www.bookspar.

com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

CHAPTER 3

LOAD FLOW ANALYSIS

[CONTENTS: Review of solution of equations, direct and iterative methods,
classification of buses, importance of slack bus and YBUS based analysis,
constraints involved, load flow equations, GS method: algorithms for finding the
unknowns, concept of acceleration of convergence, NR method- algorithms for
finding the unknowns, tap changing transformers, Fast decoupled load flow,
illustrative examples]

REVIEW OF NUMERICAL SOLUTION OF EQUATIONS

The numerical analysis involving the solution of algebraic simultaneous equations

forms the basis for solution of the performance equations in computer aided electrical
power system analyses, such as during linear graph analysis, load flow analysis
(nonlinear equations), transient stability studies (differential equations), etc. Hence, it
is necessary to review the general forms of the various solution methods with respect
to all forms of equations, as under:
1. Solution Linear equations:
* Direct methods:
- Cramer’s (Determinant) Method,
- Gauss Elimination Method (only for smaller systems),
- LU Factorization (more preferred method), etc.
* Iterative methods:
- Gauss Method
- Gauss-Siedel Method (for diagonally dominant systems)
2. Solution of Nonlinear equations:
Iterative methods only:
- Gauss-Siedel Method (for smaller systems)
- Newton-Raphson Method (if corrections for variables are small)
3. Solution of differential equations:
Iterative methods only:
- Euler and Modified Euler method,
- RK IV-order method,
- Milne’s predictor-corrector method, etc.

www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

It is to be observed that the nonlinear and differential equations can be solved only by
the iterative methods. The iterative methods are characterized by the various
performance features as under:
Selection of initial solution/ estimates
Determination of fresh/ new estimates during each iteration
Selection of number of iterations as per tolerance limit
Time per iteration and total time of solution as per the solution method selected
Convergence and divergence criteria of the iterative solution
Choice of the Acceleration factor of convergence, etc.

A comparison of the above solution methods is as under:

In general, the direct methods yield exact or accurate solutions. However, they
are suited for only the smaller systems, since otherwise, in large systems, the
possible round-off errors make the solution process inaccurate.
The iterative methods are more useful when the diagonal elements of the
coefficient matrix are large in comparison with the off diagonal elements. The
round-off errors in these methods are corrected at the successive steps of the
iterative process.
The Newton-Raphson method is very much useful for solution of non –linear
equations, if all the values of the corrections for the unknowns are very small
in magnitude and the initial values of unknowns are selected to be reasonably
closer to the exact solution.

LOAD FLOW STUDIES

Introduction: Load flow studies are important in planning and designing future
expansion of power systems. The study gives steady state solutions of the voltages at
all the buses, for a particular load condition. Different steady state solutions can be
obtained, for different operating conditions, to help in planning, design and operation
of the power system.

Generally, load flow studies are limited to the transmission system, which involves
bulk power transmission. The load at the buses is assumed to be known. Load flow

2
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

studies throw light on some of the important aspects of the system operation, such as:
violation of voltage magnitudes at the buses, overloading of lines, overloading of
generators, stability margin reduction, indicated by power angle differences between
buses linked by a line, effect of contingencies like line voltages, emergency shutdown
of generators, etc. Load flow studies are required for deciding the economic operation
of the power system. They are also required in transient stability studies. Hence, load-
flow studies play a vital role in power system studies.

Thus the load flow problem consists of finding the power flows (real and reactive)
and voltages of a network for given bus conditions. At each bus, there are four
quantities of interest to be known for further analysis: the real and reactive power, the
voltage magnitude and its phase angle. Because of the nonlinearity of the algebraic
equations, describing the given power system, their solutions are obviously, based on
the iterative methods only. The constraints placed on the load flow solutions could be:
The Kirchhoff’s relations holding good,
Capability limits of reactive power sources,
Tap-setting range of tap-changing transformers,
Specified power interchange between interconnected systems,
Selection of initial values, acceleration factor, convergence limit, etc.

Classification of buses for LFA: Different types of buses are present based on the
specified and unspecified variables at a given bus as presented in the table below:

Table 1. Classification of buses for LFA

Sl. Specified Unspecified

Bus Types Remarks
No. Variables variables
1 Slack/
|V|, δ PG, QG |V|, δ: are assumed if not
Swing Bus specified as 1.0 and 00
Generator/ A generator is present at the
2
Machine/ PV Bus PG, |V| QG, δ machine bus
About 80% buses are of PQ
3 Load/ PQ Bus PG, QG |V|, δ type
Voltage ‘a’ is the % tap change in
4
Controlled Bus PG,QG, |V| δ, a tap-changing transformer

3
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

Importance of swing bus: The slack or swing bus is usually a PV-bus with the
largest capacity generator of the given system connected to it. The generator at the
swing bus supplies the power difference between the “specified power into the system
at the other buses” and the “total system output plus losses”. Thus swing bus is
needed to supply the additional real and reactive power to meet the losses. Both the
magnitude and phase angle of voltage are specified at the swing bus, or otherwise,
they are assumed to be equal to 1.0 p.u. and 00 , as per flat-start procedure of iterative
solutions. The real and reactive powers at the swing bus are found by the computer
routine as part of the load flow solution process. It is to be noted that the source at the
swing bus is a perfect one, called the swing machine, or slack machine. It is voltage
regulated, i.e., the magnitude of voltage fixed. The phase angle is the system reference
phase and hence is fixed. The generator at the swing bus has a torque angle and
excitation which vary or swing as the demand changes. This variation is such as to
produce fixed voltage.

Importance of YBUS based LFA: The majority of load flow programs employ
methods using the bus admittance matrix, as this method is found to be more
economical. The bus admittance matrix plays a very important role in load floe
analysis. It is a complex, square and symmetric matrix and hence only n(n+1)/2
elements of YBUS need to be stored for a n-bus system. Further, in the YBUS matrix, Yij
= 0, if an incident element is not present in the system connecting the buses ‘i’ and ‘j’.
since in a large power system, each bus is connected only to a fewer buses through an
incident element, (about 6-8), the coefficient matrix, YBUS of such systems would be
highly sparse, i.e., it will have many zero valued elements in it. This is defined by the
sparsity of the matrix, as under:

Total no. of zero valued elements of YBUS

Percentage sparsity of a
=
given matrix of nth order: Total no. of entries of YBUS

S = (Z / n2) x 100 % (1)

The percentage sparsity of YBUS, in practice, could be as high as 80-90%, especially

for very large, practical power systems. This sparsity feature of YBUS is extensively

4
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

used in reducing the load flow calculations and in minimizing the memory required to
store the coefficient matrices. This is due to the fact that only the non-zero elements
YBUS can be stored during the computer based implementation of the schemes, by
adopting the suitable optimal storage schemes. While YBUS is thus highly sparse, it’s
inverse, ZBUS, the bus impedance matrix is not so. It is a FULL matrix, unless the
optimal bus ordering schemes are followed before proceeding for load flow analysis.

