POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 16 | NUMBER: 4 | 2018 | DECEMBER

Improved Optimal Power Flow for a Power System

Incorporating Wind Power Generation by Using
Grey Wolf Optimizer Algorithm

Sebaa HADDI, Omrane BOUKETIR, Tarek BOUKTIR

Department of Electrical Engineering, Faculty of Technology, University of Setif, El Bez, 19000 Setif, Algeria

[email protected], [email protected], [email protected]

DOI: 10.15598/aeee.v16i4.2883

Abstract. In this paper, an efficient Grey Wolf Opti- used for adjusting optimal settings of power systems.
mizer (GWO) search algorithm is presented for solving Therefore, it has received more attention from many
the optimal power flow problem in a power system, en- researchers throughout the world [1]. Several optimiza-
hanced by wind power plant. The GWO algorithm is tion techniques have been used to solve this problem,
based on meta-heuristic method, and it has been proven in order to find the optimal solution for operational
to give very competitive results in different optimization objective functions in a power system, such as fuel
problems. First, by using the proposed technique, the cost, voltage profile and voltage stability enhancement.
system independent variables such as the generators’ Some methods are based on nonlinear programming,
power outputs as well as the associated dependent vari- quadratic programming, Newton techniques and inte-
ables like the bus voltage magnitudes, transformer tap rior point. These methods have many drawbacks, such
setting and shunt VAR compensators values are opti- as high complexity, convergence to local optimum and
mized to meet the power system operation requirements. sensitivity to initial conditions [2].
The Optimal power flow study is then performed to
assess the impact of variable wind power generation
on system parameters. Two standard power systems Intelligent search methods such as meta-heuristic op-
IEEE30 and IEEE57 are used to test and verify the ef-timization techniques have been introduced to over-
fectiveness of the proposed GWO method. The obtained come some optimization problems encountered with
results are then compared with others given by available
classical methods. The most popular ones are; Genetic
optimization methods in the literature. The outcome of
Algorithm (GA), Particle Swarm Optimization (PSO),
the comparison proved the superiority of the GWO al- Simulated Annealing (SA), Evolutionary Programming
gorithm over other meta-heuristics techniques such as(EP), Artificial Bee Colony algorithm (ABC), Ant
Modified Differential Evolution (MDE), Enhanced Ge- Colony Optimization (ACO), Differential Evolution
netic Algorithm (EGA), Particle Swarm Optimization (DE). Based on these original methods new derived
(PSO), Biogeography Based Optimization (BBO), Ar- techniques have been obtained and used in OPF prob-
tificial Bee Algorithm (ABC) and Tree-Seed Algorithm lem as in ABC [3], EGA [4], gradient method and Gen-
(TSA). eral Algebraic Modeling System (GAMS) [5], Efficient
Evolutionary Algorithm (EEA) [5], Evolving Ant Di-
rection Differential Evolution (EADDE) [6], Differen-
Keywords tial Search Algorithm (DSA) [7], CSA [8], Krill Herd
Algorithm (KHA) [9], Simulated Annealing (SA) [10],
Interior Search Algorithm (ISA) [11], Enhanced Ge-
Grey Wolf Optimizer (GWO), grey wolves,
netic Algorithm (EGA) [12], BBO [13], PSO [14], Grav-
OPF problem.
itational Search Algorithm (GSA) [15], Genetic evolv-
ing ant direction PSODV hybrid algorithm (PSODV)
[16], Real Coded Biogeography-Based Optimization
1. Introduction RC-BBO [17] and Evolutionary Algorithm (EA) [18].
Most of these methods are recently extensively used
Optimal power flow problem has been studied for many in solving global optimization searching problems and
years and has become one of the most important means have been giving promising results beside that they

c 2018 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 471

POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 16 | NUMBER: 4 | 2018 | DECEMBER

have attractive characteristics, such as easy implemen- 2. OPF Problem Formulation

tation and fast convergence [18].
A common drawback to meta-heuristic methods is 2.1. Optimal Power Flow
that, in general, the optimization performance is highly
dependent on fine parameter tuning. However, the pro- The objective of conventional OPF problem is to min-
posed approach outperforms these methods in term of imize fuel cost for power generation by determining
convergence speed to the best solution. Moreover, the a set of control variables while satisfying system equal-
use of OPF is extended to include the study of renew- ity and inequality constraints. The OPF problem is
able energy systems like wind power, which becomes formulated by [27]:
more and more useful in recent power networks, and
many studies are made to integrate this natural power min f (x, u), (1)
efficiently to a power system. Ranjit and Jadhav in
[19], as well as Maskar et al. in [20], presented a study s · tg(x, u) = 0, (2)
of OPF problem in a system incorporating wind power
sources, using modified ABC algorithm named Gbest h(x, u) ≤ 0, (3)
guided ABC algorithm; the method showed good re-
sults for fuel cost optimization case, and voltage profile where [~x]: is the vector of dependent variables consist-
enhancement, then under wind condition the total op- ing of slack bus P G1 , load bus voltage VL , generator
erating cost is optimized efficiently, compared to other reactive power outputs QG , and transmission line load-
methods. The method presented some benefits con- ing SL . This vector is expressed by:
cerning reserve coefficient adjustment when consider-
ing imbalance cost of wind power. Meanwhile, Shanhe X T = [PG1 , VL1 , ..VN D , QG1 , ..QGN , Sl1 , ..SlN L ] , (4)
et al. [21] presented a new economic dispatch technique
based on PSO-GSA algorithm for a power system in- where N D, N G and N L are number of load buses,
cluding two wind power sources; the method was tested number of generators, and number of transmission
on a six generators’ system connected with two stochas- lines, respectively.
tic wind power sources. The test yielded good results [~u] is the vector of independent variables consisting
compared with other results found in the literature of generator voltages VG , generator real power outputs
with different methods especially for cost and emission PG except at the slack bus PG1 , transformer tap set-
reduction. Panda and Tripathy [22], and Mishra and tings TP , shunt VAR compensation QC . This vector is
Vignesh [23] introduced another OPF algorithm based expressed by:
on security constrained OPF solution of wind-thermal
generation system using modified bacteria foraging al- ~uT =[VG1 , ..VN G , .PG2 ..PGN , TP 1 , ..TPN T , QC1 , ..QCN C ],
gorithm. The method was tested on the same system (5)
stated in [18], in which the wind power variability was where: N G, N T , and N C are the number of ther-
modelled incorporating conventional thermal generat- mal generators, regulating transformers, shunt com-
ing system. Recent works in [19], [24] and [25] pre- pensators, respectively.
sented better results and faster convergence character-
istics using Grey Wolf Optimizer algorithm. Grey Wolf
Optimizer (GWO) algorithm mimics the behaviour of 1) Fuel Cost Optimization
grey wolves in nature by simulating their leadership
hierarchy, through haunting, searching for, encircling, The function f from Eq. (1) concerned in the OPF
and attacking the prey [26]. study represents the total generation cost formulation
The present paper aims to investigate the effi- and it is as:
ciency of GWO algorithm, as a new meta-heuristic XNG
population-based algorithm. It presents a solution to f (Pgi ) = 2
ai Pgi + bi Pgi + ci ($/h). (6)
the OPF problem of a power system incorporating wind i=1
power generation. The rest of the paper is organized
as follows; after the introduction, the OPF problem When considering valve effect, the function f ; is rewrit-
formulation is given in Sec 2. Subsec. 2.2. ten as:
deals with the OPF problem incorporating wind power. NPG
Section 3. presents the GWO algorithm and associ- f (Pgi ) = 2
ai Pgi + bi Pgi +
ated simulation steps for solving the OPF problem. In i=1 (7)
Sec. 4. simulation results using GWO algorithm are +c i | d i (sin(e i (P gi min − P gi ($/h),
presented and analysed. Section 5. concludes the
study. where: ai , bi , ci , di and ei are fuel cost coefficients of
ith thermal generating unit.

c 2018 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 472

POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 16 | NUMBER: 4 | 2018 | DECEMBER

2) Voltage Profile Improvement rewritten as:

N
PG lim 2
The aim of this objective function is to minimize the f (Pgi ) = fi + ηP (Pg1 − Pg1 ) + ηQ (Qg1 − Qlim
g1 )
2
i=1
load bus voltage deviations from the reference value PN L N
PB
lim 2
which is 1 per unit; this function is expressed by: +ηV i=1 (VLi − PLi ) + ηS (Sit − Pitlim )2 ,
i=1

