34

CHAPTER-4
APPLICATION OF FIREFLY ALGORITHM
4.1 INTRODUCTION
Modern engineering practices call for optimization in the use of all natural
resources and in this context, the OPF has become an important tool for power system
planning, operation and control [13-15]. In the scenario of integrated and deregulated
power industry, the OPF problem is modified to include non-linear, non-convex,
discrete and non discrete control variables. The OPF problem has as its objectives the
optimization of RPL, generation cost and voltage stability. Dommel and Tinny were
the first, to present the problem, which was dealt by many researches later on. In the
midst of many conventional techniques available like Newton method, LP, Interior
point method the FACTS technology facilitated the control of power flow, without
generator scheduling and topology changes [91].
UPFC is a popular device, of all the FACTS devices, providing shunt and
series compensation for flexibility in OPF [73-80]. However power system is not
secure till the system voltage security is not considered [112].
The conventional methods do not yield satisfactory results of obtaining a
global optimum as they are more sensitivity to the initial points, because of non-
monotonic surface of the solution surface [97-100].
This difficulty is easily overcome by the application of meta-heuristic
algorithms which are nature inspired, powerful for dealing with non-linear OPF
problems. The various techniques developed are PSO, TS, BFA, DE, GSA, recent
additions being KH and MBO [110-112].
In this work, the meta-heuristic algorithm of FA is employed, which was
developed by Xin-she-yang [113-114] after studying the behaviour of fireflies. FA
was used to solve the economic dispatch problems along with OPF with TCSC, for
power transfer capability enhancement. The FA based method for optimization, of the
various variables which are controlled during execution of the algorithm, such as
transformer taps, location of UPFC, location of OUPFC and its variables for single
objective of RPL minimization and multi-objective of RPL minimization and VSL
maximization. This chapter dwells into the FA implementation methodology with
 35

simulations for IEEE 14 bus [123], New England 39 bus [124] and practical Indian 24
bus system [57].

4.2 PROBLEM FORMULATION

The problem for real power loss and voltage stability limit can be optimized
either as a single objective or a multi objective problem. The subsequent section deals
with them.

4.2.1 Minimization of real power loss (OPF problem)

For a given set of dependent variables ‘x’ vector and control variables ‘u’
vector, the mathematical formulation is:

Minimize F(x, u)
Subject to g(x, u) =0
h(x, u) ≤0 (4.1)
Where F is the objective function i.e. RPL minimization, g and h are the set of
constraints. x, u vectors of dependent and control variables, x includes the active
power at slack bus PG1, reactive power output QG, transformer line flow sl. The
vector ‘x’ therefore can be expressed as given equation 4.1. The various defined
variables are NL, NG, nl i.e., number of load buses, the number of generators, and the
number of transmission lines. The vector ‘u’ represents the generator voltage VG,
active power of generator output PG at PV buses, transformer taps T, shunt VAR
compensation QC, excluding the slack bus PG1. NT and NC are the number of
transformers and compensators, given in equation 4.2.
x T  [ PG1 , VL1 .....VLNL , QG1 ....QGNG ....S l1 ....S ln l ] (4.2)

u T  [ PG 2 ....PGNG ,VG1 ...VGNG , QC1 ....QCNC , T1 ....TNT ] (4.3)

4.2.2 RPL minimization and VSL maximization (combined OPF-CPF problem)

The single objective of RPL minimization could be extended further with the
inclusion of VSL maximization giving rise to a new fitness function. Determination
of VSL using CPF method by identifying the load parameter λ and the resulting
equations are as given under:
PLi  PLi 0 (1   ) (4.4)
 36

QLi  QLi 0 (1   ) (4.5)

PGi  PGi 0 (1   ) (4.6)

The load parameter can be varied till the system does not exceed the limit of
stability, which, is also known as notch point (NP) of the P-V curve and so the
maximum value of the load parameter λ max indicates the voltage stability limit
(VSL).
Mathematically the multi-objective optimization problem can be formulated
as:
Optimize F(x, u λ max)
Subject to g(x, u) =0 (4.7)
h(x, u)≤0
Since both the RPL and VSL are in different range of values, the fitness function is
formulated as a weighted sum. The reciprocal of VSL is sum to original cost function
and overall cost function is can be minimized.

The fitness function is represented, now, by

F(x, u, λ max) =W1*G(x) + W2*V (λ max) (4.8)

Where G(x) =RPL
V (λ max) =1/ λ max=VSL
where w1, w2 are the weight adjustments, such that they yield similar values of RPL
and VSL respectively. The process of CPF is carried out for each solution generated
to find out this critical point in the system. Applying the process of correction and
prediction, the load increments are calculated.

4.2.3 Equality Constraints:

The power flow equations, for both single and multi objective problem are the
equality constraints, which can be written as:
NB
PGi  PDi  Vi  V j [Gij cos( i   j )  Bij sin( i   j )]  0 (4.9)
j 1

NB
QGi  Q Di  V  V j [Gij sin( i   j )  Bij cos( i   j )]  0 (4.10)
j 1
 37

NB, PG, QG, PD, QD, Gij, Bij, which are number of buses, active power generation,
reactive power generation, active power demand, reactive power demand,
conductance and susceptance respectively.

4.2.4 Inequality constraints:

The inequality constraints for both single and multi-objective optimization problems
include:
1. Generator constraints: The upper and lower limits of generator bus voltages,
active and reactive power outputs indicate the generator constraints.
VGimin  VGi  VGimax , i  1,..........NG (4.11)

PGimin  PGi  PGimax , i  1,..........NG (4.12)

QGimin  QGi  QGimax , i  1,..........NG (4.13)

2. Transformer constraints: The upper and lower limits of transformer taps

indicate this type of constraints.
Ti min  Ti  Ti max , i  1,..........NT (4.14)

3. Shunt VAR constraints: Shunt VAR upper and lower limits indicate this type
of constraints.
QCimin  QCi  QCimax , i  1,..........N C (4.15)

4. UPFC constraints:
ri min  ri  ri max i  1,......., N UPFC

(4.16)
 imin   i   imax i  1,......., N UPFC (4.17)
where NUPFC is the number of UPFCs
5. OUPFC constraints: The OUPFC settings are bounded as follows:
k imin  ki  kimax i  1,......., N OUPFC
(4.18)
 imin   i   imax i  1,......., N OUPFC (4.19)
rimin  ri  ri max i  1,......., N OUPFC
(4.20)
 imin   i   imax i  1,......., N OUPFC (4.21)
 38

where N OUPFC is the number of OUPFCs. The ratio 'k' of PST is given by tan (  ),
for it acting as a quadrature booster.

4.2.5 Security constraints: Comprises of load bus voltage and line loadings.
VLimin  VLi  VLimax , i  1,..........NL (4.22)

S limin  S limax , i  1,..........nl (4.23)

4.2.6 Limits of Control Variables Used for both single and multi-objective
problems
In this research work the control variables used for the both optimization problems
are:
(i) Transformer tap settings:
The transformer tap settings vary in between 0.85 to 1.1 in step of 0.05 for
IEEE 14 bus system, 0.85 to 1.15 in step of 0.05 for New England 39 bus
system and 0.9 to 1.05 in step of 0.125 for Indian practical 24 bus system.
(ii) UPFC control variables:
The UPFC control variables are (i)UPFC Series injected voltage magnitude
(Vse) vary within the range [0, 0.3 p.u] and (ii)UPFC series injected voltage
phase angle (γ) vary within the range [0,2π] for all the test systems.
(iii)OUPFC control variables:
The OUPFC control variables are (i) UPFC series injected voltage magnitude
(Vse) vary within the range[0, 0.15p.u.] and UPFC series injected voltage
phase angle (ρ) vary within the range [-180, 180] and phase shifting
transformer phase angle (σ)vary within the range [-20, 20] for all the test
systems.
Here the upper and lower limits of bus voltages are 1.05p.u and 0.95p.u for IEEE 14
bus, 1.1p.u and 0.9p.u for New England 39 bus system, and 0.9p.u. and 1.05 p.u. for
Indian practical 24-bus system.

