12

CHAPTER 2
LOAD FLOW AND OPTIMAL POWER FLOW ANALYSIS

Load flow and optimal power flow studies are necessary for transmission
pricing and wheeling. A brief review of load flow and optimal power flow
methodologies are presented here.
Load Flow
Load flow solution is a solution of the network under steady state
condition subjected to certain inequality constraints under which the
system operates. These constraints can be in the form of load magnitude,
bus voltages, reactive power generation of the generators, tap settings of
a tap-changing transformer etc. The load flow solution gives the bus
voltages and phase angles, hence the power injection at all the buses and
power flow through interconnecting transmission lines can be easily
calculated. Load flow solution is essential for designing a new power
system as well as for planning an extension or operation of the existing
one for varying demand. These analyses require number of load flow
solutions under both normal and abnormal (outage of transmission line
or outage of some generators) operating conditions. Load flow solution
also gives the initial state of the system when the transient behavior of
the system is to be studied.
The load flow solution of the power system mainly requires the following
calculations/steps:
1. Formulation of equations for the given network
2. Suitable mathematical technique for the solution of the equations

Under steady state condition, the network equations will be in the form
of simple algebraic equations. The loads and generations are
continuously changing in a real power system, but for solving load flow it
is assumed that loads and generations are fixed at a particular value
13
over a suitable period of time. E.g. half an hour or monthly etc depending
upon data

2.1 Bus Classification
10

In a power system each bus or node is associated with four quantities,
real and reactive powers, bus voltage magnitudes and its phase angles.
In a load flow solution two out of four quantities are specified and the
remaining two are to be calculated through the solution of the equations.
The buses are classified into the following three types depending upon
the quantities specified.
Load bus: At this bus the real and reactive components of power are
specified. It is desired to find out the voltage magnitude V and phase
angle through the load flow solution. Voltage at load bus can be
allowed to vary within a prescribed value e.g. 5%.
11

Generator bus or voltage-controlled bus: Here the voltage magnitude
corresponding to the generator voltage V and real power P
G

corresponding to its ratings is specified. It is required to find out the
reactive power generation Q
G
and the phase angle of the bus.
Slack/Swing or reference bus: Here the voltage magnitude V and
phase angle is specified. This will take care of the additional power
generation required and Transmission losses. It is required to find the
real and reactive power generations (P
G
, Q
G
) at this bus.
Load flow solution can be achieved by any iterative methods. There are
many kinds of iterative methods but as per the literature review the
Newton-Raphson method is normally applied.
In the load flow problem as explained above, two variables are specified
at each bus and the remaining variables are obtained through load flow
solutions. The additional variables to be specified for load flow solution

10 11
C.L.Wadhawa Electrical Power Systems, Second edition 1991


14
are the tap settings of regulating transformers, capacitances, resistances
etc. If the specified variables are allowed to vary in a region constrained
by practical considerations (upper and lower limits of real and reactive
generations, bus voltage limits and range of transformer tap settings),
these results in load flow solutions each pertaining to one set of values of
specified variables.

2.2 Newton-Raphson load flow (NRLF) Method
12

2.2.1 Calculation of Jacobian
For an N-bus power system there will be n equations for real power
injection
i
P and n-equations for reactive power
11
injection
i
Q .






. (1)
The number of equations to be solved depends upon the specifications
we have. If the total number of buses is n and number of generator buses
is m then the number of equations to be solved will be number of known
P
i
s and number of known Q
i
s. In the above conditions number of known
P
i
s are n-1 and the number of known Q
i
s are (n-m), therefore the total
number of simultaneous equations will be 2*n-m-1, and number of
unknown quantities are also 2*n-m-1. Unknowns to be calculated are
power angles () at all the buses except slack (i.e. n-1) and bus voltages
(V) at load bus (i.e. n-m). The following method known as Newton-

12
L..P.Singh Advanced Power System Analysis and Dynamics, Second edition 1986
Mohamed E. El-Hawary Electrical Power Systems Design and Analysis ,1983 by Reston Publishing
Company and Demand side management and dispatch report (3i network)
( ) | |
( ) | | N i Y V V Q Q Q
Y V V P P P
n
j
ij j i ij j i Di Gi i
n
j
ij j i ij j i Di Gi i
,....., 3 , 2 , 1 sin
cos
1
1
= = =
= =

=
=


15
Raphson method is used for solving the unknown quantities. The
problem formulation is as follows:
|
|
.
|

\
|

|
|
|
|
.
|

\
|

=
|
|
.
|

\
|

V
V
Q Q
V
P P
Q
P

. (2)
( )
( )
i i i
i i i
Q specified Q Q
P specified P P
=
=

(3)

