Topic 6

Optimal Dispatch of Generation

EE 605 Power Network Analysis

Dr. Muhyaddin Jamal Rawa

Contents
1. Introduction.
2. Nonlinear Function Optimization.
3. Operating Cost of a Thermal Plant.
4. Economic Dispatch Neglecting Losses and No
Generator Limits.
5. Economic Dispatch Neglecting Losses and
Including Generator Limits.
6. Economic Dispatch Including Losses.
10/02/2019 Dr. Muhyaddin Jamal Rawa 2
1. Introduction
• In practice, power plants are not located at
the same distance from the center of loads
and their fuel costs are different.
• Under normal operating conditions, the
generation capacity is more than the total
load demand and losses.
• Therefore, there are many options for
scheduling generation.

10/02/2019 Dr. Muhyaddin Jamal Rawa 3

• In an interconnected power system, the
objective is to find the real and reactive power
scheduling of each power plant in such a way
as to minimize the operating cost.
• This means that the generator’s real and
reactive powers are allowed to vary within
certain limits so as to meet a particular load
demand within minimum fuel cost.
• This is called Optimal Power Flow (OPF)
problem.

10/02/2019 Dr. Muhyaddin Jamal Rawa 4

• The OPF is used to optimize the power flow
solution of large scale power system.
• This can be done by:
 minimizing selected objective functions (also
known as cost functions) such as economic costs,
while
 maintaining an acceptable system performance in
terms of generator capability limits and the output
of the compensating devices.

10/02/2019 Dr. Muhyaddin Jamal Rawa 5

2. Nonlinear Function Optimization
2.1 Unconstrained Parameter Optimization
• Nonlinear function optimization is an
important tool in computer-aided design and
is part of a broader class of optimization called
nonlinear programming.
• The basic goal is the minimization of some
nonlinear objective cost function subject to
nonlinear equality and inequality constraints.

10/02/2019 Dr. Muhyaddin Jamal Rawa 6

• The necessary condition to minimize the cost
function

is obtained by setting derivative of f(x) with

respect to the variables equal to zero

where the gradient ∇ vector is given by

10/02/2019 Dr. Muhyaddin Jamal Rawa 7

• The term associated with the second
derivatives is given by

the above equation results in a symmetric

matrix called the Hessian matrix of the
function.

10/02/2019 Dr. Muhyaddin Jamal Rawa 8

• Once the derivative of f(x) is vanished at local
extrema , for f(x) to have a relative
minimum, the Hessian matrix H evaluated at
must be a positive definite
matrix.
• This condition requires that all the eigenvalues
of the Hessian matrix evaluated at
be positive.

10/02/2019 Dr. Muhyaddin Jamal Rawa 9

• The unconstrained minimum of a function is
found by
1. Setting its partial derivatives (with respect to the
parameters that may be varied) equal to zero and
solving for the parameter values.
2. Among the sets of parameter values obtained, those
at which the matrix of second partial derivatives of
the cost function is positive definite are local
minima.
3. If there is a single local minimum, it is also the global
minimum.
4. Otherwise, the cost function must be evaluated at
each of the local minima to determine which one is
the global minimum.
10/02/2019 Dr. Muhyaddin Jamal Rawa 10
Example 7.1
• Find the minimum of

10/02/2019 Dr. Muhyaddin Jamal Rawa 11

• Equating the first derivatives to zero gives

• Solving this equation in Matlab (X=A\B) gives

x1=3, x2=2 and x3=5.
10/02/2019 Dr. Muhyaddin Jamal Rawa 12
• The function value at this point is

• To test if this point is a minimum, the second

derivatives are evaluated to form the Hessian
matrix

10/02/2019 Dr. Muhyaddin Jamal Rawa 13

• The eigenvalues are calculated using the
Matlab function eig(H) and are found to be
1.55, 4.0 and 6.45 which are all positive.
• Hence, the Hessian matrix is a positive definite
matrix and (3, 2, 5) is a minimum point.