4.3 FIREFLY ALGORITHM BASED OPTIMIZATION

Fireflies use flash signals to attract other fireflies for potential mates. Based on
this behaviour a meta-heuristic algorithm was developed by Xin-She Yang [119-122].
The flashes of the fireflies are shown in figure 4.1.All the fireflies are considered
unisexual and their attraction is directly proportional to the intensity of their flash.
 39

Fig. 4.1 Fireflies signalling with their flash

Therefore if a firefly particle had the choice of moving towards either of two
fireflies, it will be more attracted towards the firefly with higher brightness and moves
in that direction. The movement of firefly would be random if there are no other
insects in its proximity and its movement is because of light intensity of the other
firefly. The brightness of flash is associated with the fitness function. The light
intensity also obeys the inverse square law as in equation 4.24

I s (4.24)
I r   2
r

Where I(r) is the light intensity at a distance r and Is the intensity at the source.
Considering the medium and its characteristics of absorption coefficient (γ), the light
intensity is related to the distance 'r' as given by equation 4.25


I  I0 exp  r 2  (4.25)

The initial scale for light intensity is taken as I0, γ is the absorption coefficient and r is
the distance between the fireflies. The attractiveness factor β for the other fireflies,
which is dependent on light intensity is given by

m
   0 exp( r ) m  1 (4.26)
 40

Where β0 is the attractiveness at r =0. The distance between the ith and jth firefly, r is
calculated as

d 2
rij   xk 1 i ,k  x j ,k   xi  x j (4.27)

The equation 4.28 determines the random movement given by the Gaussian
distribution randomization. For each generation the firefly is ranked based on its
brightness.

x i  x i   0 e x p   rij2  x i  x j   (4.28)

Where α is the randomization parameter and ε is the vector of random numbers

taken from Gaussian distribution. Here  controls the step size. For each generation
the firefly is ranked based on its brightness.

4.3.1 Firefly algorithm pseudo –code [114]

The basic steps of the FA can be expressed in the following pseudo-code:

objective function f(x): x = (x1... xd) T
generate initial population of fireflies xii (i = 1, 2... n)
light intensity Iii at xii is decided by f (xii)
define light absorption coefficient γ
While (t < Max Generation)
for ii = 1: n all n number of fireflies
for jj = 1: ii all n n number of fireflies
if (Ijj > Iii), move firefly ii toward jj in d-dimension
end if
attractiveness differs with distance r
evaluate new solutions and modify the light intensity
end for jj
end for ii
rank all fireflies and get the best of the current firefly
end while
Post-process and visualization
 41

4.3.2 General Flowchart of FA

The Flow chart of the firefly algorithm is shown in figure 4.2
Start

Initialization of parameters

Generate random variables

While No
Terminate
Iter < max

Yes

Evaluate light intensities based on

fitness function

Find best firefly

Move fireflies based on attraction

Iter = iter+1

Fig. 4.2 Flow chart for FA based optimization

4.3.3 Limitations of the FA

The disadvantage of FA is the result getting bagged down in local minima.

The parameters of FA are invariable over time. The algorithm does not provide for
memory of the previous situation. In future, multiple variants of FA such as Gaussian
distribution for random walk [113], chaos enhanced FA for tuning of α, β and γ are
used to solve the practical problem [114].
 42

4.4. IMPLEMENTATION OF FIREFLY ALGORITHM

The application of Firefly algorithm technique for the solution of Optimal
power flow problem and combined OPF-CPF problem is explained below:
Step1: Initialization of the firefly algorithm
(i) The dimension of the problem
(ii) The number of fireflies
(iii) The maximum number of iterations
(iv)The values of α, β, γ and δ are chosen.
(v) Iteration counter i=0
Step2: Increment the iteration counter i= i+1
Step3: Calculate the fitness of the fireflies in very iteration by using the fitness
functions as given in equations (4.1) & (4.8) and associate light intensity of each
firefly to the same.
Step4: For each iteration, the fireflies are segregated based on intensities to the best
one.
Step5: Vary the light intensity perception of all other fireflies based on the distance
between them
Step6: Move the fireflies based on attraction which depends on light intensities and
also control parameters.
Step7: If the stopping criteria in not reached go back to step2 else go to step 8
Step8: Display the results with the firefly particle of highest light intensity.

4.4.1 Flow chart of optimization strategy with FA

The flow chart of optimization strategy with FA is shown in figure 4.3.
 43

start

Obtain system data

Initialize FA parameters

Generate random firefly population

Initialize iter=0

NO
Iter < max iter

YES

Run power flow and CPF End

Evaluate fitness function

Iter = iter+1 NO
All constrains are Add penalty values to the
satisfied objective function

YES

Is
Current value YES Store value as
Better than best firefly
best firefly

NO

Evaluate light intensities according

to fitness value

Move fireflies based on the

attraction

Fig. 4.3 Optimization strategy with FA

4.5 SIMULATION RESULTS AND DISCUSSION

The effectiveness of the proposed FA method to solve optimization problem

for both single and multi-objective cases has been demonstrated using IEEE 14-bus
system [123] and New England 39-bus system [124] and practical Indian 24 bus
 44

system [57]. A software program has been developed in the MATLAB and run on
core i5, 2.5 GHz and 4GB RAM computer. The details of the test systems are given
below:

Test system 1: IEEE 14 -bus system

The single line diagram of The IEEE 14 bus test system shown in figure A.1
of appendix A. The IEEE 14 bus test system comprises of 5 generator buses, 9 load
buses, 20 transmission lines. Of this bus 1 is slack bus, lines 4-7, 4-9 and 5-6 have
discrete operating valued tap changers. Buses 9 and 14 are with discrete valued shunt
compensations. Power flow is solved for the base case with nominal taps and found
the value of RPL as 0.13394 p.u. The value of VSL obtained with CPF technique is
0.93.

Test system 2: New England 39- bus system

New England 39 bus test system consists of 10 generator buses,

12 transformers and 46 transmission lines. Single line diagram is given in figure B.1
of Appendix B. Buses from 30 to 39 are generator buses with 31 as the slack bus
and remaining are the load buses. The transformers T1 to T12 are located in the lines
2-30, 10-32, 12-11, 12-13, 19-33, 19-20, 20-34, 22-35, 23-36, 25-37, 29-38, 31-6
respectively. Power flow is run for the base case to find RPL and using CPF
technique. For this test system with nominal values of taps, the RPL is 0.4378 p.u
and VSL is 0.81.

Test system 3: 24-bus equivalent southern EHV power system network

The single line diagram of the practical Indian 24 bus test system is shown in
figure C.1 of appendix C. Practical Indian 24 bus test system consists of 4 generator
buses, 11 transformers and 27 transmission lines. Buses from 1 to 4 are generator
buses with 1 as the slack bus and remaining are the load buses. The transformers T1
to T11are located in the lines 15-1, 17-2, 24-3, 21-4, 16-5, 19-6, 20-7, 14-8, 23-9, 18-
10, 22-13, respectively. Power flow is run for the base case to find RPL and the VSL
is obtained using CPF technique. For this test system with nominal values of taps, the
RPL is 0.64642p.u and VSL is 0.0703.
 45

The FA based optimization method is developed in MATLAB environment for

optimization of both single and multi-objective problems considering, the following
cases:

Case1: Optimization with only transformer taps

Case 2: Sequential optimization of UPFC location and its parameters keeping the
optimized transformer taps fixed
Case 3: Simultaneous optimization of UPFC location and its parameters along with
taps
Next OUPFC is considered and the optimization is carried out in the same way.
Case 4: Sequential optimization of OUPFC location and its parameters keeping the
optimized transformer taps fixed.
Case 5: Simultaneous optimization of OUPFC location and its parameters along with
taps.
The simulation parameters of FA considered for the test cases are shown in Table 4.1
Table 4.1 Parameters of FA
S.NO Parameters Quantity
1 Number of fireflies 30
2 Maximum iterations 35
3 Randomness(α) 0.3
4 Absorption coefficient(γ) 1.0
5 Attractiveness( β) 0.2