NP Q
(12)
X where: ηp , ηq , ηv and ηs are penalty factors or weights
VD = | Vi − Vref |, (8)
of active power generation of slack bus, reactive power
i=1
output of generator buses, PQ bus magnitudes and
where: VD represents the voltage deviation in (p.u); transmission line loadings respectively. Their values
Vi is the ith load bus voltage; and Vref is the reference are generally taken to be 100 for the same reason in
voltage which is taken here to be 1 p.u, and thus the Eq. (9) [14], [15], [16] and [17].
objective function Eq. (6) becomes as follows:

N
2.2. OPF Problem Formulation with
PG
f (Pgi ) = 2
(ai Pgi + bi Pgi + ci )+ Wind Power
i=1
NPPQ
(9)
+w | Vi − 1 |, The fuel cost objective in Eq. (6) is augmented with
i=1 the cost associated with stochastic wind power, as in
Eq. (13) [28]
where: w represents a weighting factor selected by the
NPG
user; many works are choosing w to be 100 in order FT = 2
ai Pgi + bi Pgi + ci +
to keep the variable within the designed limits, as in i=1 (13)
[1] and [15].The OPF equality constraint such as the +F (P wj ) + C wj ($/h),
active power balance equation is expressed by: where; F (Pwj ) is the cost for generation of wind power
which is directly proportional to the wind power output
NG
X and is given by:
PGi = Pd + Pl , (10)
i=1 F (Pwj ) = dj × Pwj ($/h), (14)
where: Pd represents the load of the system, and Pl is dj : is the direct cost coefficient of non-utility service,
the total active power loss. which equals to zero for the utility services.

Cwj : represents the imbalance cost of investment in

3) OPF Incorporating Inequality j th wind power source due to two components as in
Constraints Eq. (15) [29]:
N
In order to handle the inequality constraints of depen- Pw
Cw = (Kp,j × Wj,ue )+
dent variables, including slack bus real and reactive j=1
(15)
power, load bus voltage magnitudes and transmissions N
Pw
line loading; the problem is transformed into uncon- + (KR,j × Wj,oe ) ($/h),
j=1
strained OPF problem by penalizing these quantities
using the penalty function defined as: where: Wj,ue and Wj,oe , are given by the following
expressions:

(xi − xi max ) if x > xi max ,
    kj  
 vr,j
h(xi ) = (xi min − xi )2 if x < xi max , (11)  (Pwr,j − Pwj ) exp − ckj 
    kj   i 

0 if xi min ≤ xi ≤ xi max .
 v P v 
 − exp − o,j wr,j in,j 
+ vr,j −vin,j 
 ckj
   i   kj  
 vr,j 
where: h(xi ) is the penalty function of variable xi , here 
 +Pwj exp − ckj 

the xi represents dependent variables, xi min and xi max    kj   i 
are the upper and lower limits of xi variable, respec- W j,ue =
 − exp − 1,j v P v  (16)
wr,j in,j  ,
 ckj
+ vr,j −vin,j 
tively. 

( " i
 kj kj # 

1 v1,j
The value of the penalty function grows with · Γ 1 + ki , ckj
 
 
i
a quadratic form when the constraints are violated,
 
 " #) 
 kj
 kj
and equals to zero if the constraints are not violated,
 
1 vr,j
−Γ 1 + ki , ckj
 
while the extended objective function Eq. (6) can be i

c 2018 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 473

POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 16 | NUMBER: 4 | 2018 | DECEMBER

    kj  
v
 (P wr,j ) 1 − exp − cin,j kj  1) System Equality Constraints with Wind
   kj   i  Energy
 − exp − vo,j Pwr,j vin,j 

 ckj
+ 
vr,j −vin,j 
  i
  kj   The equality constraints for the case of wind power are
 vr,j 

 +Pwj exp − ckj 
 expressed by [26]:
 i 
  kj  
Wj,oe =
 v Pwr,j vin,j  , (17) NG Nw
 − exp − ckj + vr,j −vin,j 
1,j X X
 ( " i

 PGi + PW j = Pd + P − l. (19)
  kj kj #  i=1 j=1
1 v
· Γ 1 + ki , c1,j
 
 kj 

 "
i
# )

 The active power losses are given by the formula:
  kj kj 
1 vr,j

−Γ 1 + ki , ckj
 Nl 
Gnij | Vi |2 + | Vj |2 −2 | Vi || Vj |
P
i Ploss =
n=1 (20)
cos(δi − δj )] ,
where: v1 = vin,j + (vr,j − vin,j )PW,j /PW r,j ; k > 0,
where: i and j are the sending and receiving ends of
c > 0 are the shape factor and scale factor, respectively.
particular line n. N l; is the number of lines. The equal-
PW r ; is the available active power for the j th wind
ity constraints from Eq. (8) and Eq. (9) are rewritten
turbine. PW r,j , is the rated wind power output, PW,j
for the wind node j as:
is the actual wind power output of j th wind turbine.
Vin , V0 and Vr are the cut-in, cut-off and rated wind PW j − Pdj − Pj,cal (V, δ) = 0, (21)
speed, respectively.
QW j − Qdj − Qj,cal (V, δ) = 0. (22)
Equation (15) represents the stochastic nature of
wind power output for which the following parameters The control variables vector is modified as:
are associated:
~uT = [VG1 , ..VN G , .PG2 ..PGN , Pw1 , ..PNw ,
(23)
Tp1 , ..TpN T , QC1 , ..QCN C ] ,
• Kp,j : penalty cost coefficient for not using all where: NW represents the number of wind generators
available power from j th wind turbine due to in the power system network.
under-generation estimated from j th wind turbine,

2) Wind Generators Constraints

• KR,j : reserve cost coefficient due to the reserve
capacity used to compensate the over-estimated In addition to the precedent inequality constraints, we
wind power of j th wind turbine. can write;

0 ≤ PW i ≤ PW r,i , i = 1, ..Nw , (24)

• Wj,ue and Wj,oe , are the expected value of j th wind where: PW r , is the rated active power output of the
turbine for over-estimated and under-estimated ith wind turbine unit.
energy output which was calculated using Eq. (16)
and Eq. (17) [2].
3) Spinning Reserve Constraints Model for
OPF with Wind Energy
To deal with wind speed variations of wind turbine,
the generated power from wind can be approximated The spinning reserve is the reserve capacity used for
with respect to particular wind speed V , as follows [2]: sudden load increase, unpredictable fall in wind power
output or forced outage of thermal generators units.
 The spinning reserve has two limits which are the upper

 0 V ≤ Vin , and lower limits that represent system up and system
aV 3 + bV 2 + cV + d V > V > V ,

r in down spinning reserves USR and DSR; given by the
Pw (V ) = (18)
P V > V ≥ V , following expressions: [2] and [30]:
 we off r


0 V ≥ Voff .

Nw
X
PU S ≥ RU SR + r% × PW,j , (25)
j=1
Pw (V ) is the available wind power output, a, b, c,
and d; are constants, in this study the generated wind Nw
power output is used as negative real power load con-
X
PDS ≥ RDSR × s% + r% × PW,j , (26)
nected at special bus in the test system. j=1

c 2018 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 474

POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 16 | NUMBER: 4 | 2018 | DECEMBER

where; r is the influence coefficient that gives the per-

centage of wind power contributing to USR and DSR.
The USR can be represented with respect to the total
load and total wind power by: (α)

(β)
N
X
PU Si ≥ Pd × s% + r% × PW T , (27) (δ)
i=1

(ω)
where: U Si represents the maximum up spinning re-
serve limit of ith thermal unit, and s is the percent- Kappa (κ) and lambda (λ)
age of load contributing to USR, these constraints will
be considered during the implementation of GWO al- Fig. 1: Hierarchy levels of grey wolves.
gorithm. As the rate of wind power penetration in-
creases, it becomes more difficult to predict the exact
amount of power injected by all generators into the (a)
power grid. This added more uncertainty when ac-
counting the spinning reserve requirements.

