Professional Documents
Culture Documents
13 - Chapter 3 PDF
13 - Chapter 3 PDF
CHAPTER – 3
METHODOLOGIES
3.0 INTRODUCTION
Power Systems.
power flow equations and system and equipment operating limits. The
security requirements.
transmission losses)
2. Environmental dispatch
General goals
Among the above the following objectives are most commonly used:
NG
F ( PG ) i i PGi i pG2i (3.1)
i 1
Min F ( PG ) f ( x, u) (3.2)
g ( x, u) 0 (3.3)
h ( x, u) 0 (3.4)
Net interchange
Minimization
minimum.
42
equations [18]:
where:
N
Pi (V , ) |Vi | |Vi ||Yij | cos(i j ij ) (3.9)
i 1
N
Qi (V , ) |Vi | |Vi ||Yij | sin(i j ij ) (3.10)
i 1
=0 (3.12)
flow constraints,
i min i i max
(3.20)
line losses
Nl
PL Plk
k 1
(3.21)
phase angles as
Nl
Min PL g
i 1
k
(Vi 2 V j2 2ViV j cos(i j ) (3.23)
3.1.1 for cost minimization, with voltage and phase angle expressed in
rectangular form.
Minimization
OPF problems.
(EMSs).
the load demand for a power system while up keeping the security of
security requires keeping each device in the power system within its
marginal cost data. This marginal cost data can aid in the pricing of
phase shift angles as well. The OPF also is able to monitor system
problems. If any security problems occur, the OPF will modify its
have the same type of devices and operating requirements. The model
flow. The next generation OPF has been greater as power systems
The demand for an OPF tool has been increasing to assess the
state and recommended control actions both for off line and online
studies, since the first OPF paper was presented in 60’s. The thrust
4. How well the future OPF provide local or global control measures to
survey [33] and classified the OPF algorithms based on their solution
2. Intelligent methods.
conventional methods is further sub divided into the following [2, 3]:
simulation run, they become too slow if number of variables are large
problems. In most cases, they can find the global optimum solution.
been covered with a lucid presentation in Tabular form. This helps the
problem [29] with the main motivation being the existence of the
concept of the state and control variables, with load flow equations
providing a nodal basis for the elimination of state variables. With the
in the space of the control variables with the load flow equations and
F (PG )
all gen
Fi (PGi ) (3.24)
i
on each PQ bus
x Vi (3.25)
on each PV bus
i
k
on the Slack bus / reference
Vk bus
P net
y net
k
on each PQ bus
Qk
(3.26)
net
Pk
sch on each PV
Vk bus
the parameters are fixed, such as P and Q at each load bus in respect
u
y (3.27)
p
53
Now with this we can define a set of m equations that govern the
PGi V , PGinet
for each PQ (load) bus
g ( x, y ) QGi V , QGi net
(3.28)
PGk V , PGknet
for each PV (generator) bus
not including reference bus
Newton Power Flow. It may be noted that the reference bus power
function of the control variables and state variables. For this, cost
F (PG ) F (P
gen
i Gi ) Fref (PG ref ) (3.29)
And the first summation does not include the reference bus. The PGi s
as
Pi (v, ) P inet
( x, u, p) Fi ( PGi ) Fref [ Pref (| v |, )] [1, 2 ,.., N ]
Qi (v, ) Q i
net
gen
(3.33)
function and N Lagrange multipliers one for each of the N power flow
equations.
= 0 (3.34)
It is represented as
T
L f g
Lx 0 (3.35)
x x x
T
L f g
Lu = 0 (3.36)
u u u
L
L = g (x, u, p ) 0 (3.37)
55
function w.r.t the state variables x. Since the objective function itself
is not a function of the state variable except for the reference bus, this
becomes:
Pref
Fref (Pref )
Pref 1
f Pref
Fref (Pref )
x Pref |V1 | (3.38)
g
The term in equation (3.35) is actually the Jacobian matrix
x
for the Newton Power flow which is already known. That is:
This matrix has to be transposed for use in Eq. (3.35). Eq. (3.36)
f
Here the vector is vector of the derivatives of the objective function
u
P F1(P1 )
1
f
F2 (P2 ) (3.40)
u P
2
g
The other term in Eq. (3.36), actually consists of a matrix of all
u
Step 2: Solve for Power flow. This guarantees Eq. (3.33) is satisfied.