10/02/2019 Dr. Muhyaddin Jamal Rawa 14

2.2 Constrained Parameter Optimization:
2.2.1 Equality Constraints
• This type of problems arises when there are
functional dependencies among the
parameters to be chosen.
• The problem is to minimize the cost function

subject to the equality constraints

10/02/2019 Dr. Muhyaddin Jamal Rawa 15

• Such problems may be solved by the Lagrange
multiplier method.
• This method provide an augmented cost
function by introducing k-vector λ of
undetermined quantities.
• The unconstrained cost function becomes

10/02/2019 Dr. Muhyaddin Jamal Rawa 16

• The resulting necessary conditions for
constrained local minima of ( ) are the
following

• Note that equation (7.10) is simply the original

constraints.

10/02/2019 Dr. Muhyaddin Jamal Rawa 17

Example 7.2
• Use the Lagrange multiplier method for
solving constrained parameter optimizations
to determine the minimum distance from
origin of the xy plane to a circle described by

The minimum distance is obtained by

minimization of the distance square, given by

10/02/2019 Dr. Muhyaddin Jamal Rawa 18

• Matlab plot command is used to plot the circle
as shown in Fig. 7.1.

• The graph clearly shows that the minimum

distance is 5, corresponding to point (4, 3).
10/02/2019 Dr. Muhyaddin Jamal Rawa 19
• Lagrange multiplier can be used to minimize
f(x, y) subject to the constraint described by
the circle equation.
• The Lagrange function is

• The necessary conditions for the extrema are

10/02/2019 Dr. Muhyaddin Jamal Rawa 20

• The solution of the above three equations will
provide optimal points.
• For this example, a direct solution can be
obtained as follows.
• Eliminating λ from the first two equations
results in

• Substituting for y in the third equation yields

10/02/2019 Dr. Muhyaddin Jamal Rawa 21

• The solution of the above quadratic equations
are (x1=4 and y1=3) or (X2=12 and y2=9).
• Thus, the corresponding extrema are at points
(4, 3) with λ=1 and point (12, 9) with λ=-3.
• It is clear from Fig. 7.1 that the minimum
distance is at point (4, 3) and the maximum
distance is at point (12, 9).

10/02/2019 Dr. Muhyaddin Jamal Rawa 22

• To distinguish these points, the second
derivatives are obtained and the Hessian
matrices evaluated at these points are
formed.
• The matrix with positive eigenvalues is a
positive definite matrix and the parameters
correspond to the minimum point.

10/02/2019 Dr. Muhyaddin Jamal Rawa 23

• In many problems, a direct solution is not
possible and the above equations are solved
iteratively.
• A significant superior method applicable to
continuous function is the Newton-Raphson
method.

10/02/2019 Dr. Muhyaddin Jamal Rawa 24

2.2 Constrained Parameter Optimization:
2.2.2 Inequality Constraints
• Practical optimization problems contain
inequality constraints and equality constraints.
• The problem is to minimize the cost function

subject to the equality constraints

and the inequality constraints

10/02/2019 Dr. Muhyaddin Jamal Rawa 25

• The Lagrange is extended to include the
inequality constraints by introducing m-vector
μ of undetermined quantities.
• The unconstrained cost function becomes

• The resulting necessary conditions for

constrained local minima of ( ) are the
following.

10/02/2019 Dr. Muhyaddin Jamal Rawa 26

• Note that equation (7.18) is simply the original
equality constraints.

10/02/2019 Dr. Muhyaddin Jamal Rawa 27

• Suppose is a relative minimum.
• The inequality constraints in (7.19) is said to
be inactive if strict inequality holds at
and μj=0.
• On the other hand, when strict equality holds,
i.e. if the constraint

• This is known as the Kuhn-Tucker necessary

condition.

10/02/2019 Dr. Muhyaddin Jamal Rawa 28

Example 7.3
• Solve Example 7.2 with an additional
inequality constraint defined below. The
problem is to find the minimum value of

subject to one equality constraint

and one inequality constraint

10/02/2019 Dr. Muhyaddin Jamal Rawa 29

• The unconstrained cost function from (7.16) is

• The resulting necessary conditions for

constrained local minima of ( ) are

10/02/2019 Dr. Muhyaddin Jamal Rawa 30

• Eliminating μ from the first two equations
result in

• From the fourth condition, we have

• Substituting for y in the above equation yields

10/02/2019 Dr. Muhyaddin Jamal Rawa 31

• Substituting for x in the previous equation to
get

• Substituting for x and y in the third condition

(equality constraint) results in an equation in
terms of λ

10/02/2019 Dr. Muhyaddin Jamal Rawa 32

• From which the following equation is obtained

• Roots of the above equation are

• Substitute for these values of λ in the

expression for x and y, the corresponding
extrema are

10/02/2019 Dr. Muhyaddin Jamal Rawa 33

• The minimum distance from the cost function
is 5.385, located at point (5, 2), and the
maximum distance is 6.71 located at point (3,
6).
• Adding the inequality constraint

to the graphs in Fig. (7.1), the solution is

verified graphically as shown in Fig. (7.2).