3. Used Algorithm
(b)
3.1. GWO Algorithm

Grey Wolf Optimizer (GWO) is a new algorithm pro-

posed by Mirjalili et al. in 2014 [31]. This algorithm
mimics the leadership hierarchy and hunting technique
used by grey wolves to catch their prey until stopping
its movement. GWO is similar to other population-
based meta-heuristic algorithms, by simulating the nat-
ural behavior of grey wolves in their social life when
searching for food; they follow hierarchy structure in
the group (Fig. 1). The first level representing the lead-
ers of the group is called (alpha), the second level in the
hierarchy of grey wolves is (beta) which helps alpha to
make decisions. The next levels are delta and omega;
they are the lowest ranks in the group; they have to eat
after all levels. In fact, these wolves are group-hunting Fig. 2: (a) attacking prey (b) hunting prey by wolves.
that take three main steps; chasing, encircling and at-
tacking. The algorithm starts with a given number of
wolves whose positions are randomly generated. They can be considered as the fittest solution. The
next level in the chain is called beta (β), the wolves
of this level help the alpha ones in supervising other
groups’ actions. They can replace the alpha wolves
when they die or become aged and begin to be the best
3.2. Steps of GWO Algorithm candidate solution. The lowest ranking grey wolves
are delta (δ) wolves and omega (ω) wolves [27] and
[32]. Therefore types α, β, and δ leading the opti-
Four types of wolves groups can be used to simulate
mization (hunting) process, while ω group is to track
the leadership hierarchy of grey wolves. This hierarchy
them. Kappa (κ) and lambda (λ) wolves are directed
is represented in Fig. 1, respecting the social dominant
by omega in the hierarchy.
degree, the high class is named alpha (α), mostly re-
sponsible for making decisions about hunting and order The main steps involved in the original GWO algo-
the other wolves in the pack. rithm are as follows:

c 2018 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 475

POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 16 | NUMBER: 4 | 2018 | DECEMBER

• Initialize the search agents. 3.3. Pseudo Code for GWO

Algorithm
• Assign Alpha, Beta and Gamma by fitness.

• Encircling the prey: represent the circular area Initialize the grey wolf population; Xi; i=1. . . n
around the best solution (prey). This step can be Initialize parameters; a, A, and C
represented by the following equations: Calculate the fitness of each Search_Agent;
Xa =the best search agent;
Xβ =the second best search agent;
~ p (t) − X(t) |,
D =| C · X (28) Xδ the third best search agent;
~ p (t) − A · D |,
X(t + 1) =| X (29) While Iter≤ Max_Iter
For j∈{search space}
where: X~ p is the prey’s position vector. (A)
~ and (C),
~ Sort the population of grey wolves according to their
are vectors given by the following equations: fitness
Update the Update the position of the current Search
a = 2(1 − t/Tmax ), (30) Ahent using Eq. (39);
endfor % search space
~ = 2 · ar1 − a,
A (31) Update a, A and C
Calculate the fitness of the new search agents;
~ = 2r2 ,
C (32) Update Xa , Xβ and Xδ
where: t is the current iteration and Tmax , total itera- Iter=Iter+1;
tions. End; Return, Best solution found so far Xa ;

The parameter a decreases linearly in the range

of [2, 0] for successive iterations using Eq. (30); that
model wolfs behaviour approaching the prey; r1 and r2 4. Case Study and Simulation
are random vectors in the range [0, 1]. Results
• Hunting step: the encircling process comes to the In this section, the optimal power flow problem is im-
second step involving hunting guided by the alpha plemented using GWO algorithm and two case stud-
wolf group. The following equations represent this ies are considered. For the first case study, the simu-
step: lation is carried out on IEEE30- bus system as used
in [34], by solving conventional OPF and consider-
Dα =| C1 · Xα (t) − X(t) |, (33) ing quadratic model of thermal generators cost using
Eq. (6). Then, the OPF problem is implemented con-
Dβ =| C2 · Xβ (t) − X(t) |, (34) sidering wind power for a given wind speed and cost
Dδ =| C3 · Xδ (t) − X(t) |, (35) profiles. Later, the OPF problem is implemented con-
sidering different wind speed profiles.
X 1 = X α − A1 · D α X 2 , (36)
In the second case study, the simulation is carried
X 2 = X β − A2 · D β X 3 , (37) out on IEEE57-bus system. The purpose of these
studies is to validate the results obtained using
X 3 = X δ − A3 · D δ , (38) GWO algorithm by comparing them with the results
X(t + 1) = (x1 + X2 + X3 )/3. (39) available in the literature.

• Attacking the prey: Firstly, r1 and r2 are ran-

domly selected for mutation (A and C), then the
4.1. Case Study N◦ 1: IEEE30 Bus
base vector (X) is randomly selected within the
range [r1 , r2 ], that is to drive the algorithm to Test System
global solution and avoid local optima. The fact
that “a” decreases from 2 to 0 makes the explo- 1) Case 1.1: OPF with Quadratic Fuel Cost
ration more efficient, but slows down the GWO
convergence characteristics. So, the final step of The objective function for this case study is given by
attacking the prey is done by decreasing linearly Eq. (6), for all thermal generators units, the numerical
the value of “a” from 2 to 0 [33]. data and parameters are taken from [35], the PQ bus
voltages are between 0.95 and 1.05 p.u, the shunt Var
• Steps 2 to 5 are then repeated until the maximum Compensator are not considered in this case study, ex-
number of iterations is reached. cept for the two shunt capacitors banks, at nodes 10

c 2018 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 476

POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 16 | NUMBER: 4 | 2018 | DECEMBER

and 24 of 19 and 4.3 Mvars respectively. The optimum Tab. 2: Comparison of quadratic fuel cost case 1.1.
control settings obtained by using GWO algorithm are Methods Fuel cost ($/h)
presented in Tab. 1. MDE [35] 802.376
ABC [3] 802.305
Tab. 1: Optimal power flow without considering dependent EGA [4] 802.060
variables. GAMS [5] 801.519
GWO 801.176
Control Lower/up Case Case Case
variables per limits 1.1 1.2 1.3
P1(MW) 50 200 176.1721 176.472 199.988
P2 20 80 48.0926 48.795 20.0000 that obtained by many other algorithms as depicted in
P5 15 35 21.1376 21.506 15.0152 Tab. 2. The corresponding convergence graph is shown
P8 10 30 23.3591 21.799 10.0000 in Fig. 3.
P11 10 30 11.3591 11.993 10.0000
P13 12 40 12.0000 12.000 12.0000
Tab. 3: Optimal power flow considering dependent variables.
V1 0.95 - 1.05 1.0600 1.0600 1.06000
V2 0.95 - 1.10 1.0512 1.0512 1.0512
Control Lower/up Case Case Case
V5 0.95 - 1.10 1.0224 1.0224 1.0224
variables per limits 1.1 1.2 1.3
V8 0.95 - 1.10 1.0333 1.0333 1.0333
P1 (MW) 50 200 176.9340 176.953 199.636
V11 0.95 - 1.10 1.0820 1.0820 1.0820
P2 20 80 48.7328 48.8151 20.0000
V13 0.95 - 1.10 1.0910 1.0910 1.0910
P5 15 35 21.2692 21.2488 22.2126
T11 0.90 - 1.10 1.0150 1.0150 1.0170
P8 10 30 21.0177 21.0724 25.1402
T12 0.90 - 1.10 0.9070 0.9070 0.9070
P11 10 30 11.8525 11.7632 13.2466
T13 0.90 - 1.10 0.9680 0.9680 0.9680
P13 12 40 12.0000 12.0000 12.2392
T14 0.90 - 1.10 0.9550 0.9550 0.9550
V1(p.u) 0.95 - 1.05 1.0999 1.0999 1.0999
Fuel cost 801 804 910
- V2 0.95 - 1.10 1.0885 1.0885 1.0885
$/h .1769 .4726 .6575
V5 0.95 - 1.10 1.0631 1.0631 1.0631
Power loss - 9.1528 9.202 12.709
V8 0.95 - 1.10 1.0712 1.0712 1.0712
Voltage
- 0.10 0.1082 - V11 0.95 - 1.10 1.0998 1.0998 1.0998
deviations
V13 0.95 - 1.10 1.0733 1.0733 1.0733
C10
0.00 - 5.00 4.1669 4.1669 4.1669
In order to assess the potential of the proposed ap- (Mvars)
proach, a comparison between the obtained results of C15 0.00 - 5.00 0.2398 0.2398 0.2398
C17 0.00 - 5.00 4.2017 4.2017 4.2017
fuel cost and those reported in the literature has been C20 0.00 - 5.00 0.1489 0.1489 0.1489
carried out. The results of this comparison are given C21 0.00 - 5.00 0.6478 0.6478 0.6478
in Tab. 2. It is worth mentioning that the comparison C22 0.00 - 5.00 4.2499 4.2499 4.2499
has been carried out with the same test system data. C23 0.00 - 5.00 1.3886 1.3886 1.3886
C24 0.00 - 5.00 2.1815 2.1815 2.1815
Different OPF results of active generation powers C29 0.00 - 5.00 2.0780 2.0780 2.0780
and losses for different case studies are given in Tab. 1. T11 0.90 - 1.10 1.0461 1.0150 1.0170
T12 0.90 - 1.10 0.9000 0.9070 0.9070
T13 0.90 - 1.10 0.9997 0.9680 0.9680
T14 0.90 - 1.10 0.9642 0.9550 0.9550
801.5 Fuel cost
GWO - 798.3107 806.1530 916.6968
($/h)
Power loss
801 - 8.4061 8.4526 9.0762
(MW)
Best fuel cost ($/h)

