Energy 124 (2017) 720e740

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

A modified bacteria foraging based optimal power flow framework for

Hydro-Thermal-Wind generation system in the presence of STATCOM
Ambarish Panda a, *, M. Tripathy b, A.K. Barisal b, T. Prakash c
a
Department of Electrical and Electronics Engineering, Sambalpur University Institute of Information Technology, Burla, India
b
Veer Surendra Sai University of Technology, Odisha, Burla, India
c
National Institute of Technology, Raipur, India

a r t i c l e i n f o a b s t r a c t

Article history: Considering the importance of clean energy, the combined operation of hydro-thermal-wind (HTW)
Received 18 July 2016 system is formulated in optimal power flow (OPF) framework. The objective is to find an optimal gen-
Received in revised form eration schedule for the HTW system where the system will work economically and in a voltage secure
28 January 2017
manner with reduced loss during normal as well as stressed system operation. As system voltage may be
Accepted 16 February 2017
Available online 20 February 2017
vulnerable especially during under estimation (UE) situation, provision of additional reactive power (Q)
support is essential as a possible solution. This is achieved by installing shunt facts devices i.e. (STAT-
COM) at the weak nodes of the power network. A comparative assessment between wind-thermal (WT)
Keywords:
Optimal power flow
and HTW system operation with STATCOM at different wind penetration levels is also depicted. The
Modified bacteria foraging algorithm optimum operational paradigms are obtained by optimizing the objective with Genetic Algorithm (GA),
Hydro-Thermal-Wind system Hybrid Algorithm (HA) and modified bacteria foraging algorithm (MBFA). After several tests, superiority
Cost-effective operation of MBFA optimization over HA and GA is revealed so that the IEEE30-bus system operates in a voltage
secure and cost-effective manner.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction on the hydraulic and wind-thermal power system network. The

Economic and voltage secure operation of interconnected power
systems necessitates the solution of intricate optimization prob-
lems that involve superior computational tools. Evolutionary
computation is one of such tool that has proven its capability in
solving complex problems. In this paper, the MBFA is applied in the
solution of optimal generation scheduling in a HTW generation
system. The necessity of HTW system is that, during the operation
of WT system, to compensate for meeting the deficit of wind power
during the over prediction scenario, the system operator has to
depend on thermal powered units (TPU). But increased burden on
TPU may be a threat to environment due to increased emission.
Therefore, some other forms of renewable energy resources may be
considered being integrated into the power system. In this context,
the hydro-electric energy is chosen as an alternative. The proposed
HTW generation scheduling aims at determining optimal hydro,
thermal and wind generations in order to meet the load demands
over a scheduled horizon of one day, satisfying various constraints
* Corresponding author.
E-mail address: ambarish101@gmail.com (A. Panda).
goal is to minimize total operation costs of the hybrid (HTW) sys-
tem. The problem therefore becomes an optimization problem with
highly nonlinear characteristics.
Solving the hydrothermal scheduling problem has been the
theme of exhaustive exploration for years [1e8] with optimization
techniques like differential dynamic programming [1], fuzzy based
evolutionary programming [2], modified differential evolution [3],
enhanced particle swarm optimization [4], improved PSO [5], ge-
netic algorithm [6], self adaptive PSO [7], modified cuckoo search
algorithm [8]. However, very few researchers [9,10] have concen-
trated on the combined operation and optimum dispatch of hydro-
thermal -wind units. In both the work the problem of economic-
emission dispatch is focussed using NSGA-III [9], and ant-lion
optimization [10]. As per the survey made by the authors, no
system-based study has yet addressed the potential of increased
wind penetration on security, reliability and operational economy
of HTW system in OPF framework with incorporation of shunt
FACTS devices. Investigation of these topics require models that can
simulate the effects of wind power integration on hydro-thermal
power system under a variety of operating conditions, while also
satisfying the constraints of HTW generation system. Incorporation

http://dx.doi.org/10.1016/j.energy.2017.02.090
0360-5442/© 2017 Elsevier Ltd. All rights reserved.
 A. Panda et al. / Energy 124 (2017) 720e740 721

of emission issue as a contributory objective in OPF framework is
addressed in Ref. [11]. Authors in Refs. [12e14] have developed a
multi objective WT system operation strategy that considers cost
minimization, loss minimization and voltage secure system
operation.
Thus, in this work an attempt is made to model and demonstrate
the above issue in on optimal power flow framework while opti-
mizing the suitably formulated operational objectives with MBFA,
GA, and HA algorithms. The optimized generating schedules are
tested for their competence in obtaining a secure power system
operation, when the system is subjected to numbers of (N-1) con-
tingencies in the form of line outages (LO), load increase (LI). All the
simulations are carried out in MATLAB/SIMULINK environment.
The primary contributions of this work are given as follows.
violate any of the operating constraints and be able to operate the
system in a secured manner even with some changes in the oper-
ating conditions. Fig. 1 depicts the schematic diagram of the hydro-
thermal-wind generation system.
Mathematically, the objective function may be represented as.
Minimize
FS ¼ FT þ FW þ pfc (1)
In the above expression, FS is the total system operation cost.
FT is the cost of thermal power generation.
FW is the cost of wind power generation.
pfc is the penalty function for handling the constraint violation.
Subject to constraints:

▪ Formulation, modelling and optimal scheduling of hydro- X

Ng X
Nh X
Nw
Pgt þ Pgh þ Pwr ¼ Ploss þ Pload (2)
thermal generation system in the presence of wind power. t t r
▪ Implementation of MBFA, HA and GA algorithm and their vali-
dation in an existing IEEE30 bus power network in a multi
objective OPF framework with wind integration. The existing X
Ng X
Nw
Qgt þ Qwr ¼ Qloss þ Qload (3)
thermal units in the test system are replaced with wind and
t r
hydro units with proper modifications in the power flow
models.
min max
▪ The real power generation schedules of all types of units are Pgt  Pgt  Pgt (4)
optimized using MBFA, HA and GA. The operating efficiencies
and voltage profiles given by different optimal schedules are min max
Qgt  Qgt  Qgt (5)
then evaluated and compared under different operating
scenarios.
▪ Apart from considering the modelling aspects of wind power Vtmin  Vt  Vtmax (6)
uncertainty in the OPF formulation, the focus is also towards the
necessity of the inclusion of shunt FACTS devices into the OPF
max
framework, to improve the overall system voltage profile. Pwr  Pwr (7)
▪ With different wind penetration levels and different combina-
tion of generation resources.i.e. WT and HTW system, a min
Qwr max
 Qwr  Qwr (8)
comparative analysis of voltage secure operation is presented.

min max
The paper has been organized in the following manner. The PHk  PHkj  PHk (9)
main problem has been formulated in Section 2. The modelling of
wind intermittency has been discussed in Section 3. A brief over-
view of GA, HA and MBFA technique applied in this study has been
presented in Section 4. In Section 5, procedural details of simula-
tions and results have been presented. In the same section few
pertinent observations are made regarding the results obtained
which has led to the conclusion in Section 6.