Real power terms will be calculated for all the buses except slack bus
and reactive power terms will be calculated for all load buses. In the
above equation
vector correction the is
vector mismatch the is
|
|
.
|

\
|

|
|
.
|

\
|

V
Q
P

and
matrix Jacobian the is
|
|
|
|
.
|

\
|

=
V
Q Q
V
P P
J

.. (4)
The elements of the Jacobian matrix can be calculated using the
following equations

16

.. (5)


Procedure for this iterative method is for the given system first the Y-bus
matrix has to be formed.
B j G Y + =
Where
Y is a bus admittance matrix
G is real part of Y-bus matrix
B is imaginary part of Y-bus matrix
The resistance and reactance of each line have been given for any system
from which the admittance matrix can be formed.


( )
( )
( )
( )
( )
( )
ij j i ij i
j
i
ii i ij j i ij j
n
j i
i
ij j i ij j i
j
i
ii i i
i
i
ij j i ij i
j
i
ii i ij j i ij j
n
j i
i
ij j i ij j i
j
i
ii i i
i
i
Y V
V
Q
B V Y V
V
Q
Y V V
Q
G V P
Q
Y V
V
P
G V Y V
V
P
Y V V
P
B V Q
P

+ =

+ =

=
=
sin
sin
cos
cos
cos
sin
1
2
1
2
17
2.2.2 Iterative Algorithm for N-R Method
1. With voltage and angle (usually 0 = ) at slack bus fixed, assume
voltage magnitude and power angles at PQ buses and at all PV
buses. Generally flat voltage start will be used.
2. Compute
i
P for all buses except slack bus and
i
Q for all PQ
buses using Eq. (3). If all the values are less than the prescribed
tolerance, stop the iterations.
3. If the convergence criterion is not satisfied, evaluate elements of
the jacobian using Eq. (5)
4. Solve the Eq. (2) for correction vector.
5. Update voltage angles and magnitudes by adding the
corresponding changes to the previous values and return to step 2.
2.2.3 Adjustments in NRLF Solution
In the present work some constraints have been applied on voltages at
the PQ buses, reactive power at the PV buses transformer taps so as to
keep the system healthy.
2.2.3.1 Constraints on reactive power
Since Q
i
at generator buses is not given, it is calculated after each
iteration at all generator buses and it is been checked for the condition

max min i i i
Q Q Q < < (6)
If the above condition is satisfied then the i
th
bus will remain as
generator bus and there will be no change in the procedure. If the
condition violates then Q
i
will be set at the limit values it can be either
lower limit or upper limit depending upon the violation and i
th
bus will be
treated as load bus from the next iteration. Because of this change of bus
type one additional equation corresponding to Q
i
will be added to NRLF
scheme.

18
2.2.3.2 Constraints on bus voltages:
After each iteration voltages and angles are updated with the obtained
correction matrix. Voltages have to satisfy the following condition

max min i i i
V V V < < . (7)

If it satisfies the condition then there will be no change in the process, if
it violates then corresponding voltage has to be fixed at its lower or upper
limits depending upon the violation and the bus type is then changed to
generator type. Because of this, additional equation corresponding to Q
i
will be deleted from NRLF scheme, so the size of Jacobian matrix will be
decreased.
2.2.3.3 Constraints on transformer taps
The transformer taps can have the values as follows
max min i i i
t t t < < . (8)

In brief the important step in power flow and loss calculation is as
follows--
1 Read data Line data, T/F data, Gen. Data, Load data
2 Form ] [
BUS
Y
3 Specify the initial value of voltages
0 1
0 0 0
= =
i i
V V Flat voltage start at PQ bus
4 Initialize iteration count to 0 and Compute
i
P at known value of
voltage at all the buses except slack bus and
i
Q at known value of
voltage at all the buses except PQ bus
5 Form the mismatch vector
19

specified
i
Calculated
i
specified
i
Calculated
i i
Q Q Q
P P P
=
=

6 Check for convergence



1 1
1 1
Q
P
i

Normally 00004 . = pu on 100 MVA
If the solution is obtained then go to step 11 else go to step 7.
7 Advance count K=K+1
8 Form Jacobian at
) 1 (
=
K
v v at the latest value and its inverse
| |
1
J .
9 Find | |
(

=
(

Q
P
J
v
1

10 Update

+ =
+ =
Old New
Old New
v v v

Go to step 4
11 Compute
Di i Gi
Q Q Q + = and
Di i Gi
P P P + = for slack bus
12 Do PQ to PV type bus switching according to the rules.
13 Find the power flow over individual line by following formula

ij j i j i ij
Z V V P / ) ( =
14 Real losses in an individual line are calculated as follows