Voltage
800.5 - 0.422 0.077 0.078
deviations

800
For the methods EADDE in [6], GABC in [7], EEA
799.5 in [5], CSA in [8], KHA in [9], SA in [10], and ISA in
[11], the PQ bus voltages are between 0.95 and 1.1 p.u,
799 the transformers tap setting and shunt Var compen-
sators are considered in the same case study, and the
798.5
generator voltages are taken close to their high per-
50 100 150 200 250 300 350 400 missible limit. Table 3 shows the corresponding op-
Iteration timal power flow results when using the optimal set-
Fig. 3: Convergence characteristic of IEEE30 bus system case
tings of dependent variables. It can be observed from
1.1. Tab. 4 that GWO algorithm gives better results. The
system reactive generation powers for this case study
are within their specified limits as in Tab. 5. Table 6
The best fuel cost calculated by the proposed algo- presents a comparison of optimal power flow results of
rithm for this case is 801.1769 $/h, which is better than the proposed algorithm with other methods found in

c 2018 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 477

POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 16 | NUMBER: 4 | 2018 | DECEMBER

Tab. 4: Comparison when optimizing dependent variables. Tab. 7: Comparison when optimizing dependent variables.

Methods Fuel cost ($/h) Methods Fuel cost ($/h)

EADDE [6] 800.204 BBO [13] 804.998
DSA [7] 800.388 PSO [14] 806.380
EEA [5] 800.083 DE [1] 805.262
CSA [8] 799.707 GWO 806.1530
EGA [12] 799.5600
BBO [13] 799.1116
KHA [9] 799.0310
loading amounts are within acceptable limits. As we
MFPA [40] 799.1592
GSA [15] 798.675 can see from Fig. 6(a), the voltage magnitude is en-
GWO 798.3107 hanced after the improvement by GWO, and all the
load bus voltages are within the permissible range.
Tab. 5: Comparison when optimizing dependent variables.
806
React. Power Gen. Limits Qg GWO
Q1 -20 200 -18.7646
G2 -40 50 23.1157 805.5
Q5 -40 40 27.3300
Q8 -15 40 33.7790

Best fuel cost ($/h)

805
Q11 -6 24 17.9905
Q13 -6 24 2.55070
804.5
Tab. 6: GWO-OPF results comparison for case 1.1.
804
Pgi
SA ISA KHA GSO GWO
(MW)
P1 173.15 177.124 177.04 174.920 176.9046 803.5
P2 48.54 48.933 48.690 44.150 48.7226
P5 19.23 21.3175 21.300 21.760 21.2697 803
P8 12.81 21.0006 21.080 25.730 21.0509
100 200 300 400 500 600 700
P11 11.64 11.8605 11.880 11.120 11.8556
Iteration
P13 12.00 11.860 12.020 13.810 12.0000
Tot. Gn
277.37 292.095 292.01 291.49 291.8034 Fig. 4: Convergence characteristic of IEEE30 bus system case
(MW)
1.2.
Cost
799.45 799.277 799.03 799.06 798.3106
($/h)
Losses
9.200 8.695 8.610 8.48 8.4034 140
(MW)
Normal case
120 Load flow limits

the literature as in [3] and [19]. Figure 6(a) shows the 100
voltage profile of case 1.1, without improvement.
load flow (MW)

80

60
2) Case 1.2: OPF with Voltage Profile
Improvement 40

20
Minimizing only the total fuel cost using OPF problem
as in case 1.1; can result in a feasible solution, but
0
voltage profile may not be acceptable. Thus, in this
second case, the objective here is to minimize the fuel 0 10 20 30 40 50
cost and improving the voltage profile at the same time line number
by minimizing the voltage deviation of PQ buses from Fig. 5: Transmission Load flows obtained by GWO.
the unity 1.0. [36].
The results obtained using the proposed approach
are compared with other methods in the literature as
3) Case 1.3: OPF for Fuel Cost Including
shown in Tab. 7 where the total cost found by GWO,
Valve Point Effect
in this case, is better than that obtained before.
Figure 4 shows the convergence graph. Figure 5 Considering the same system data as in [23], the valve
presents the transmission load flow of the system, from point effect is incorporated and the fuel cost is evalu-
this figure, we can see that the obtained transmission ated using the Eq. (7). Simulation of power flow results

c 2018 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 478

POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 16 | NUMBER: 4 | 2018 | DECEMBER

1 1
30
29 1,1 2 3 29 2 normal_case
Voltage 28 1,10 3
28 4 27 4
27 1,05 5 improvement
26 6 26 1,00 5
25 1 7 25 6
24 8 24 0,90 7
0,95
23 9 23 8
22 10 0,80
21 11 22 9
20 12 21 10
19 13
18 17 15 14 20 11
16 19 12
18 13
(a) 17 16 15 14

1 Fig. 7: Voltage profile for normal and contingency conditions.

Vbefore
29 30
1,100 2 3
28 4 Vafter 120
27 1,050 5
26 6 load flow for normal conditions
100
25 1,000 7 LF for contengency conditions
24 8
active load flow (MW)

0,950 80
23 9
22 10 60
21 11
20 12 40
19 13
18 17 15 14
16 20

(b)
0

Fig. 6: a) Bus voltage magnitude case 1.2, b) Comparison of 0 10 20 30 40 50

voltage profile of IEEE30 bus case 1.1 & case 1.2.
line number

Fig. 8: Load flow profile for normal and contingency conditions.

of this case study is compared with other available re-
sults as in Tab. 8.
The load-bus voltages of contingency case are below
Tab. 8: Obtained results comparison case 1.3. their normal limit (deviated from their normal lim-
its).To alleviate this problem we apply Eq. (11) and
Methods Fuel cost ($/h) Eq. (12), to bring the voltage at these load buses within
PSO [14] 932.7642 0.95 and 1.05 p.u. Figure 9 shows the corrected voltage
ABC [3] 945.4495
GSA [15] 929.7240
profile.
GABC [19] 931.7450
BBO [13] 919.7647
MFPA [40] 917.8298 4.2. Case 2: OPF with Wind Energy
GWO 916.6968
Case Study

1) Case 2.1: OPF with Stochastic Wind

4) Case 1.4: System Analysis Under (N-1) Power Modelling
Contingency
In this section, GWO algorithm is used to solve OPF
To investigate the efficiency of GWO under contin- problem for system including stochastic wind power in
gency, a line outage conditions are created on the test addition to conventional thermal generators. In this
system as in [23], in which four contingency conditions case, the system has been modified by replacing con-
are considered (lines: 12–15, 10–20, 15–23 and 6–28). ventional generators by wind farms located at buses
For these four conditions, the voltage profile for nor- 5, 11 and 13; each with a total capacity of 60 MW.
mal and contingency conditions is shown in Fig. 7, and Two case studies are considered here: in the first case,
corresponding load flow profile is in Fig. 8. the wind power is modelled using Weibull distribution

c 2018 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 479

POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 16 | NUMBER: 4 | 2018 | DECEMBER

1.2 1
voltage profile during contengency 30
1,10
29 2 3
1.15 Voltage profile after correction 28 4 Vwind
27 1,05 5
26 6
Voltage magnitude (p.u)

1.1
25 1,00 7
1.05 24 8
23 0,95 9
1
22 10
0.95
21 11
20 12
19 13
0.9 1817 1514
16
0.85
Fig. 11: Bus voltage magnitude for wind case.
0.8
0 5 10 15 20 25 30
Bus number Tab. 9: Simulation results for wind case study.