2. Problem formulation

2.1. Problem

The random behaviour of wind power generation makes the

scheduling problem more complex, making the system vulnerable
to insecure voltage conditions. In this context, to mitigate the po-
wer imbalance, additional costs are added to the objective function
to manage the power imbalance during the under estimation (UE)
and over estimation (OE) scenarios [12e14]. However, during OE
scenario in WT system operation, additional reserve units of the
thermal systems may have to be summoned due to shortage in the
available wind power from its original scheduled value. To reduce
the dependency on thermal generating units and simultaneously to
aim for emission reduction, hydro powered units may be incorpo-
rated to fulfil the demand.
The objective of the hydro-thermal-wind scheduling problem is
to minimize the total system operational cost to meet the load
demands during the intervals of the generation scheduling. The
optimal schedules from all the three types of units should not Fig. 1. Schematic diagram of Hydro-Thermal-Wind generation system.
722 A. Panda et al. / Energy 124 (2017) 720e740
Vkmin  Vkj  Vkmax
qmin
k  qkj  qmax
k (11)
Ru h
X i
Vkðjþ1Þ ¼ Vkj þ quðjtÞ þ SuðjtÞ  qkðjþ1Þ  Skðjþ1Þ þ rkðjþ1Þ
u¼1
 
2 2
PHkj ¼ c1k Vkj þ c2k Vkj þ c3k Vkj qkj þ c4k Vkj þ c5k qkj þ c6k
Equations (2) and (3) denotes the load balance equation for both
real and reactive power respectively. Equations (4) and (5) refer to
the limiting values (min and max) of real and reactive power limits
of thermal generating units. (6) Denotes the limits within which the
system voltage magnitude varies. Variation of real and reactive
power output of wind powered units are indicated by (7) and (8).
The operating limits of hydro powered units are restricted by lower
and upper bounds which is signified by (9). Similarly reservoir
capacity constraint and water discharge constraint of equivalent
hydro powered units are considered in (10) and (11). Finally the
hydraulic continuity constraint and hydro power generation con-
straints are depicted by (12) and (13) respectively. The abbrevia-
tions used and their meaning for all the equations are mentioned in
Appendix A.4.
Mathematical expressions of the terms described in (1) are
discussed as below.
2.1.1. Cost of thermal power generation (FT )
Extensively commended quadratic cost function representing
the cost of thermal power generation for the thermal generators are
mathematically represented by (14) as follows. It is depicted in
Table A3 in the appendix.
  2
FT ¼ Ct Pgt ¼ at Pgt þ bt Pgt þ ct
2.1.2. Cost of intermittent wind generation (FW )
This cost consists of three components.
FW ¼ FD þ FP þ FR
First one deals with cost of wind power purchase from wind
power producer (WPP), second one deals with expected availability
of surplus wind power during UE scenario, third one deals with
expected availability of deficit wind power during OE scenario.
These components are named as direct cost, penalty cost and
reserve cost respectively [13]. The mathematical interpretations of
the above components are depicted as below.
Direct Cost.(FD ).
FD ¼ Cwr ðPwr Þ ¼ dr Pwr
Penalty Cost.(FP ).
 
FP ¼ Cp;wr Pwr;av  Pwr ¼ KPr ðPw av  Pwr Þ
ZPro
¼ Kpr ðw  Pwr Þfw ðwÞdw
Pwr
Reserve Cost.(FR ).
(10) FR ¼ CR wr ðPwr  Pwr av Þ ¼ KRr ðPwr  Pw av Þ
ZPwr
¼ KRr ðPwr  wÞfw ðwÞdw (18)
0
2.1.3. Incorporation of penalty functions for constraint violation
(12) Principally in an optimization procedure, only those solutions
are considered to be optimal which are obtained without violating
any of the constraints. To ensure that the variable has not violated
the limits during the optimization process, penalty functions are
(13) added to the objective function. In this work to achieve the above
facts, the inequality constraints for real power line flow and bus
voltage limits are incorporated as penalty functions represented as
Pfc1 and Pfc2, which can be defined by the mathematical expressions
(19) and (20). The functioning of the penalty factors can be
explained as follows. As long as any variable Lk remains within its
defined minimum and maximum limits of Lmin and Lmax, the pen-
alty factor Pfk for the same shall become zero, else it will take a high
value proportional to pf [21].
Pfc1 ¼ abs½signðPmax  Pk Þ  1*pf þ abs½signðPmin  Pk Þ þ 1*pf
(19)
Pfc2 ¼ abs½1 þ signðVk  Vmax Þ*pf þ abs½1  signðVk  Vmin Þ*pf
(20)
2.1.4. Methodology of solving the scheduling problem
The prime objective of the HTW scheduling problem is to
minimize the total system operation cost such that the load de-
mand supplied from hydro plant, thermal plant and wind powered
units in the intervals of the generation scheduling horizon can be
met and simultaneously, all the equality and inequality operational
constraints are satisfied.
(14) The HTW scheduling approach may be described as
 The hydro discharge of kth unit at jth scheduling horizon qkj is
calculated by using the hydraulic continuity equation (12).
During this, the spillage is assumed to be zero.
 Knowing the amount of water discharge, the reservoir volumes
at different intervals are determined and hydro power genera-
(15) tion is calculated using (13).
 After finding the hydro power, the wind power available is
evaluated.
 Finally after obtaining the hydro and wind power generation for
satisfying the load demand, the thermal power generation is
evaluated using (2).
3. Operational issues with wind speed variability
3.1. Modelling of wind speed intermittency
(16)
The power output of the wind farm can be thought of as the
output of an equivalent wind turbine generator (WTG). The rela-
tionship between the wind power output [13] and the wind speed
is deterministic for a given wind turbine generator. Hence,
modelling of the wind speed in the farm is very important. Owing
(17) to the limitations of availability of statistical methods and vari-
ability of wind speed, some studies [15e17] have tried to find out
the probability distribution functions (PDF) which shall approxi-
mate the randomness with considerable accuracy. This work
 A. Panda et al. / Energy 124 (2017) 720e740 723

applies Weibull PDF as explained in (21), in order to model the Two discrete probability events occur when there is no wind
variability of wind flow. power output and rated power output. Mathematically this may be
represented as (22) and (23) as discussed below.
  ðk1Þ vk Probability of event Pw ¼ 0 is
k v  c
fv ðvÞ ¼ e ;0<v<∞ (21)
c c
Pr ðPw ¼ 0Þ ¼ Pr ðV < vi Þ þ Pr ðV  vo Þ ¼ FV ðvi Þ þ ð1  FV ðvo ÞÞ
Here c and k are the scale and shape parameters respectively,   k    k 
v vo
which together control the extent of variability and pattern of wind ¼ 1  exp  i þ exp 
c c
flow. Scale parameter has the unit of m/sec while shape parameter
is a dimensionless quantity. The effect of variation of c and k on (22)
behaviour of wind pattern is depicted in Fig. 2. Probability of event Pw ¼ Pr is
When wind speed is low, the blades of wind turbine would not
rotate due to the friction and large inertia, and no electricity is Pr ðPw ¼ Pr Þ ¼ Pr ðvr < V < vo Þ ¼ FV ðvo Þ  FV ðvr Þ
generated. Only when wind speed is fast enough and the blades   k    k 
vr vo
have enough torque to overcome the resistance, the wind plant will ¼ exp  þ exp  (23)
start to generate electricity. The speed at which the turbine begins c c
to rotate and generate power is referred as cut-in speed and this From Fig. 3, the discrete and continuous portions of the wind
speed is typically between 3 and 5 m/s. If the wind speed keeps power output probability function based on the Weibull wind
going up to a certain level that may damage the rotor, the self- speed pdf with k ¼ 2 and c factors of 10, 15, and 20 are plotted.
protection system of the turbine will stop the rotor. This wind
speed, at which the braking system is activated, is called cut-out
speed [29].
As the WECS power output has a constant zero value when the 4. Overview of optimization methodologies
speed is below the cut-in wind speed and also above the cut-out
wind speed, the power output random variable will be discrete in For the purpose of optimization MBFA, GA and HA are employed
these ranges. The WECS power output is a mixed random variable, to optimize both the objective function and their respective per-
which is continuous between values of zero and rated power, and is formances are compared. Brief overviews of the algorithms are
discrete at values of zero and rated power output. discussed below.