The simulation results for the all the cases for the test systems considered are
discussed below:
4.5.1 Case 1: Optimization with only transformer taps
4.5.1.1 Case 1.1: Optimization of only RPL (single objective case)
In this case, the transformer taps are optimized with FA technique for single
objective case. The optimized transformer taps along with RPL and VSL obtained
from the proposed technique are given in tables 4.2, 4.3 and 4.4 for the test systems
considered. From the results, it may be noted that the minimum RPL of 0.13343p.u. ,
0.4203 p.u., and 0.54368 p.u. as obtained using FAS is less by 0.38076%, 3.9972 %
 46

and 15.8926 % compared to base case result 0.13394p.u., 0.4378p.u., and 0.64642
p.u. for IEEE 14 bus, New England 39 bus and practical Indian 24 bus systems
respectively. In the single objective case, VSL is not included in the fitness function.
With the optimized taps the CPF is run for estimating the voltage stability limit in the
single objective case and found that 0.96, 0.810 and 0.13710 for the test systems
considered.
4.5.1.2 Case 1.2: Optimization of both RPL and VSL (multi-objective case)
When the optimization of both RPL and VSL is considered in the multi-
objective case, the cost function is modified i.e., the reciprocal of VSL is added to the
real power loss and the optimization is carried out with FA technique. The optimized
transformer taps along with RPL and VSL obtained with proposed technique for multi
objective case are given in the tables 4.2, 4.3 and 4.4 for the test systems considered.
From the results it is seen that the VSL is improved but RPL is increased marginally
with FA technique for all the test systems. Even RPL is increased slightly; the overall
multi-objective function is reduced compared to single objective case. In this case
RPL has increased by a margin of 0.2323% (from 0.13343 p.u to 0.13374 p.u.),
2.4982 % (from 0.4203p.u. to 0.4308p.u.) and 0.3108% (form 0.54368 p.u. to
0.54537p.u.) and VSL has improved to 6.25 % (form 0.96 to 1.02), 3.7037 % (from
0.810 to 0.840), and 1.4659% ( from 0.137110 to 0.139120) for IEEE 14, New
England 39 and Indian 24 bus test systems respectively compared to single objective
case. But, the combined fitness function obtained with FA for multi-objective case
has reduced to 5.804 % (from 10.5500 to 9.9376), 3.3714 %( from 12.7659 to
12.3355) and 1.3230 %( from 7.83708 to 7.7334) for the corresponding test systems,
compared to single objective case. This shows that the FA is able to optimize both
RPL and VSL simultaneously for stable and secure operation of power system.
 47

Table 4.2: Results of IEEE 14 bus test system for case study 1
Optimization of both
Optimization of only
Control variable RPL and VSL
RPL
Settings
FAS FAM
T1 (4-7) 1.00 1.05
T2 (4-9) 0.90 1.05
T3 (5-6) 0.95 1.05
RPL, p.u. 0.13343 0.13374
VSL 0.96 1.02
Combined fitness function
10.5500 9.9376
(W1*RPL+W2*1/VSL)

Note 1: FA is applied to optimize only RPL denoted as FAS

Note 2: FA is applied to optimize both RPL and VSL denoted as FAM

Table 4.3: Results of New England 39 bus test system for case study 1
Optimization of Optimization of both
Control variable only RPL RPL and VSL
Settings
FAS FAM
T1 (2-30) 1.15 1.10
T2 (10-32) 1.10 1.05
T3 (12-11) 1.05 1.05
T4 (12-13) 1.05 1.05
T5 (19-33) 1.15 1.10
T6(19-20) 1.10 1.10
T7 (20-34) 1.10 1.05
T8 (22-35) 1.10 1.10
T9(23-36) 1.10 1.15
T10(25-37) 1.10 1.05
T11 ( 29-38) 1.15 1.10
T12 (31-6) 1.10 1.10
RPL, (p.u.) 0.4203 0.4308
VSL 0.810 0.840
Combined fitness function
(W1*RPL+W2* 12.7659 12.3355
1/VSL)
 48

Table 4.4 Results of the practical Indian 24- bus system for case study 1

Optimization of Optimization of both

only RPL RPL and VSL (multi-
Control variable
(single objective objective case)
Settings
case)
FAS FAM
T1 (15-1) 1.0500 1.0500
T2 (17-2) 1.0250 1.0500
T3 (24-3) 1.0500 1.0500
T4 (21-4) 1.0250 1.0375
T5 (16-5) 0.9250 1.0500
T6 (19-6) 1.0125 1.0500
T7 (20-7) 1.0000 1.0500
T8 (14-8) 0.9250 1.0500
T9 (23-9) 1.0125 1.0375
T10 (18-10) 0.9750 1.0500
T11 (22-13) 0.9875 1.0500
RPL(p.u.) 0.54368 0.54537
VSL 0.137110 0.139120
Combined fitness function
7.83709 7.7334
(W1*RPL+W2*1/VSL)

Fig. 4.4 Convergence characteristic of IEEE14- bus system for case study 1.1
 49

Fig. 4.5 Convergence characteristic of New England 39-bus system for case study 1.1

Fig. 4.6 Convergence characteristic of practical Indian 24-bus system for case
study 1.1

The figures 4.4 to 4.6 show the single objective RPL minimization, depicting
the convergence. From the results it is found that the performance of the proposed FA
is better in many parameters such as minimization of loss, computational time, speed
of convergence and accuracy. The minimum number of iterations taken by this
algorithm to reach the global minimum proves the superiority of the algorithm.
 50

Fig. 4.7 P-V curve of the weakest bus of IEEE14- bus system with FAS for case study
1.1

Fig. 4.8 P-V curve of the weakest bus of New England 39- bus system with FAS for
case study 1.1
 51

Fig. 4.9 P-V curve of the weakest bus of practical Indian 24-bus system with FAS for
case study 1.1

Figures 4.7, 4.8 and 4.9 show the p-v curves for the weakest bus of the test
systems considered, are from the FA with single objective under optimal control
variables of transformer tap settings excluding VSL in the fitness function. The CPF
for the voltage stability limit resulted in values 0.96, 0.810 and 0.13710 for the test
systems considered.

Fig. 4.10 P-V curve of weakest bus of IEEE 14- bus system with FAM for case

study 1.2.
 52

Fig. 4.11 P-V curve of weakest bus of New England 39- bus system with FAM for
case study 1.2

Fig. 4.12 P-V curve of weakest bus of practical Indian 24-bus system with FAM for
case study 1.2
 53

Figure 4.10, 4.11 and 4.12 show the p-v curves of weakest bus of the test
systems considered with optimal transformer tap settings obtained using FA for multi-
objective case. It is clear from the results that VSL has improved to 9.6774 %( from
0.93 to 1.02), 3.7037 % ( from 0.81 to 0.84), and 97.8947 %( from 0.0703 to
0.139120) for IEEE 14, New England 39 and Indian 24 bus test systems respectively
compared to base case . Also It is observed from the figures that the VSL has
improved 6.25 % (form 0.96 to 1.02), 3.7037 % (from 0.810 to 0.840), and 1.4659%
(from 0.137110 to 0.139120) for the test systems considered, compared to single
objective case. It can be concluded that voltage stability margin was improved using
CPF, to ensure the feasibility of the optimal transformer tap settings for secure
operation of power system.

Fig. 4.13 Voltage profiles of practical Indian 24-bus system with FAS for case

study 1.1
 54

Fig. 4.14 Voltage profiles of practical Indian 24- bus system with FAM for case
study 1.2

Figures 4.13 and 4.14 show the voltage profiles at different buses for base case
power flow and for FA based optimization for single objective case with optimum
transformer taps for practical Indian 24 bus system. Similarly the voltage profiles for
multi-objective case are given in figure 4.14. From the figures, the improved voltage
profiles obtained from FA at all load buses are observed for both the cases. Similar
kind of voltage profiles are observed for IEEE 14-bus and New England 39 bus test
systems.