THE LOAD FLOW PROBLEM

Here, the analysis is restricted to a balanced three-phase power system, so that the
analysis can be carried out on a single phase basis. The per unit quantities are used for
all quantities. The first step in the analysis is the formulation of suitable equations for
the power flows in the system. The power system is a large interconnected system,
where various buses are connected by transmission lines. At any bus, complex power
is injected into the bus by the generators and complex power is drawn by the loads. Of
course at any bus, either one of them may not be present. The power is transported
from one bus to other via the transmission lines. At any bus i, the complex power Si
(injected), shown in figure 1, is defined as
Si = SGi – SDi (2)

Fig.1 power flows at a bus-i

where Si = net complex power injected into bus i, SGi = complex power injected by
the generator at bus i, and SDi = complex power drawn by the load at bus i. According
to conservation of complex power, at any bus i, the complex power injected into the

5
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

bus must be equal to the sum of complex power flows out of the bus via the
transmission lines. Hence,
Si = Sij ∀ i = 1, 2, ………..n (3)
where Sij is the sum over all lines connected to the bus and n is the number of buses in
the system (excluding the ground). The bus current injected at the bus-i is defined as
Ii = IGi – IDi ∀ i = 1, 2, ………..n (4)
where IGi is the current injected by the generator at the bus and IDi is the current drawn
by the load (demand) at that bus. In the bus frame of reference
IBUS = YBUS VBUS (5)
where
I1
I2
IBUS = . is the vector of currents injected at the buses,
.
In

YBUS is the bus admittance matrix, and

V1
V2
VBUS = . is the vector of complex bus voltages.
.
Vn

Equation (5) can be considered as

n
Ii = YijV j ∀ i = 1, 2, ………..n (6)
j =1

The complex power Si is given by

Si = Vi I i*
*
n
= Vi Yij V j
j =1

n
= Vi Yij* V j* (7)
j =1

Let V i ∆ V i ∠ δ i = V i (cos δ i + j sin δ i )

δ ij = δ i − δ j

6
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

Yij = Gij + jBij

Hence from (7), we get,

Vi V j (cos δ ij + j sin δ ij ) (Gij − j Bij )

n
Si = (8)
j =1

Separating real and imaginary parts in (8) we obtain,

Vi V j (Gij cos δ ij + Bij sin δ ij )

n
Pi = (9)
j =1

Vi V j (Gij sin δ ij − Bij cos δ ij )

n
Qi = (10)
j =1

An alternate form of Pi and Qi can be obtained by representing Yik also in polar form
as Yij= Yij ∠θij (11)

Again, we get from (7),

n
Si = Vi ∠ δ i Yij ∠ − θ ij V j ∠ − δ j (12)
j =1

The real part of (12) gives Pi.

Yij V j cos (− θ ij + δ i − δ j )
n
Pi = Vi
j =1

n
= Vi Yij V j cos − (θ ij − δ i + δ j ) or
j =1

n
Pi = Vi V j Yij cos(θij − δ i + δ j ) ∀ i = 1, 2,………..n, (13)
j =1

Similarly, Qi is imaginary part of (12) and is given by

n
Qi = Vi Yij V j sin − (θ ij − δ i + δ j ) or
j =1

n
Qi = − Vi V j Yij sin (θ ij − δ i + δ j ) ∀ i = 1, 2,………..n (14)
j =1

Equations (9)-(10) and (13)-(14) are the ‘power flow equations’ or the ‘load flow
equations’ in two alternative forms, corresponding to the n-bus system, where each
bus-i is characterized by four variables, Pi, Qi, |Vi|, and δi. Thus a total of 4n
variables are involved in these equations. The load flow equations can be solved for

7
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

any 2n unknowns, if the other 2n variables are specified. This establishes the need for
classification of buses of the system for load flow analysis into: PV bus, PQ bus, etc.

DATA FOR LOAD FLOW

Irrespective of the method used for the solution, the data required is common for any
load flow. All data is normally in pu. The bus admittance matrix is formulated from
these data. The various data required are as under:
System data: It includes: number of buses-n, number of PV buses, number of
loads, number of transmission lines, number of transformers, number of shunt
elements, the slack bus number, voltage magnitude of slack bus (angle is generally
taken as 0o), tolerance limit, base MVA, and maximum permissible number of
iterations.
Generator bus data: For every PV bus i, the data required includes the bus
number, active power generation PGi, the specified voltage magnitude Vi , sp , minimum

reactive power limit Qi,min, and maximum reactive power limit Qi,max.
Load data: For all loads the data required includes the the bus number, active
power demand PDi, and the reactive power demand QDi.
Transmission line data: For every transmission line connected between buses
i and k the data includes the starting bus number i, ending bus number k,.resistance of
the line, reactance of the line and the half line charging admittance.
Transformer data:
For every transformer connected between buses i and k the data to be given includes:
the starting bus number i, ending bus number k, resistance of the transformer,
reactance of the transformer, and the off nominal turns-ratio a.
Shunt element data: The data needed for the shunt element includes the bus
number where element is connected, and the shunt admittance (Gsh + j Bsh).

GAUSS – SEIDEL (GS) METHOD

The GS method is an iterative algorithm for solving non linear algebraic equations.
An initial solution vector is assumed, chosen from past experiences, statistical data or
from practical considerations. At every subsequent iteration, the solution is updated

8
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

till convergence is reached. The GS method applied to power flow problem is as

discussed below.