wind speed characterization(k=1)

0.4
c=5
weibull pdf

c=10
0.2 c=15
c=20

0
0 5 10 15 20 25 30 35 40
wind speed
wind speed characterization(k=2)
0.4
weibull pdf

0.2

0
0 5 10 15 20 25 30 35 40
wind speed
wind speed characterization(k=3)
0.4
weibull pdf

0.2

0
0 5 10 15 20 25 30 35 40
wind speed
Fig. 2. Weibull probability density function for different values of shape and scale parameter.
724 A. Panda et al. / Energy 124 (2017) 720e740

0.9
c=10
0.8 c=15
c=20
0.7

0.6
wind power pdf

0.5

0.4

0.3

0.2

0.1

0
0 0.2 0.4 0.6 0.8 1
ratio of wind power(ro)
Fig. 3. Weibull mixed probability function for different wind speed.

4.1. Genetic algorithm and l are initialized before the first iteration.

Genetic Algorithm (GA) [18,19] is an established parallel search 1) Begin elimination dispersal loop.
algorithm, which has been applied in varieties of optimization 2) Begin reproduction loop.
problems in power systems. In GA, the initial set of randomly 3) Begin chemotaxis loop.
generated population evolves through several generations to a) For c i ¼ 1, 2 … S, cost function F (i, j, k, l) is calculated.
reproduce fitter candidate solutions using crossover and mutation b) Find the optimum bacterium qob , from all those evaluated
operators. As the algorithm is based on idea of the survival of the until that point of evolution. c i ¼ 1, 2 … S, determine the
fittest, the weaker off-springs get eliminated. During the search swarm attractant cost FCC, that is added with F(i, j, k, l) to
process, as there is no restriction on the search space, it enhances obtain Fsw(i, j, k, l) [20].
the robustness of GA. c) If j ¼ 1
i) Randomly select two parents of bacteria for reproduction,
4.2. Hybrid algorithm out of the pool of all S bacteria.
ii) Generate two random numbers t1 and t2.
The Hybrid Algorithm (HA) is synthesized by implementing the a) If t1  Pmutation, do mutation at the decimal crossover point
mutation strategies of GA along with the BFA that was first pro- decided by t2
posed in Ref. [20], so that the optimization efficiencies of both the b) Else if, Pmutation < t1  PCrossover, do Crossover at the decimal
algorithms may be further improved in some specific problems. The crossover point decided by t2. It is to be noted that, after the
steps involved in HA can be explained as follows. above GA operators, qi ðj; k; lÞ has generated bacterial off
At the outset, variables like number of control parameters (p), springs qiCross ðj; k; lÞ
bacteria (S), chemotactic process (Nc), reproduction events (G) and iii) Carry out swimming or tumbling, using the following
elimination & dispersal events (D) for the algorithm, are initialized. equation
Further, the control parameters like maximum swimming length
Ns. (Nc), the probability of elimination and dispersal Ped,, the
swarming coefficients i.e.dattract ; uattract ; hrepelent and urepelent , run DS ðiÞ
length unit (C(i)) and swim length SL are all chosen judiciously. qi ðj þ 1; k; lÞ ¼ qiCross ðj; k; lÞ þ CðiÞ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi (24)
P(p,S,1), specifies the location of the initial set of S bacteria, each DTS ðiÞDS ðiÞ
consisting of p random numbers. After scaling up, each of the
random numbers represents a possible solution of the control where,DS ðiÞ2<p is a predefined random direction vector, with each
variables. element DSm ðiÞ, m ¼ 1,2,..p.
The iterative steps of HA proceeds as follows. The cost function
for the initial bacterial population inside the inner most chemotaxis d) For, j > 1
loop, is evaluated. Any ith bacterium and its corresponding cost i) Repeat steps 3. c.(i) & (ii) and reorienting the qiCross ðj; k; lÞ
function in the jth chemotactic, kth reproduction and lth elimina- through swimming/tumbling
tion stages is identified by, qi(j,k,l) and F(i, j, k, l) respectively. j, k, ii) If, Fsw(i, j, k, l) < Fsw(i, j-1, k, l)
 A. Panda et al. / Energy 124 (2017) 720e740
Convergence Characteristics
3580
3560
3540
3520
3500
Cost
3480
3460
3440
3420
3400
3380
0 1000 2000 3000 4000 5000 6000
No of generation
Fig. 4. Convergence characteristics of HTW system with GA, HA and MBFA techniques.
4. Increment ‘j’ i. e j ¼ j þ 1. Go to step 3, if j < Nc (Continue
Chemotaxis loop)
5. Carry out the process of reproduction.
a) For the given k and l, let Jhealth ¼ Sort fJsw ðj; k; lÞg . Sort
j2f1/N g
bacteria in order of ascending cost Jhealth. Ac higher cost of any
bacterium means lower health for minimization.
b) From the total S bacteria, the better half upholds the evolu-
tion process replacing the other less healthier half of bacteria,
who are eliminated following elitism.
c) Increment the reproduction loop counter i. e k ¼ k þ 1. Go to
step 2 if k < G.
6. In the process of elimination & dispersal, generate a random
number t3.
a) If t3 <¼ Ped, then the existing entire set of bacteria gets
eliminated and dispersed in a new random direction. Incre-
ment l ¼ l þ 1. Go to step 1, if l < D.
b) Else continue with Reproduction again, i.e Go to step 2.
It can be noticed that in the steps 3.c.(i)&(ii) and 3.d.(i)&(ii), the
hybridization of GA operators in the BFA, is applied. In addition to
above, the use of optimum bacterium (qob ) in evaluation of swarm
attractant cost (FCC) in step 3.b, as suggested earlier in Ref. [30],
improves its ability to swarm.
4.3. Modified bacteria foraging algorithm
Bacteria Foraging Algorithm, similar to numbers of such algo-
rithms is based on the evolution principle of E. coli bacteria present
in human intestines. The algorithm is based on the principle that
natural selection tends to eliminate animals with poor foraging
strategies compared to others and favour those having successful
foraging strategies. During the process of evolution, bacteria having
725
GA
HA
MBFA
7000 8000 9000 10000
poor foraging strategies are either eliminated or mutated into
better ones. The foraging strategy of E. coli bacterium is governed by
four processes, namely, chemotaxis, swarming, reproduction, and
elimination-dispersal.
The detailed steps involved in BFA can be obtained from
Ref. [20]. In MBFA [12], the original version of the algorithm pro-
posed in Ref. [20] is modified inside both chemotaxis and swarming
stages. The proposed modifications which expedite the conver-
gence have also applied in this work. Two major modification
points adopted in this improved version of BFA i.e. MBFA can be
referred from Ref. [12]. The flow chart of the algorithm with above
modifications is given in Section. A2 in the appendix. Moreover the
parameters used in all the above mentioned optimization algo-
rithms are mentioned in the appendix.
5. Simulation results and discussion
For simulation of the work, the IEEE-30 bus system [22] is
considered as the system under study. Bus 1 is the slack bus which
is connected to the largest generating unit (Thermal unit). Gener-
ator at bus 2 is a thermal generating unit while that installed at 8th
bus is a hydro powered unit with an installed capacity of 40 MW.
The test system is modified by replacing conventional generators
with wind farms located at fifth, eleventh and thirteenth bus. Each
wind farm (WF) consists of several wind turbine generators (WTG)
equipped with DFIGs. In this work, WF at bus no.5 consists of ten
WTG (each of 3 MW) having a total capacity of 30 MW. Similarly
WF at bus numbers11 and 13 each consist of eight WTG of 3 MW
capacities with total capacities of 24 MW each. The simulation for
the proposed objective function is carried out in MATLAB/SIMU-
LINK environment. The program was run on a 2.30 GHz, Intel(R)
Core-i3, with 4 GB RAM notebook configuration. The analysis of
726 A. Panda et al. / Energy 124 (2017) 720e740