4.5.2 Case 2: Sequential optimization of UPFC location and its variables keeping
the transformer taps fixed
4.5.2.1 Case 2.1: Optimization of only RPL (single objective case)

With the optimized transformer taps fixed (obtained from case 1.1), the UPFC
variables are then optimized using the proposed FA technique so that objective
function could still be reduced. The optimized UPFC location and its parameters
along with RPL and VSL obtained with proposed technique are given in tables 4.5,
4.6 and 4.7 for the test systems considered. The power injection model as discussed in
 55

chapter-3 is used for modelling the UPFC. For IEEE 14- bus system only 7 lines are
considered for connecting the UPFC as the remaining 13 lines consists of
transformers and feeding generator powers to the network. For New England 39- bus
system 32 lines are considered for connecting the UPFC as the remaining 14 lines
consists of transformers and feeding generator powers to the network. Similarly for
practical Indian 24- bus system 16 lines are considered for connecting the UPFC as
the remaining 11 lines consists of transformers and feeding generator powers to the
network. It is clear from the results that with the incorporation of UPFC device, the
RPL has reduced to 17.9344 %( from 0.13343 to 0.1095), 15.6554 % (from 0.4203 to
0.3545), 4.69577 %( from 0.54368 to 0.51815) compared to case 1.1 for IEEE 14
bus, New England 39 bus and practical Indian 24 bus system respectively . It is seen
that when more nonlinear equipment like UPFC is inducted into the system, the
fitness function has successfully converged with FA. This shows the potentiality of
the proposed algorithm in solving the non liner optimization problem particularly the
RPL minimization problem. In the single objective case VSL is not included in the
fitness function. With the optimized taps the CPF is run for estimating the voltage
stability limit in the single objective case and found to be 0.990, 0.900 and 0.14481
for IEEE 14- bus, New England 39- bus and practical Indian 24- bus systems
respectively.
4.5.2.2 Case 2.2: Optimization of both RPL and VSL (multi-objective case)
When both RPL and VSL are optimized in the multi-objective case, the fitness
function is modified. The reciprocal of VSL is added to the real power loss and the
optimization is carried out with FAM technique. The optimized UPFC parameters
along with RPL and VSL obtained with proposed technique are given in tables 4.5,
4.6 and 4.7 for the test systems considered. From the results it is seen that the VSL is
improved but RPL is increased marginally with FAM technique for all the test
systems. Even RPL is increased slightly; the overall multi-objective function that is
the sum of real power loss and reciprocal of VSL is reduced compared to single
objective case. RPL has increased by a margin of 8.9497% (from 0.1095 p.u to 0.1193
p.u.), 5.5571 % ( from 0.3545 p.u. to 0.3742 p.u.) and 1.0498% (form 0.51815 p.u. to
0.52359 p.u.) and VSL has improved to 9.0909 % (form 0.990 to 1.08), 13.333 %
(from 0.900 to 1.020), and 5.4485 % ( from 0.14481 to 0.1527) for IEEE 14, New
 56

England 39 and Indian 24 bus test systems respectively compared to single objective
case. But, the combined fitness function obtained with FAM has reduced to 8.1455 %(
from 10.2103 to 9.37862), 11.2292 %( from 11.4656 to 10.1781) and 4.8116 %( from
7.42237 to 7.068523) for IEEE 14- bus, New England 39- bus and practical Indian
24- bus systems respectively compared to single objective case. Also it is clear that
when UPFC is inducted to the system the overall combined fitness function is further
reduced to 5.62489 % (from 9.9376 to 9.37862), 17.4893 %( from12.3355 to
10.1781) and 8.5974 % (from 7.7334 to 7.068523) for the test systems considered,
compared to case 1.2. It is found that with FAM, the UPFC parameters can be
optimized that the system could be loaded beyond double (is more than “1”), its
nominal loading before a voltage collapse could occur for all the test systems. So FA
is best succeeded in reducing the combined objective of loss minimization and VSL
maximization.

Table 4.5 Results of the IEEE 14- bus test system for case study 2
Optimization of both
Optimization of only
RPL and VSL
Control variable RPL
(multi-objective case)
Settings (single objective case)
FAS FAM
UPFC series injected
0.024695 0.02443
voltage, Vse, (p.u.)
UPFC series injected
1.16420 1.453079
voltage angle, γ , (rad)
UPFC location 9-10 9-10
RPL, p.u. 0.1095 0.1193
VSL 0.990 1.08
Combined fitness
function 10.2103 9.37862
(W1*RPL+W2*1/VSL)
 57

Table 4.6 Results of New England 39- bus system for case study 2

Optimization of both
Optimization of only
RPL and VSL
Control variable RPL
(multi-objective case)
Settings (single objective case)
FAS FAM
UPFC series injected
0.0294 0.0096
voltage, Vse, (p.u.)
UPFC series injected
1.9330 0.9134
voltage angle,( γ) , (rad)
UPFC location 26-29 26-27
RPL, (p.u.) 0.3545 0.3742
VSL 0.900 1.020
Combined fitness function
(W1*RPL+W2* 11.4656 10.1781
1/VSL)

Table 4.7 Results of the practical Indian 24- bus system for case study 2

Optimization of
Optimization of
both RPL and VSL
only RPL
Control variable (multi-objective
(single objective
Settings case)
case)
FAS FAM
UPFC series injected
0.123501 0.060143
voltage ,Vse , (p.u.)
UPFC series injected 1.085670
4.58378
voltage angle ,γ , (rad.)
UPFC location 22-23 23-20
RPL (p.u.) 0.51815 0.52359
VSL 0.14481 0.15279
Combined fitness
function 7.42237 7.068523
(W1*RPL+W2*1/VSL)
 58

Fig. 4.15 Convergence characteristic of IEEE 14- bus system for case study 2.1

Fig. 4.16 Convergence characteristics of New England 39- bus system for case
study 2.1
 59

Fig. 4.17 Convergence characteristic of practical Indian 24 bus system for case
study 2.1

The figures 4.15 to 4.17 show the single objective RPL minimization,
depicting the convergence. From the results it is found that the performance of the
proposed FA is better in many parameters such as minimization of loss,
computational time, speed of convergence and accuracy. The minimum number of
iterations taken by this algorithm to reach the global minimum proves the superiority
of the algorithm.
 60

Fig. 4.18 P-V curve of the weakest bus of IEEE 14- bus system with FAS for
case study 2.1

Fig. 4.19 P-V Curve of the weakest bus of the New England 39- bus system with FAS
for case study 2.1
 61

Fig. 4.20 P-V Curve of the weakest bus of the practical Indian 24- bus system with
FAS for case study 2.1

The figures 4.18, 4.19 and 4.20 show the p-v curves for the weakest bus of the
test systems considered, are for the FA with single objective under optimal control
variables of UPFC location and its parameters excluding VSL in the fitness function.
The CPF for voltage stability limit resulted in values 0.99, 0.9 and 0.14481 for the test
systems considered.
 62

Fig. 4.21 P-V curve of the weakest bus of IEEE 14- bus system with FAM for case
study 2.2

Fig. 4.22 P-V Curve of the weakest bus of the New England 39-bus system with FAM
for case study 2.2
 63

Fig. 4.23 P-V Curve of the weakest bus of the practical Indian 24- bus system with
FAM for case study 2.2

Figures 4.21, 4.22 and 4.23 show the p-v curves of weakest bus of IEEE 14-
bus, New England 39- bus and Indian 24- bus systems respectively when variables are
sequentially optimized with FA for multi-objective case. It is clear from the results
that VSL has improved to 16.1290 %( from 0.93 to 1.08), 25.9259 % ( from 0.81 to
1.02), and 117.339 %( from 0.0703 to 0.15279) for IEEE 14, New England 39 and
Indian 24 bus test systems respectively compared to base case . Also It is observed
from the figures that the VSL has improved 6.25 % (form 0.96 to 1.02), 3.7037 %
(from 0.810 to 0.840), and 1.4659% ( from 0.137110 to 0.139120) for IEEE 14, New
England 39 and Indian 24 bus test systems respectively compared to single objective
case. It can be concluded that voltage stability margin was further improved using
CPF, to ensure the feasibility of the optimal position and parameter settings of UPFC
for secure operation of power system.
 64

Fig. 4.24 Voltage profiles of practical Indian 24- bus system with FAS for case
study 2.1

Fig. 4.25 Voltage profiles of practical Indian 24- bus system with FAM for case
study 2.2
 65

Figures 4.24 and 4.25 relate the voltage profiles corresponding to the base case
PF and with FA of all buses of the test system considered. It is clear from the figures
that FA has helped to improve voltage profile in all the load buses. Similar
characterises can be observed for the test systems considered.
4.5.3 Case 3: Simultaneous Optimization of UPFC location and parameters along
with taps.
4.5.3.1 Case 3.1: Optimization of only RPL (single objective case)

