Professional Documents
Culture Documents
ifs2F20172F32 32fifs 32 3 Ifs1523722fifs 32 Ifs152372 2
ifs2F20172F32 32fifs 32 3 Ifs1523722fifs 32 Ifs152372 2
net/publication/305627778
CITATIONS READS
130 1,599
3 authors:
Gevork B. Gharehpetian
Amirkabir University of Technology
1,100 PUBLICATIONS 21,535 CITATIONS
SEE PROFILE
All content following this page was uploaded by Hamid Reza Baghaee on 28 February 2017.
Abstract. In this paper, a multi-objective algorithm is presented for optimal power management and design of a hybrid
Wind/Photovoltaic/ generation system with hydrogen energy storage system including electrolyzer, fuel cell and hydrogen
tank to supply power demand in a microgrid system. The generation units are intrinsically non-dispatchable and moreover,
the major components of the system i.e. wind turbine generators, photovoltaic arrays and DC/AC converter may be subjected
to failure. Also, solar radiation, wind speed and load data are assumed to be entirely deterministic. The goal of this design is
to use a novel multi-objective optimization algorithm to minimize the objective functions i.e. annualized cost of the system,
loss of load expected and loss of energy expected and provide optimal energy management in the microgrid. The system
costs involve investment, replacement and operation and maintenance costs. Prices are all empirical and components are
commercially available. The simulation results for different cases reveal the impact of components outage on the reliability
and cost of the system. So, they are directly depends on component’s reliabilities, i.e. outages lead to need for a larger and
more expensive generation system to supply the load with the acceptable level of reliability. In addition, an approximate
method for reliability evaluation of hybrid system is presented which lead to reduce computation time. Simulation results
show the effectiveness of proposed multi-objective algorithm to solve optimal sizing problem in contrast with traditional
single objective methods.
Keywords: Microgrid, wind energy, photovoltaic, fuel cell, reliability, optimal power management and sizing, multi-objective
particle swarm optimization
1064-1246/17/$35.00 © 2017 – IOS Press and the authors. All rights reserved
1754 H.R. Baghaee et al. / Multi-objective optimal power management and sizing of a reliable wind
Fig. 1. Block diagram f a hybrid wind/Photovoltaic generation microgrid unit with hydrogen based ESS.
WG/PV units connected to a common DC bus and are successfully implemented for optimal sizing of
a combination of a Fuel Cell (FC) stack, an elec- hybrid stand-alone systems, assuming continuous
trolyzer, and a hydrogen storage tank as the Energy and reliable supply of load [7, 18, 26–28]. GA
Storage System (ESS). In recent years, hydrogen as a and preference-inspired co-evolutionary algorithm
suitable storage medium in renewable energy systems (PICEA) have been used in [3] and [17] to find
has become a matter of grave challenge [5–8]. optimal size of a wind/PV/battery system subject
Unlike diesel generator, hydrogen-based ESS are to reliability index of LPSP. However, they do not
emission free and they do not need any supply of fuel. consider outage probabilities of components such as
Moreover, hydrogen-based systems will be econom- wind turbines and PV arrays. In [14], a methodology
ically reasonable in future considering expectation of for calculation of economic costs of power inter-
increase in fuel price and extreme reductions in FC ruptions for different user sectors and interruption
costs [9–11]. Also, in hybrid PV/WG/diesel systems, durations has been developed.
it is not possible to store surplus during good sea- Reliability assessment is relevant for any engi-
sons. In contrary, in proposed hydrogen-based ESS, neering system [29]. But, previous researches do not
electrolyzer converts the excess energy into chemi- consider reliability issues in depth. For instance, some
cal form, i.e. produces hydrogen, and stores in the very given phenomena that may extremely affect sys-
hydrogen tank [12, 13]. tem’s reliability and cost, such as failures and outages
Because of intermittent characteristic of wind of generating units are usually ignored.
speed and radiation, most important challenge in Impacts of uncertainties in operating parameters
design of such systems is reliable supply of demand and reliability of a Proton Exchange Membrane Fuel
so, the goal is optimal design of a hybrid system for Cell (PEMFC) has been studied and it has been noted
reliable and economical supply of the load [7]. Litera- that, uncertainty and reliability shall be considered
ture offers a variety of methods for optimal designing in designing stage of any robust and applicable sys-
of hybrid PV/WG generating systems [1–4, 7, 10, tem and finally, it suggests that a stochastic modeling
12–28]. framework should be interfaced with a numerical
In [15], non-Linear programming (NLP) is used to optimization scheme [30]. This suggestion was a
find the optimal size and location of grid-connected great motivation to consider the impact of component
wind turbines. Simple iterative algorithms have been reliabilities, on economical design of stand-alone
used for optimal sizing of hybrid WG/PV system renewable systems. So, in [4], the outage probabili-
with battery ESS [12, 19–25]. Deficiency of power ties of PV arrays and WGs have been considered. But,
supply probability (DPSP) [19] and Loss of power further studies revealed that the availability of DC/AC
supply probability (LPSP) [23–25] have been used converter, as the only single cut-set [29] in reliability
as the technical criteria in iterative solution for opti- diagram, has an extreme influence over the system’s
mal sizing of hybrid WG/PV systems with battery reliability. Later, [28] investigated the problem more
ESS. In [16], Genetic Algorithm (GA) and in some carefully as a Single Objective Optimization (SOO)
other works, Particle Swarm Optimization (PSO) problem.
H.R. Baghaee et al. / Multi-objective optimal power management and sizing of a reliable wind 1755
α
hhub where, Ptank FC is the transferred power from the
vW = vmeasure
W × (3) hydrogen tank to the FC. Storage efficiency (hstorage )
hmeasure
may present losses resulted from leakage or pump-
where, α is the exponent law coefficient [3]. ing, and assumed to be equal to 95% [28, 36]. The
mass of stored hydrogen, at time step t, is calculated
2.3. Power generated by renewable Units as:
Etan k (t)
By adding WG and PV power outputs, injected mtan k (t) = (7)
power from renewable sources to DC bus is: HHVH2
fail fail fail where, the Higher Heating Value (HHV) of hydrogen
Pren nWG , nPV = NWG − nWG × PWG
is equal to 39.7 kWh/kg [28, 37]. Hydrogen pressure
drop, a small fraction of the hydrogen (here, 5%) may
fail
+ NPV − nPV × PPV (4)
not be extracted. This fraction is the lower limit of the
fail fail stored energy [28]. Therefore,
Where, NWG , NPV , nWG and nWG are number
of installed and failed WG turbines and PV panels, Etank,min ≤ Etank (t) ≤ Etank,max (8)
respectively [32]. At this point, generated power by
renewable sources distributes through two streams, 2.6. Fuel cell
one stream goes to DC/AC inverter to meet the load
(Pren−inv ), and other, transfers the extra power to A PEM fuel cell has reliable performance under
electrolyzer for hydrogen production (Pren−el ). intermittent supply and is commercially available
[33]. Its output power can be defined as a function
2.4. Electrolyzer of its input and efficiency (ηFC ) which is considered
to be constant (here, 50%) [9].