3580
GA
3560 HA
MBFA
3540

3520

3500
Cost

3480

3460

3440

3420

3400

3380
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
No of generation 4
x 10
Fig. 5. Convergence characteristics of HTW system with GA, HA and MBFA techniques.

system behaviour and optimal operating configuration is evaluated
under different cases as described below.
5.1. Cost effective system operation
In this case the objective function (1) is optimized for the
considered test system with three optimization techniques at the
nominal operating conditions. The optimization algorithms are
MBFA, HA and GA as discussed in sections 4.1, 4.2 and 4.3 respec-
tively. In order to determine the best operating conditions in terms
of operational economy, voltage security and fastness in operation a
comparative analysis is carried out with the three above mentioned
techniques. The details of parameters chosen for these algorithms
are mentioned in the Appendix.
Figs. 3 and 4 represents the convergence characteristics found
with the three different algorithms plotted at 10000 and 20000
generations respectively. In Fig. 4 the GA optimized curve con-
verges at 3396.76 $/hr and HA optimized curve converges
approximately at 3389.42 $/hr. But the MBFA optimized scenario
provides the superior characteristics with a final solution of
3386.92$/hr. In order to find any further improvement in the
Table 1
Optimized generation schedule of the Hydro-Thermal-Wind system in p.u. with
MBFA, HA and GA during normal operation.
characteristics, the number of generations during the optimization
process is continued till 2  104 generations. Still, MBFA retained its
superiority over GA and HA in providing the most economic solu-
tion. This is depicted in Fig. 5 and presented in Table 1. It may be
seen that when GA and HA converged at 3395.24$/hr and 3389.34$/
hr respectively, MBFA converged at 3386.25$/hr. Different compo-
nents of the final optimized solution obtained with the three al-
gorithms are tabulated in Table 1.
In this table, PL represents the real power transmission loss and
TC is the total system operational cost. As seen from the table, the
loss is minimum (0.0854 p.u.) with MBFA optimized schedule while
it is maximum (0.0922 p.u.) with GA optimized schedule. The loss
comes within these values i.e. 0.0900 p.u. with HA optimized
schedule.
Besides these aspects, the fastness of operation while achieving
the optimal operational paradigm is provided by MBFA technique.
CPU run time
80
70
60
50
Time in minutes.

Number Generator MBFA

40
MBFA HA GA HA
Pg Q Pg Q Pg Q 30 GA
1 1.2010 0.9372 1.2640 0.9537 1.3169 0.9616 20
2 0.6184 0.3270 0.6200 0.3270 0.5804 0.3270
5 0.3000 0.1100 0.3000 0.1100 0.2995 0.1100 10
8 0.3686 0.4500 0.3100 0.4500 0.2999 0.4500
11 0.2210 0.3000 0.2200 0.3000 0.2198 0.3000 0
13 0.2096 0.3000 0.2100 0.3000 0.2098 0.3000
10000 g en 20000 g en
PL(p.u.) 0.0854 0.0900 0.0922
TC($/hr) 3386.25 3389.34 3395.24
Fig. 6. Time elapsed during the different optimization processes.
 A. Panda et al. / Energy 124 (2017) 720e740 727

Convergence Characteristics
2775

2770

2765

2760
Cost

2755

2750

2745

2740
0 2000 4000 6000 8000 10000
No of generation
Fig. 7. Cost convergence of HTW with 45% wind penetration up-to 10000 generations.

This is depicted in Fig. 6 where, in getting the results shown by
Figs. 4 and 5., MBFA takes minimum computational time.
The above findings presented by Table 1 and Figs. 4e6 depicts
the superiority of MBFA optimized schedule over GA and HA in
terms of
 Operating the Hydro-Thermal-Wind (HTW) generation system
with minimum cost.
 Operating the power system with minimum transmission loss.
 Obtaining the optimal convergence characteristics within min-
imum time.
But all these findings are subjected to 25% wind power pene-
tration (WPPN) in the HTW power network. By increasing the wind
power penetration level up-to 45%, the impact of wind on overall
system operation is studied.

24
stage1
22 stage2
stage3

20

18
Hydro discharge

16

14

12

10

8
5 10 15 20 25
Hour
Fig. 8. Hourly discharge (  104 m3) of plant during different stages.
728 A. Panda et al. / Energy 124 (2017) 720e740

45

Reservoir storage volume 40

35

30

stage1
stage2
stage3
25
5 10 15 20 25
Hour
Fig. 9. Variation of reservoir volume during the scheduling horizon.

In this context, MBFA is used to optimize the objective function
(1) and the cost convergence characteristic of the modified system
with 45% WPPN is depicted in Fig. 7.
Comparing the characteristics presented by MBFA technique for
the HTW system with two different WPPN levels as depicted in
Figs. 4 and 7, a notion may be drawn that with 20% increment in
WPPN the system operation cost is decremented approximately by
15%. The real power transmission loss reduces to 0.0544 p.u. which
was 0.0854 p.u. with 25% WPPN. Fig. 8 depicts hourly hydro
discharge from the reservoir. After knowing the water discharges,
the reservoir volumes at different scheduling intervals are found
out. Then, the hydro power generations are calculated using (13).
Fig. 9 depicts the trajectories of the reservoir storage volume during
different optimization stages i.e. after 1000, 3000 and 5000
generations.

