreephase Power Flow Program for the PC, Conference
presented at 1987 Frontiers Power
Conference, Stillwater, OK, October 1987.

puper,

IC21 IEEE Distribution Planning Working Group Report,


Radial Distribution
Test Feeders, IEEE
lEansactions on Power Systems, Vol. 6, No. 3,
August 1991, pp. 975-985.
Manuscript received April 10, 1995.

You might also like