Simultaneous optimization of UPFC location and its variables along with the
transformer taps with FAS could still reduce the objective function. For optimizing
with the FAS, the number of variables now becomes 6 i.e., 3 transformer tap
positions, and 3 UPFC variables, 15, i.e., 12 transformer tap positions, and three
UPFC variables 14, i.e,11 transformer tap positions, and 3 UPFC variables for IEEE
14- bus, New England 39- bus and practical Indian 24 -bus system respectively. The
optimized transformer taps, UPFC location and its parameters along with RPL and
VSL obtained with proposed technique are given in tables 4.8, 4.9 and 4.10 for the
test systems considered. It is clear from the results that, the RPL has reduced
furthermore to 3.5433 %( from 0.1095 to 0.10562), 2.3131 % (from 0.3545 to 0.3463
), 5.4385 %( from 0.51815 to 0.48997) compared to case 2.1 for IEEE 14- bus, New
England 39- bus and practical Indian 24- bus system respectively when variables are
simultaneously optimized . With the optimized taps the CPF is run for estimating the
voltage stability limit in the single objective case and found to be 1.11, 0.930 and
0.15097 for the test systems considered.
4.5.3.2 Case 3.2: Optimization of both RPL and VSL (multi-objective case)
When both RPL and VSL are optimized in the multi-objective case, the cost
function is modified. The reciprocal of VSL is added to the real power loss and the
optimization is carried out with FAM technique. The optimized transformer taps,
UPFC parameters along with RPL and VSL obtained with proposed technique are
given in tables 4.8, 4.9 and 4.10 for the test systems considered. From the results it is
seen that the VSL is improved but RPL is increased marginally with FAM technique
for all the test systems. Even RPL is increased slightly; the overall multi-objective
function that is the sum of real power loss and reciprocal of VSL is reduced compared
 66

to single objective case. RPL has increased by a margin of 4.3457% (from 0.10562 to
0.11021), 4.470 % (from 0.3463 to 0.3617) and 0.042859 % (form 0.48997 to
0.49018) and VSL has improved to 5.40505 % (form 1.11to 1.17), 16.1290 % (from
0.930 to 1.08), and 6.0873 % (from 0.15097 to 0.16016) for IEEE 14, New England
39 and practical Indian 24 bus test systems respectively compared to single objective
case. But the combined fitness function obtained with FAM has reduced to 5.01853
%( from 9.11462 to 8.65721), 13.3166 %( from 11.0989 to 9.6209) and 5.3397 %(
from 7.11380to 6.73394) for IEEE 14 bus, New England 39 bus and practical Indian
24 bus systems respectively compared to single objective case. Also it is clear that
when variables are simultaneously optimized the overall combined fitness function is
further reduced to 7.6920 % (from 9.37862 to 8.65721), 5.4744 %( from 10.1781 to
9.6209), 4.73342 %( 7.068523 to 6.73394) for the test systems considered, compared
to case 2.2. It is found that with FAM, when variables are simultaneously optimized
that the system could be more loaded before a voltage collapse could occur for all the
test systems. So FA is best succeeded in reducing the combined objective of loss
minimization and VSL maximization.

Table 4.8 Results of IEEE 14- bus system for case study 3
Optimization of both
Optimization of only
RPL and VSL
Control variable RPL
(multi-objective case)
Settings (single objective case)
FAS FAM
T1 (4-7) 0.95 1.05
T2 (4-9) 0.95 1.00
T3 (5-6) 0.95 1.00
UPFC series injected
0.02454 0.02130
voltage ,Vse , (p.u.)
UPFC series injected
1.3498 1.34192
voltage angle, γ , (rad.)
UPFC location 9-10 9-10
RPL, (p.u.) 0.10562 0.11021
VSL 1.11 1.17
Combined fitness function
9.11462 8.65721
(W1*RPL+W2*1/VSL)
 67

Table 4.9 Results of the New England 39- bus system for case study 3
Optimization of both
Optimization of only RPL RPL and VSL
Control variable
(single objective case) (multi-objective case)
Settings
FAS FAM
T1 (2-30) 1.10 1.00
T2 (10-32) 1.00 1.05
T3 (12-11) 1.00 1.05
T4 (12-13) 1.00 1.05
T5 (19-33) 1.00 1.10
T6(19-20) 1.00 1.10
T7 (20-34) 1.00 1.05
T8 (22-35) 1.05 1.05
T9(23-36) 1.00 1.00
T10(25-37) 1.05 1.10
T11 ( 29-38) 1.05 1.05
T12 (31-6) 1.00 1.05
UPFC series injected voltage,
0.0225 0.0168
Vse, (p.u.)

UPFC series injected

1.2759 1.2544
voltage angle ,γ, (rad)

UPFC location 16-24 16-19

RPL, (p.u.) 0.3463 0.3617
VSL 0.930 1.080
Combined fitness
function
11.0989 9.6209
(W1*RPL+W2*
1/VSL)
 68

Table 4.10 Results of practical Indian 24- bus system for case study 3

Optimization of both
Optimization of only
RPL and VSL
Control variable RPL
(multi-objective case)
Settings (single objective case)
FAS FAM
T1 (15-1) 0.9875 1.0375
T2 (17-2) 1.0250 1.0250
T3 (24-3) 1.0125 1.0375
T4 (21-4) 1.0125 1.0375
T5 (16-5) 0.9000 0.9875
T6 (19-6) 0.9250 0.9250
T7 (20-7) 0.9250 0.9250
T8 (14-8) 1.0500 0.9125
T9 (23-9) 0.9875 0.9750
T10 (18-10) 0.9750 1.0125
T11 (22-13) 1.0375 1.0250
UPFC series injected
0.059694
voltage, Vse , (p.u.) 0.048454
UPFC series injected
1.059780 1.322957
voltage angle ,γ, (rad.)
UPFC location 22-23 12-14
RPL (p.u.) 0.48997 0.49018
VSL 0.15097 0.16016
Combined fitness
function 7.11380 6.73394
(W1*RPL+W2*1/VSL)
 69

Fig. 4.26 Convergence characteristic of IEEE 14- bus system for case study 3.1

Fig. 4.27 Convergence characteristic of New England 39-bus system for case

study 3.1
 70

Fig. 4.28 Convergence characteristic of practical Indian 24 bus system for case study
3.1

The figures 4.26 to 4.28 show the single objective RPL minimization,
depicting the convergence. From the results it is found that the proposed algorithm
performs better in terms of loss minimization, convergence speed and accuracy. The
solution started with higher value but finally reached further minimum value in the
objective values and it clearly shows that global solution is reached with minimum
number of iterations. Analyzing the simulation results, from the convergence point of
view, it is clear that when variables are simultaneously optimized, it can be noted that
the performance of FA is better than the performance of case 1.1 and case 1.2.
 71

Fig.4.29 P-V curve of the weakest bus of IEEE14- bus system with FAS for case
study 3.1

Fig. 4.30 P-V curve of the weakest bus of New England 39- bus system with FAS for
case study 3.1
 72

Fig. 4.31 P-V curve of the weakest bus of practical Indian 24 bus system with FAS for
case study 3.1
Figures 4.29, 4.30 and 4.31 show the p-v curves for the weakest bus of the
test systems considered, are for the FA with single objective under optimal control
variables of UPFC parameters along with taps excluding VSL in the fitness function.
The CPF for voltage stability limit resulted in values 1.11, 0.930 and 0.15097 for the
test systems considered.
 73

Fig.4.32 P-V curve of the weakest bus of IEEE14- bus system with FAM for case
study 3.2

Fig. 4.33 P-V curve of the weakest bus of New England 39- bus system with FAM for
case study 3.2
 74

Fig .4.34 P-V curve of the weakest bus of practical Indian 24-bus system with FAM
for case study 3.2

Figures 4.32, 4.33 and 4.34 show the p-v curves of weakest bus of IEEE 14-
bus system, New England 39- bus system and Indian 24- bus system when variables
are simultaneously optimized with FA for multi-objective case. It is clear from the
results that VSL has improved to 25.8064 %( from 0.93 to 1.17), 33.333 % (from 0.81
to 1.08), and 127.8236 %( from 0.0703 to 0.16016) for the test systems considered,
compared to base case. Also It is observed from the figures that the VSL has
improved to 5.40505 % (form 1.11to 1.17), 16.1290 % (from 0.930 to 1.08), and
6.0873 % (from 0.15097 to 0.16016) ) for IEEE 14, New England 39 and Indian 24
bus test systems respectively compared to single objective case. It can be concluded
that voltage stability margin significantly improved using CPF, to ensure the
feasibility of the optimal position and parameter settings of UPFC for secure
operation of power system.
 75

Fig.4.35 Voltage Profiles of practical Indian 24- bus system with FAS for case

study 3.1

Fig. 4.36 Voltage Profiles of practical Indian 24- bus system with FAM for case
study 3.2
 76

Figures 4.35 and 4.36 relate the voltage profiles corresponding to the base case
PF and with FA of all buses of the test system considered. It is clear from the figures
that FA has helped to improve voltage profile in all the load buses which leads to
significant reduction of losses and improvement in stability margin. Similar
characterises can be observed for the test systems considered.