1.06
HA
1.04 GA
MBFA

1.02

1
Voltage (p.u.)

0.98

0.96

0.94

0.92

0.9
5 10 15 20 25 30
Bus No
Fig. 10. HTW System voltage profile during normal operating condition.
 A. Panda et al. / Energy 124 (2017) 720e740 729

5.2. Voltage secure system operation
In order to test the competence of MBFA optimized generation
schedule in providing an improved voltage response during normal
and stressed operating condition, a comparison is made between
the three optimization techniques. This is presented as
Case 1: Normal operating Condition
Case 2: Stressed operating Condition
5.2.1. Case 1
The system voltage response during Case.1 is depicted in Fig. 10.
It shows that though a minor improvement in bus voltage is shown
by HA schedule over GA, remarkable enhancement in voltage in
most of the buses is presented by MBFA optimized schedule. The
numerical analysis of this fact is depicted in Table 2.
As seen from the table, the minimum and mean voltage
magnitude obtained with MBFA is 0.9098 p.u. and 0.9510 p.u.
respectively. These are higher than the corresponding voltages
obtained with HA and GA. Moreover, comparing the standard de-
viations (SD) obtained with the three techniques, MBFA presents
the minimum value i.e. 0.0323 p.u. as compared to SD obtained
with HA and GA which are 0.0328 p.u. and 0.0330 p.u. From these
above findings, the superiority of MBFA optimized schedule in
Table 2
Analysis of improved system voltage during normal operating condition.
Sr No Bus voltage characteristics Voltage during normal
Operating conditions (p.u.)
Mean Minimum
Case-1 MBFA 0.9510 0.9098
HA 0.9485 0.9068
GA 0.9474 0.9055
depicting a superior voltage during normal system operation is
established.
5.2.2. Case-2
In this case, the operational competency of MBFA is tested with
HA and GA during stressed operating conditions.
To verify this, the system operation is subjected to stress situ-
ation by intentionally introducing load increase (LI) scenario in one
of the randomly chosen bus in the power network. In this case bus-
12 is the chosen one which is subjected to load increase at 20%
interval. Thus bus-12 is gradually subjected to a series of 20% load
increase. i.e. 20%, 40%, 60%, 80% and 120%. These are depicted in
Figs. 11e15. From the figures it may be seen that in every load in-
crease scenario, the voltage obtained with MBFA schedule provides
relatively superior performance. The mathematical analysis of the
voltage improvement is presented in Table 3.
From Table 3, it may be summarized that
 For 20% load increase, the mean and minimum system voltage as
found from MBFA schedule is 0.9660 p.u. and 0.9294 p.u. while
those with HA optimized schedule is 0.9642 p.u. and 0.9271 p.u.
respectively and from GA optimized schedule 0.9633 p.u. and
0.9261 p.u. It implies that MBFA schedule depicts superior
results.
 Besides these findings, the standard deviation (SD) with 20% LI
for MBFA, HA, GA optimized schedule is 0.0292 p.u, 0.0296 p.u
and 0.0297 p.u. respectively.
 Thus the voltage deviation is minimum with MBFA case while it
is moderate with HA and maximum with GA, though the dif-
ference is not significant.
 For each LI case, the mean and minimum system as depicted
Maximum SD
from Table 3 bus voltage with MBFA optimized schedule is
1.0500 0.0323 higher than that of the HA and GA schedule and the SD is
1.0500 0.0328 consistently less in magnitude in MBFA case as compared to HA
1.0500 0.0330
and GA even during severe stressed situation of 120% LI.

20% LI scenario
1.06
MBFA
HA
1.04
GA

1.02
voltage mag(p.u.)

0.98

0.96

0.94

0.92
5 10 15 20 25 30
Bus no
Fig. 11. System voltage response when bus-12 is subjected to 20% LI.
730 A. Panda et al. / Energy 124 (2017) 720e740

40% LI scenario
1.06
MBFA
HA
1.04
GA

1.02
voltage mag(p.u.)

0.98

0.96

0.94

0.92
5 10 15 20 25 30
Bus no
Fig. 12. System voltage response when bus-12 is subjected to 40% LI.

60% LI scenario
1.06
MBFA
1.04 HA
GA

1.02
voltage mag(p.u.)

0.98

0.96

0.94

0.92

0.9
5 10 15 20 25 30
Bus no
Fig. 13. System voltage response when bus-12 is subjected to 60% LI.

5.3. Comparative assessment of system voltage of HTW system system operation.

versus WT system The analysis is carried out in four operational environments as
mentioned below.
In this scenario, the operational efficiency of Hydro-Thermal-
Wind (HTW) generation system with respect to Wind-Thermal Scenario1: With 25% wind power penetration
(WT) system is compared in terms of achieving voltage secure Scenario2: With 45% wind power penetration
 A. Panda et al. / Energy 124 (2017) 720e740 731

80% LI scenario
1.06
MBFA
1.04 HA
GA
voltage mag(p.u.) 1.02

0.98

0.96

0.94

0.92

0.9
5 10 15 20 25 30
Bus no
Fig. 14. System voltage response when bus-12 is subjected to 80% LI.

1.06
MBFA 20% LI
HA 20% LI
1.04
GA 20% LI
MBFA 120% LI
1.02 HA 120% LI
GA 120% LI
voltage mag(p.u.)

0.98

0.96

0.94

0.92

0.9
5 10 15 20 25 30
Bus no
Fig. 15. Comparative analysis between 20% and 120% LI.

Scenario3: With 25% wind power penetration in the presence of
STATCOM.
Scenario4: With 45% wind power penetration in the presence of
STATCOM.
In this context, details of STATCOM [23] along with modelling
issues in power flow frame work may be referred from Ref. [12].
5.3.1. Scenario 1
In this case the optimized generation schedule with MBFA for
both WT and HTW generation system is obtained. Taking this into
consideration, the system voltage during normal operating
732 A. Panda et al. / Energy 124 (2017) 720e740

Table 3 The numerical analysis of voltage improvement is tabulated in

Relative system voltage improvement analysis during LI condition. Table 4.
Sr No Bus voltage Voltage during stressed (Load increased) From the above results it may be depicted that
characteristics
Operating conditions (p.u.)
 With 25% wind penetration, the mean and minimum WT system
20% 40% 60% 80% 120%
voltage is 0.9523 p.u. and 0.9113 p.u. respectively whereas for
Case-2 MBFA Mean 0.9660 0.9626 0.9595 0.9568 0.9527 HTW system these are 0.9510 p.u. and 0.9098 p.u. respectively.
Min 0.9294 0.9252 0.9213 0.9179 0.9126
 SD as found with 25% wind penetration is 0.0320 p.u. for WT
Max 1.0500 1.0500 1.0500 1.0500 1.0500
SD 0.0292 0.0298 0.0304 0.0309 0.0318 system while that is 0.0323 p.u. for HTW system.
HA Mean 0.9642 0.9607 0.9575 0.9546 0.9503  Thus, in mean and minimum voltage evaluation, WT system
Min 0.9271 0.9228 0.9188 0.9152 0.9097 shows marginal improvement over HTW system, though the
Max 1.0500 1.0500 1.0500 1.0500 1.0500 improvement is not momentous.
SD 0.0296 0.0302 0.0308 0.0314 0.0322
GA Mean 0.9633 0.9597 0.9565 0.9536 0.9493
 It is observed that for less wind penetration, marginal
Min 0.9261 0.9217 0.9177 0.9141 0.9085 improvement in system response can be shown by WT system
Max 1.0500 1.0500 1.0500 1.0500 1.0500 over HTW system.
SD 0.0297 0.0303 0.0310 0.0315 0.0324  For 45% wind penetration, the % of mean voltage and % of
improvement in the magnitude of minimum voltage improve-
ment for WT system noticeably increases over HTW system.
 The SD of WT system is found to be less than that of the HTW
condition is evaluated and the comparative analysis is presented in system.
Fig. 16.
From the above figure, it may be pointed out that WT generation
system with 25% wind penetration has provided a relatively better
voltage response as compared to HTW system though the Table 4
Comparative analysis of system voltage with different wind penetration.
improvement in voltage at majority of buses is very minor. Statis-
tical analysis of these results are enumerated in Table 4. Therefore, Sr No Bus voltage Analysis of system voltage (p.u.) in the
to study the performance of both the systems (WT and HTW) in the characteristics presence of
presence of increased wind penetration, contribution of wind po- 25% wind 45% wind
wer to the overall generation system is increased to 45%. This leads penetration penetration
to the next section. WT HTW WT HTW