Case (a): Systems with PQ buses only:

Initially assume all buses to be PQ type buses, except the slack bus. This means that
(n–1) complex bus voltages have to be determined. For ease of programming, the
slack bus is generally numbered as bus-1. PV buses are numbered in sequence and PQ
buses are ordered next in sequence. This makes programming easier, compared to
random ordering of buses. Consider the expression for the complex power at bus-i,
given from (7), as:
*
n
Si = Vi Y ij V j
j = 1

This can be written as

n
Si* = Vi* Yij V j (15)
j =1

Since S i* = Pi – jQi, we get,

Pi − jQi n
= Yij V j
Vi* j =1

So that,
n
Pi − jQi
= Y ii V i + Y ij V j (16)
Vi* j =1
j ≠ i

Rearranging the terms, we get,

1 Pi − jQ n
Vi = i
− Yi jV j ∀ i = 2, 3,………..n (17)
Y ii V i* j =1
j≠ i

Equation (17) is an implicit equation since the unknown variable, appears on both
sides of the equation. Hence, it needs to be solved by an iterative technique. Starting
from an initial estimate of all bus voltages, in the RHS of (17) the most recent values
of the bus voltages is substituted. One iteration of the method involves computation of
all the bus voltages. In Gauss–Seidel method, the value of the updated voltages are
used in the computation of subsequent voltages in the same iteration, thus speeding up

9
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

convergence. Iterations are carried out till the magnitudes of all bus voltages do not
change by more than the tolerance value. Thus the algorithm for GS method is as
under:

Algorithm for GS method

1. Prepare data for the given system as required.
2. Formulate the bus admittance matrix YBUS. This is generally done by the
rule of inspection.
3. Assume initial voltages for all buses, 2,3,…n. In practical power systems,
the magnitude of the bus voltages is close to 1.0 p.u. Hence, the complex
bus voltages at all (n-1) buses (except slack bus) are taken to be 1.0 ∠ 00.
This is normally refered as the flat start solution.
4. Update the voltages. In any (k + 1)
st
iteration, from (17) the voltages are
given by

1 Pi − jQi i −1 n
Vi ( k +1) = − Yi j V j(k +1) − Yij Vj(k) ∀ i=2,3,…n (18)
Yii (Vi ( k ) ) * j =1 j = i +1

Here note that when computation is carried out for bus-i, updated values
are already available for buses 2,3….(i-1) in the current (k+1)st iteration.
Hence these values are used. For buses (i+1)…..n, values from previous,
kth iteration are used.
5. Continue iterations till
∆Vi ( k +1) = Vi ( k +1) − Vi ( k ) < ε ∀ i = 2,3,…n (19)

Where, ε is the tolerance value. Generally it is customary to use a value of

0.0001 pu.
6. Compute slack bus power after voltages have converged using (15)
[assuming bus 1 is slack bus].
n
S1* = P1 – jQ1 = V1* Y1 j V j (20)
j =1

7. Compute all line flows.

8. The complex power loss in the line is given by Sik + Ski. The total loss in
the system is calculated by summing the loss over all the lines.

10
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

Case (b): Systems with PV buses also present:

At PV buses, the magnitude of voltage and not the reactive power is specified. Hence
it is needed to first make an estimate of Qi to be used in (18). From (15) we have
n
Qi = – Im Vi* Yij V j
j =1

Where Im stands for the imaginary part. At any (k+1)st iteration, at the PV bus-i,
i −1 n
Qi( k +1) = – Im (Vi ( k ) ) * Yij V j( k +1) + (Vi ( k ) ) * Yij V j( k ) (21)
j =1 j =i

The steps for ith PV bus are as follows:

1. Compute Qi( k +1) using (21)

2. Calculate Vi using (18) with Qi = Qi( k +1)

3. Since Vi is specified at the PV bus, the magnitude of Vi obtained in step 2

has to be modified and set to the specified value Vi , sp . Therefore,

Vi ( k +1) = Vi , sp ∠δ i( k +1) (22)

The voltage computation for PQ buses does not change.

Case (c): Systems with PV buses with reactive power generation limits specified:
In the previous algorithm if the Q limit at the voltage controlled bus is violated during
any iteration, i.e Qi( k +1) computed using (21) is either less than Qi, min or greater than
Qi,max, it means that the voltage cannot be maintained at the specified value due to
lack of reactive power support. This bus is then treated as a PQ bus in the (k+1)st
iteration and the voltage is calculated with the value of Qi set as follows:
If Qi < Qi,min If Qi > Qi,max
Then Qi = Qi,min. Then Qi = Qi,max.
(23)
If in the subsequent iteration, if Qi falls within the limits, then the bus can be switched
back to PV status.

Acceleration of convergence
It is found that in GS method of load flow, the number of iterations increase with
increase in the size of the system. The number of iterations required can be reduced if

11
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

the correction in voltage at each bus is accelerated, by multiplying with a constant ,

called the acceleration factor. In the (k+1)st iteration we can let
(
Vi ( k + 1) ( accelerate d ) = Vi ( k ) + α Vi ( k + 1) − Vi ( k ) ) (24)

where is a real number. When =1, the value of Vi ( k +1) is the computed value. If
1< < 2, then the value computed is extrapolated. Generally is taken between 1.2
to 1.6, for GS load flow procedure. At PQ buses (pure load buses) if the voltage
magnitude violates the limit, it simply means that the specified reactive power
demand cannot be supplied, with the voltage maintained within acceptable limits.

Examples on GS load flow analysis:

Example-1: Obtain the voltage at bus 2 for the simple system shown in Fig 2, using
the Gauss–Seidel method, if V1 = 1 ∠ 00 pu.

Fig : System of Example 1

Solution:
Here the capacitor at bus 2, injects a reactive power of 1.0 pu. The complex power
injection at bus 2 is
S2 = j1.0 – (0.5 + j 1.0) = – 0.5 pu.
V 1 = 1 ∠ 00
− j2 j2
YBUS =
j2 − j2

1 P2 − jQ2
V2( k +1) = − Y21 V1
Y22 ( )
V2( k )
*

Since V1 is specified it is a constant through all the iterations. Let the initial voltage at
bus 2, V20 = 1 + j 0.0 = 1 ∠ 00 pu.