T 1
g f
(3.41)
x x
Step 4: Substitute from Eq. (3.41) into Eq. (3.36) and compute the
gradient.
T
L f g
L = (3.42)
u u u
where u
size.
Merits
Demerits
1) The higher the dimension of the gradient, the higher the accuracy
flow.
constraints.
algorithm for the power flow problem for a long time The Newton
possible.
61
historical data, and transformer tap ratios will be near 1.0 p.u.
L( z ) f ( x) T h( x) T g ( x) (3.44)
and g(x) only includes the active (or binding) inequality constraints.
L( z )
Gradient = L( z ) = a vector of the first partial derivatives of
zi
2 L( z ) 2 L( z ) 2 L( z )
xi x j xi j xi j a matrix of
the second
2 L( z ) 2 L( z )
0 partial
2
Hessian = L ( z ) H 0
i j i x j
z z derivatives
2 of the
L( z )
x 0 0 Lagrangian
i j
(3. 46)
62
algorithm.
x L( z* ) x L([ x* , * , * ]) 0 (3.47)
L( z* ) L([ x* , * , * ]) 0 (3.48)
L( z* ) L([ x* , * , * ]) 0 (3.49)
i * = 0 Real (3.52)
loop of the flow chart in Fig. 3.2 performs this search for the binding
or active constraints.
The Merits and Demerits of Newton Method are summarized and given
below.
Merits
applications.
6) With this method efficient and robust solutions can be obtained for
inequality constraints.
8) There is no need of user supplied tuning and scaling factors for the
optimisation process.
Demerits
1) The penalty near the limit is very small by which the optimal
grounds that
(c) The accuracy and scope of linearised model is adequate for most
engineering purposes.
It may be noted that point (a) is certainly true while point (b)
quasi linear, but it needs to be checked out for any given system and
application.
Minimise F (x x , u u )
0 0
(3.53)
Subject to g (x x , u u ) 0
0 0
(3.54)
70
h (x 0 x , u 0 u ) 0 (3.55)
g , h are the linear approximations to the original are the non linear
constraints.
The basic steps required in the L.P based OPF algorithm is as follows
[53]:
conditions.
algorithm solves the network and test operating limits in sparse form
However the power flow solution may be performed using either the
P
Q J V or u PQ J x (3.56)
or a decoupled formulation based on the modified Fast decoupled
power flow equations given as under
B P (3.57)
B V Q (3.58)
In most applications of L.P based OPF, the later model is used.
Merits
VAR compensation
reliability and very high speed enables it , suitable for real time or
Demerits
2) Although LP methods are fast and reliable, but they have some
approximations.
80
and vectors.
81
respectively, as [5]:
1
f ( x ) xT R x a T x (3.59)
2
1 T
fi ( x) x H i x biT x i 1, 2,...m (3.60)
2
S FxT x U
Let A , y Z , b K (3.62)
Fx 0
Ay b (3.63)
m m
S R i H i , w a i bi (3.65)
i 1 i 1
and U Sx w (3.66)
and
To Compute K :
1 T
f ( x x) f ( x) f x x x f x x x (3.69)
2
x U
A , (3.70)
z qJ
which consists of
S x FxT x U
and Fx x k qJ .
FxT (z qv) (q 1)U where the last equation is satisfied in the sense
u u
A (3.71)
v J
1
f ( x x) f ( x) qf x u q 2 N
2
where N uT f xx u (3.72)
f xu
q (3.73)
N
T min qJ i ,( Di Ki ) (3.74)
Using the K , x and z are solved from Eq. (3.63) and then x and z
are updated.
mentioned below.