10/02/2019 Dr. Muhyaddin Jamal Rawa 34

10/02/2019 Dr. Muhyaddin Jamal Rawa 35
3. Operating Cost of a Thermal Plant
• The factors influencing power generation at a
minimum cost are
 operating efficiencies of generators,
 fuel cost, and
 transmission losses.
• The most efficient generator in the system
does not guarantee minimum cost as it may
be located in area where fuel cost is high.

10/02/2019 Dr. Muhyaddin Jamal Rawa 36

• Also, if the plant is located far from the load
centre, transmission losses may be
considerably higher and hence the plant may
be overly uneconomical.
• Therefore, the problem is to determine the
generation of different plants such that the
total operating cost is minimum.
• The economic cost plays an important role in
the economic scheduling.

10/02/2019 Dr. Muhyaddin Jamal Rawa 37

• The input to the thermal plant is generally
measured in Btu/h, and the output is
measured in MW.

10/02/2019 Dr. Muhyaddin Jamal Rawa 38

• A simplified input-output curve of a thermal
unit known as heat-rate curve is given in Fig.
7.3a. Converting the ordinate of heat-rate
curve from Btu/h to $/h results in the fuel-cost
curve shown in Fig. 7.3b.

10/02/2019 Dr. Muhyaddin Jamal Rawa 39

• In all practical cases, the fuel cost of generator
i can be represented as a quadratic function of
real generation.

• An important characteristic is obtained by

plotting the derivative of the fuel-cost curve
versus the real power.

10/02/2019 Dr. Muhyaddin Jamal Rawa 40

• This is known as the incremental fuel cost
curve shown in Fig. 7.4.

10/02/2019 Dr. Muhyaddin Jamal Rawa 41

• The incremental fuel-cost is a measure of how
costly it will be to produce the next increment
of power.
• The total operating cost includes
 the fuel cost,
 the cost of labour,
 supplies, and
 maintenance.

10/02/2019 Dr. Muhyaddin Jamal Rawa 42

• These costs are assumed to be a fixed
percentage of the fuel cost and are generally
included in the incremental fuel-cost curve.

10/02/2019 Dr. Muhyaddin Jamal Rawa 43

4 Economic Dispatch Neglecting Losses
and No Generator Limits
• The simplest economic dispatch problem is
the case when the transmission line losses are
neglected.
• This means that the problem does not
consider system configuration and line
impedances.

10/02/2019 Dr. Muhyaddin Jamal Rawa 44

• Hence, the model assumes that the system is
only one bus with all generation and loads
connected to it as shown in Fig. 7.5.

10/02/2019 Dr. Muhyaddin Jamal Rawa 45

• Since transmission line losses are neglected,
the total demand PD is the sum of all
generation.
• A cost function Ci is assumed to be known for
each plant.

10/02/2019 Dr. Muhyaddin Jamal Rawa 46

• The problem is to find the real power
generation for each plant such that the
objective function (i.e. total production cost)
as defined by the equation

is minimum, subject to the constraint

10/02/2019 Dr. Muhyaddin Jamal Rawa 47

 Where
Ct is the total production cost.
Ci is the production cost of the ith plant.
Pi is the generation of ith plant.
PD is the total load demand.
ng is the total number of dispatchable
generating plants.

10/02/2019 Dr. Muhyaddin Jamal Rawa 48

• A typical approach is to augment the
constraints into objective function by the
Lagrange multipliers.

• The minimum of this unconstrained function

is found at the point where the partials of the
function to its variables are zero.