4.5.4 Case 4: Sequential optimization of OUPFC location and its parameters

keeping the optimized transformer taps fixed
4.5.4.1 Case 4.1: Optimization of only RPL (single objective case)
With the optimized transformer tap values, the OUPFC location and its
parameters are optimized with FA technique for single objective of RPL
minimization. The optimized OUPFC location and its parameters along with RPL and
VSL obtained with proposed technique are given in tables 4.11, 4.12 and 4.13 for
IEEE 14- bus, New England 39- bus and practical Indian 24-bus systems respectively.
The power injection model as discussed in chapter-3 is used for modelling the
OUPFC. For IEEE 14 bus system only 7 lines are considered for connecting the
OUPFC as the remaining 13 lines consists of transformers and feeding generator
powers to the network. For New England 39 bus system 32 lines are considered for
connecting the OUPFC as the remaining 14 lines consists of transformers and feeding
generator powers to the network. Similarly for practical Indian 24 bus system 16 lines
are considered for connecting the OUPFC as the remaining 11 lines consists of
transformers and feeding generator powers to the network. It is clear that with the
incorporation of OUPFC the RPL has reduced to 6.5296 % (from 0.1095 to
0.102350), 2.5867 %( from 0.3545 to 0.34533), and 12.8539 %( from 0.51815 to
0.451547) for the test systems considered, compared to case 2.1 i.e., UPFC is
considered in the system. From the results it is clear that OUPFC gives better
performance compared to UPFC. Also it is seen that when more nonlinear equipment
like OUPFC is inducted into the system, the fitness function has successfully
converged with FA. This shows the potentiality of the proposed algorithm in solving
the non liner optimization problem particularly the RPL minimization problem when
OUPFC is introduced. In the single objective case VSL is not included in the fitness
function. With the optimized taps the CPF is run for estimating the voltage stability
 77

limit in the single objective case and found to be 0.960, 0.930 and 0.147330 for the test
systems considered.

4.5.4.2 Case 4.2: Optimization of both RPL and VSL (multi-objective case)

When both RPL and VSL are optimized in the multi-objective case, the cost
function is modified. The reciprocal of VSL is added to the real power loss and the
optimization is carried out with FAM technique. The optimized OUPFC parameters
along with RPL and VSL obtained with proposed technique are given in table 4.11,
4.12 & 4.13 for IEEE 14, New England 39 and Indian 24 bus test systems
respectively. From the results it is seen that the VSL is improved but RPL is increased
marginally with FAM technique for all the test systems. Even RPL is increased
slightly; the overall multi-objective function that is the sum of real power loss and
reciprocal of VSL is reduced compared to single objective case. RPL has increased by
a margin of 4.3702% (from 0.102350 p.u to 0.106823 p.u), 6.5936 % (from0.34533
p.u. to 0.36810 p.u.) and 8.4578 % (form 0.451547 p.u to 0.489738 p.u.) and VSL has
improved to 15.625 % (form 0.960 to 1.11), 12.9032 % (from 0.93 to 1.05), and
10.9790% ( from 0.14300 to 0.15870) for the test systems considered, compared to
single objective case. But, the combined fitness function obtained with FAM has
reduced to 13.3393 %( from 10.5190 to 9.115832), 11.4942 %( from 11.1010 to
9.891915) and 4.5542 %( from 7.23903 to 6.90935) for IEEE 14- bus, New England
39 -bus and practical Indian 24-bus systems respectively compared to single objective
case. It is noticed that the combined fitness function reduced to 2.8019 %( from
9.37862 to 9.115832), 2.8117 % (from 10.1781 to 9.891915) and 3.92710% (from
7.068523 to 6.790935) for the test systems considered, compared to case 2.2. From
the results it is clear that the percentage reduction of combined fitness function is
more when OUPFC is considered compared to UPFC considered in the system. This
shows the effectiveness of OUPFC for further increment of loadability limit. It is
found that with FAM, the OUPFC parameters can be so optimized that the system
could be loaded further beyond double its nominal loading before a voltage collapse
could occur for all the test systems. So FA is best succeeded in reducing the combined
objective of loss minimization and VSL maximization with OUPFC device.
 78

Table 4.11 Results of IEEE 14- bus system for case study 4
Optimization of Optimization of both
only RPL RPL and VSL
Control variable
(single objective (multi-objective case)
Settings
case)
FAS FAM
UPFC series injected
0.065360 0.024399
voltage ,Vse , (p.u.)
UPFC series injected
27.712677 -51.399738
voltage angle, ρ, (deg.)
-12.813490
PST phase angle, σ , (deg.) 2.880455
UPFC location 9-10 4-5
0.102350 0.106823
RPL (p.u.)
VSL 0.96 1.110000
Combined fitness function
10.5190 9.115832
(W1*RPL+W2*1/VSL)

Table 4.12 Results of New England 39- bus system for case study 4
Optimization of
Optimization of both RPL and
only RPL VSL
Control variable
(single objective (multi-objective
Settings
case) case)

FAS FAM
UPFC series injected voltage,
0.007521 0.007066
Vse , (p.u.)
UPFC series injected voltage
-18.547949 -27.692616
angle, ρ, (deg.)
PST phase angle, σ , (deg.) -5.277385 -8.943183
UPFC location 7-8 5-8
RPL (p.u.) 0.348356(0.345336) 0.368106
VSL 0.93 1.05
Combined fitness function
11.1010 9.891915
(W1*RPL+W2*1/VSL)
 79

Table 4.13 Results of practical Indian 24-bus system for case study 4

Optimization of Optimization of both

only RPL RPL and VSL
Control variable
(single objective (multi-objective case)
Settings
case)
FAS FAM
UPFC series injected voltage
0.083953 0.010565
Vse, , (p.u.)
UPFC series injected voltage
84.102762 20.546615
angle, ρ, (deg.)
PST phase angle , σ , (deg.) -19.959175 -6.789774
UPFC location 21-20 21-19
RPL (p.u.) 0.451547 0.489738
VSL 0.147330 0.158700
Combined fitness function
7.23903 6.790935
(W1*RPL+W2*1/VSL)

Fig. 4.37 Convergence characteristic IEEE 14- bus system for case study 4.1
 80

Fig. 4.38 Convergence characteristic of New England 39- bus system for case

study 4.1

Fig. 4.39 Convergence characteristic of practical Indian 24 bus system for case

study 4.1
 81

The Figures 4.37 to 4.39 show the single objective RPL minimization,
depicting the convergence. From the results it is found that the proposed FA with
OUPFC performs better in terms of loss minimization, convergence speed and
accuracy. The solution started with higher value but finally reached further minimum
value in the objective values and it clearly shows that global solution is reached with
minimum number of iterations. Analyzing the simulation results, from the
convergence point of view, it is clear that with the incorporation of OUPFC, it can be
noted that the performance of FA with OUPFC is better than the performance of FA
with UPFC (case 2.1).

Fig. 4.40 P-V curve of weakest bus of IEEE 14- bus system with FAS for case
study 4.1
 82

Fig. 4.41 P-V curve of the weakest bus of New England 39- bus system with FAS for
case study 4.1

Fig. 4.42 P-V curve of the weakest bus of practical Indian 24 bus system with FAS for
case study 4.1
 83

Figures 4.40, 4.41 and 4.42 show the p-v curves for the weakest bus of the
test systems considered, are for the FA with single objective under optimal control
variables of OUPFC location and its parameters excluding VSL in the fitness
function. The CPF for voltage stability limit resulted in values 0.96, 0.93 and
0.147330 for the test systems considered.