Reserved
Fig. 9: Voltage profile for normal and contingency conditions. Pgi GABC BFA Excess
GWO real
(MW) [19] [23] power
power
P1 50.219 56.530 50.524
function in form of imbalance costs of wind power in P2 20.581 34.285 20.461
the main cost objective Eq. (15), which is minimized PWIND1 60.000 50.729 59.995 40.411 0.001
P8 35.000 65.956 34.976
subject to all given constraints. While, in the second
PWIND2 60.000 40.405 60.000 26.783 0
case study, the OPF problem is solved considering dif- PWIND3 59.999 39.162 59.904 25.550 0.029
ferent wind speeds. Total.
Gen. 285.80 287.06 285.86
The test system data given in [23] are taken for this (MW)
study. The simulation convergence curve and voltage Cost
819.293 947.50 826.82
at different buses of the system are given in Fig. 10 and ($/h)
Losses
Fig. 11, OPF schedule is given in Tab. 9; the optimal - - 2.4144
(MW)
results are then compared with GABC [19] and BFA
[23].
As seen from Fig. 10, the total fuel cost is decreased
960
GWO
by the integration of wind power source in the system.

940
2) Case 2.2: OPF Study with Wind Energy
Considering Reserve Constraints
Best fuel cost ($/h)

920

900 Case 2.2.1: OPF without Wind Power

880
In this case study, we used the same configuration as
in [2], by considering the nodes 1, 2, 13, 22, 23 and 27
860
as generator buses and total system load of 189.2 MW.
First, we proceeded for optimal power flow without
840
wind energy; the simulation results of this case are
compared to those reported in [2], as shown in Tab. 10.
20 40 60 80 100 120 140 160 180 200 It can be noticed that the obtained GWO cost is
Iteration better comparing with the case without wind power.
Fig. 10: Convergence graph of fuel cost for wind case. Case 2.2.2: Wind Energy with Zero Cost

The obtained results show that the GWO method Two scenarios of wind power integration levels are con-
performs better when compared with other methods sidered in this study; 10 %, and 20 % of the system
for the same case study. The reserved power is higher load. These levels are connected to bus 8. Using
than the surplus power in Tab. 9, which justifies the Eq. (25), Eq. (26) and Eq. (27), we calculated the spin-
fact that the utility service is to purchase an important ning reserve under different wind speeds at the second
amount of reserve for covering any unavailable wind hour, assuming the wind speed at the first hour was
energy. 3 m·s−1 .

c 2018 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 480

POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 16 | NUMBER: 4 | 2018 | DECEMBER

Tab. 10: Simulation results for wind case with spinning reserve.

Wind UP/Down Spinning reserve requirements UP/Down Spinning reserve capacity Total cost
speed for wind power Conditions (MW) supplied by thermal units (MW) ($/h)
Scenario 1 USR DSR Pw (MW) USR DSR
4 28.964 0.584 1.169 55.000 47.008 569.8760
5 29.829 1.449 2.899 55.000 46.575 563.1616
6 30.993 2.613 5.227 55.000 45.993 554.1656
7 32.292 3.912 7.824 55.000 45.344 544.1806
8 33.635 5.255 10.511 55.000 44.672 533.9034
Scenario 2 USR DSR Pw (MW) USR DSR
4 29.614 1.234 2.469 55.000 46.683 564.8280
5 31.325 2.945 5.897 55.000 45.827 551.5847
6 33.678 5.225 10.45 55.000 44.687 534.1380
7 36.265 7.885 15.77 55.000 43.357 513.9586
8 38.899 10.51 21.02 55.000 42.045 494.2616

1 The wind speeds values were respectively 4, 5, 6, 7,

30
1,1 2 3 Bus voltage and 8 m·s−1 ; different computation results of scenarios
2829 4 1 and 2 are presented in Tab. 12 and the system
27 1,05 5
26 1 6 voltage profile is shown in Fig. 13.
25 0,95 7
24 8 Tab. 12: Simulation results for wind case study.
23 0,9 9 Wind
22 10 speed 4 5 6 7 8
21 11 (m·s−1 )
20 12 Scenario 1(10 % of wind penetration)
1918 13 Cost
17 1514 ($/h)
571.24 571.565 581.487 605.395 644.38
16 Scenario 2 (20 % of wind penetration)
Cost
Fig. 12: Voltage profile at different nodes for modified system. 570.92 586.359 643.340 762.651 936.10
($/h)

Tab. 11: OPF results of modified IEEE30 bus system. Case 2.2.3: OPF Considering Wind Power Cost
Pgi (MW) GWO Without wind EPSO [2]
P1 43.4397 43.425 In this case, we assume that the wind power has the
P2 57.7903 55.785 same direct cost of [19] d1 = 1 $/h, without considering
P13 17.4824 17.716
P22 23.0944 23.131
the imbalance cost. Simulation results for wind case
P22 17.2086 18.241 study.
P27 32.6450 33.307
W1 - -
The simulation result is shown in Tab. 12, we can
Total gen. (MW) 191.6604 191.605 see that when wind speed increases, the total operation
Cost ($/h) 574.7271 574.766 cost increases too, due to the wind direct cost impact
Losses (MW) 2.4604 2.408 on the total operating cost.
Voltage div 1.0572

4.3. Case 3: OPF with Stochastic

The spinning reserve of the system was s = 15 % of Wind Speed
the total demand, and the up-spinning reserve was set
to improve the safety of the power system operation In this case study and in order to check the effect of
under wind intermittent conditions or uncertain wind uncertain wind power on the test power system, two
power. Simulation results for wind case study. wind farms each with capacity of 30 MW have been
connected at two separate locations; at nodes 26 and
This was achieved by using Eq. (27), in which this
node 30 as in [19]. The results obtained are then com-
reserve constraint of wind generation was r = 50 % of
pared with the case without wind energy.
the system load. After computing the wind power us-
ing Eq. (25), we run the OPF program to calculate the Two cases are considered here; the first one where
cost associated with this wind injection then we pro- the scale factor “c” takes the values of 3 to 30 while
ceeded to the calculation of different spinning reserve keeping the shape factor at k = 2, then by keeping
constraint limits, the obtained results are depicted in the scale factor constant at the value 10 and varying
Tab. 10. the reserve coefficient (Krw) from its base value of 4,

c 2018 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 481

POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 16 | NUMBER: 4 | 2018 | DECEMBER

and with the installed wind power capacity for each Tab. 13: Simulation results for wind case study.
wind farm of 20 MW instead of 30 MW by applying With With
the proposed approach taking into consideration these With
With wind wind
Pgi wind
conditions, we find the results as shown in Fig. 13(a). (MW)
-out
(c=2,
(c=10, (c=10,
wind k=10, k=2,
For the second case study, we maintained the val- k=2)
Krw = 4) Krw = 30)
ues of wind turbines Weibull model factors constant, P1 176.1721 143.002 156.945 156.87
vi n = 4 m·s−1 , vr = 12 m·s−1 , vout = 25 m·s−1 , c = 3, P2 48.0926 40.799 44.029 44.053
P5 21.1376 18.943 19.957 19.984
k = 2, Kpw = 1, Krw = 4, but considering the direct P8 23.3591 10.000 10.000 10.018
costs of the two wind farms d1 = d2 = 1.3 $/h. Simu- P11 11.3591 10.019 10.000 10.013
lation results are presented in Tab. 13, the convergence P13 12.0000 12.026 12.000 12.014
characteristics for different values of reserve coefficient W1 - 29.942 20.000 19.958
“Krw ” is given in Fig. 13(b). W2 - 30.000 20.989 20.000
Total.
Generally, the direct cost of wind power is less than Gen. 292.1205 294.731 292.931 292.61
(MW)
the average cost of thermal power, and the penalty
Cost
cost of not using all the available wind power is consid- 801.176 741.514 744.821 744.82
($/h)
ered less than the direct cost. From Fig. 13(b), it can Losses
9.180 11.327 9.5230 9.5123
(MW)
Voltage
0.108 0.108 0.1084
div.
Wind
Krw=4 Over_E 26.69 25.321 25.302
747.5 Krw=10 MW
Krw=20
Krw=30
operating fuel cost ($/h)

747 Krw=40
be seen that, the larger the value of c the higher the
746.5 value of wind speed and hence wind power penetration
amount. However, the amount of wind power injected
746 at bus 26 remains, less than that injected at bus 30,
due to the thermal loading limit of the transmission
745.5
line at this section.