1 MBFA Mean 0.9523 0.9510 0.9679 0.9648

5.3.2. Scenario 2 Min 0.9113 0.9098 0.9278 0.9240
Fig. 17 demonstrates the impact of increased wind penetration Max 1.0500 1.0500 1.0500 1.0500
SD 0.0320 0.0323 0.0296 0.0302
on the system voltage.

1.06
Wind-Thermal
Hydro-Thermal-Wind
1.04
0.95
1.02
0.94

1
voltage mag(p.u.)

0.93

0.98
0.92

0.96
0.91
15 20 25

0.94

0.92

0.9
5 10 15 20 25 30
Bus no
Fig. 16. System voltage response with 25% wind penetration. (Scenario1).
 A. Panda et al. / Energy 124 (2017) 720e740 733

Increased wind penetration

1.06
HTW
WT
1.04

1.02
voltage mag(p.u.)

0.98

0.96

0.94

0.92
5 10 15 20 25 30
Bus no
Fig. 17. System voltage response with 45% wind penetration (scenario.2).

1.06
WT during scenario1
HTW during scenario1
1.04 HTW during scenario3
WT during scenario3

1.02

1
voltage mag(p.u.)

0.98

0.96

0.94

0.92

0.9
5 10 15 20 25 30
Bus no
Fig. 18. System voltage response with 25% wind penetration with STATCOM.
734 A. Panda et al. / Energy 124 (2017) 720e740

Table 5 system in terms of maintaining a secure voltage in the presence of

Comparative analysis of system voltage with different wind penetration level in the STATCOM is analysed when wind power penetration level is 25% of
presence of STATCOM.
system generation capacity. The STATCOM is installed at the
Sr No Bus voltage Analysis of system voltage (p.u.) with weakest bus, i.e. the 30th bus [27] in the considered IEEE30 bus
characteristics STATCOM for system.
25% wind 45% wind From the above figure, it may be depicted that significant
penetration penetration improvement in the voltage magnitude is observed for both the
WT HTW WT HTW systems after the incorporation of STATCOM in the system. But,
1 MBFA Mean 0.9734 0.9725 0.9850 0.9829 here also the WT system shows marginal improvement in voltage
Min 0.9440 0.9430 0.9569 0.9545 as compared to HTW system.
Max 1.0500 1.0500 1.0500 1.0500
SD 0.0272 0.0273 0.0248 0.0253

 Also, while increasing the wind penetration levels it is found
that
 The mean system voltage increases from 0.9523 p.u. to 0.9679
p.u. for WT system while that increases from 0.9510 p.u. to
0.9648 p.u. for HTW system.
 The minimum system voltage increases from 0.9113 p.u. to
0.9278 p.u. for WT system while that increases from 0.9098
p.u. to 0.9240 p.u. for HTW system.
 The SD in voltage reduces from 0.0320 p.u. to 0.0296 p.u. for
WT system while that reduces from 0.0323 p.u. to 0.0302 p.u.
for HTW system.
In order to figure out any further improvement in the system
voltage for both the system as discussed above, STATCOM [24e26]
is incorporated at the weakest bus [27] in the network. This is
depicted in the next section.
5.3.3. Scenario 3
In this scenario, relative performance of both the WT and HTW
5.3.4. Scenario 4
In this scenario, relative performance of both the WT and HTW
system in terms of maintaining a secure voltage in the presence of
STATCOM is analysed when wind power penetration level is
increased from 25% to 45%.
Similar to previous results as illustrated by Fig. 18, WT system
with STATCOM, considerably improves the system voltage as
compared to WT system without STATCOM and HTW system with
STATCOM. The statistical analyses of the above findings are
enumerated in Table 5.
On comparison between the data obtained from Tables 4 and 5,
the salient findings may be summarized as below.
The presence of STATCOM for 25% wind penetration has.
 Enhanced the mean WT system voltage by 0.0211 p. u. and
minimum system voltage by 0.0327 p.u.
 Reduced the standard deviation of voltage by 0.0048 p.u.
 Similarly upgraded the mean HTW system voltage by 0.0215 p.u.
and minimum voltage by 0.0332 p.u.
 Decreased the SD for HTW system by 0.005 p.u.

Increased wind penetration

1.06
HTW during scenario2
WT during scenario2
1.04 HTW during scenario4
WT during scenario4

1.02
voltage mag(p.u.)

0.98

0.96

0.94

0.92
5 10 15 20 25 30
Bus no
Fig. 19. System voltage response with 44% wind penetration with STATCOM.
 A. Panda et al. / Energy 124 (2017) 720e740 735

1.06
GA
HA
1.04
MBFA

1.02

1
Voltage (p.u.)

0.98

0.96

0.94

0.92

0.9
5 10 15 20 25 30
Bus No
Fig. 20. Comparative analysis of HTW system voltage response during Scenario-1 for outage of line 8e28 with GA, HA and MBFA optimized schedule.

1.04 GA
HA
1.02 MBFA

0.98
Voltage (p.u.)

0.96

0.94

0.92

0.9

0.88

5 10 15 20 25 30
Bus No
Fig. 21. Comparative analysis of HTW system voltage response during Scenario-3 for outage of line 10e20 with GA, HA and MBFA optimized schedule.

Similar notion may be drawn for incorporation of STATCOM is presented by WT system compared to HTW generation
with 45% wind penetration. system.
 Considering all the scenarios, it may be summarized that with
 It is important to note that, even after the installation of STAT- less wind penetration, approximately 20%e25%, WT system
COM in both the systems i.e. WT and HTW, the evaluating pa- behaves almost similar to HTW system in terms of maintaining a
rameters have significantly shown improvement but still the voltage secure operation.
superiority of maintaining a relatively enhanced system voltage
736 A. Panda et al. / Energy 124 (2017) 720e740

1.06
HTW during scenario1
HTW during scenario2
1.04 HTW during scenario3
HTW during scenario4
1.02

1
Voltage (p.u.)

0.98

0.96

0.94

0.92

0.9

5 10 15 20 25 30
Bus No
Fig. 22. Comparative analysis of HTW system voltage response during different Scenarios for outage of line 15e18 with MBFA.

1.06
HTW during scenario1
HTW during scenario2
1.04
HTW during scenario3
HTW during scenario4
1.02

1
Voltage (p.u.)

0.98

0.96

0.94

0.92

0.9

0.88

5 10 15 20 25 30
Bus No
Fig. 23. Comparative analysis of HTW system voltage response during different Scenarios for outage of line 10e20 with MBFA.