12
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

− 0.5
V21 =
1
− j 2 1∠0 0
(
− j 2 × 1∠0 0 )
= 1.0 – j0.25 = 1.030776 ∠ – 14.0360

− 0.5
V22 =
1
− j 2 1.030776∠14.036 0
(
− j 2 × 1∠0 0 )
= 0.94118 – j 0.23529 = 0.970145 ∠ –14.0360

− 0.5
V23 =
1
− j 2 0 .970145∠14.036 0
(
− j 2 × 1 ∠0 0 )
= 0.9375 – j 0.249999 = 0.970261 ∠ –14.9310

− 0.5
V24 =
1
− j 2 0 .970261∠14.9310
(
− j 2 × 1∠0 0 )
= 0.933612 – j 0.248963 = 0.966237 ∠ –14.9310

− 0.5
V25 =
1
− j 2 0 .966237∠14.9310
(
− j 2 × 1∠0 0 )
= 0.933335 – j 0.25 = 0.966237 ∠ –14.9950

Since the difference in the voltage magnitudes is less than 10-6 pu, the iterations can
be stopped. To compute line flow
V1 − V2 1∠0 0 − 0.966237 ∠ − 14.995 0
I 12 = =
Z 12 j 0.5
= 0.517472 ∠ −14.9310
0 0
S12 = V1 I 12* = 1 ∠ 0 × 0.517472 ∠ 14.931
= 0.5 + j 0.133329 pu
V2 − V1 0.966237 ∠ − 14.995 0 − 1∠0 0
I 21 = =
Z 12 j 0.5
= 0.517472 ∠ −194.930

S 21 = V2 I 21
*
= – 0.5 + j 0.0 pu
The total loss in the line is given by
S12 + S21 = j 0.133329 pu
Obviously, it is observed that there is no real power loss, since the line has no
resistance.

13
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

Example-2:
For the power system shown in fig. below, with the data as given in tables below,
obtain the bus voltages at the end of first iteration, by applying GS method.

Power System of Example 2

Line data of example 2

R X BC
SB EB
(pu) (pu) 2
1 2 0.10 0.40 -
1 4 0.15 0.60 -
1 5 0.05 0.20 -
2 3 0.05 0.20 -
2 4 0.10 0.40 -
3 5 0.05 0.20 -

Bus data of example 2

PG QG PD QD VSP
Bus No.
(pu) (pu) (pu) (pu) (pu)
1 - - - - 1.02 0o
2 - - 0.60 0.30 - -
3 1.0 - - - 1.04 -
4 - - 0.40 0.10 - -
5 - - 0.60 0.20 - -

Solution: In this example, we have,

• Bus 1 is slack bus, Bus 2, 4, 5 are PQ buses, and Bus 3 is PV bus
• The lines do not have half line charging admittances
P2 + jQ2 = PG2 + jQG2 – (PD2 + jQD2) = – 0.6 – j0.3

14
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

P3 + jQ3 = PG3 + jQG3 – (PD3 + jQD3) = 1.0 + jQG3

Similarly P4 + jQ4 = – 0.4 – j0.1, P5 + jQ5 = – 0.6 – j0.2
The Ybus formed by the rule of inspection is given by:
2.15685 -0.58823 0.0+j0.0 -0.39215 -1.17647
-j8.62744 +j2.35294 +j1.56862 +j4.70588
-0.58823 2.35293 -1.17647 -0.58823 0.0+j0.0
+j2.35294 -j9.41176 +j4.70588 +j2.35294
0.0+j0.0 -1.17647 2.35294 0.0+j0.0 -1.17647
Ybus =
+j4.70588 -j9.41176 +j4.70588
-0.39215 -0.58823 0.0+j0.0 0.98038 0.0+j0.0
+j1.56862 +j2.35294 -j3.92156
-1.17647 0.0+j0.0 -1.17647 0.0+j0.0 2.35294
+j4.70588 +j4.70588 -j9.41176

The voltages at all PQ buses are assumed to be equal to 1+j0.0 pu. The slack bus
voltage is taken to be V10 = 1.02+j0.0 in all iterations.

1 P2 − jQ2
V21 = o*
− Y21 V1o − Y 23 V30 − Y24 V40 − Y25 V50
Y22 V2

− 0.6 + j 0.3
=
1
Y22 1.0 − j 0.0
{
− (− 0.58823 + j 2.35294) × 1.02∠0 o }
{ } {
− (− 1.17647 + j 4.70588) × 1.04∠0 o − (− 0.58823 + j 2.35294) × 1.0∠0 0 }]
= 0.98140 ∠ −3.0665o = 0.97999 – j0.0525
Bus 3 is a PV bus. Hence, we must first calculate Q3. This can be done as under:
Q3 = V3 V1 (G31 sin δ 31 − B31 cos δ 31 ) + V3 V2 (G32 sin δ 32 − B32 cos δ 32 )

+ V3
2
(G33 sin δ 33 − B33 cos δ 33 ) + V3 V4 (G34 sin δ 34 − B34 cos δ 34 )

+ V3 V5 (G35 sin δ 35 − B35 cos δ 35 )

We note that 1 = 0o; 2 = –3.0665o; 3 = 0o; 4 = 0o and 5 = 0o

∴ 31 = 33 = 34 = 35 = 0o ( ik = i – k); 32 = 3.0665o
Q3 = 1.04 [1.02 (0.0+j0.0) + 0.9814 {–1.17647 × sin(3.0665o) – 4.70588
×cos(3.0665o)}+1.04{–9.41176 ×cos(0o)}+1.0 {0.0 + j0.0}+1.0{–4.70588×cos(00)}]
= 1.04 [–4.6735 + 9.78823 – 4.70588] = 0.425204 pu.

1 P3 − jQ3
V31 = − Y31 V1o − Y 32 V21 − Y34 V40 − Y35 V50
Y33 V3o *

15
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

1 1.0 − j 0.425204
=
Y33 1.04 − j 0.0
{
− (− 1.7647 + j 4.70588) × (0.98140∠ − 3.0665 o ) }
{
− (− 1.17647 + j 4.70588) × (1∠0 o )}]
= 1.05569 ∠ 3.077o = 1.0541 + j0.05666 pu.
Since it is a PV bus, the voltage magnitude is adjusted to specified value and V31 is

computed as: V31 = 1.04 ∠3.077 0 pu

1 P4 − jQ4
V41 = − Y41 V1o − Y 42 V21 − Y43 V31 − Y45 V50
Y44 V4o *

1 − 0.4 + j 0.1
=
Y44 1.0 − j 0.0
{
− (− 0.39215 + j1.56862 ) × 1.02∠0 o }
{
− (− 0.58823 + j 2.35294 ) × (0.98140∠ − 3.0665 o ) }]
0.45293 − j 3.8366
= = 0.955715 ∠ –7.303o pu = 0.94796– j0.12149
0.98038 − j 3.92156

1 P5 − jQ5
V51 = o*
− Y51 V1o − Y 52 V21 − Y53 V31 − Y54 V41
Y55 V5

1 − 0.6 + j 0.2
=
Y55 1.0 − j 0.0
{
− (− 1.17647 + j 4.70588) × 1.02∠0 o }
− {(− 1.17647 + j 4.70588) × 1.04∠3.077 o } ]
= 0.994618 ∠ −1.56o = 0.994249 – j0.027
Thus at end of 1st iteration, we have,
V1 = 1.02 ∠ 00 pu V2 = 0.98140 ∠ –3.0660 pu
V3 = 1.04 ∠ 3.0770 pu V4 = 0.955715 ∠ –7.3030 pu
and V5 = 0.994618 ∠ –1.560 pu

Example-3:
Obtain the load flow solution at the end of first iteration of the system with data as
given below. The solution is to be obtained for the following cases
(i) All buses except bus 1 are PQ Buses
(ii) Bus 2 is a PV bus whose voltage magnitude is specified as 1.04 pu
(iii) Bus 2 is PV bus, with voltage magnitude specified as 1.04 and 0.25 Q2 1.0
pu.