745

10 20 30 40 50 4.4. Case Study N◦ 2: IEEE57 Bus

Iteration Test System
(a)
This system consists of 7 thermal generators, with bus
785 1 is considered as slack bus; 2, 3, 6, 8, 9 and 12 as
GWO
c=4 PV buses, 50 load buses and 80 lines, among which 17
780 c=10 lines are equipped with tap changing transformers. In
c=20 addition, three shunt Var compensators are installed at
775 c=30
buses 18, 25 and 53. The system data are taken from
total cost ($/h)

770
[37]. Two cases are investigated in this case study:

765

760 1) Case 1: OPF for Quadratic Fuel Cost

755 In this case study, the objective function to be opti-

mized is represented by the quadratic fuel cost, related
750
to thermal generators unit described by the Eq. (6).
50 100 150 200 250 300 350 400
Iterations The optimal power flow for the first case study us-
(b) ing GWO takes the settings of the algorithm as the
followings: search agent number equals to 30, and the
Fig. 13: Convergence characteristic for a) different values of re- number of runs equals to 300. These are the same
serve coefficient (Krw ), b) different values of scale fac- system settings used for the other methods, and the
tor “c”. obtained simulation results are shown in Tab. 14.

c 2018 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 482

POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 16 | NUMBER: 4 | 2018 | DECEMBER

Tab. 14: Optimal control variables settings for case 1. 10 4

Lower/ GWO
Control Control 4.25
upper Case 1 Case 1
variables variables
limits
P1(MW) 0 576 143.7886 T24-25 1.0125 4.24

Best fuel cost ($/h)

P2 0 150 89.7403 T25-26 1.0000
P3 0 120 45.1711 T7-29 1.0125 4.23
P6 0 100 72.1034 T34-32 0.9125
P8 0 300 459.8802 T11-41 0.9000 4.22
P9 0 120 94.9161 T15-45 1.0125
P12 0 300 360.4463 T14-46 0.9875
4.21
V1 0.95–1.05 1.0499 T10-51 1.0000
V2 0.95–1.10 1.0479 T13-49 0.9625
V3 0.95–1.10 1.0408 T11-43 0.9625 4.2
V6 0.95–1.10 1.0493 T40-56 0.9625
V8 0.95–1.10 1.0342 T39-57 0.9625 4.19
V9 0.95–1.10 1.0332 T9-55 0.9875
50 100 150 200
V12 0.95–1.10 1.0406 Qsc1 1.0170
Iteration
T4-18 0.90–1.10 0.9375 Qsc1 0.9070
T4-18 0.90–1.10 1.0500 Qsc2 0.9680
Fig. 15: Convergence curve for IEEE57 case 1.
T21-20 0.90–1.10 0.9750 - -
Fuel cost
- 41683.5076 -
($/h)
Power loss 2) Case 2: OPF with Voltage Profile
- 15.2460 -
(MW) Improvement of IEEE57 Test System

Bus voltage enhancement is one of the most significant

The results obtained by the proposed method were safety and service qualification indices. In order to as-
compared with others available methods, this compari- sess this case, a two-fold objective function is consid-
son shows that the GWO algorithm gives better results ered to minimize the operating fuel cost and enhancing
when compared to many algorithms found in the liter- the voltage profile at the same time by minimizing all
ature as shown in Tab. 15. the load bus deviations from the reference value. Volt-
age profile, in this case, is compared to that of the
Tab. 15: Comparison of fuel costs case 1. precedent one as shown in Fig. 16 and the operating
cost curve is shown in Fig. 17.
Methods Fuel cost ($/h)
TSA [41] 41685.07
HS [38] 41693.358 1 bus voltage
ABC [3] 41693.958 555657
1,15 234
case1
54 5
BBO [13] 41721.246 5253 1,10 67
MATPOWER [37] 41737.790 51 8 bus voltage
50 1,05 9 case2
EADDE [6] 41713.620 49 10
48 1,00 11
GSA [15] 41695.8717 47 12
KHA [9] 41709.2647 46 0,95 13
GWO 41683.5076 45 0,90 14
44 0,85 15
43 16
42 17
41 18
40 19
39 20
1,1 38 21
37 22
36 23
1,05 3534 2524
333231 2726
30 2928
1
Fig. 16: Voltage improvement profile comparison.
0,95
0,9 It is clear that the voltage profile is enhanced effi-
ciently compared with that in case 1. This could be
achieved by the optimal tuning of the control param-
eters within the constraints range as given in Tab. 16
using the proposed GWO technique.
It can be seen that the proposed GWO method con-
verges to a better result than EADDE [7] method
Fig. 14: Bus voltage profile for IEEE57 case 1. by decreasing the fuel cost from 42051.44 $/h to

c 2018 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 483

POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 16 | NUMBER: 4 | 2018 | DECEMBER

Tab. 16: Performances measures for the TFC ($/h) in both cases.

System IEEE 30-bus system IEEE 57-bus system

Method GWO EADDE [6] MDE [35] PSO [14] GWO TSA [41] ABC [3] PSO [14]
Min 798.2934 800.204 802.376 800.409 41,684.00 41,685.07 41,781.00 41,688.68
Mean 798.6380 800.241 802.382 800.450 41,686.00 41,687.78 41,840.00 41,697.58
Max 800.1367 800.278 802.404 801.231 41,688.29 41,689.05 41,927.00 41,727.86
runs 40 30 40 20 50 50 20 20

10 4 putational time. Figure 18 presents the convergence

GWO curve for IEEE30 bus system after 40 runs and Fig. 19
4.25
the convergence curve for IEEE57-bus system after 50
runs.
4.24
Best fuel cost ($/h)

801
4.23 GWO

4.22 800.5

Best fuel cost ($/h)

4.21
800

4.2
799.5
4.19
50 100 150 200 799
Iteration

Fig. 17: Convergence curve for IEEE57 in case 2. 798.5

5 10 15 20 25 30 35 40
41817.3826 $/h and voltage deviation from 0.7882 to Iteration
0.74. Fig. 18: Convergence curves for IEEE57 with 50 runs.
Tab. 17: Optimal control variables settings for case 2.

Control Control 10 4
Case 2 Case 2
variables variables GWO
P1 142.189 T24-25 0.9033 4.28
P2 89.894 T24-25 0.9767
P3 45.148 T24-26 1.0279
4.26
P6 72.928 T7-29 0.9861
Best fuel cost ($/h)

P8 459.393 T34-32 0.9210

P9 84.089 T11-41 0.9368 4.24
P12 363.088 T15-45 0.9713
V1 1.0212 T14-46 0.9720
4.22
V2 1.0740 T10-51 0.9933
V3 1.0646 T13-49 0.9327
V6 0.9913 T11-43 0.9397 4.2
V8 1.0519 T40-56 1.0269
V9 1.0808 T39-57 0.9504
V12 1.0103 T9-55 0.9976 4.18
T4-18 1.0760 Qsc1 1.1419
T4-18 0.9313 Qsc1 0.2719 5 10 15 20 25 30 35 40 45 50
T21-20 1.0032 Qsc2 0.4971 Iteration
Fuel cost
41817.382
($/h) Fig. 19: Convergence curves for IEEE57 with 50 runs.
Power loss
16.1146 -
(MW)
Voltage div.
0.74
(p.u)
5. Conclusion
From the comparison of the results shown in Tab. 16,
it can be concluded that the solution quality of the This paper presents an optimal power flow study us-
GWO algorithm is very competitive and challenging ing a new meta-heuristic population-based search algo-
because it converges to the best solution with less com- rithm called Grey Wolf Optimizer (GWO). Considering

c 2018 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 484

POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 16 | NUMBER: 4 | 2018 | DECEMBER