 But with enhancement in wind power penetration level nearly
45%, considerable improvement in the system voltage may be
observed compared to HTW system.
In order to further evaluate the robustness of the optimization
techniques on improving the relative system voltage performances
of the previously mentioned scenarios during stressed operating
conditions, a comparative analysis is conducted between them. In
this context, stressed operating conditions are created in the sys-
tem by intentionally introducing line outages (LO) at few of the
randomly chosen lines one at a time. This is presented in Figs. 20
and 21.
 A. Panda et al. / Energy 124 (2017) 720e740 737

1.05 HTW during Scenario1

HTW during Scenario2
WT during Scenario2

1
Voltage (p.u.)

0.95

0.9

0.85

5 10 15 20 25 30
Bus No
Fig. 24. Comparative analysis of WT and HTW system voltage response during Scenario-1 and Scenario-2 for outage of line 27e30 with MBFA.

1.06
HTW during scenario.3
HTW during scenario.4
WT during scenario.4
1.04

1.02
Voltage (p.u.)

0.98

0.96

0.94
5 10 15 20 25 30
Bus No
Fig. 25. Comparative analysis of WT and HTW system voltage response during Scenario-3 and Scenario-4 for outage of line 14e15 with MBFA.
738 A. Panda et al. / Energy 124 (2017) 720e740
1.05
1
Voltage (p.u.)
0.95
0.9
HTW during scenario.3
HTW during scenario.4
WT during scenario.4
0.85
5 10
Bus No
Fig. 26. Comparative analysis of WT and HTW system voltage response during Scenario-3 and Scenario-4 for outage of line 25e27 with MBFA.
From Figs. 20 and 21 it may be depicted that
 MBFA optimized schedule presents relatively superior voltage
performance during different LO conditions as compared to GA
and HA optimized schedules.
 Therefore, MBFA is selected for evaluating and finding the best
operating performance in terms of maintaining an improved
system voltage among all the four above mentioned scenarios
during LO situations. This is applied for the HTW system.
In this context, two separate LO events are created in the HTW
system one after another in the form of outage of line 15e18 and
10e20. This is depicted in Figs. 22 and 23 respectively. It may be
clearly pointed out that scenario.4 presents a significant improve-
ment in the system voltage even during different stressed operating
conditions.
It is already demonstrated from Figs. 16e19 that, during normal
operating condition, the WT system portrayed superior voltage
profile than HTW system in all the four scenarios. Therefore, to
validate the occurrence of similar consequences, both WT and HTW
system behaviour is tested during LO conditions. The results are
presented in Figs. 24e26.
The salient findings drawn from Figs. 22e24 is enumerated
below.
 During stressed operating condition (i.e. outage of line 27e30),
HTW system for Scenario-2 depicts an improved voltage
response than Scenario-1 (as shown in Fig. 24).
 But the voltage response as shown by WT system for Scenario-2
during the above mentioned LO condition is superior to HTW
system (as shown in Fig. 24).
15 20 25 30
 Thus a notion may be drawn that even in the absence of shunt
FACTs devices, WT system presents comparatively a better
voltage secure operation than HTW system under stressed
operating environment.
In the presence of STATCOM, a comparison is made between
HTW system operation under Scenario-3 and Scenario- 4 during
outage of randomly chosen lines (14e15) and line (25e27). This is
shown in Figs. 25 and 26 respectively.
 It is found that HTW system while operating in Scenario- 4
presents an improved voltage response than Scenario-3. But as
previously found for Scenario-2, WT system in Scenario- 4 de-
picts a superior voltage performance compared to HTW system
though the improvement is trivial.
 Therefore it may be concluded that, as far as maintaining the
voltage security issue is concerned, WT operation may be a
preferred option than HTW operation during substantial wind
penetration.
This leads us to reach at a concluding remark as mentioned
below.
6. Conclusion
Integration of hydro power plants into the existing wind-
thermal system would reduce the burden on thermal plant due
to the uncertainty posed by wind power. But inclusion of hydro
power generation with WT system in OPF framework, introduces
further complexities and real time constraints in the generation
scheduling. Under such an intricate operating scenario, the
 A. Panda et al. / Energy 124 (2017) 720e740 739

optimal operational paradigm has been evaluated using three

intelligent optimization techniques which can fulfill different ob- start
jectives. The results are validated in IEEE30 bus power network.
Among the three techniques i.e. GA, HA and MBFA, MBFA has
shown its superiority in demonstrating better and faster conver-
gence in achieving the operational objectives. Additionally, MBFA
initialize variable
optimized schedule has depicted supremacy in minimizing the
cost of operation, loss minimization and voltage security
enhancement as compared to other intelligent approaches. ‘El’ and ‘Di’
Therefore, with MBFA generation schedule, different wind pene- loop counter,
tration level and different combination of generation resources are m=m+1
considered so that the advantage of incorporating STATCOM in
operating the system in a voltage secure manner can clearly be
depicted. In order to further corroborate the analysis, under both
normal and stressed operating scenario the voltage security aspect m>E Terminate
of WT and HTW generation system is evaluated so that the best Yes
operating paradigm may be presented. As real time operational No
issues and constraints have been considered in the formulation
and evaluation of the objectives, it may assist the power system ‘Rpd’ loop
operator during planning and operation of multi source power counter,
system operation. The work may further be extended by inclusion s=s+1
of transient and dynamic stability aspects with other environ-
mental and spatial planning concerns.

APPENDIX s>R
Yes
A.1) No

a). Best set of parameters of MBFA used to obtain conver-

‘chemotactics’
gence solution. A loop counter,
1. No of bacteria ¼ 20. i=i+1
2. Run length unit ¼ 0.04.
3. No of chemotactic step ¼ 4.
Yes
4. Magnitude of attractant by a bacterium ¼ 1.08
5. Probability of elimination of bacteria ¼ 0.25
i > Nc
b). Parameters used in GA model.
1. No of individual in population ¼ 40. No
2. Probability of crossover ¼ 0.95.
3. Probability of mutation ¼ 0.05. Objective function
c) Parameters of HA, that synthesises BFA with the mutation F(k,i)
strategy of GA.
1. No of bacteria ¼ 20.
2. Run length unit ¼ 0.04.
3. No of chemotactic step ¼ 4.
F(k,i) < F(k,i-1)
4. Magnitude of attractant by a bacterium ¼ 1.08
No
5. Probability of elimination and dispersal of bacteria ¼ 0.2 Yes
6. Probabilities of Crossover and Mutation are same as those
of GA Sl=Sl+1

No
A.2) Flow chart of proposed optimization technique (MBFA).
Sl< Sml
Abbreviations used in the flow chart of MBFA.
Yes Tumble
El: Elimination loop No
Di: Dispersal loop k > NB
E: Number of elimination events
R: Number of reproduction events
Nc: Number of chemotactic processes
Sl: Swim length
A
Sml: Maximum swim length
NB: Number of bacteria
740 A. Panda et al. / Energy 124 (2017) 720e740