1. Input Data
(a) n, m, p and (to replace zero, usually lies between 10-3 and 10-5)
2. Initialization
Otherwise go to step 4.
and f x u 0 . Go to part
5. Determining K
(a) Calculate q by Eq. (3.73) and then find K i from Eq. (3.74) &
Merits
very fast.
4) The method can solve both the load flow and economic dispatch
problems.
possible.
established methods.
Demerits
algorithms.
interior point rarely visits too many extreme points before an optimal
and successfully applied to optimal power flow [37, 84, and 85].
98
optimization problems in the late 1980s and early 1990s and as can
The Interior Point Method (IPM) can solve a large scale linear
In the simplex method the solution goes from corner point to corner
reaches the optimal solution after few iterations (less than the
simplex method).
the first few iterations. This feature is very important, because for
x3
x2 x13
x12
x1
x11
x0
c
x10
iterative manner until the optimal solution has been found. The
Method
Objective Function
Equality Constraints
Eq. (3.9) and (3.10) are nonlinear and can be linearized by the Taylor’s
expansion using
P (V , ) J11 J12
(3.76)
Q (V , ) J 21 J 22 V
101
where
J 11 J 12
is the Jacobian matrix
J 21 J 22
Transmission loss (PL) given in the Eq. (3.12) can be directly calculated
Inequality Constraints
1 T
Minimize x Qx c T x (3.77)
2
Subject to Ax = b, x ≥ 0 (3.78)
1 T
Maximize x Qx bT W (3.79)
2
where
( X k )T S k
k
(3.84)
n
X = diag ( …… ) (3.85)
S = diag ( …… ) (3.86)
using the Newton’s method which yields the following system Eq.
A 0 0 d xk Ax k b
Q
AT I dwk Qx A w s c
k T k k
(3.87)
S k 0 X k d zk
X kS k e k e
The right hand side of Eq. (3.87) is so-called slackness vectors and
t k b Ax k (3.88)
u k Qx k c AT wk s k (3.89)
v k ke X kS ke (3.90)
Adxk t k (3.91)
With Eq. (3.91) and Eq. (3.94), a dual search direction can be derived
1
dwk A(S k X kQ )1 X k AT A(S k X kQ )1 (X k u k v k ) t k (3.95)
The equation for a primal search direction can be derived from Eq.
(3.94).
With the primal search direction and Eq. (3.93), a slack search
To find appropriate step lengths while keeping the primal and dual
j
Pk min x kj /dxk | dxk 0 j
(3.98)
j
Dk min s kj /dsk | dsk 0 j
(3.99)
max
k
min Pk , Dk (3.100)
k 0.99 k
max (3.101)
x k 1 x k k dxk (3.102)
w k 1 w k kdwk (3.103)
s k 1 s k k dsk (3.104)
104
by-step as follows.
subroutine.
Step 10: Check if there are any violations in Eq. (3.15) and Eq.
step 4.
Merits
Demerits
ways [8]:
GAs work with a coding of the parameter set, not the parameters
variables.
trapping.
genetic operator to evolve the current state into the next state with
minimum computations.
Pgi would be coded as a binary string of ‘0’s and ‘1’s with length 4
digits.
117
active power generations PGi for i=1, 2,..7 with a length of 4-Digits
Each of PGi is bounded with in PGimax and PGi min. The choice of
following equation:
118
and
6.66, 9.33, 6.66, 36.66, 6.66, 27.33 MW) and with corresponding bit
lengths Li for i =1,2,..,7. The parameter domain of PGi for i=1, 2,..,7 is
Operation.
The process, that starts from the present population and leads
[111, 113].
fmax -
Initial Population PG1 PG2 PG3 PG4 PG5 PG6 PG7 f(PGi )
f(PGi )
0010001100101100110111010000 76.784 19.98 18.66 79.92 476.58 86.58 0.00 1067300 293200
1010011111111010000100100000 383.92 46.62 140.00 66.6 36.66 13.32 0.00 1360500 0.00
0110010101100100110000111110 230.352 33.3 55.98 26.64 439.92 19.98 382.62 30509.80 1329991
Max 1360500
Min 30509.80
locus,
Locus = 3
Table 3.4.4 The result after crossover and mutation of the first
population [111, 113].