Most electrolyzers produce hydrogen at pressure
around 30 bars which can then be further compressed PFC− inv = Ptank−FC × ηFC (9)
for storage [28, 33]. But, the reactant pressures within
a Proton Exchange Membrane FC (PEMFC) are 2.7. DC/AC converter
around 1.2 bar (a bit higher than atmosphere pres-
sure) [34]. So, the electrolyzer output is directly The inverter’s losses can be presented by its effi-
injected to a hydrogen tank [8, 9, 18, 34, 35]. In ciency (ηinv ). The efficiency is roughly supposed to
this paper, the electrolyzer is directly connected to be constant (here, 90%) [9].
hydrogen tank, although, the developed software is
flexible to handle compressor model. When the tank Pinv− load = (PFC−inv + Pren−inv ) × ηinv (10)
is fully charged, compressors pump the hydrogen into
a second high-pressure tank. So, compressor does not 2.8. Operation strategy
work continuously and, it consumes lower energy.
Transferred power from electrolyzer to hydrogen is: Strategy of system operation is governed by the
Pel− tank = Pren−el × ηel (5) following rules:
(t)
where, ηel is the electrolyzer efficiency [9, 28]. • If Pren (t) = Pload
ηinv , then whole power gener-
ated by renewable sources is injected to the load
2.5. Hydrogen tank through inverter.
(t)
• If Pren (t) > Pload
ηinv then the surplus power is
The energy of hydrogen stored in the tank at time transferred to electrolyzer. If the injected power
step t can be obtained by the following equation: exceeds electrolizer’s rated power and the sur-
plus energy will circulate in a dump resistor.
(t)
Etan k (t) = Etan k (t − 1) • If Pren (t) < Pload
ηinv then shortage power will be
supplied by fuel cell. If shortage power exceeds
Ptan k−FC (t)
+ Pele−tan k (t) − × t fuel cell’s rated power or stored hydrogen cannot
ηstorage afford the shortage, some fraction of the load
(6) must be shaded. This fact leads to loss of load.
H.R. Baghaee et al. / Multi-objective optimal power management and sizing of a reliable wind 1757
Under all above conditions, components limits are • Loss of power supply probability [4, 32]
regarded and Equations (1–10) rein the system.
H
LOEE
LPSP = H
= E [LOE (h)] (15)
3. Reliability/cost assessment D (h) h=1
h=1
The microgrid system is simulated over a year with
where, D(h) is the load demand (kWh) at time step h.
1-h time steps and reliability/cost assessment studies
are carried out. Then, using economic factors, results • Equivalent loss factor [4, 32]
are expanded to the 20-year period of a microgrid
lifetime. Load growth and uncertainty in load, solar H
radiation and wind speed are neglected. 1 E [Q (h)]
ELF = (16)
H D (h)
h=1
3.1. Reliability Indices
In all above equations, H is the number of time
Several reliability indices introduced before [17, steps in which system’s reliability is evaluated (here,
32, 33, 38, 39]. LOLE, LOEE or Expected Energy not H = 8760). The ELF is the ratio of the effective
Supplied (EENS), LPSP, and Equivalent Loss Factor forced outage hours to the total number of hours
(ELF) are some of the most common used indices in [33]. Therefore, ELF is chosen as the main reliability
the reliability evaluation of generating systems. From constraint of the optimization problem. In developed
these, LOLE is a loss of load index, whereas others countries, electricity suppliers aim at ELF <0.0001.
belong to category of loss of energy indices. On the other hand, in rural areas and stand-alone
applications (this study), ELF <0.01 is acceptable
• Loss of load expected [4, 32] [33].
H
3.2. System’s reliability model
LOLE = E [LOL (h)] (11)
h=1
In this study, outages of PV arrays, wind turbine
where, E[LOL(h)] is the expected value of loss of generators, and DC/AC converter are considered.
load at hth time step defined by [4, 32]: Forced Outage Rate (FOR) of PVs and WGs is
assumed to be 4% [28, 39] and they will be available
E [LOL] = T (s) × f (s) (12) with a probability of 96%. Probability of encoun-
s∈S tering each state is calculated through binomial
fail
In the above equation, f (s) is the probability of distribution function [40]. For example, given nWG
fail
encountering the state s, and T (s) is the loss of load out of total NWG installed WGs, and nPV out of total
duration (h), given that the occurring state s and S is NPV installed PV arrays are failed, the probability of
the set of all possible states. encountering this state is calculated as follows:
• Loss of energy expected (expected energy not fail fail
fren nWG , nPV
supplied [4, 32]
NWG fail
NWG −nWG fail
H = fail × AWG × (1 − AWG ) nWG
nWG
LOEE = EENS = E [LOE (h)] (13)
h=1
NPV NPV −nPV
fail fail
where, E[LOE(h)] is the expected value of loss of × fail × APV ×(1 − APV )
nPV
(17)
nPV
energy, or energy not supplied, at time step h:
where, AWG and APV are availabilities of each WG
E [LOE] = Q (s) × f (s) (14)
and PV array. Since, other components do not have
s∈S
moving parts and are installed indoors, their out-
Here, Q(s) is the amount of loss of energy (kWh) age probabilities, in comparison with WGs and PV
when system encounters state s. arrays, are much lower and therefore negligible. But,
1758 H.R. Baghaee et al. / Multi-objective optimal power management and sizing of a reliable wind
0.04
(21)
0.02
Ultimately, it can be proved that:
15
0 10
eq
Pren = NWG × PWG × AWG + NPV × PPV × APV
200
150 5
(22)
100
50
0
# of failed Wind Turbines However, Equation (22) could be logically antic-
0
# of failed PV arrays
a) ipated. Results, presented in the following section,
probability will validate the acceptability of the proposed approx-
15
imate method. By using this method, number of all
possible states for outages of 14 WGs and 199 PV
arrays at each time step, reduces from 6000 to only
10
2 states.