16
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

Fig. System for Example 3

Table: Line data of example 3

R X
SB EB
(pu) (pu)
1 2 0.05 0.15
1 3 0.10 0.30
2 3 0.15 0.45
2 4 0.10 0.30
3 4 0.05 0.15

Table: Bus data of example 3

Pi Qi Vi
Bus No.
(pu) (pu)
1 – – 1.04 ∠ 00
2 0.5 – 0.2 –
3 – 1.0 0.5 –
4 – 0.3 – 0.1 –

Solution: Note that the data is directly in terms of injected powers at the buses. The
bus admittance matrix is formed by inspection as under:

3.0 – j9.0 –2.0 + j6.0 – 1.0 + j3.0 0

–2.0 + j6.0 3.666 – j11.0 – 0.666 + j2.0 – 1.0 + j3.0
YBUS =
–1.0 + j3.0 –0.666 + j2.0 3.666 – j11.0 –2.0 + j6.0
0 –1.0 + j3.0 – 2.0 + j6.0 3.0 – j9.0

Case(i): All buses except bus 1 are PQ Buses

Assume all initial voltages to be 1.0 ∠ 00 pu.

1 P2 − jQ2
V21 = o*
− Y21 V1o − Y 23 V30 − Y24 V40
Y22 V2

17
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

1 0.5 + j 0.2
=
Y22 1.0 − j 0.0
{ (
− (− 2.0 + j 6.0 ) × 1.04∠0 o )}
{ ( )} {
− (− 0.666 + j 2.0 ) × 1.0∠0 o − (− 1.0 + j 3.0 ) × 1.0∠0 0 ( )}]
= 1.02014 ∠ 2.605o

1 P3 − jQ3
V31 = o*
− Y31 V1o − Y 32 V21 − Y34 V40
Y33 V3

1 − 1.0 − j 0.5
=
Y33 1.0 − j 0.0
{
− (− 1.0 + j 3.0 ) × (1.04∠0.0 o ) }
{ } {
− (− 0.666 + j 2.0 ) × (1.02014∠2.605o ) − (− 2.0 + j 6.0 ) × 1.0 ∠0 0 ( )}]
= 1.03108 ∠ – 4.831o

1 P4 − jQ4
V41 = o*
− Y41 V1o − Y 42 V21 − Y43 V31
Y44 V4

1 0.3 + j 0.1
=
Y44 1.0 − j 0.0
{ (
− (− 1.0 + j3.0 ) × 1.02014∠2.605 o )}
− {(− 2.0 + j 6.0) × (1.03108∠ − 4.831o )} ]
= 1.02467 ∠ −0.51o
Hence
V11 = 1.04 ∠ 00 pu V21 = 1.02014 ∠ 2.6050 pu

V31 = 1.03108 ∠ –4.8310 pu V41 = 1.02467 ∠ –0.510 pu

Case(ii): Bus 2 is a PV bus whose voltage magnitude is specified as 1.04 pu

We first compute Q2.
Q2 = V2 [ V1 (G21 sin δ 21 − B21 cos δ 21 ) + V2 (G22 sin δ 22 − B22 cos δ 22 )

+ V3 (G23 sin δ 23 − B23 cos δ 23 ) + V4 (G24 sin δ 24 − B24 cos δ 24 ) ]

= 1.04 [1.04 {–6.0} + 1.04 {11.0}+1.0{– 2.0} + 1.0 {–3.0}= 0.208 pu.

1 0.5 − j 0.208
V21 =
Y22 1.04 ∠0 0
{ (
− (− 2.0 + j 6.0) × 1.04∠0 o )}
{ ( )} {
− (− 0.666 + j 2.0 ) × 1.0∠0 o − (− 1.0 + j3.0) × 1.0∠0 0 ( )}]
= 1.051288 + j0.033883
The voltage magnitude is adjusted to 1.04. Hence V21 = 1.04 ∠ 1.8460

18
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

V31 =
1 − 1.0 − j 0.5
Y33 1.0 ∠0 0
{
− (− 1.0 + j 3.0 ) × (1.04∠0.0 o ) }
{ } { (
− (− 0.666 + j 2.0 ) × (1.04∠1.846 o ) − (− 2.0 + j 6.0 ) × 1.0 ∠0 0 )}]
= 1.035587 ∠ – 4.951o pu.

1 0.3 + j 0.1
V41 =
Y44 1.0 − j 0.0
{ (
− (− 1.0 + j 3.0 ) × 1.04∠1.846 o )}
{
− (− 2.0 + j 6.0 ) × (1.035587∠ − 4.951o ) }]
= 0.9985 ∠ – 0.178o
Hence at end of 1st iteration we have:
V11 = 1.04 ∠ 00 pu V21 = 1.04 ∠ 1.8460 pu

V31 = 1.035587 ∠ –4.9510 pu V41 = 0.9985 ∠ –0.1780 pu

Case (iii):Bus 2 is PV bus, with voltage magnitude specified as 1.04 & 0.25 Q2 1 pu.
If 0.25 Q2 1.0 pu then the computed value of Q2 = 0.208 is less than the lower
limit. Hence, Q2 is set equal to 0.25 pu. Iterations are carried out with this value of Q2.
The voltage magnitude at bus 2 can no longer be maintained at 1.04. Hence, there is
no necessity to adjust for the voltage magnitude. Proceeding as before we obtain at
the end of first iteration,
V11 = 1.04 ∠ 00 pu V21 = 1.05645 ∠ 1.8490 pu

V31 = 1.038546 ∠ –4.9330 pu V41 = 1.081446 ∠ 4.8960 pu

Limitations of GS load flow analysis:

GS method is very useful for very small systems. It is easily adoptable, it can be
generalized and it is very efficient for systems having less number of buses.
However, GS LFA fails to converge in systems with one or more of the features as
under:
• Systems having large number of radial lines
• Systems with short and long lines terminating on the same bus
• Systems having negative values of transfer admittances
• Systems with heavily loaded lines, etc.