Fig. 4.43 P-V curve of weakest bus of IEEE 14- bus system with FAM for case
study 4.2
 84

Fig. 4.44 P-V curve of the weakest bus of New England 39- bus system with FAM for
case study 4.2

Fig. 4.45 P-V curve of the weakest bus of practical Indian 24-bus system with FAM
for case study 4.2
 85

Figures 4.43, 4.44 and 4.45 show the p-v curves of weakest bus of IEEE 14-
bus system, New England 39-bus system and Indian 24-bus system when OUPFC
variables are sequentially optimized with FA for multi-objective case. It is clear from
the results that VSL has improved to 19.3548 %( from 0.93 to 1.11), 29.6296 % (
from 0.81 to 1.05), and 125.7467 %( from 0.0703 to 0.158700) for IEEE 14, New
England 39 and Indian 24 bus test systems respectively compared to base case . Also
It is observed from the figures that the VSL has improved to VSL has improved to
2.777 % (form 1.08 to 1.11), 2.9411 % (from 1.02 to 1.05), and 3.8680% ( from
0.15279 to 0.158700) for IEEE 14, New England 39 and Indian 24 bus test systems
respectively compared to case 2.2. Also It is observed from the figures that the VSL
has improved to VSL has improved to 15.625 % (form 0.960 to 1.11), 12.9032 %
(from 0.93 to 1.05), and 10.9790% (from 0.14300 to 0.15870) for the test systems
considered, compared to single objective case. It can be concluded that voltage
stability margin was further more improved using CPF, to ensure the feasibility of the
optimal position and parameter settings of OUPFC for secure operation of power
system

Fig. 4.46 Voltage Profiles of practical Indian 24- bus system with FAS for case
study 4.1
 86

Fig. 4.47: Voltage Profiles of practical Indian 24- bus system with FAM for case
study 4.2

Figures 4.46 and 4.47 show the voltage profiles corresponding to the base case
PF and with FA of all buses of the test system considered. It is clear from the figures
that FA has helped to improve voltage profile in all the load buses which leads to
further reduction of losses and improvement in stability margin. Similar characterises
can be observed for the test systems considered.
4.5.5 Case 5: Simultaneous optimization of OUPFC location and its parameters
along with taps
4.5.5.1 Case 5.1: Optimization of only RPL (single objective case)

Simultaneous optimization of OUPFC location and its variables along with the
transformer taps with FAS could still reduce the objective function. For optimizing
with the FAS, the number of variables now becomes 6 i.e., 3 transformer tap
positions, and 3 UPFC variables, 15, i.e., 12 transformer tap positions, and three
UPFC variables 14, i.e,11 transformer tap positions, and 3 UPFC variables for IEEE
14- bus, New England 39- bus and practical Indian 24 -bus system respectively. The
optimized transformer taps, OUPFC location and its parameters along with RPL and
 87

VSL obtained with proposed technique are given in tables 4.14, 4.15 and 4.16 for
IEEE 14 bus, New England 39 bus and practical Indian 24 bus system respectively.
Also the RPL is reduced to 3.78716 % (from 0.10562 to 0.10162), 1.3254 %(from
0.3463 to 0.341710), and 8.0794 % (from 0.48997 to 0.450383) for IEEE 14 bus,
New England 39 bus and practical Indian 24 bus system respectively compared to
case 3.1.With the optimized taps the CPF is run for estimating the voltage stability
limit in the single objective case and found to be 1.11, 0.960 and 0.164980 for the test
systems considered.
4.5.5.2 Case 5.2: Optimization of both RPL and VSL (multi-objective case)
When both RPL and VSL are optimized in the multi-objective case, the cost
function is modified. The reciprocal of VSL is added to the real power loss and the
optimization is carried out with FAM technique. The optimized transformer taps,
UPFC parameters along with RPL and VSL obtained with proposed technique are
given in tables 4.8, 4.9 & 4.10 for IEEE 14, New England 39 bus and practical Indian
24 bus test systems respectively. From the results it is seen that the VSL is improved
but RPL is increased marginally with FAM technique for all the test systems. Even
RPL is increased slightly; the overall multi-objective function that is the sum of real
power loss and reciprocal of VSL is reduced compared to single objective case. RPL
has increased by a margin of 3.9454% (from 0.101612 to 0.105621), 2.6762 % (from
0.341710 to 0.350855) and 7.6499 % (form 0.450383 to 0.484837) and VSL has
improved to 8.1081 % (form 1.11to 1.20), 15.625 % (from 0.960 to 1.11), and
24.8636 % (from 0.164980 to 0.20600) for IEEE 14, New England 39 and practical
Indian 24 bus test systems respectively compared to single objective case. But the
combined fitness function obtained with FAM has reduced to 7.3723 %( from
9.11062 to8.438954), 12.9986 %( from 10.7583 to 9.359864) and 18.006 %( from
6.51172 to 5.33905) for the test systems considered, compared to single objective
case. It is noticed from the results that the combined fitness function reduced to
2.521% (from 8.65721 to 8.438954), 15.6685 (from 11.0989 to 9.359864), and
24.9458% (7.1138 to 5.339205) for the test systems considered, compared to case 3.2.
It is found that with FAM, when OUPFC variables are simultaneously optimized that
the system could be more loaded before a voltage collapse could occur for all the test
 88

systems compared to case 3.2. So FA is best succeeded in reducing the combined

objective of loss minimization and VSL maximization considering with OUPFC.

Table 4.14 Results of IEEE 14- bus system for case study 5

Optimization of both
Optimization of only
RPL and VSL
Control variable RPL
(multi-objective case)
Settings (single objective case)
FAS FAM
T1 (4-7) 1.0 1.00
T2 (4-9) 1.0 0.90
T3 (5-6) 1.0 0.95
UPFC series injected
0.058625 0.096260
voltage ,Vse , (p.u.)
UPFC series injected
6.857240 57.295169
voltage angle ,ρ, (deg.)
PST phase angle, σ , (deg.) 2.338319 16.218863
UPFC location 9-10 12-13
RPL (p.u.) 0.101612 0.105621
VSL 1.110000 1.200
Combined fitness function
9.11062 8.438954
(W1*RPL+W2*1/VSL)

Table 4.15 Results of New England 39- bus system for case study 5

Optimization of both
Optimization of only
RPL and VSL
Control variable RPL
(multi-objective case)
Settings (single objective case)
FAS FAM
T1 (2-30) 0.95 0.85
T2 (10-32) 1.00 0.95
T3 (12-11) 1.00 0.90
T4 (12-13) 0.90 0.95
T5 (19-33) 0.95 1.00
T6(19-20) 1.00 0.95
T7 (20-34) 1.00 1.00
T8 (22-35) 0.95 0.95
T9(23-36) 0.95 0.90
T10(25-37) 0.95 0.90
T11 ( 29-38) 0.90 0.95
T12 (31-6) 1.00 0.90
 89

UPFC series injected

voltage ,Vse , (p.u.) 0.014989 0.011831
UPFC series injected
-0.006453 -28.949538
voltage angle ,ρ, (deg.)
PST phase angle, σ , (deg.) -0.000955 -7.832045
UPFC location 8-9 3-4
RPL (p.u.) 0.341710 0.350855
VSL 0.96 1.11
Combined fitness function
(W1*RPL+W2* 10.7583 9.359864
1/VSL)

Table 4.16 Results of practical Indian 24- bus test system for case study 5

Optimization of both
Optimization of only
RPL and VSL
Control variable RPL
(multi-objective case)
Settings (single objective case)
FAS FAM
T1(15-1) 0.9375 1.0375
T2 (17-2) 1.0000 1.0500
T3 (24-3) 0.9875 1.0500
T4 (21-4) 0.9750 1.0250
T5 (16-5) 1.0000 0.9375
T6 (19-6) 0.9625 0.9375
T7 (20-7) 0.9375 0.9000
T8 (14-8) 0.9625 0.9000
T9 (23-9) 0.9625 1.0250
T10 (18-10) 1.0125 1.0375
T11 (22-13) 1.0000 0.9875
UPFC series injected
0.040771 0.039189
voltage Vse , (p.u.)
UPFC series injected
40.412440 26.363203
voltage angle ,ρ, (deg.)
PST phase angle , σ, (deg.) 14.911315 18.413796
UPFC location 19-6 22-23
RPL (p.u.) 0.450383 0.484837
VSL 0.164980 0.206000
Combined fitness function
(W1*RPL+W2* 6.51172 5.339205
1/VSL)
 90