both wind and thermal power generators, in order to pp. 229–236. ISSN 0885-8950. DOI: 10.1109/tp-
evaluate the effectiveness of the proposed technique; wrs.2002.1007886.
three case studies are considered in this work.
[5] SURENDER, S. R., P. R. BIJWE and A. R.
By adding to the normal operation condition, the ABHYANKAR. Faster evolutionary algorithm
N-1 contingency condition represented by lines outage base optimal power flow using incremental vari-
and the uncertainty of wind power, which is modelled ables. Electrical Power and Energy System. 2014,
using Weibull distribution function is investigated. vol. 54, iss. 1, pp. 198–210. ISSN 0142-0615.
Simulations results obtained by OPF analysis for two DOI: 10.1016/j.ijepes.2013.07.019.
standard test systems IEEE-30, and IEEE-57 bus sys-
tems without considering wind power are compared [6] VAISAKH, K. and L. R. SRINIVAS. Evolv-
with results of other methods available in the litera- ing ant direction differential evolution for
ture. The outcome of the comparison confirms the ef- OPF with non-smooth cost functions. Engineer-
fectiveness and robustness of the proposed algorithm. ing Applications of Artificial Intelligence. 2011,
vol. 24, iss. 3, pp. 426–436. ISSN 0952-1976.
Similarly, the results obtained in presence of wind en- DOI: 10.1016/j.engappai.2010.10.019.
ergy system were compared with those of other meth-
ods reported in the literature using the IEEE 30 bus [7] ABACI, K. and V. YAMACLI. Differential
system. By increasing the value of reserve coefficient, search algorithm for solving multi-objective op-
the value of the injected amount in the system can timal power flow problem. International Jour-
be limited by the transmission system permissible ca- nal of Electrical Power & Energy Systems. 2016,
pacity of the existing network. On the other hand; vol. 79, iss. 1, pp. 1–10. ISSN 0142-0615.
when increasing the wind penetration level by increas- DOI: 10.1016/j.ijepes.2015.12.021.
ing wind speed, the total operating cost decreases.
[8] GHASEMI, M., S. GHAVIDEL, M. GITI-
The method presents compromising performances ZADEH and E. AKBARI. An improved
measures compared to other methods found in the lit- teaching–learning-based optimization algorithm
erature. This analysis will be extended in the future using Levy mutation strategy for non-smooth
to include spinning reserve in the main optimal power optimal power flow. International Journal of
flow problem. Electrical Power & Energy Systems. 2015,
vol. 65, iss. 1, pp. 375–384. ISSN 0142-0615.
DOI: 10.1016/j.ijepes.2014.10.027.
References
[9] ROY, P. K. and C. PAUL. Optimal power
flow using krill herd algorithm. International
[1] ABOU, A. A., M. A. ABIDO and S. R. SPEA. Op- Transactions on Electrical Energy Systems. 2015,
timal power flow using differential evolution algo- vol. 25, iss. 8, pp. 1397–1419. ISSN 2050-7038.
rithm. Electrical Engineering. 2009, vol. 91, iss. 2, DOI: 10.1002/etep.1888.
pp. 69–78. ISSN 0948-7921. DOI: 10.1007/s00202-
009-0116-z. [10] ROA-SEPULVEDA, C. A. and B. J. PAVEZ-
LAZO. A solution to the optimal power flow using
[2] CHANG, Y. C., T. Y. LEE, C. L. CHEN and simulated annealing. In: IEEE Porto Power Tech
R. M. JAN. Optimal power flow of a wind- Proceedings. Porto: IEEE, 2001, pp. 5–9. ISBN 0-
thermal generation system. International Jour- 7803-7139-9. DOI: 10.1109/PTC.2001.964733.
nal of Electrical Power & Energy Systems. 2014,
vol. 55, iss. 1, pp. 312–320. ISSN 0142-0615. [11] BENTOUATI, B., S. CHETTIH, L. CHAIB
DOI: 10.1016/j.ijepes.2013.09.028. and V. SREERAM. Interior search algo-
rithm for optimal power flow with non-
[3] REZAEI, M. and A. KARAMI. Artificial bee smooth cost functions. Cogent Engineering.
colony algorithm for solving multi-objective op- 2017, vol. 4, iss. 1, pp. 1–17. ISSN 2331-1916.
timal power flow problem. International Jour- DOI: 10.1080/23311916.2017.1292598.
nal of Electrical Power & Energy Systems. 2013,
vol. 53, iss. 1, pp. 219–230. ISSN 0142-0615. [12] KUMARI, M. S. and S. MAHESWARAPU.
DOI: 10.1016/j.ijepes.2013.04.021. Enhanced Genetic Algorithm based compu-
tation technique for multi-objective Optimal
[4] BAKIRTZIS, A. G., P. N. BISKAS, C. E. Power Flow solution. International Journal of
ZOUMAS and V. PETRIDIS. Optimal power flow Electrical Power & Energy Systems. 2010,
by enhanced genetic algorithm. IEEE Transac- vol. 32, iss. 6, pp. 736–742. ISSN 0142-0615.
tions on Power Systems. 2002, vol. 17, iss. 2, DOI: 10.1016/j.ijepes.2010.01.010.

c 2018 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 485

POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 16 | NUMBER: 4 | 2018 | DECEMBER

[13] BHATTACHARYA, A. and P. K. CHATTOPAD- optimization algorithm for wind–thermal eco-

HYAY. Application of biogeography-based optimi- nomic emission dispatch problem considering
sation to solve different optimal power flow prob- wind power availability. International Journal
lems. IET Generation, Transmission & Distribu- of Electrical Power & Energy Systems. 2015,
tion. 2011, vol. 5, iss. 1, pp. 70–80. ISSN 1751- vol. 73, iss. 1, pp. 1035–1050. ISSN 0142-0615.
8687. DOI: 10.1049/iet-gtd.2010.0237. DOI: 10.1016/j.ijepes.2015.06.014.

[14] ABIDO, M. A. Optimal power flow using par- [22] PANDA, A. and M. TRIPATHY. Security
ticle swarm optimization. International Journal constrained optimal power flow solution of
of Electrical Power & Energy Systems. 2002, wind-thermal generation system using modi-
vol. 24, iss. 7, pp. 563–571. ISSN 0142-0615. fied bacteria foraging algorithm. Energy. 2015,
DOI: 10.1016/S0142-0615(01)00067-9. vol. 93, iss. 1, pp. 816–827. ISSN 0360-5442.
DOI: 10.1016/j.energy.2015.09.083.
[15] DUMAN, S., U. GUVENC, Y. SONMEZ
and N. YORUKEREN. Optimal power flow [23] MISHRA, S., Y. MISHRA and S. VIGNESH.
using gravitational search algorithm. En- Security constrained economic dispatch consider-
ergy Conversion and Management. 2012, ing wind energy conversion systems. In: IEEE
vol. 59, iss. 1, pp. 86–95. ISSN 0196-8904. Power and Energy Society General Meeting. De-
DOI: 10.1016/j.enconman.2012.02.024. troit: IEEE, 2011, pp. 1–8. ISBN 978-1-4577-1000-
1. DOI: 10.1109/PES.2011.6039544.
[16] VAISAKH, K., L. R. SRINIVAS and K. MEAH.
Genetic evolving ant direction PSODV hy- [24] SULAIMAN, M. H., Z. MUSTAFFA, M. R. MO-
brid algorithm for OPF with non-smooth HAMED and O. ALIMAN. Using the gray wolf
cost functions. Electrical Engineering. 2013, optimizer for solving optimal reactive power dis-
vol. 95, iss. 3, pp. 185–199. ISSN 0948-7921. patch problem. Applied Soft Computing. 2015,
DOI: 10.1007/s00202-012-0251-9. vol. 32, iss. 1, pp. 286–292. ISSN 1568-4946.
DOI: 10.1016/j.asoc.2015.03.041.
[17] KUMAR, R. A. and L. PREMALATHA. Op-
[25] EL-FERGANY, A. A. and H. M. HASANIEN.
timal power flow for a deregulated power sys-
Single and Multi-objective Optimal Power
tem using adaptive real coded biogeography-
Flow Using Grey Wolf Optimizer and Differ-
based optimization. International Journal of
ential Evolution Algorithms. Electric Power
Electrical Power & Energy Systems. 2015,
Components and Systems. 2015, vol. 43,
vol. 73, iss. 1, pp. 393–399. ISSN 0142-0615.
iss. 13, pp. 1548–1559. ISSN 1532-5008.
DOI: 10.1016/j.ijepes.2015.05.011.
DOI: 10.1080/15325008.2015.1041625.
[18] KHAMEES, K. A., A. E. RAFEI, N. M. BADRA [26] IAHKALI, H. and M. VAKILIAN. Stochastic unit
and A. Y. ABDELAZIZ. Solution of optimal commitment of wind farms integrated in power
power flow using evolutionary-based algorithms. system. Electric Power Systems Research. 2010,
International Journal of Engineering, Science vol. 80, iss. 9, pp. 1006–1017. ISSN 0378-7796.
and Technology. 2017, vol. 9, no. 1, pp. 55–68. DOI: 10.1016/j.epsr.2010.01.003.
ISSN 2141-2839.
[27] MIRJALILI, S., S. M. MIRJALILI and
[19] ROY, R. and H. T. JADHAV. Optimal power A. LEWIS. Grey Wolf Optimizer. Ad-
flow solution of power system incorporating vances in Engineering Software. 2014,
stochastic wind power using Gbest guided arti- vol. 69, iss. 9, pp. 46–61. ISSN 0965-9978.
ficial bee colony algorithm. International Jour- DOI: 10.1016/j.advengsoft.2013.12.007.
nal of Electrical Power & Energy Systems. 2015,
vol. 64, no. 1, pp. 562–578. ISSN 0142-0615. [28] MIRJALILI, S. and S. Z. M. HASHIM. A new
DOI: 10.1016/j.ijepes.2014.07.010. hybrid PSOGSA algorithm for function optimiza-
tion. In: International Conference on Computer
[20] MASKAR, M. B., A. R. THORAT, P. D. BA- and Information Application. Tianjin: IEEE,
MANE and I. KORACHGAON. Optimal power 2010, pp. 374–377. ISBN 978-1-4244-8598-7.
flow incorporating thermal and wind power DOI: 10.1109/ICCIA.2010.6141614.
plant. In: International Conference on Cir-
cuit, Power and Computing Technologies. Kollam: [29] BAI, W., D. LEE and K. LEE. Stochas-
IEEE, 2017, pp. 1–6. ISBN 978-1-5090-4967-7. tic Dynamic Optimal Power Flow Integrated
DOI: 10.1109/ICCPCT.2017.8074265. with Wind Energy Using Generalized Dy-
namic Factor Model. IFAC-Papers OnLine. 2016,
[21] JIANG, S., Z. JI and Y. WANG. A novel grav- vol. 49, iss. 27, pp. 129–134. ISSN 2405-8963.
itational acceleration enhanced particle swarm DOI: 10.1016/j.ifacol.2016.10.731.