# of failed PV array s
4. Problem statement
5
Also, CRF and K are capital recovery factor [3] Minimize: f (x) = [f1 (x) , f2 (x) , . . . , fM (x)]T (36)
and single payment present worth [16], respectively:
gj (x) ≤ 0; i = 1, 2, . . . J
ir × (1 + ir)R Subject to: (37)
CRF (ir, R) = (25) hk (x) = 0; i = 1, 2, . . . K
(1 + ir)R − 1
where, fi (x) is the ith objective function, gj (x) is the
yi jth inequality constraint and hk (x) is the kth equality
1
Ki (ir, Li , yi ) = (26) constraint [45, 47, 48]. The constraints given by (33)
(1 + ir)n×Li
n=1 define feasible region and any point x in defines a
where, L and y are useful lifetime and number of feasible solution. The vector f (x) is a function, which
replacements of the component during useful lifetime maps the set into the set and represents all pos-
of the project, respectively [2]. Using Equation (13), sible values of the objective functions. The notion
annual cost of loss of load is calculated by: of an optimum solution in MOO is different than
SOO, the concept of Pareto dominance formulated
ACloss = LOEE × Closs (27) by Vilfredo Pareto is used as [45, 47, 48]: A vector
where, Closs is cost of customer’s dissatisfaction u = [u1 , u2 , . . . , uM ]T is said to dominate a vec-
(US$/kWh) described in Section III. Finally, the tor v = [v1 , v2 , . . . , vM ]T (denoted by u≺v), for a
objective functions are defined as: MOO problem, if:
∀i ∈ {1, . . . , M} , ui ≤ vi
JCost = min ACi (28)
X ∧ ∃i ∈ {1, . . . , M} : ui < vi (38)
i
JRe liability1 = min {LOEE} (29) where, M is the dimension of the objective space. A
X
solution u ∈ U, where U is the universe, is said to be
JRe liability2 = min {LOLE} (30) Pareto Optimal if and only if, there exists no other
X
solution v ∈ U, that u is dominated by v. Such solu-
(31) tions (u) are called non-dominated solutions. The set
i is the component indicator, and X is a seven dimen- of all non-dominated solutions constitutes the Pareto
sional vector consisting of optimization variables. set or non-dominated set [45, 47, 48].
The optimization problem is subjected to:
5.2. Particle swarm optimization
E [ELF ] ≤ ELFmax (32)
Ni ≥ 0 (33) The PSO is a population-based algorithm that
π exploits a population of individuals to probe promis-
0 ≤ θPV ≤ (34) ing region of the search space [45, 50–53]. Suppose
2
that the search space is n-dimensional, then ith
Etan k (0) ≤ Etan k (8760) (35) particle is a n-dimensional vector, Xi = [xi1 , xi2 ,
where, θP V is array’s installation angle, and con- . . . , xin ]T and velocity Vi = [vi1 , vi2 , . . . , vin ]T ,
straint (34) ensures that the amount of stored energy where i = 1, 2, . . . , N and N is size of popula-
in the hydrogen tank at the end of first year will be tion. In PSO, the particle i remembers the best
more than its initial amount (worst condition). position it visited so far (Pbest ), referred to as
Pi = [pi1 , pi2 , . . . , pin ]T , and best position of best
particle in the swarm (Gbest ) is referred as Gi =
5. Multi Objective Particle swarm [g1 , g2 , . . . , gn ]T . Each particle i adjusts its position
optimization (MOPSO) in the next iteration t + 1, with respect to following
equations [45, 50, 53]:
5.1. Multi-objective optimization
Vi (t + 1) = ω (t) × Vi (t) + c1 × r1
A general minimization problem of M objectives × (Pi (t) − Xi (t)) + c2 × r2
can be stated as: given x = [x1 , x2 , . . . , xd ]T , where
× (Gi (t) − Xi (t)) (39)
d is the dimension of the decision variable space
[45, 46]: Xi (t + 1) = Xi (t) + χ × Vi (t + 1) (40)
H.R. Baghaee et al. / Multi-objective optimal power management and sizing of a reliable wind 1761
where ω(t) is inertia coefficient. The factor χ, is used such a problem, the calculation for the distance of the
to limit velocity (here, χ = 0.7). c1 , and c2 denote nearest neighbor is done in an order such that, the dis-
the cognitive and social parameters and r1 and r2 are tance of a solution to its nearest neighbor which has
random real numbers drawn from interval [0, 1] [45, not already been considered, is taken. For example,
50]. More details have been provided in [45]. if the nearest neighbor distance is computed in the
order a, b, c, d, e, f, g and h then, for the nearest
5.3. Multi-objective particle swarm optimization neighbor of solution f , solution g should be consid-
ered since, its distance to solution e has already been
To make the PSO algorithm capable of dealing with considered [45, 46].
MOO problems, some modifications become neces-
sary because PSO is an inherent SOO algorithm. In 5.4. Best tradeoff solution
[46], the personal best performance (Pbest ) of each
individual particle is replaced with the new solution Because none of Pareto optimal set solutions is
if and only if it dominates the former Pbest . Also, the absolute global optimal, the designer should
two major issues should be considered in the updat- choose the most proper solution according. To han-
ing process of the global best performance (Gbest ). dle this dilemma fuzzy sets are proposed [45]. In this
First, the fitness assignment and selection should be approach, a linear membership function is defined for
addressed such that a search can move towards the each objective function (Ji ) as follows:
Pareto optimal set. Second, the diversity of swarm
Ji − Jimin
should be maintained to prevent premature conver- ui = (41)
gence and obtain an evenly distributed Pareto optimal Jimax − Jimin
front [45, 46]. Here, the authors employ an archiving It is evident that the lower values of the member-
mechanism proposed in [45] to form a repository, ship function indicate more degrees of achievement
which may contain only a limited number of solu- of the objective function. For every non-dominated
tions. The density parameter (deni ) of the solution solution (k), the aggregate membership function can
i defined as its distance to the nearest neighbor in be defined as follows:
the archive is introduced as a measure of the diver-
M
sity of archive. However, it may happen that two
Uk = uki (42)
solutions are the closest to each other. For instance,
i=1
for a two-objective (f1 and f2 ) minimization prob-
lem illustrated in Fig. 5, solutions e and f are closest The solution with the minimum membership, U k
to each other and therefore, will get the same value can be considered as the best compromise solution
for the distance to their nearest neighbor. To prevent (Best Tradeoff I). Here, the authors propose another
criterion to choose the best compromised solution.