Fig.4.48 Convergence characteristic of IEEE 14- bus system for case study 5.1

Fig. 4.49: Convergence characteristic of New England 39- bus system for case
study 5.1
 91

Fig. 4.50 Convergence characteristic of practical Indian 24-bus system for case
study 5.1

The Figures 4.48 to 4.50 show the single objective RPL minimization,
depicting the convergence. From the results it is found that the proposed algorithm
performs better in terms of loss minimization, convergence speed and accuracy. The
solution started with higher value but finally reached further minimum value in the
objective values and it clearly shows that global solution is reached with minimum
number of iterations. Analyzing the simulation results, from the convergence point of
view, it is clear that when variables are simultaneously optimized, it can be noted that
the performance of FA is better than the performance of case 3.1.
 92

Fig. 4.51 P-V curve of weakest bus of IEEE 14- bus system with FAS for case
study 5.1

Fig. 4.52 P-V Curve of weakest bus of New England 39-bus system with FAS for
case study 5.1
 93

Fig. 4.53 P-V Curve of weakest bus of practical Indian 24-bus system with FAS for
case study 5.1

Figures 4.51, 4.52 and 4.53 show the p-v curves for the weakest bus of the test
systems considered, are for the FA with single objective under optimal control
variables of OUPFC location along with taps excluding VSL in the fitness function.
The CPF for voltage stability limit resulted in values 1.11, 0.96 and 0.164980 for the
test systems considered.
 94

Fig .4.54 P-V curve of weakest bus of IEEE 14 bus system with FAM for case
study 5.2

Fig. 4.55 P-V Curve of weakest bus of New England 39- bus system with FAM for
case study 5.2
 95

Fig. 4.56 P-V Curve of weakest bus of practical Indian 24- bus system with FAM for
case study 5.2

Figures 4.54, 4.55 and 4.56 show the p-v curves of weakest bus of IEEE 14
bus system, New England 39 bus system and Indian 24 bus system when OUPFC
variables along with taps are simultaneously optimized obtained with FA for multi-
objective case. It is clear from the results that VSL has improved to 29.0322 %( from
0.93 to 1.20), 37.037 % ( from 0.81 to 1.11), and 193.02 %( from 0.0703 to 0.20600)
for IEEE 14, New England 39 and Indian 24 bus test systems respectively compared
to base case . Also It is observed from the figures that the VSL has improved to VSL
has improved to 2.5641 % (form 1.17 to 1.20), 2.777 % (from 1.08 to 1.11), and
28.6213 % (from 0.16016 to 0.20600) for the test systems considered, compared to
case 3.2. Also It is observed from the figures that the VSL has improved to VSL has
improved to 8.1081 % (form 1.11to 1.20), 15.625 % (from 0.960 to 1.11), and
24.8636 % (from 0.164980 to 0.20600) for the test systems considered. compared to
single objective case. It can be concluded that voltage stability margin significantly
improved using CPF, to ensure the feasibility of the optimal position and parameter
settings of OUPFC for secure operation of power system.
 96

Fig. 4.57 Voltage Profiles of practical Indian 24- bus system with FAS for case
study 5.1

Fig. 4.58 Voltage Profiles of practical Indian 24- bus system with FAM for case
study 5.2
 97

Figures 4.57 and 4.58 show the voltage profiles corresponding to the base case
PF and with FA of all buses of the test system considered. It is clear from the figures
that FA has helped to improve voltage profile in all the load buses which leads to
significant reduction of losses and improvement in stability margin. Similar
characterises can be observed for test systems considered.

4.6 PARAMETRS TUNING OF FA

Determination of the parameters such as population size, randomization

parameter (α), attractiveness (β) and absorption coefficient (γ) are important to
implement FA successfully, because they affect the performance of the algorithm for
various problems. Under a fixed iteration number of 30, the parameter settings for
the algorithm are varied with several trials as shown in tables 4.1 7 and 4.18 for IEEE
14-bus and New England 39-bus systems and the optimal settings for best result are
chosen. From the tables, the trial 6 gives the best values for RPL over the other trials.
Based on the trials the population size of 25 is chosen and the values of α, β, and γ are
set as 0.3, 0.2, and 1.0 respectively. For practical Indian 24- bus system also the same
parameters are applied.

Table 4.17 Influence of control parameters on the performance of FA for IEEE 14-bus system

Randomizat RPL (MW)

Attractive Absorption
Population ion
Trial ness coefficient Case
size parameter Case 2.1 Case 3.1
(β) (γ) 1.1
(α)
1 1 0.1 0.1 0.9 13.3579 11.1601 11.0706
2 5 0.1 0.1 0.9 13.3579 11.1321 11.0462
3 10 0.2 0.1 0.95 13.3572 11.1221 11.0528
4 15 0.2 0.2 0.95 13.3578 11.0392 10.9831
5 20 0.3 0.2 1.0 13.3521 10.9729 10.6428
6 25 0.3 0.2 1.0 13.3430 10.9576 10.5620
7 30 0.4 0.2 1.0 13.3430 10.9576 10.5632
 98

Table 4.18 Influence of control parameters on the performance of FA for New

England 39- bus system

Randomizat RPL (MW)

Attractive Absorption
Population -ion
Trial ness coefficient Case Case Case
size parameter
(α) (β) (γ) 1.1 2.1 3.1
1 1 0.1 0.1 0.9 43.717 42.031 40.770
2 5 0.1 0.1 0.9 42.668 41.093 39.924
3 10 0.2 0.1 0.95 42.668 40.088 36.801
4 15 0.2 0.2 0.95 42.118 38.568 36.564
5 20 0.3 0.2 1.0 42.10 35.462 35.630
6 25 0.3 0.2 1.0 42.03 35.451 34.630
7 30 0.4 0.2 1.0 42.03 35.468 34.630

4.7 ROBUSTNESS OF FA
To test the robustness of FA, 10 trial runs were performed for IEEE 14-bus
and New England 39-bus test systems. Tables 4.19 and 4.20 show the results of RPL
values and computational time for case studies 1.1, 2.2 and 3.1 of IEEE 14 bus test
system and New England 39 bus system. It can be seen here that the optimal RPL
obtained by the proposed FA for all the three cases are always nearer to the average
value for the test systems, showing the robustness of the proposed FA method for
the OPF problem of RPL minimization. Same observations are made for the practical
Indian 24 bus system.

Table 4.19 RPL values and computational time for IEEE 14-bus system for 10 trials
RPL (p.u.)
Simulation
S.no Case study
Best value worst value Average value time(s)

1 1.1 0.133431 0.133573 0.133529. 5.18

2 2.1 0.109576 0.113095 0.111705 5.97
3 3.1 0.109576 0.113095 0.111705 5.97
 99

Table 4.20 RPL values and computational time for New England
39- bus system for10 trials

RPL (p.u.)
Simulation
S.no Case study
Best value worst value Average value time(s)

1 1.1 0.42503
0.4203 0.43177 29.96
2 2.1 0.3545 0.38934 0.37124 31.36
3 3.1 0.3463 0.37744 0.36953 65.73

From tables 4.19 and 4.20, it is also clear that the proposed FA method is
computationally efficient and the time requirement is vey less.

4.8 CONCLUSION

FA method has been successfully implemented to find the optimum control

variables for the problem of RPL minimization (single objective case) and for
simultaneous optimization of RPL and VSL (multi-objective case). Transformers tap
settings, UPFC and OUPFC parameters are considered as control variables. The
UPFC and OUPFC devices are considered separately during the optimization process.
The effectiveness of the proposed method is demonstrated on the test systems
considered. In multi-objective case even though there is slight increase in the real
power loss, the VSL has improved and the combined fitness function has reached
optimal solution with the proposed algorithm. From the simulation results it can be
concluded that the proposed algorithm is capable of finding optimum control
variables for both single and multi-objective optimization problems and so helpful for
stable and secure operation of the power system. The results also indicate the better
convergence characteristics, computational efficiency and robustness of the proposed
FA.