The result after crossover of the The result after mutation of the
Chromosome
first population first population
1 0110010101100100110000111110 0110010101100100111000111110
2 0111100010111111000010110111 0111100011111111000010110111
3 0110010101100100110000111110 0110010101010100110010111111
4 0111100010111111000010110111 0111100010111111111010110111
120
i.e. Single Point Crossover (Table 3.4.3). The segments of bits from the
parents behind the crossover point are exchanged with each other to
point of the strings randomly and switching the left segments of this
to their scores using the roulette wheel [8]. Thus, the strings with
because all the copies in the roulette have the same probability to be
selected.
All strings and bits have the same probability of mutation. For
439.92 to 513.24.
PG1 PG2 PG3 PG4 PG5 PG6 PG7 mutation PG1 PG2 PG3 PG4 PG5 PG6 PG7
121
[111, 113].
fmax -
Second generation PG1 PG2 PG3 PG4 PG5 PG6 PG7 f(PGi )
f(PGi )
0110010101100100111000111110 230.35 33.3 55.98 26.6 513.2 19.98 382.62 5503.1 882116.9
0111100011111111000010110111 268.74 53.28 140.0 100.00 0.00 73.26 163.98 887620 0.00
0111100010111111111010110111 268.74 53.28 102.63 100.00 513.24 73.26 191.31 8015.6 879604.4
Max 887620
Min 4985.56
122
NG
F ( PG ) i i PGi i pG2i PGimin PGi PGimax
i 1
The objective is to search (PG1, PG2, PG3, PG4, PG5, PG6, PG7) in their
function f (PGi) takes a chromosome (a possible (PG1, PG2, PG3, PG4, PG5,
PG6, PG7)) and returns a value. The value of the cost is then mapped
into a fitness value fit (PG1, PG2, PG3, PG4, PG5, PG6, PG7) so as to fit in the
genetic algorithm.
min and the max reactive rate of the generators buses and the min
and the max of the voltage levels of all buses. All these require a fast
and robust load flow program with best convergence properties. The
123
NG
Applied genetic
algorithm operator
f max fi (PGi ), if f max fi (PGi ) ; i 1, NG ,
fitness
Coding reproduction-crossover- i
0, otherwise.
mutation.
and then test the fitness of best members in the last population. If no
search initiated.
124
Contribution
furnished below:
Journal /
Author Title of
Sl.No Publication Significant Contributions / Salient
[Ref. No] Topic
Details Features
1 A. Bakritzs,Genetic IEE Proc,- Solved Economic dispatch problems
V. Algorithm Generation (two) using Genetic Algorithm
Perirtridis Solution to Transmission method. Its merits are, the non
and the Distribution, restriction of any convexity
S. Kazarlis Economic vol. 141, limitations on the generator cost
[101] Dispatch no. 4, pp. function and effective coding of GAs
Problem 377-382, to work on parallel machines.
July 1994. GA is superior to Dynamic
programming, as per the
performance observed in Economic
dispatch problem.
The run time of the second GA
solution (EGA method)
proportionately increases with size
of the system.
2 Po-Hung Large- IEEE Solved Large Scale Economic
Chen and Scale Transactions Dispatch problem by Genetic
Hong-ChanEconomic on Power Algorithm.
Chang Dispatch Systems, Designed new encoding technique
[14] by Genetic Vol. 10, no. where in, the chromosome has only
Algorithm 4, pp. 1919 an encoding normalized
Discover – 1926, increamental cost.
Nov. 1995. There is no correlation between
total number of bits in the
chromosome and number of units.
The unique characteristic of Genetic
Approach is significant in big and
intricate systems which other
approaches fails to accomplish.
Dispatch is made more practical by
125
Merits
optimisation problems.
De Merits
problem size.
trajectory towards its own previous best position and towards the
global best position attained till then. PSO is easy to implement and
PSO, the potential solutions, called particles, fly through the problem
particle makes it s decision using its own experience together with its
neighbor’s experience.