) (
Ji ij Loss
P P P =
20
2.3 Optimal Power flow
Planning and operational requirements very often ask for an adjustment
of the generated powers according to certain criteria. One such very
common operational criterion is to operate the generators at minimum
generating costs. The application of such a criterion immediately
suggests variable input powers and bus voltages, which need to be
determined so that a minimum generation cost is achieved. At this point
it is not only the voltages at nodes where the loads are supplied but also
the input powers together with the corresponding voltages at the
generator nodes, is also to be determined. The degree of freedom for the
choice of inputs seems to be exceedingly large, but due to the presence of
an objective, namely to reach the minimum of the generating costs the
problem is well defined and the aim still being the same, i.e. the
determination of the nodal voltages in the system. As long as the power
flow model stays the same it is considered to be the optimal power flow
problem where the objective is a scalar function of the state variables. In
essence, any optimal power flow problem can be reduced to such a form.
But practical requirements ask for a more realistic definition, the main
addition being the statement of constraints. Fortunately, the theory
developed by Kuhn and Tucker is able to provide the optimality
conditions, which guarantees the correctness of the result in the end.
However, the optimality conditions do not offer a solution method.
2.3.1 Objective Function
Cost Minimization Function
The major component of generator operating cost is the fuel input/hour,
while maintenance contributes only to a small extent. The input-output
curve of a unit can be expressed in million kilocalories per hour or
directly in terms of rupees per hour versus output in megawatts. For
every generator, ( )
min
MW is the minimum-loading limit below which it is
uneconomical to operate the unit and ( )
max
MW is the maximum output
21
limit. The analytical operating cost can be written as ) (
Gi i
P C Rs/hour at
output ( )
Gi
P , where the suffix i stands for the unit number. It generally
Suffices to fit a second-degree polynomial, i.e.,

i Gi i
Gi
i i
d P b P a C + + =
2
Rs/hour

The slope of the cost curve i.e.
Gi i
dP dC / is called the incremental fuel cost
(IC), and is expressed in units of rupees per megawatt hour (Rs / MWh).

2.4 Problem with optimal power flow in India
As for as the applicability of the OPF in Indian condition/UPPCL network
is concerned following points to be noted
1. For optimal scheduling of generator it is necessary that the
generation should always be more than the demand so that the
optimal scheduling can be done to give the over all minimum
generation cost. This case is not true with India as generation is
rarely more than demand.
2. Non availability of the generator cost coefficient (a,b,d) required for
evaluating fuel cost and used to predict the optimal scheduling of
the generator.
3. The cost equation shows that a, b, d should be positive as for more
amount of generation more cost will be involved. After plotting the
various heat rate curves of some of the generators in UPPCL
network it is found that a, b, d are not positive. The reason behind
it is that the quality of the coal or fuel is not uniform and differs a
lot throughout the year as well as the data provided for the fuel
consumptions does not exactly reflect the amount of power
generated (manipulated data). So the exact generator cost cannot
be predicted in the form of the given equation.
22
4. The past data of generation and cost for heat rate curve for each
generator is not available.
5. An independent transmission company has nothing to do with the
optimal scheduling of the generators of generating company, where
the rate of the power purchase is already fixed.

2.5 Application of OPF in Locational Marginal Pricing
(LMP)
The LMP method is a two-part method and most advanced model for
transmission pricing. It is based on the idea that the short run marginal
cost for transmission services is the difference in the Locational prices
between the source and the destination as given by the formula-

Hourly Transmission Charges = P
Transaction
(MP
Receiving node
MP
Sending node
)


The marginal prices are calculated by running the OPF with some
constraints. The transmission charge includes the congestion charges
and an access charge, so that the transmission utility is able to recover
its total transmission ARR. In order to calculate the access charge, first,
the total congestion cost is estimated. The access charge is then set so
that the transmitting utility is able to recover the remaining revenues
after the congestion revenues have been subtracted. The access charges
can be calculated by the following formula,

Access Charge =
( )
( ) MW Peak System
Losses of Cost Rv Congestion Trm ARR . .) (


Effectively, the LMP method sets a two parts tariff with the fixed charge
being the access charge in Rs per MW-month. The variable charge
(congestion etc.) is the difference in nodal prices averaged over the month
in Rs per MWh.
23
In nodal pricing models, the first and foremost condition is to optimize
the generation cost to meet the future load demand. Due to the
limitations of non-availability of desired data OPF studies cannot be done
and hence the Nodal/ Locational marginal pricing method is not feasible.
However if an effort is made to supply the desired data, transmission
pricing using LMP can be done after taking suitable assumptions.