First, the distance of each non-dominated solution
(k) from the origin in the fuzzified coordinate, is
calculated by the following equation:
1/2
M 2
Vk = uki (43)
i=1
6. Simulation results
600
6.1. Base case
500
Table 1
Components’ specifications [2, 9, 26, 28, 29]
Component Capital cost Replacement Cost O&M Life (yr) Efficiency Availability Unit
(US$/unit) (US$/unit) (US$/unit-yr)
WG 19400 1500 75 20 – 96 7.5 kW
PV panel 7000 6000 20 20 – 96 1 kW
electrolyzer 2000 1500 25 20 0.75 100 1 kW
Hydrogen storage tank 1300 1200 15 20 0.95 100 1 kg
Fuel cell 3000 2500 175 5 0.50 100 1 kW
inverter 800 750 8 15 0.90 99.89 1 kW
Table 2
Optimal combination in base case for hybrid, Only WG and Only PV Systems
Type of the system Tradeoff solution Optimal Combination
NWG NPV Pel (kW) Mtank (kg) PFC (kW) Pinv (kW) θPV
Hybrid System Best tradeoff I 9 251 126.760 192.235 48.152 49.832 29.994
Best tradeoff II 9 251 127.019 193.063 49.030 50.504 29.970
Minimum Cost 9 251 126.155 191.222 45.489 48.939 30.075
Minimum LOEE 9 251 127.533 193.817 50.9084 51.486 29.909
Minimum LOLE 9 251 127.533 193.817 50.9084 51.486 29.909
Only WG Best tradeoff I 204 – 354.584 1009.295 55.993 53.902 –
Best tradeoff II 204 – 354.584 1009.295 55.993 53.902 –
Minimum Cost 204 – 354.610 1009.212 55.943 53.872 –
Minimum LOEE 204 – 354.584 1009.295 55.993 53.902 –
Minimum LOLE 204 – 354.584 1009.295 55.993 53.902 –
Only PV Best tradeoff I – 340.016 167.665 200.981 49.019 49.816 57.230
Best tradeoff II – 340.040 167.642 201.871 49.683 50.146 57.193
Minimum Cost – 340.298 167.687 193.405 45.540 47.930 57.418
Minimum LOEE – 339.936 167.003 212.213 55.778 53.037 56.862
Minimum LOLE – 339.936 167.003 212.213 55.778 53.037 56.862
been solved assuming 100% availability of all com- 6.3. Impact of DC/AC convertor on the system’s
ponents. As shown in Tables 5 and 6, comparing reliability
to the base case, in this case a smaller system
can supply the load with lower costs and higher To understand the impact of inverter on the sys-
reliability. In fact, as a consequence of failure prob- tem’s reliability, in this section the optimization
abilities of WGs, PV arrays, and DC/AC converter, problem has been subject to reliability constraint of
annual cost of the system increases from 3,134,902 ELF <0.0001, i.e. reliability constraint in developed
to US$3,183,155/yr, for hybrid WG/PV system, from countries [33]. The program has been run and actu-
7077828 to US$7222327/yr, for only WG system, ally, this huge fitness value is resulted from the heavy
and from 3703402 to US$3792828/yr, for only PV penalty term of fitness function assigned to reliabil-
system, respectively. Additionally, components’ out- ity inequality constraint. Taking a closer look at the
age probability results in increase about US$64809 result reveals that the ELF of the system can fall even
for hybrid WG/PV, US$182075 for only WG, and to zero. This fact, directly results from the vital role
US$89930 for only PV systems, respectively, in of the inverter, as the only single cut-set, in reliabil-
investment cost of the system that might be con- ity diagram of the hybrid system. Given Psupply , Ainv ,
siderable. Besides, it can be observed that failure and Aothers are, respectively, probability of supplying
probabilities cause deterioration in system’s reliabil- the load, availability of the inverter, and availability
ity, i.e. ELF index from 0 to 0.00005850 for hybrid of all other components as well as energy resources,
WG/PV system, and from 0.00001486 to 0.00005385 the upper limit of the probability of supplying the
for only PV system, respectively. Also, since in this load is:
case all the components are assumed 100% available,
Psup ply = Aothers × Ainv Lim Psup ply = Ainv
both accurate and approximate methods provide same ⇒
results. In fact, at each time step, system can work Aother→1
only in one feasible state. (44)
1764
Table 3
Optimal costs and reliability indices in base case for Hybrid, Only WG and Only PV Systems
Type of the system Tradeoff solution Reliability Indices Costs Etank
LOLE (h/yr) LOEE (MWh/yr) ELF LPSP Investment Cost (MUS$) JCOST (US$/yr) ACloss (US$/yr) Etank (0) Etank (8760)
Hybrid System Best tradeoff I 34 25 0.00006196 0.00008813 3183155 2592947.76 228.7388 3761.05 6923.076
Best tradeoff II 9 5.848345 0.0000147 0.00002173 3195813 2597714.82 32.73018 3777.48 6960.692
Minimum Cost 154.937751 229.4364754 0.00060050 0.00085257 3149071 2581715 1285.336 3740.933 6660.267
Minimum LOEE 0.518042 0.13885044 3.4087e-13 5.1596e-13 3220598 2606141 0 3792.438 6777.453
Minimum LOLE 0.518042 0.13885044 3.4087e-13 5.1596e-13 3220598 2606141 0 3792.438 6777.453
Only WG Best tradeoff I 0 0 0 0 7222327 6170553.45 0 19979.7 31875.62
Best tradeoff II 0 0 0 0 7222327 6170553.45 0 19979.7 31875.62
Minimum Cost 0 0 0 0 7221651 6170322 0 19978.03 31872.99
Minimum LOEE 0 0 0 0 7222327 6170553 0 19979.69 31876.28
Minimum LOLE 0 0 0 0 7222327 6170553 0 19979.69 31876.28
Only PV Best tradeoff I 30 21.36 0.00005043 0.000075 3792828 3.16E+06 210.4686 3934.65 7454.67
Best tradeoff II 15 10.2 0.00002523 0.00003214 3802057 3166882.15 82.49374 3952.31 7490.008