Appendix A.3) References

[1] Chang S, Chen C, Fung I, Luh PB. Hydroelectric generation scheduling with an
effective differential dynamic programming. IEEE Trans Power Syst 1990;5:
737e43.
Table A3
[2] Basu M. An interactive fuzzy satisfying method based on evolutionary pro-
Cost coefficients of thermal generating units.
gramming technique for multiobjective short-term hydrothermal scheduling.
Generator No Cost coefficients Electr Power Syst Res 2004;69:277e85.
[3] Lakshminarasimman L, Subramanian S. Short-term scheduling of hydro-
at bt ct Min limit (MW) Max limit (MW) thermal power system with cascaded reservoirs by using modified differential
evolution. IEE Proceedings Generation. Transm Distrib 2006;153:693e700.
1 0.00975 2.5 0 50 200 [4] Yuan X, Wang L, Yuan Y. Application of enhanced PSO approach to optimal
2 0.0175 1.75 0 20 80 scheduling of hydro system. Energy Convers Manage 2008;49:2966e72.
[5] Hota PK, Barisal AK, Chakrabarti R. An improved PSO technique for short term
optimal hydro-thermal scheduling. Electr Power Syst Res 2009;79:1047e53.
Appendix A 4) [6] Kumar VS, Mohan MR. A genetic algorithm solution to the optimal short-term
hydrothermal scheduling. Electr Power Energy Syst 2011;33:827e35.
[7] Mahor A, Rangnekar S. Short term generation scheduling of cascaded hydro
Nomenclature: electric system using novel self adaptive inertia weight PSO. Int J Electr Power
Energy Syst 2012;34:1e9.
[8] Nguyen TT, Vo DN. Modified cuckoo search algorithm for short-term hydro-
t, g Conventional/Thermal generating units thermal scheduling. Electr Power Energy Syst 2015;65:271e81.
r, w Renewable/Wind powered units. [9] Yuan X, Tian H, Yuan Y, Huang Y, Ikram RM. An extended NSGA-III for solution
Pgt Power output of tth thermal generator (MW). of multi-objective hydro-thermal-wind scheduling considering wind power
cost. Energy Convers Manag 2015;96:568e78.
WT Wind-Thermal [10] Dubey HM, Pandit M, Panigrahi BK. Hydro-thermal-wind scheduling
HTW Hydro-Thermal-Wind employing novel ant lion optimization technique with composite ranking
at, bt, ct Cost coefficients of tth thermal unit. index. Renew Energy 2016;99:18e34.
[11] Niknam T, Narimani MR, Jabbari M, et al. A modified shuffle frog leaping al-
Ct ðPgt Þ Cost of thermal power generation ($/h).
gorithm for multi objective optimal power flow. Energy 2011;53:6420e32.
dr Direct cost coefficient of the rth wind generator($/MWh). Elsevier.
Pwr Scheduled power output of rth wind unit. [12] Panda A, Tripathy M. Security constrained optimal power flow solution of
wind-thermal generation system using modified bacteria foraging algorithm.
Cwr ðPwr Þ Cost of purchase of wind power (Direct cost).
Energy 2015;95:816e27.
Pwr Scheduled power output of rth wind unit. [13] Panda A, Tripathy M. Optimal power flow solution of wind integrated power
Pwr;av Available wind power from rth wind powered generator. system using modified bacteria foraging algorithm. Elect Power Energy Syst
Pro Rated wind power. 2014;54:306e14. Elsevier.
[14] Jabr RA. BC Pal. Intermittent wind generation in optimal power flow dis-
Cp;wr Penalty cost function for under estimating the available patching. IET Gen. Trans Distrib 2009;3:66e74.
wind power. [15] Tsikalakis AG, Katsigiannis YA, Georgilakis PS, et al. Determining and
CR;wr Reserve cost function for over estimating the wind power. exploiting the distribution function of wind power forecasting error for the
economic operation of autonomous power system. IEEE Power Eng Soc
KPr Penalty cost coefficient for the rth wind generator ($/MWh). General Meet 2006.
KRr Reserve cost coefficient of rth wind generator ($/MWh). [16] Seguro JV, Lambert TW. Modern estimation of the parameters of the Weibull
fw ðwÞ Wind power probability density function (PDF). wind speed distribution for wind energy analysis. J Wind Eng Ind Aerodyn
2000;85:15e84.
k Shape parameter.(unit less) [17] Akpinar EK, Akpinar S. An assessment on seasonal analysis of wind energy
c Scale parameter.(unit same as wind speed) characteristics and wind turbine characteristics. Energy Convers Manag
v Wind speed. 2005;46:1848e67.
[18] Yanik S, Surer O, Oztaysi B. Designing sustainable energy regions using genetic
Pfc1 Penalty function to meet the real power line flow constraint.
algorithms and location-allocation approach. Energy 2016;97:161e72.
Pfc2 Penalty function to meet the bus voltage magnitude [19] Dizqah AM, Maheri A, Busawon K. An accurate method for the PV model
constraint. identification based on a genetic algorithm and the interior-point method.
Renew Energy 2014;72:212e22.
Pk Actual magnitude of real power for the kth line under
[20] Passino KM. Biomimicry of bacterial foraging for distributed optimization and
consideration. control. IEEE Control Syst Mag 2002;22:52e67.
Pmax , Pmin Maximum and minimum limiting values of real power [21] Tripathy M, Mishra S. Bacteria Foraging based solution to optimize both real
(p.u.). power loss and voltage stability limit. IEEE Trans Power Syst 2007;22:240e8.
[22] Pai MA. Computer methods in power system analysis. TMH Publishers; 1996.
Vk Voltage magnitude of the kth bus (p.u.). [23] Acha E, Claudio R. Fuerte-esquivel, Ambriz Perez, facts modeling and simu-
Vmax , Vmin Maximum and minimum values of bus voltage (p.u.). lation in power network. John Wiley & Sons Ltd; 2004.
PHkj Hydro power output PH of kth unit during jth scheduling [24] Arulampalam A, Barnes M, Jenkins N, et al. Power quality and stability
improvement of wind farm using STATCOM supported with hybrid battery
horizon. energy storage. IEEE Trans Gener Transm Distrib 2006;153:701e10.
min , P max The operating limit (upper and lower bounds)of
PHk Hk [25] Abd-Elazim SM, Ali ES. Optimal location of STATCOM in multi-machine power
hydro plant. system for increasing loadability by cuckoo search algorithm. Electr Power
Energy Syst 2016;80:240e51.
Vkj The operating volume of kth reservoir during jth scheduling [26] Abd-Elazim SM, Ali ES. Imperialist competitive algorithm for optimal STAT-
horizon. COM design in a multi-machine power system. Electr Power Energy Syst
qkj Water discharge of kth turbine. 2016;76:136e46.
[27] Acharjee P, Mallick S, Thakur SS, Ghoshal SP. Detection of maximum load-
t The water delay time between reservoir k and its upstream u at ability limits and weak buses using chaotic PSO considering security con-
interval j. straints. Chaos, Solit Fractals 2011;44:600e12.
Ru The set of upstream units directly above hydro-plant k. [29] Patel R. Wind and solar power systems. Boca Raton, FL: CRCPress; 1999.
[30] Ackermann T. Wind power in power system. John Wiley &Sons; 2005.
PHkj Hydro power generation from unit k at j th interval.
c1k ; c2k …:c6k Hydro power Coefficients.