10/02/2019 Dr. Muhyaddin Jamal Rawa 49

• First condition, given by (7.26), results in

• Since

then

10/02/2019 Dr. Muhyaddin Jamal Rawa 50

• Therefore, the condition for optimum dispatch
is

or

Second condition, given by (7.27), results in

10/02/2019 Dr. Muhyaddin Jamal Rawa 51

• Equation (7.30) is the equality constraint that
was to be imposed.
• In conclusion, when losses are neglected with
no generator limits, for most economic
operation,
all plants must operate at equal incremental
production cost while satisfying the equality
constraint given by (7.30).

10/02/2019 Dr. Muhyaddin Jamal Rawa 52

• In order to find the solution, (7.29) is solved
for Pi

• The relations given by (7.31) are known as the

coordination equations.
• They are function of λ.

10/02/2019 Dr. Muhyaddin Jamal Rawa 53

• An analytical solution can be obtained for λ by
substituting for Pi in (7.30), i.e.

or

The value of λ found from (7.33) is substituted in

(7.31) to obtain the optimal scheduling of
generation.
10/02/2019 Dr. Muhyaddin Jamal Rawa 54
• The solution for economic dispatch neglecting
losses was found analytically.
• However, when losses are considered, the
resulting equations as seen in section 6 are
nonlinear and must be solved iteratively.
• Thus, an iterative procedure is introduce here
and (7.31) is solved iteratively.

10/02/2019 Dr. Muhyaddin Jamal Rawa 55

• In an iterative search technique, starting with
two values of λ, a better value of λ is obtained
by extrapolation, and the process is continued
until Δpi is within a specified accuracy.
• However, as mentioned before, a rapid
solution is obtained by the use of the gradient
method.
• To do this, (7.32) is written as

10/02/2019 Dr. Muhyaddin Jamal Rawa 56

• Expanding the left hand side of (7.34) using
Taylor’s series about an operating point λ(k),
and neglecting all the higher order terms
results in

10/02/2019 Dr. Muhyaddin Jamal Rawa 57

 or

and therefore

where

The process is continued until ΔP(k) is less than a

specified accuracy.
10/02/2019 Dr. Muhyaddin Jamal Rawa 58
Example 7.4
• The fuel-cost function for three thermal plants
in $/h are given by

where P1, P2 and P3 are in MW. The total load

PD is 800MW. Neglecting line losses and
generator limits, find the optimal dispatch and
the total cost in $/h by

10/02/2019 Dr. Muhyaddin Jamal Rawa 59

a. Analytical method using (7.33).
b. Graphical demonstration.
c. Iterative technique using the gradient
method.

10/02/2019 Dr. Muhyaddin Jamal Rawa 60

a. From (7.33), λ is found to be

10/02/2019 Dr. Muhyaddin Jamal Rawa 61

 Substituting for λ in the coordination
equation, given by (7.31), the optimal
dispatch is

10/02/2019 Dr. Muhyaddin Jamal Rawa 62

b. From (7.28), the necessary conditions for
optimal dispatch are

subject to

10/02/2019 Dr. Muhyaddin Jamal Rawa 63

 to demonstrate the concept of equal
incremental cost for optimal dispatch,
MATLAB plot command can be used to plot
the incremental cost of each plant on the
same graph as shown in Fig. 7.6.
To obtain a solution, various values of λ could
be tried until one is found which produces

10/02/2019 Dr. Muhyaddin Jamal Rawa 64

10/02/2019 Dr. Muhyaddin Jamal Rawa 65
 For each λ, if

λ is increased; otherwise, if

λ is reduced.

Therefore, the horizontal dashed-line shown

in the graph is moved up or down until the
optimum point is obtained at which

10/02/2019 Dr. Muhyaddin Jamal Rawa 66

 For this example, with PD=800 MW, the
optimal dispatch is
P1=400 MW.
P2=250 MW.
P3=150 MW.
=8.5 $/MWh.

10/02/2019 Dr. Muhyaddin Jamal Rawa 67

c. For the numerical solution using the gradient
method, assume the initial value of λ(1)=6.0.
From coordination equations, given by (7.31),
P1, P2 and P3 are

10/02/2019 Dr. Muhyaddin Jamal Rawa 68

 Since PD=800 MW, the error ΔP from (7.39) is

From (7.37)

10/02/2019 Dr. Muhyaddin Jamal Rawa 69

 Therefore, the new value λ is

continuing the process, for the second

iteration, to get

10/02/2019 Dr. Muhyaddin Jamal Rawa 70

 Since ΔP(2)=0, the equality constraint is met
in two iterations. Therefore, the optimal
dispatch are

and the total fuel cost is

10/02/2019 Dr. Muhyaddin Jamal Rawa 71

Example 7.5
• Neglecting generator limits and system losses,
use dispatch program to obtain the optimal
dispatch of generation for thermal plants
specified in Example 7.4.