GS method successfully converges in the absence of the above problems. However,

convergence also depends on various other set of factors such as: selection of slack
bus, initial solution, acceleration factor, tolerance limit, level of accuracy of results
needed, type and quality of computer/ software used, etc.

19
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

NEWTON –RAPHSON METHOD

Newton-Raphson (NR) method is used to solve a system of non-linear algebraic

equations of the form f(x) =0. Consider a set of n non-linear algebraic equations given
by
f i ( x1 , x 2 .......x n ) = 0 i = 1,2......n (25)
0 0 0
Let x1 , x2 ...........xn , be the initial guess of unknown variables and
0 0 0
∆x1 , ∆x2 .......∆x n be the respective corrections. Therefore,
0 0 0 0 0 0
f i ( x1 + ∆x1 , x 2 + ∆x2 .........xn + ∆x n ) = 0 i = 1,2......n (26)
The above equation can be expanded using Taylor’s series to give
0
0 0 0 ∂f i 0 ∂f i 0 ∂f i 0
f i ( x1 , x2 .......xn ) + ∆x1 + ∆x 2 + ..... + ∆x n
∂x1 ∂x 2 ∂x n

+ Higher order terms = 0 ∀ i = 1,2......n (27)

0 0 0
∂f i ∂f i ∂f i
Where, , , ……..… are the partial derivatives of fi with respect
∂x1 ∂x 2 ∂x n
0 0 n
to x1 , x 2 .......x n respectively, evaluated at ( x1 , x2 ...........x0 ). If the higher order terms
are neglected, then (27) can be written in matrix form as
0 0 0
∂f 1 ∂f 1 ∂f 1
. . .
f1
0 ∂x1 ∂x 2 ∂x n ∆x1
0
0 0 0
f2
0 ∂f 2 ∂f 2 ∂f 2 ∆x 2
0
. . .
. ∂x1 ∂x 2 ∂x n .
+ . . . . . . =0 (28)
. .
. . . . . .
. .
0 . . . . . . 0
fn ∆x n
∂f n ∂f n ∂f n
. . .
∂x1 ∂x 2 ∂x n

In vector form (28) can be written as

F 0 + J 0 ∆X 0 = 0
Or F 0 = − J 0 ∆X 0
Or ∆X 0 = −[ J 0 ] −1 F 0 (29)

And X 1 = X 0 + ∆X 0 (30)

20
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

Here, the matrix [J] is called the Jacobian matrix. The vector of unknown variables is
updated using (30). The process is continued till the difference between two
successive iterations is less than the tolerance value.

NR method for load flow solution in polar coordinates

In application of the NR method, we have to first bring the equations to be solved, to
the form f i ( x1 , x 2 ,...x n ) = 0 , where x1 , x 2 ,...x n are the unknown variables to be

determined. Let us assume that the power system has n1 PV buses and n2 PQ buses.
In polar coordinates the unknown variables to be determined are:
(i) δ i , the angle of the complex bus voltage at bus i, at all the PV and PQ buses. This

gives us n1 + n2 unknown variables to be determined.

(ii) Vi , the voltage magnitude of bus i, at all the PQ buses. This gives us n2 unknown

variables to be determined.
Therefore, the total number of unknown variables to be computed is: n1 + 2n2 , for

which we need n1 + 2n2 consistent equations to be solved. The equations are given
by,
∆Pi = Pi , sp − Pi ,cal = 0 (31)

∆Qi = Qi , sp − Qi ,cal = 0 (32)

Where Pi , sp = Specified active power at bus i

Qi , sp = Specified reactive power at bus i

Pi ,cal = Calculated value of active power using voltage estimates.

Qi ,cal = Calculated value of reactive power using voltage estimates

∆P = Active power residue

∆Q = Reactive power residue
The real power is specified at all the PV and PQ buses. Hence (31) is to be solved at
all PV and PQ buses leading to n1 + n2 equations. Similarly the reactive power is

specified at all the PQ buses. Hence, (32) is to be solved at all PQ buses leading to n2
equations.

21
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

We thus have n1 + 2n2 equations to be solved for n1 + 2n2 unknowns. (31) and (32)
are of the form F(x) = 0. Thus NR method can be applied to solve them. Equations
(31) and (32) can be written in the form of (30) as:
∆P J J2 ∆δ
= 1 (33)
∆Q J3 J4 ∆V

Where J 1 , J 2 , J 3 , J 4 are the negated partial derivatives of ∆P and ∆Q with respect

to corresponding δ and V . The negated partial derivative of ∆P , is same as the partial

derivative of Pcal, since Psp is a constant. The various computations involved are
discussed in detail next.

Computation of Pcal and Qcal:

The real and reactive powers can be computed from the load flow equations as:
n
Pi ,Cal = Pi = Vi Vk (Gik cos δ ik + Bik sin δ ik )
k =1

n
Vi Vk (Gik cos δ ik + Bik sin δ ik )
2
= Gii Vi + (34)
k =1
k ≠i

n
Qi ,Cal = Qi = Vi Vk (Gik sin δ ik − Bik cos δ ik )
k =1

n
Vi Vk (Gik sin δ ik − Bik cos δ ik )
2
= − Bii Vi + (35)
k =1
k ≠i

The powers are computed at any (r + 1)st iteration by using the voltages available from
previous iteration. The elements of the Jacobian are found using the above equations
as:

Elements of J1
∂Pi n
= Vi Vk {Gik (− sin δ ik ) + Bik cos δ ik }
∂δ i k =1
k ≠i

2
= −Qi − Bii Vi

∂Pi
= Vi Vk (Gik (− sin δ ik )(−1) + Bik (cos δ ik )(−1) )
∂δ k

22
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

Elements of J3
∂Qi n
Vi Vk (Gik cos δ ik + Bik sin δ ik ) = Pi − Gii Vi
2
=
∂δ i k =1
k ≠i

∂Qi
= − Vi Vk (Gik cos δ ik + Bik sin δ ik )
∂δ k
Elements of J2
∂Pi n
Vk (Gik cos δ ik + Bik sin δ ik ) = Pi + Vi Gii
2 2
Vi = 2 Vi Gii + Vi
∂ Vi k =1
k ≠i