Minimum Cost 227.081825 361.30623035 0.00094232 0.00134260 3739635 3141763 2026.47 3784.262 7150.989
Minimum LOEE 0 0 0 0 3881517 3192644 0 4157.607 7897.704
Minimum LOLE 0 0 0 0 3881517 3192644 0 4157.607 7897.704
Table 4
Optimal Combination when all components are 100% available for Hybrid, Only WG and Only PV Systems
Type of the system Tradeoff solution Optimal Combination
NWG NPV Pel (kW) Mtank (kg) PFC (kW) Pinv (kW) θPV
Hybrid System Best tradeoff I 7 246 127.259 187.027 50.692 50.012 33.0616
Best tradeoff II 7 246 127.259 187.027 50.692 50.012 33.0616
Minimum Cost 7 246 127.362 186.989 50.931 50.180 33.025
Minimum LOEE 7 245 127.199 189.839 53.896 51.141 32.251
Minimum LOLE 7 245 127.199 189.839 53.896 51.141 32.251
Only WG Best tradeoff I 193 – 355.709 1002.344 62.006 54.744 –
Best tradeoff II 193 – 355.709 1002.344 62.006 54.744 –
Minimum Cost 193 – 353.766 999.374 56.303 52.548 –
Minimum LOEE 193 – 355.709 1002.344 62.006 54.744 –
H.R. Baghaee et al. / Multi-objective optimal power management and sizing of a reliable wind
7000
3.35
6000
3.3
Etank [k W]
5000
3.25
3.2 4000
3.15
3000
3.1
300 2000 E0tank
200
200 Etank
150
100 100 1000
50 0 1000 2000 3000 4000 5000 6000 7000 8000 9000
Loss of Energy Expected (MWh/yr) 0 time [hour]
0
Loss of Load Expected (h/yr)
a)
Fig. 8. 3D Pareto optimal sets and best tradeoffs in base case for x 10
4
7.36
Etank [k W]
2.5
7.34
2
Annual Cos t (M$)
7.32
7.3 1.5
7.28
1
7.26
0.5 E0tank
7.24
Etank
7.22
0
1 0 1000 2000 3000 4000 5000 6000 7000 8000 9000
0.5 1 time [hour]
0
0
0.5
b)
-0.5 -0.5
Loss of Energy Expected (MWh/yr) -1 -1
Loss of Load Expected (h/yr) 8000
Fig. 9. 3D Pareto optimal sets and best tradeoffs in base case for 7000
only WG system.
6000
3D Pareto Optimal Set
Etank [k W]
5000
3.95
4000
3.9
Annual Cos t (M$)
3.85 3000
E0tank
Etank
3.8
2000
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
3.75 time [hour]
3.7 c)
400
300 250
200
200 150
100 100 Fig. 11. Hourly expected amount of stored energy in the hydrogen
50
Loss of Energy Expected (MWh/yr) 0 0 tank during a year for a) hybrid WG/PV b) only Wg and c) only
Loss of Load Expected (h/yr)
PV systems.
Fig. 10. 3D Pareto optimal sets and best tradeoffs in base case for
only PV system.
1766
Table 5
Optimal costs and reliability indices when all components are 100% available for Hybrid, Only WG and Only PV Systems
Type of the system Tradeoff solution Reliability Indices Costs Etank
LOLE (h/yr) LOEE (MWh/yr) ELF LPSP Investment Cost (MUS$) JCOST (US$/yr) ACloss (US$/yr) Etank(0) Etank(8760)
Hybrid System Best tradeoff I 0 0 0 0 3134902 2528139.39 0 3657.67 6853.478
Best tradeoff II 0 0 0 0 3134902 2528139.39 0 3657.67 6853.478
Minimum Cost 0 0 0 0 3130699 2522146 304.5668 3656.91 6630.147
Minimum LOEE 0 0 0 0 3182522 2547590 0 3713.481 6731.608
Minimum LOLE 0 0 0 0 3182522 2547590 0 3713.481 6731.608
Only WG Best tradeoff I 0 0 0 0 7077828 5988477.93 0 19841.7 31542.21
Best tradeoff II 0 0 0 0 7077828 5988477.93 0 19841.7 31542.21
Minimum Cost 0 0 0 0 7001743 5961866 0 19782.75 31422.36
Minimum LOEE 0 0 0 0 7077828 5988478 0 19841.7 31542.21
Minimum LOLE 0 0 0 0 7077828 5988478 0 19841.7 31542.21
Only PV Best tradeoff I 9.1 6.2 0.00001484 0.000023 3703402 3073583.2 228.61105 3416.88 6420.145
Best tradeoff II 7 4.9 0.00001221 0.00002008 3706363 3075210.28 28.15963 3429.23 6444.847
Minimum Cost 6 5.02850497 0.00001243 0.00001886 3691224 3065754 639.0887 3442.451 6468.211
Minimum LOEE 6 4.86651195 0.00001203 0.00001808 3706752 3075339 135.3972 3428.188 6439.889
Minimum LOLE 5 3.02194714 0.00000075 0.00001123 3710460 3076082 149.1336 3409.787 6403.086
H.R. Baghaee et al. / Multi-objective optimal power management and sizing of a reliable wind
Table 6
Optimal Combination when Ainv = 1 and ELFmax = 0.0001 for Hybrid, Only WG and Only PV Systems
Type of the system Tradeoff solution Optimal Combination
NWG NPV Pel (kW) Mtank (kg) PFC (kW) Pinv (kW) θPV
Hybrid System Best tradeoff I 7 252 128.598 190.624 47.931 49.635 31.305
Best tradeoff II 7 252 128.6 190.635 47.969 49.658 31.305
Minimum Cost 7 252 128.586 190.62 47.884 49.612 31.302
Minimum LOEE 7 252 128.599 190.637 47.971 49.659 31.304
Minimum LOLE 7 252 128.598 190.637 47.97 49.659 31.304
Only WG Best tradeoff I 200 – 357.205 1000.669 49.013 49.8 –
Best tradeoff II 200 – 357.439 1000.507 49.307 50.27 –
Minimum Cost 200 – 357.050 1000.460 47.745 49.578 –
Minimum LOEE 200 – 357.598 1000.699 52.039 50.253 –
Minimum LOLE 200 – 357.594 1000.667 52.038 50.327 –
Only PV Best tradeoff I – 346 172.158 167.281 50.744 50.147 57.224
Best tradeoff II – 346 172.601 166.431 51.67 50.487 57.162
Minimum Cost – 345 170.38 173.483 48.925 49.063 57.314
Minimum LOEE – 347 173.19 164.526 52.331 50.872 57.125
Minimum LOLE – 347 173.19 164.526 52.331 50.872 57.125
H.R. Baghaee et al. / Multi-objective optimal power management and sizing of a reliable wind
1767
1768
Table 7
Optimal costs and reliability indices when Ainv = 1 and ELFmax = 0.0001 for Hybrid, Only WG and Only PV Systems
Type of the system Tradeoff solution Reliability Indices Costs Etank
LOLE (h/yr) LOEE (MWh/yr) ELF LPSP Investment Cost (MUS$) JCOST (US$/yr) ACloss (US$/yr) Etank(0) Etank(8760)