10/02/2019 Dr. Muhyaddin Jamal Rawa 72

10/02/2019 Dr. Muhyaddin Jamal Rawa 73
5. Economic Dispatch Neglecting Losses
and Including Generator Limits
• The power output of any generator
 should not exceed its rating
 nor should it be below that necessary for
stable boiler operation.
• Thus, the generator are restricted to lie within
given minimum and maximum limits.

10/02/2019 Dr. Muhyaddin Jamal Rawa 74

• The problem is to find the real power
generation for each plant such that
 the objective function (i.e. total production cost) as
defined by (7.23) is minimum,

 Subject to the constraint given by

 Subject to the inequality constraints given by

Where Pi(min) and Pi(max) are the minimum and maximum

generating limits for plant i, respectively.
10/02/2019 Dr. Muhyaddin Jamal Rawa 75
• The Kuhn-Tucker conditions complement the
Lagrangian conditions to include the
inequality constraints as additional terms.
• The necessary conditions for the optimal
dispatch with losses neglected become

10/02/2019 Dr. Muhyaddin Jamal Rawa 76

• The numerical solution is the same as before.
• That is, for an estimated λ, Pi are found from
the coordination equation (7.31) and iteration
is continued until

• As soon as any plant reaches a maximum or

minimum, the plant becomes pegged at the
limit.

10/02/2019 Dr. Muhyaddin Jamal Rawa 77

• In effect, the plant output becomes a constant
and only the unviolated plants must operate
at equal incremental cost.

10/02/2019 Dr. Muhyaddin Jamal Rawa 78

Example 7.6
• Find the optimal dispatch and the total cost in
$/h for thermal plants of Example 7.4 when
the total load is 975 MW with the following
generator limits in MW.

10/02/2019 Dr. Muhyaddin Jamal Rawa 79

• Assume the initial value of λ(1)=6.0. From
coordination equations given by (7.31), P1, P2
and P3 are

• Since PD=975 MW, the error ΔP from (7.39) is

10/02/2019 Dr. Muhyaddin Jamal Rawa 80

• From (7.37)

• Therefore, the new value of λ is

10/02/2019 Dr. Muhyaddin Jamal Rawa 81

• Continuing the process, for the second
iteration, to get

10/02/2019 Dr. Muhyaddin Jamal Rawa 82

• Since

the equality constraint is met in two

iterations.
• However, P1 exceeds its upper limit.
• Thus, this plant is pegged at its upper limit.
• Hence, P1=450 MW and is kept at this value.

10/02/2019 Dr. Muhyaddin Jamal Rawa 83

• Thus, the new imbalance in power is

• From (7.37)

• Therefore, the new value of λ is

10/02/2019 Dr. Muhyaddin Jamal Rawa 84

• The results after the third iteration are

and

10/02/2019 Dr. Muhyaddin Jamal Rawa 85

• Since
 ΔP(3)=0
 the equality constraint is met.
 P2 and P3 are within their limits.
• Thus, the optimal dispatch is

10/02/2019 Dr. Muhyaddin Jamal Rawa 86

• The total fuel cost is

10/02/2019 Dr. Muhyaddin Jamal Rawa 87

• The following commands can be used to
obtain the optimal dispatch of generation
including generator limits.

10/02/2019 Dr. Muhyaddin Jamal Rawa 88

10/02/2019 Dr. Muhyaddin Jamal Rawa 89
6. Economic Dispatch Including Losses
• When
 transmission distances are very small, and
 load density is very high,
transmission losses may be neglected and the
optimal generation is achieved with all plants
operating at equal incremental production
cost.

10/02/2019 Dr. Muhyaddin Jamal Rawa 90

• However, in a large interconnected network
where
 power is transmitted over long distances with
 low load density areas,
transmission losses are a major factor and
affect the optimal dispatch of generation.