c 2018 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 486

POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 16 | NUMBER: 4 | 2018 | DECEMBER

[30] XIE, L., H. D. CHIANG and S. H. LI. Op- [38] SINSUPAN, N., U. LEETON and T. KUL-
timal power flow calculation of power sys- WORAWANICHPONG. Application of har-
tem with wind farms. In: IEEE Power mony search to optimal power flow prob-
and Energy Society General Meeting. Detroit: lems. In: International Conference on Ad-
IEEE, 2011, pp. 1–6. ISBN 978-1-4577-1000-1. vances in Energy Engineering. Beijing: IEEE,
DOI: 10.1109/PES.2011.6039105. 2010, pp. 219–222. ISBN 978-1-4244-7831-6.
DOI: 10.1109/ICAEE.2010.5557575.
[31] MONDAL, S., A. BHATTACHARYA and
S. H. NEE-DEY. Multi-objective economic [39] REDDY, S. S., and C. SRINIVASA RATH-
emission load dispatch solution using gravita- NAM. Optimal Power Flow using Glowworm
tional search algorithm and considering wind Swarm Optimization. International Journal of
power penetration. International Journal of Electrical Power & Energy Systems. 2016,
Electrical Power & Energy Systems. 2013, vol. 80, iss. 1, pp. 128–139. ISSN 0142-0615.
vol. 44, iss. 1, pp. 282–292. ISSN 0142-0615. DOI: 10.1016/j.ijepes.2016.01.036.
DOI: 10.1016/j.ijepes.2012.06.049.
[40] REGALADO, J. A., B. E. EMILIO and
[32] MOHAMED, A. A., A. M. EL-GAAFARY, E. CUEVAS. Optimal power flow solution
Y. S. MOHAMED and A. M. HEMEIDA. Multi- using Modified Flower Pollination Algorithm.
objective Modified Grey Wolf Optimizer for In: IEEE International Autumn Meeting on
Optimal Power Flow. In: Eighteenth Interna- Power, Electronics and Computing. Ixtapa:
tional Middle East Power Systems Conference IEEE, 2015, pp. 1–6. ISBN 978-1-4673-7121-6.
(MEPCON). Cairo: IEEE, 2016, pp. 982–990. DOI: 10.1109/ROPEC.2015.7395073.
ISBN 978-1-4673-9063-7. DOI: 10.1109/MEP- [41] EL-FERGANY, A. A. and H. M. HASANIEN.
CON.2016.7837016. Tree-seed algorithm for solving optimal power
flow problem in large-scale power systems incorpo-
[33] KAPOOR, S., I. ZEYA, C. SINGHAL and S. J.
rating validations and comparisons. Applied Soft
NANDA. A Grey Wolf Optimizer Based Auto-
Computing. 2018, vol. 64, iss. 1, pp. 307–316.
matic Clustering Algorithm for Satellite Image
ISSN 1568-4946. DOI: 10.1016/j.asoc.2017.12.026.
Segmentation. Procedia Computer Science. 2017,
vol. 115, iss. 1, pp. 415–422. ISSN 1877-0509.
DOI: 10.1016/j.procs.2017.09.100.
About Authors

[34] ANANTASATE, S. and P. BHASAPUTRA. Sebaa HADDI was born in Setif, Algeria. He
A multi-objective bees algorithm for multi- received his M.Sc. from University of Setif in 1997.
objective optimal power flow problem. In: The 8th follows his study in the University of Ferhat Abbes
Electrical Engineering/ Electronics, Computer, Setif 1, has got his B.Sc. degree in Electrical Engi-
Telecommunications and Information Technology neering Power system from Setif University (Algeria)
(ECTI) Association of Thailand. Khon Kaen: in 1997, and his M.Sc. degree in 2009 in the field of
IEEE, 2011, pp. 852–856. ISBN 978-1-4577-0425- electrical network, now he prepares for the Doctorate
3. DOI: 10.1109/ECTICON.2011.5947974. degree in the Department of Electrical Engineering, of
the university of Setif, His research interests include
[35] SAYAH, S. and K. ZEHAR. Modified dif- the optimization in power system, optimal integration
ferential evolution algorithm for optimal of renewable sources, Facts device.
power flow with non-smooth cost functions.
Energy Conversion and Management. 2008, Omrane BOUKETIR was born in Setif, Algeria.
vol. 49, iss. 11, pp. 3036–3042. ISSN 0196-8904. He received his M Eng. in Electrical Automation from
DOI: 10.1016/j.enconman.2008.06.014. Setif University (Algeria) in 1995. In 1999 he obtained
his M.Sc. degree from University Putra Malaysia
[36] BOUCHEKARA, H. R. E. H. Optimal power in the field of Automation and Robotics and Ph.D.
flow using black-hole-based optimization degree in power electronics systems from the same
approach. Applied Soft Computing. 2014, university in 2005. His research areas include power
vol. 24, iss. 1, pp. 879–888. ISSN 1568-4946. electronics and drive systems, SiC switching devices,
DOI: 10.1016/j.asoc.2014.08.056. CAD tools in electrical engineering and trends and
methods in tertiary education.
[37] ZIMMERMAN, R., C. MURILLO-SANCHEZ
and D. GAN. Matlab power System Simulation Tarek BOUTKIR was born in Setif, Algeria.
Package. New York: School of Electrical Engineer- He received his M.Sc. in Electrical Engineering Power
ing, Cornell University, 2007. system from Setif University (Algeria) in 1994, his

c 2018 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 487

POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 16 | NUMBER: 4 | 2018 | DECEMBER

M.Sc. degree from Annaba University in 1998, his electric power systems, Multi-Objective Optimization
Ph.D. degree in power system from Batna University for power systems, and Voltage Stability and Security
(Algeria) in 2003. His areas of interest are the Analysis. He is the Editor-In-Chief of Journal of
application of the meta-heuristic methods in opti- Electrical Systems (Algeria), the Co-Editor of Journal
mal power flow, FACTS control and improvement in of Automation & Systems Engineering (Algeria).

c 2018 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 488