Hybrid System Best tradeoff I 44.30959 32.15181 0.00008182 0.000119 3176238 2588307 182.5857 3729.073 6712.936
Best tradeoff II 42.30947 30.68983 0.000078068 0.000114 3176722 2588459 174.055 3729.278 6713.378
Minimum Cost 48.55628 34.04078 0.000086681 0.000126 3175652 2588120 193.934 3728.985 6712.557
Minimum LOEE 42.30947 30.60183 0.000077843 0.000114 3176750 2588467 173.693 3729.318 6713.428
Minimum LOLE 42.30947 30.6298 0.000077915 0.000114 3176741 2588464 173.895 3729.323 6713.429
Only WG Best tradeoff I 9 13.806811 4.0083175e-05 5.1305238e-5 7069553 6082159 58.973 19808.46 31442.89
Best tradeoff II 6 8.828806 2.190914e-05 3.280728e-5 7073804 6083675 48.193 19805.23 31436.44
Minimum Cost 46 37.401437 0.00010329 0.000139 7054331 6077595 204.306 19804.3 31434.57
Minimum LOEE 1 0.9480587 2.254001e-06 3.5229256e-6 7105097 6092423 5.309 19809.06 31444.09
Minimum LOLE 1 0.9487957 2.255754e-06 3.5256647e-6 7105132 6092430 5.313 19808.42 31442.81
Only PV Best tradeoff I 4 2.12976 5.246075e-06 7.91405e-06 3815387 3176132 11.969 3265.707 6115.158
Best tradeoff II 0 0 0 0 3825983 3178964 0 3248.838 6081.447
Minimum Cost 33. 26.94925 6.68782e-05 0.0001 3791526 3167315 153.9 3388.825 6361.156
Minimum LOEE 0 0 0 0 3839682 3186956 0 3211.028 6006.041
Minimum LOLE 0 0 0 0 3839682 3186956 0 3211.028 6006.041
H.R. Baghaee et al. / Multi-objective optimal power management and sizing of a reliable wind
Table 8
Comparison of the accurate method with the approximate method for Hybrid, Only WG and Only PV Systems
Type of the system Tradeoff solution Method Reliability Indices Costs
LOLE (h/yr) LOEE (MWh/yr) ELF LPSP Investment Cost (MUS$) JCOST (US$/yr) ACloss (US$/yr)
Base case
Hybrid System Best tradeoff I Accurate 29 21.5 5.38e-05 7.99e-05 3792827.535 3163513.154 203.6610
Approx. 30 36.4 0.00011 0.000135 3792827.535 3160000 120.4686
Error (%) –3.48 –69.06 –105.08 –69.06 0 0 40.848472
Best tradeoff II Accurate 9.12 5.84 1.47e-05 2.17e-05 3195813.245 2597714.822 32.7507328
Approx. 9 5.85 1.47e-05 2.17e-05 3195813.245 2600000 32.73017597
Error (%) 1.29 –0.063 –0.0619 –0.06281 0 0 0.06276754
Only WG Best tradeoff I Accurate 0 0 0 0 7222326.555 6170553.454 0
Approx. 0 0 0 0 7222326.555 6170000 0
Error (%) 0 0 0 0 0 0 0
Best tradeoff II Accurate 0 0 0 0 7222327 6170553 0
Approx. 0 0 0 0 7222327 6170553 0
Error (%) 0 0 0 0 0 0 0
Only PV Best tradeoff I Accurate 29 21.5 5.38e-05 7.99e-05 3792827.535 3163513.154 203.661
Approx. 30 36.4 0.000110 0.00013 3792827.535 3160000 120.4686
Error (%) –3.45 –69.06 –105.08 –69.06 0 0 40.8485
Best tradeoff II Accurate 14 9.37 2.33e-05 3.48e-05 3802057.042 3166882.15 90.634
Approx. 15 16.2 4.92e-05 6.01e-05 3802057.042 3170000 52.494
Error (%) –7.14 –72.66 –111.23 –72.66 0 0 42.081
Approx.
100% Available Components
Hybrid System Best tradeoff I Accurate 0 0 0 0 3134902 2528139 0
Approx. 0 0 0 0 3134902 2528139 0
Error (%) 0 0 0 0 0 0 0
Best tradeoff II Accurate 0 0 0 0 3134902 2528139 0
Approx. 0 0 0 0 3134902 2528139 0
Error (%) 0 0 0 0 0 0 0
Only WG Best tradeoff I Accurate 0 0 0 0 7077828 5988478 0
Approx. 0 0 0 0 7077828 5988478 0
Error (%) 0 0 0 0 0 0 0
Best tradeoff II Accurate 0 0 0 0 7077828 5988478 0
Approx. 0 0 0 0 7077828 5988478 0
Error (%) 0 0 0 0 0 0 0
H.R. Baghaee et al. / Multi-objective optimal power management and sizing of a reliable wind
Table 9
Comparison of optimal combination between proposed multi-objective PSO and single objective algorithm presented in [28] for Hybrid
System
Optimization Method Tradeoff solution Optimal Combination
NWG NPV Pel (kW) Mtank (kg) PFC (kW) Pinv (kW) θPV
Base Case
Proposed MOPSO Best tradeoff I 7 246 127.259 187.027 50.692 50.0121 33.0616
Best tradeoff II 7 246 127.259 187.027 50.692 50.0121 33.0616
SOO – 8 224 119.44 144.19 43.431 43.725 34.129
100% Available Components
Proposed MOPSO Best tradeoff I 7 246 127.259 187.027 50.692 50.0121 33.0616
Best tradeoff II 7 246 127.259 187.027 50.692 50.0121 33.0616
SOO – 9 214 117.27 127.1 43.736 46.464 33.993
Table 10
Comparison of results between proposed multi-objective PSO and single objective algorithm presented in [28] for Hybrid System
Optimization Method Tradeoff solution Reliability Indices Costs
LOLE LOEE ELFE LPSP JCOST Investment Cost ACloss
(h/yr) (MWh/yr) (US$/yr) (MUS$) (US$/yr)
Base Case
Proposed MOPSO Best tradeoff I 34 25 0.00006196 0.00008813 3183155 2592947.76 228.7388
Best tradeoff II 9 5.848345 0.00001473 0.00002173 3195813 2597714.82 32.73018
SOO – 335.96 2.372667 0.00836036 0.00881527 250410 2321000 13287
100% Available Components
Proposed MOPSO Best tradeoff I 0 0 0 0 3134902 2528139.39 0
Best tradeoff II 0 0 0 0 3134902 2528139.39 0
SOO – 303 2202768.7 0.0078850 0.00818404 242648 2247700 12335
Usually, there are two remedies for improving the the approximate method estimates reliability of the
reliability limit of such a hybrid system. The first, base case system somehow optimistically.
using a more reliable (more available) inverter, and
the second, using two or more inverters in parallel. For 6.5. Comparison of proposed MOPSO with SOO
example, in case of using a 100% available inverter, algorithm
the software yields Tables 6 and 7.