10/02/2019 Dr. Muhyaddin Jamal Rawa 91

• One common practice for including the effect
of transmission losses is to
 express the total transmission loss as a quadratic
function of the generator power outputs.
• The simplest quadratic form is

10/02/2019 Dr. Muhyaddin Jamal Rawa 92

• A more general formula containing a linear
term and a constant term, referred to as
Kron’s loss formula, is

• Bij are called loss coefficients or B-coefficients.

10/02/2019 Dr. Muhyaddin Jamal Rawa 93

• B-coefficients are assumed constant and
reasonable accuracy can be expected provided
the actual operating conditions are close to
the base case where the B-constants were
computed.

10/02/2019 Dr. Muhyaddin Jamal Rawa 94

• The economic dispatching problem is to
minimize the overall generating cost Ci, which
is the function of plant output.

subject to the constrain that generation

should equal total demands plus losses, i.e.

10/02/2019 Dr. Muhyaddin Jamal Rawa 95

 satisfying the inequality constraints, expressed
as follow:

where Pi(min) and Pi(max) are the minimum and

maximum generating limits, respectively, for
plant i.

10/02/2019 Dr. Muhyaddin Jamal Rawa 96

• Using the Lagrange multiplier and adding
additional terms to include the inequality
constraints to get

10/02/2019 Dr. Muhyaddin Jamal Rawa 97

• The constraints should be understood to mean
that
μi(max)=0 when Pi<Pi(max)
μi(min)=0 when Pi>Pi(min)
• In other words, if the constraint is not
violated, its associated μ variable is zero and
the corresponding term in (7.47) does not
exist.

10/02/2019 Dr. Muhyaddin Jamal Rawa 98

• The constraint only becomes active when
violated.
• The minimum of this unconstrained function
is found at the point where the partials of the
function to its variables are zero.

10/02/2019 Dr. Muhyaddin Jamal Rawa 99

• Equations (7.50) and (7.51) imply that Pi
should not be allowed to go beyond its limit,
and when Pi is within its limits

and Kuhn-Tucker function becomes the same

as the Lagrangian one.
10/02/2019 Dr. Muhyaddin Jamal Rawa 100
• First condition, given by (7.48), results in

• Since

then

and therefore the condition for optimum

dispatch is

10/02/2019 Dr. Muhyaddin Jamal Rawa 101

• The term is known as the incremental
transmission loss.
• Second condition, given by (7.49), results in

• Equation (7.53) is precisely the equality

constraint that was to be imposed.

10/02/2019 Dr. Muhyaddin Jamal Rawa 102

• Classically, Equation (7.52) is arranged as

or

where Li is known as the penalty factor of

plant i and is given by

10/02/2019 Dr. Muhyaddin Jamal Rawa 103

• Hence, the effect of transmission loss is to
introduce a penalty factor with a value that
depends on the location of the plant.
• Equation (7.55) shows that the minimum cost
is obtained when the incremental cost of each
plant multiplied by its penalty factor is the
same for all plants.

10/02/2019 Dr. Muhyaddin Jamal Rawa 104

• The incremental production cost is given by
(7.22),

and the incremental transmission loss is

obtained from the loss formula (7.43) which
yields

10/02/2019 Dr. Muhyaddin Jamal Rawa 105

• Substituting the expression for the
incremental production cost and the
incremental transmission loss in (7.52) results
in

or

10/02/2019 Dr. Muhyaddin Jamal Rawa 106

• Expanding (7.58) to all plants results in the
following linear equations in a matrix form

or in a compact form

10/02/2019 Dr. Muhyaddin Jamal Rawa 107

• To find the optimal dispatch for an estimated
value of λ(1), the simultaneous linear equation
given by (7.60) is solved.
• In MATLAB, the following command is used.

10/02/2019 Dr. Muhyaddin Jamal Rawa 108

• Then, the iterative process is continued using
the gradient method.
• To do this, from (7.58), Pi at the kth iteration is
expressed as

10/02/2019 Dr. Muhyaddin Jamal Rawa 109

• Substituting for Pi from (7.61) into (7.53)
results in

or

10/02/2019 Dr. Muhyaddin Jamal Rawa 110

• Expanding the left-hand side of the above
equation in Taylor’s series about an operating
point λ(k), and neglecting the higher-order
terms results in

or

10/02/2019 Dr. Muhyaddin Jamal Rawa 111

 where

and therefore,

where

10/02/2019 Dr. Muhyaddin Jamal Rawa 112

• The process is continued until ΔP(k) is less than
a specified accuracy.
• If an approximate loss formula expressed by

is used, Bij=0, B00=0, and solution of the

simultaneous equation given by (7.61)
reduces to the following simple expression

10/02/2019 Dr. Muhyaddin Jamal Rawa 113

 and (7.66) reduces to

10/02/2019 Dr. Muhyaddin Jamal Rawa 114

Example 7.7
• The fuel cost in $/h of three thermal plants of
a power system are

where P1, P2 and P3 are in MW.