132
Objective Function
NG NG
J FT F (PG ) Fi (PGi ) ( i i PGi i PG2i )
i 1 i 1
Subject to: g (x , u ) 0
h (x, u ) 0
uT [VG1 ... VGN , PG2 .... PGN , T1 ... TNT , Qc1 ....QcNC ] (3.109)
G G
as given below.
and reactive power outputs are restricted by their lower and upper
ND
J aug J P (PG1 PG 1lim ) V (VL i VG i lim )2
2
i 1
NG Nl
Q (QG i QG i lim )
2
S (Sl i Sl i max )2 (3.113)
i 1 i 1
x max
x x max
x lim (3.114)
x min
x x min
……xj, m (t)], where xs are the optimised parameters and xj, k(t) is the
position of the jth particle with respect to the kth dimension, i.e. the
[Xi(t),…. Xn(t)T.
vj,,k(t) is the velocity component of the jth particle with respect to kth
dimension.
influences the trade off, between the global and local exploration
compares its fitness value at the current position, to the best fitness
value it has ever attained at any time up to the current time. The best
called the individual best, X* (t). In this way, the best position X* (t).for
function J, the individual best of the jth particle X*j (t) is determined
such that J(X*j (t)) J(X*j ( )), t. For simplicity it is assumed that
Jj* = J(X*j (t)).For the jth particle, individual best can be expressed as X*j
Global best X** (t): It is the best position among all individual
best positions ( i.e. the best of all) achieved so far .Therefore ,the
global best can be determined as such that J(X**j (t)) J(X*j ( )),
process will terminate. In the present case, the search will terminate if
a) The number of iterations since, the last change of the best solution
or
considered here.
dimension is given as :
Step 1 (Initialization)
objective function J.
Set the particle associated with J best as the global best, X ** (0) , with
an objective function of J ** .
Using the global best and individual best of each particle, the jth
following equation:
c2 r2 ( x**
j , k (t 1) x j , k (t 1)) (3.116)
second term represents the cognitive part of PSO where the particle
changes its velocity based on its own thinking and memory. The third
term represents the social part of PSO where the particle changes its
a particle violates the velocity limits, set its velocity equal to the limit.
x j , k (t ) v j , k (t ) x j , k (t 1) (3.117)
Search for the minimum value J min among J *j , where min is the
step 2.
furnished below:
Merits
computational efficiency.
5. PSO can easily deal with non differentiable and non convex
objective functions.
6. PSO has the flexibility to control the balance between the global
Demerits
significant errors.
ability).
approaches for OPF solution that can overcome the disadvantages and
The methods may satisfy necessary conditions but not all the
constraints.
constraints.
The process may converge slowly due to the requirement for the
These methods are difficult to apply for the problems with discrete
to be the optimum.
Algorithm (EGA) for the solution of OPF problem. The EGA method has
following features:
control variables.
control variable.
chromosome.
The test results presented in [103] are quite attractive. However the
EGA method:
that can avoid all the difficulties in the various approaches and
the following objectives that will improve genetic algorithm for OPF
solution.
as infinite.
Need for a suitable local search method that can achieve a right
accuracy.
and efficiency.
Need for the proper genetic operators that will resolve some of the
Need for reducing time for searching for a global optimum solution
which can help to combat the genetic drift problem caused by the
obstruct any genetic drift. Due to its limited population size, a genetic
premature convergence.
method can speed up the search to locate the exact global optimum
minimum time.
the memory size and convergence speed and affects the search speed
search over continuous space difficult [112]. The other benefit that
genetic algorithms
computational complexities.
3. 5 CONCLUSIONS
explaining the objective function along with non linear equality and
in Power Systems.
reduction of the costs of meeting the load demand for a power system
Economic dispatch.
form. This helps the reader to quickly get to know the significant
method and Interior Point method. Among these methods, the Interior
convergence.
mitigate the OPF problem. This chapter explores the advantages and