To show the efficiency of proposed algorithm,
6.4. Comparison of the accurate method with the the multi-objective particle swarm optimization pre-
approximate method sented in this paper has been compared with the
conventional single objective particle swarm opti-
To compare accurate method with the approx- mization algorithm proposed in [28]. The results have
imate method, the approximate method has been been presented in Tables 9 and 10. It can be seen
applied to the problem. In this case, optimization that the proposed MOPSO algorithm offers better
process has been terminated in about 7 h which is combination with lower reliability indices.
considerably less than 20 h. Results are presented in
Table 8. Since, the deference between approximate
and accurate method is only in reliability calculations, 7. Conclusion
both of them result in same investment, replacement,
and operation and maintenance costs. Also, Table 8 The main goal to propose optimal power man-
indicates that the approximate reliability indices are agement and sizing design of hybrid wind–solar
acceptably close to actual indices provided by the generating microgrid systems is a reliable supply
accurate method. It is observed that, the magnitudes of the load under varying weather conditions, with
of the percentage errors are almost below 10%. Addi- minimum cost. In this paper, a hybrid WG/PV/FC
tionally, negative percentage errors are evidences that microgrid system with hydrogen ESS is designed
H.R. Baghaee et al. / Multi-objective optimal power management and sizing of a reliable wind 1771
for a 20-year period of operation. Optimal combina- [3] Z. Shi, R. Wang and T. Zhang, Multi-objective opti-
tion of components is achieved by a novel MOPSO mal design of hybrid renewable energy systems using
preference-inspired coevolutionary approach, Solar Energy
algorithm, which acceptably converges to the opti- 118 (2015), 96–106.
mum combination. The MOO problem involves cost [4] A. Kashefi Kaviani, G.H. Riahy and S.H.M. Kouhsari,
and reliability indices as objective functions and also Optimal design of a reliable hydrogen-based stand-alone
subjected to reliability constraint. Results indicate wind/PV generation system, In: Proceeding of 11th Inter-
national Conference on Optimization of Electrical and
that costs of the system, directly, depend on its com- Electronic Equipment (OPTIM’08), Brasov, Romania,
ponents’ reliabilities and the outage probabilities of 2008.
three major components of the system, i.e. WG, [5] M.H. Athari and M.M. Ardehali, Operational performance
PV and DC/AC converter are taken into account. of energy storage as function of electricity prices for on-grid
hybrid renewable energy system by optimized fuzzy logic
Comprehensive reliability/cost assessment of such a controller, Renewable Energy 85 (2016), 890–902.
system, considering failures of other components as [6] C. Bussar, P. Stöcker, Z. Cai, L. Moraes Jr, R. Alvarez, H.
well as uncertainty in wind speed, solar radiation, and Chen, C. Breuer, A. Moser, M. Leuthold and D.U. Sauer,
load data, demands computationally intensive and Large-scale integration of renewable energies and impact
on storage demand in a european renewable power system
time consuming algorithms like Monte Carlo simu- of 2050, Energy Procedia 73 (2015), 145–115.
lations which is beyond the aspects of this paper and [7] A. Kashefi Kaviani, H.R. Baghaee and G.H. Riahy, Design
may be subject of future studies. Versatile software, and optimal sizing of a photovoltaic/wind-generator sys-
developed in MATLAB programming environment, tem using particle swarm optimization, In: Proceedings
of the 22nd power system conference (PSC), Tehran, Iran,
carries out all these huge computations, including 2007.
yearly simulation of the microgrid system with 1- [8] M.J. Khan and M.T. Iqbal, Dynamic modeling and sim-
h time steps, accurate and approximate evaluations ulation of a small wind-fuel cell hybrid energy system,
Renewable Energy 30 (2005), 421–439.
of reliability indices, and MOO algorithm. The soft-
[9] B. Zakeri and S. Syri, Electrical energy storage systems: A
ware is capable of integrating any component model comparative life cycle cost analysis, Renewable and Sus-
and therefore, quite flexible to be implemented to tainable Energy Reviews 42 (2015), 569–596.
any application. It is just needed to input the wind [10] R. Belfkira, L. Zhang and G. Barakat, Optimal sizing
speed, solar radiation, and load demand data, as well study of hybrid wind/PV/diesel power generation unit, Solar
Energy 85(1) (2011), 100–110.
as specifications of the microgrid components and [11] R. Dufo-López and J.L. Bernal-Agustı́n, Multi-objective
then, running the software. Results of a case study, design of PV–wind–diesel–hydrogen–battery systems,
based on the empirical data, testify that component Renewable Energy 33(12) (2008), 2559–2572.
outages can extremely impact system’s reliability and [12] A. Mills and S. Al-Hallaj, Simulation of hydrogen-based
hybrid systems using hybrid2, Int J Hydrogen Energy 29
economy. Moreover, this study reveals the signifi- (2004), 991–999.
cant impact of DC/AC converter, as the only single [13] A. Rajabi-Ghahnavieh and S.A. Nowdeh, Optimal PV–FC
cut-set in the reliability diagram of the hybrid sys- hybrid system operation considering reliability, Interna-
tional Journal of Electrical Power & Energy Systems 60
tem. It is observed that the inverter’s reliability is
(2014), 325–333.
an upper limit for the system’s reliability. Also, an [14] W.D. Kellogg, M.H. Nehrir, G. Venkataraman and V. Gerez,
approximate method for reliability evaluation of the Generation unit sizing and cost analysis for stand alone
hybrid microgrid system considerably reduces the wind, photovoltaic and hybrid wind/PV systems, IEEE
Trans Energy Convers March 13(1) (1998), 70–75.
time and computations. Since the accurate method
[15] S. Roy, Optimal planning of wind energy conversion sys-
is time-consuming, the approximate method may be tems over an energy scenario, IEEE Trans Energy Convers
useful in rough calculations like design/sensitivity September 12(3) (1997).
analysis. [16] G. Merei, C. Berger and D.U. Sauer, Optimization of an
off-grid hybrid PV–Wind–Diesel system with different bat-
tery technologies using genetic algorithm, Solar Energy 97
(2013), 460–473.