Plant outputs are subject to the following limits

10/02/2019 Dr. Muhyaddin Jamal Rawa 115

• For this problem, assume the real power loss
is given by the simplified expression

where the loss coefficients are specified in per

unit on a 100-MVA base.
• Determine the optimal dispatch of generation
when the total system load is 150 MW.

10/02/2019 Dr. Muhyaddin Jamal Rawa 116

• In the cost function Pi is expressed in MW.
• Therefore, the real power loss is terms of MW
generation is

10/02/2019 Dr. Muhyaddin Jamal Rawa 117

• For the numerical solution using the gradient
method, assume the initial value of

• From coordination equation, given by (7.70),

P1(1), P2(1) and P3(1) are

10/02/2019 Dr. Muhyaddin Jamal Rawa 118

• The real power loss is

• Since PD=150 MW, the error ΔP(1) from (7.68)

is

• From (7.71)

10/02/2019 Dr. Muhyaddin Jamal Rawa 119

• From (7.65)

• Therefore, the new value of λ is

10/02/2019 Dr. Muhyaddin Jamal Rawa 120

• Continuing the process, for the second
iteration, to get

• The real power loss is

10/02/2019 Dr. Muhyaddin Jamal Rawa 121

• Since PD=150 MW, the error ΔP(2) from (7.68)
is

• From (7.71)

10/02/2019 Dr. Muhyaddin Jamal Rawa 122

• From (7.65)

• Therefore, the new value of λ is

10/02/2019 Dr. Muhyaddin Jamal Rawa 123

• For the third iteration

• The real power loss is

10/02/2019 Dr. Muhyaddin Jamal Rawa 124

• Since PD=150 MW, the error ΔP(3) from (7.68)
is

• From (7.71)

10/02/2019 Dr. Muhyaddin Jamal Rawa 125

• From (7.65)

• Therefore, the new value of λ is

10/02/2019 Dr. Muhyaddin Jamal Rawa 126

• Since Δλ(3) is small, the equality constraint is
met in four iterations, and the optimal
dispatch for λ=7.6789 are

• The real power loss is

10/02/2019 Dr. Muhyaddin Jamal Rawa 127

• The total fuel cost is

10/02/2019 Dr. Muhyaddin Jamal Rawa 128

• The dispatch program can be used to find the
optimal dispatch for generation.
• The program is designed for the loss
coefficients to be expressed in per unit.
• The loss coefficients are arranged in a matrix
form with the variable name B.
• The base MVA must be specified by the
variable name basemva.
• If base MVA is not specified, it is set to 100
MVA.
10/02/2019 Dr. Muhyaddin Jamal Rawa 129
10/02/2019 Dr. Muhyaddin Jamal Rawa 130
10/02/2019 Dr. Muhyaddin Jamal Rawa 131
Example 7.8
• Fig. 7.7 shows the one-line diagram of a
power system described in Example 7.9. the B
matrices of the loss formula for this system
are found in Example 7.9. they are given in per
unit on a 100 MVA base as follows.

10/02/2019 Dr. Muhyaddin Jamal Rawa 132

• Cost functions, generators limits and total
loads are given in Example 7.7.
• Use dispatch program to obtain the optimal
dispatch of generation.

10/02/2019 Dr. Muhyaddin Jamal Rawa 133

10/02/2019 Dr. Muhyaddin Jamal Rawa 134
10/02/2019 Dr. Muhyaddin Jamal Rawa 135
Reference
• H. Saadat, Power System Analysis, 3rd ed., PSA
Publishing, 2010.
• J. Grainger and W.D. Stevenson, Power System
Analysis, Mc Graw-Hill, 1994.
• [Link]

10/02/2019 Dr. Muhyaddin Jamal Rawa 136