∂Pi
Vk = Vi Vk (Gik cos δ ik + Bik sin δ ik )
∂ Vk

Elements of J4
∂Pi n
Vi Vk (Gik sin δ ik − Bik cos δ ik ) = Qi − Vi Bii
2 2
Vi = −2 Vi Bii +
∂ Vi k =1
k ≠i

∂Qi
Vk = Vi Vk (Gik sin δ ik − Bik cos δ ik )
∂ Vk

Thus, the linearized form of the equation could be considered again as:
∆δ
∆P H N ∆V
=
∆Q M L V

The elements are summarized below:

∂Pi 2
(i) H ii = = −Qi − Bii Vi
∂δ i

∂Pi
(ii) H ik = = a k f i − bk e i = Vi V k (Gik sin δ ik − Bik cos δ ik )
∂δ k

∂Pi 2
(iii) N ii = Vi = Pi + Gii Vi
∂ Vi

∂Pi
(iv) N ik = Vk = a k ei + bk f i = Vi Vk (Gik cos δ ik + Bik sin δ ik )
∂ Vk

∂Qi 2
(v) M ii = = Pi − Gii Vi
∂δ i

23
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

∂Qi
(vi) M ik = = −(a k ei + bk f i ) = − N ik
∂δ k

∂Qi 2
(vii) Lii = Vi = Qi − Bii Vi
∂ Vi

∂Qi
(viii) Lik = Vk = a k f i − bk ei = H ik
∂ Vk

In the above equations,

Yik = Gik + jBik

ek + jf k = Vk (cos δ k + j sin δ k )

And a k + jbk = (Gik + jBik )(ek + jf k ) (36)

If Yik = 0.0 + j 0.0 (if there is no line between buses i and k ) then the corresponding
off-diagonal elements in the Jacobian matrix will also be zero. Hence, the Jacobian is
also a sparse matrix.

Size of the sub-matrices of the Jacobian: The dimensions of the various sub-
matrices are as per the table below:

Matrix size
H (n1+n2) × (n1+n2)
N (n1+n2) × (n2)
M (n2) × (n1+n2)
L (n2) × (n2)
J (n1+2n2) × (n1+2n2)
P (n1+n2) × 1
Q n2 × 1
(n1+n2) × 1
∆V / V n2 × 1

24
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

ALGORITHM FOR NR METHOD

IN POLAR COORDINATES

1. Formulate the YBUS

2. Assume initial voltages as follows:
Vi = Vi , sp ∠0 0 (at all PV buses)

Vi = 1∠0 0 (at all PQ buses)

( r +1) ( r +1)
3. At (r+1)st iteration, calculate Pi at all the PV and PQ buses and Qi at all the
(r )
PQ buses, using voltages from previous iteration, Vi . The formulae to be used are
n
Vi Vk (Gik cos δ ik + Bik sin δ ik )
2
Pi ,Cal = Pi = Gii Vi +
k =1
k ≠i

n
Vi Vk (Gik sin δ ik − Bik cos δ ik )
2
Qi ,Cal = Qi = − Bii Vi +
k =1
k ≠i

4. Calculate the power mismatches (power residues)

(r ) ( r +1)
∆Pi = Pi , sp − Pi ,cal (at PV and PQ buses)
(r ) ( r +1)
∆Qi = Qi , sp − Qi ,cal (at PQ buses)
(r )
5. Calculate the Jacobian [ J (r ) ] using Vi and its elements spread over H, N, M, L
sub- matrices using the relations derived as in (36).
6. Compute

∆δ ( r )
∆ V (r ) ∆P ( r )
= [ J ( r ) ] −1
∆Q ( r )
V

7. Update the variables as follows:

δ i ( r +1) = δ i ( r ) + ∆δ i ( r ) (at all buses)
( r +1) (r ) (r )
Vi = Vi + ∆ Vi

8. Go to step 3 and iterate till the power mismatches are within acceptable tolerance.

25
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

DECOUPLED LOAD FLOW

In the NR method, the inverse of the Jacobian has to be computed at every iteration.
When solving large interconnected power systems, alternative solution methods are
possible, taking into account certain observations made of practical systems. These
are,

• Change in voltage magnitude Vi at a bus primarily affects the flow of reactive

power Q in the lines and leaves the real power P unchanged. This observation
∂Qi ∂Pi
implies that is much larger than . Hence, in the Jacobian, the elements
∂Vj ∂Vj

of the sub-matrix [N ] , which contains terms that are partial derivatives of real
power with respect to voltage magnitudes can be made zero.

• Change in voltage phase angle at a bus, primarily affects the real power flow P
over the lines and the flow of Q is relatively unchanged. This observation implies
∂Pi ∂Q
that is much larger than i . Hence, in the Jacobian the elements of the sub-
∂δ j ∂δ j

matrix [M ] , which contains terms that are partial derivatives of reactive power
with respect to voltage phase angles can be made zero.

These observations reduce the NRLF linearised form of equation to

∆δ
∆P H 0
= ∆V (37)
∆Q 0 L V

From (37) it is obvious that the voltage angle corrections ∆δ are obtained using real
power residues ∆P and the voltage magnitude corrections ∆V are obtained from

reactive power residues ∆Q . This equation can be solved through two alternate
strategies as under:

26
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

Strategy-1
(i) Calculate ∆P ( r ) , ∆Q (r ) and J (r )

∆δ (r )
∆P (r )
(ii) Compute ∆ V
(r )
[ ]
= J (r )
−1

∆Q (r )
V (r )

(iii) Update δ and V .

(iv) Go to step (i) and iterate till convergence is reached.

Strategy-2

(i) Compute ∆P (r ) and Sub-matrix H ( r ) . From (37) find ∆δ (r ) = H (r ) [ ]

−1
∆P (r )

(ii) Up date δ using δ (r +1) = δ (r ) + ∆δ (r ) .

(iii) Use δ (r +1) to calculate ∆Q (r ) and L( r )

∆ V (r )
(iv) Compute (r )
[ ]
= L( r )
−1
∆Q ( r )
V

(v)Update, V (r +1) = V (r ) + ∆V (r )

(vi) Go to step (i) and iterate till convergence is reached.

In the first strategy, the variables are solved simultaneously. In the second strategy the
iteration is conducted by first solving for ∆δ and using updated values of δ to
calculate ∆ V . Hence, the second strategy results in faster convergence, compared to

the first strategy.

FAST DECOUPLED LOAD FLOW

If the coefficient matrices are constant, the need to update the Jacobian at every
iteration is eliminated. This has resulted in development of fast decoupled load Flow
(FDLF). Here, certain assumptions are made based on the observations of practical
power systems as under:

• Bij >>Gij (Since the X ratio of transmission lines is high in well designed
R
systems)

27
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

• The voltage angle difference (δ i − δ j ) between two buses in the system is very

small. This means cos (δ i − δ j ) ≅ 1 and sin (δ i − δ j ) = 0.0

2
• Qi << Bii Vi

With these assumptions the elements of the Jacobian become

H ik = Lik = − Vi Vk Bik (i ≠ k )
2
H ii = Lii = − Bii Vi

The matrix (37) reduces to

[∆P] = [Vi V j Bij′ ][∆δ ]

[∆Q] = [Vi V j Bij′′ ] ∆VV (38)

″
Where Bij′ and Bij are negative of the susceptances of respective elements of the

bus admittance matrix. In (38) if we divide LHS and RHS by Vi and assume V j ≅ 1 ,

we get,

∆P
V
[ ]
= Bij′ [∆δ ]

∆Q
V
[ ] ∆VV
= Bij′′ (39)

Equations (39) constitute the Fast Decoupled load flow equations. Further
simplification is possible by:
• Omitting effect of phase shifting transformers
• Setting off-nominal turns ratio of transformers to 1.0
• In forming Bij′ , omitting the effect of shunt reactors and capacitors which

mainly affect reactive power

• Ignoring series resistance of lines in forming the Ybus.

28
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

With these assumptions we obtain a loss-less network. In the FDLF method, the
matrices [B ′] and [B ′′] are constants and need to be inverted only once at the
beginning of the iterations.

REPRESENTATION OF TAP CHANGING TRANSFORMERS

Consider a tap changing transformer represented by its admittance connected in series

with an ideal autotransformer as shown (a= turns ratio of transformer)

Fig. 2. Equivalent circuit of a tap setting transformer

Fig. 3. -Equivalent circuit of Fig.2 above.

By equating the bus currents in both the mutually equivalent circuits as above, it can
be shown that the -equivalent circuit parameters are given by the expressions as
under:
(i) Fixed tap setting transformers (on no load)
A = Ypq/ a
B = 1/a (1/a -1) Ypq
C = (1-1/a) Ypq

29
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

(i) Tap changing under load (TCUL) transformers (on load)

A = Ypq
B = (1/a -1) (1/a + 1 – Eq/Ep) Ypq
C = (1-1/a) (Ep/Eq) Ypq

Thus, here, in the case of TCUL transformers, the shunt admittance values are
observed to be a function of the bus voltages.

COMPARISON OF LOAD FLOW METHODS

The comparison of the methods should take into account the computing time required
for preparation of data in proper format and data processing, programming ease,
storage requirements, computation time per iteration, number of iterations, ease and
time required for modifying network data when operating conditions change, etc.
Since all the methods presented are in the bus frame of reference in admittance form,
the data preparation is same for all the methods and the bus admittance matrix can be
formed using a simple algorithm, by the rule of inspection. Due to simplicity of the
equations, Gauss-Seidel method is relatively easy to program. Programming of NR
method is more involved and becomes more complicated if the buses are randomly
numbered. It is easier to program, if the PV buses are ordered in sequence and PQ
buses are also ordered in sequence.

The storage requirements are more for the NR method, since the Jacobian elements
have to be stored. The memory is further increased for NR method using rectangular
coordinates. The storage requirement can be drastically reduced by using sparse
matrix techniques, since both the admittance matrix and the Jacobian are sparse
matrices. The time taken for a single iteration depends on the number of arithmetic
and logical operations required to be performed in a full iteration. The Gauss –Seidel
method requires the fewest number of operations to complete iteration. In the NR
method, the computation of the Jacobian is necessary in every iteration. Further, the
inverse of the Jacobian also has to be computed. Hence, the time per iteration is larger
than in the GS method and is roughly about 7 times that of the GS method, in large
systems, as depicted graphically in figure below. Computation time can be reduced if

30
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

the Jacobian is updated once in two or three iterations. In FDLF method, the Jacobian
is constant and needs to be computed only once. In both NR and FDLF methods, the
time per iteration increases directly as the number of buses.

Figure 4. Time per Iteration in GS and NR methods

The number of iterations is determined by the convergence characteristic of the

method. The GS method exhibits a linear convergence characteristic as compared to
the NR method which has a quadratic convergence. Hence, the GS method requires
more number of iterations to get a converged solution as compared to the NR method.
In the GS method, the number of iterations increases directly as the size of the system
increases. In contrast, the number of iterations is relatively constant in NR and FDLF
methods. They require about 5-8 iterations for convergence in large systems. A
significant increase in rate of convergence can be obtained in the GS method if an
acceleration factor is used. All these variations are shown graphically in figure below.
The number of iterations also depends on the required accuracy of the solution.
Generally, a voltage tolerance of 0.0001 pu is used to obtain acceptable accuracy and
the real power mismatch and reactive power mismatch can be taken as 0.001 pu. Due
to these reasons, the NR method is faster and more reliable for large systems. The
convergence of FDLF method is geometric and its speed is nearly 4-5 times that of
NR method.

31
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

Figure 5. Total time of Iteration in

GS and NR methods

Figure 6. Influence of acceleration factor

on load flow methods

32
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS

FINAL WORD

In this chapter, the load flow problem, also called as the power flow problem, has
been considered in detail. The load flow solution gives the complex voltages at all the
buses and the complex power flows in the lines. Though, algorithms are available
using the impedance form of the equations, the sparsity of the bus admittance matrix
and the ease of building the bus admittance matrix, have made algorithms using the
admittance form of equations more popular.
The most popular methods are the Gauss-Seidel method, the Newton-Raphson
method and the Fast Decoupled Load Flow method. These methods have been
discussed in detail with illustrative examples. In smaller systems, the ease of
programming and the memory requirements, make GS method attractive. However,
the computation time increases with increase in the size of the system. Hence, in large
systems NR and FDLF methods are more popular. There is a trade off between
various requirements like speed, storage, reliability, computation time, convergence
characteristics etc. No single method has all the desirable features. However, NR
method is most popular because of its versatility, reliability and accuracy.

33
www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | RESULTS | FORUMS