[17] D. Xu, L. Kang, L. Chang and B. Cao, Optimal sizing of
References
standalone hybrid wind/PV power systems using genetic
algorithms, In: Canadian Conference on Electrical and
[1] S. Sinha and S.S. Chandel, Review of recent trends in Computer Engineering, 2005, 2005, pp. 1722–1725.
optimization techniques for solar photovoltaic–wind based [18] S.M. Hakimi, S.M. Tafreshi and A. Kashefi Kaviani, Unit
hybrid energy systems, Renewable and Sustainable Energy sizing of a stand-alone hybrid power system using particle
Reviews 2015 50 (2015), 755–769. swarm optimization (PSO), In: Proceeding of the Inter-
[2] A.S.A. Busaidi, H.A. Kazem, A.H. Al-Badi and M.F. Khan, national Conference on Automation and Logistics, Jinan,
A review of optimum sizing of hybrid PV–Wind renew- China, 2007.
able energy systems in Oman, Renewable and Sustainable [19] A. Kaabeche, M. Belhamel and R. Ibtiouen, Siz-
Energy Reviews 53 (2016), 185–193. ing optimization of grid-independent hybrid photo-
1772 H.R. Baghaee et al. / Multi-objective optimal power management and sizing of a reliable wind
voltaic/wind power generation system, Energy 36(2) (2011), power generation system, Renewable Energy 31 (2006),
1214–1222. 1641–56.
[20] F.O. Hocaoğlu, Ö.N. Gerek and M. Kurban, A novel hybrid [36] M.Y. El-Sharkh, M. Tanrioven, A. Rahman and M.S. Alam,
(wind–photovoltaic) system sizing procedure, Solar Energy Cost related sensitivity analysis for optimal operation of a
83(11) (2009), 2019–2028. grid-parallel PEM fuel cell power plant, J Power Sources
[21] P. Nema, R.K. Nema and S. Rangnekar, A current and future 161 (2006), 1198–1207.
state of art development of hybrid energy system using wind [37] K. Strunz and E.K. Brock, Stochastic energy source access
and PV-solar: A review, Renewable and Sustainable Energy management: Infrastructure- integrative modular plant for
Reviews 13(8) (2009), 2096–2103. sustainable hydrogen-electric cogeneration, Int J Hydrogen
[22] R. Belfkira, C. Nichita, P. Reghem and G. Barakat, Mod- Energy 31 (2006), 1129–1141.
eling and optimal sizing of hybrid renewable energy [38] I. Bagen and R. Billinton, Evaluation of different operating
system, Power Electronics and Motion Control Confer- strategies in small standalone power systems, IEEE Trans
ence, 2008 EPE-PEMC 2008. 13th Publication Year, 2008, Energy Convers September 20(3) (2005), 654–660.
pp. 1834–1839. [39] R. Karki and R. Billinton, Reliability/cost implications of
[23] S. Diaf, G. Notton, M. Belhamel, M. Haddadi and A. PV and wind energy utilization in small isolated power
Louche, Design and techno-economical optimization for systems, IEEE Trans Energy Convers Dec 16(4) (2001),
hybrid PV/wind system under various meteorological con- 368–373.
ditions, Applied Energy 85(10) (2008), 968–998. [40] S. Nomura, Y. Ohata, T. Hagita, H. Tsutsui, S. Tsuji-Iio and
[24] A. Maleki and A. Askarzadeh, Artificial bee swarm opti- R. Shimada, Wind farms linked by SMES systems, IEEE
mization for optimum sizing of a stand-alone PV/WT/FC Trans Appl Supercond (2005), 1951-1954.
hybrid system considering LPSP concept, Solar Energy 107 [41] M.K. Khairil and S. Javanovic, Reliability modeling of unin-
(2014), 227–235. terruptible power supply systems using fault tree analysis
[25] A. Maleki and F. Pourfayaz, Optimal sizing of autonomous method, Eur Trans Electr Power (2007).
hybrid photovoltaic/wind/battery power system with LPSP [42] M. Marchesoni and S. Savio, Reliability analysis of a fuel
technology by using evolutionary algorithms, Solar Energy cell electric city car, In: IEEE 2005 European Conference
115 (2015), 471–483. on Power Electronics and Applications, 2005, p. 10.
[26] C. Shang, D. Srinivasan and T. Reindl, An improved parti- [43] R. Karki and R. Billinton, Cost-effective wind energy uti-
cle swarm optimisation algorithm applied to battery sizing lization for reliable power supply, IEEE Trans Energy
for stand-alone hybrid power systems, International Jour- Convers June 19 (2004), 435–440.
nal of Electrical Power & Energy Systems 74 (2016), [44] G. Tina, S. Gagliano and S. Raiti, Hybrid solar/wind power
104–117. system probabilistic modeling for long-term performance
[27] S.M. Hakimi and S.M. Moghaddas-Tafreshi, Optimal siz- assessment, Sol Energy 80 (2006), 578–588.
ing of a stand-alone hybrid power system via particle swarm [45] H.R. Baghaee, M. Mirsalim, G.B. Gharehpetian and A.
optimization for Kahnouj area in south-east of Iran, Renew- Kashefi Kaviani, Security/cost-based optimal allocation of
able Energy 34(7) (2009), 1855–1862. multi-type FACTS devices using multi-objective particle
[28] A. Kashefi Kaviani, G.H. Riahy and S.H.M. Kouh- swarm Optimization, Simulation: International Transac-
sari, Optimal design of a reliable hydrogen-based tions of the Society for Modeling and Simulation 88(8)
stand-alone wind/PV generating system, considering com- (2012), 999–1010. DOI: 10.1177-0037549712438715
ponent outages, Renewable Energy 34(11) (2009), 2380– [46] L. Wang and C. Singh, Environmental/economic power
2390. dispatch using a fuzzified multi-objective particle swarm
[29] R. Billinton and R.N. Allan, Reliability evaluation of optimization algorithm, Electric Power System Research 77
engineering systems: Concepts and techniques. 2nd ed. Ger- (2007), 1654–1664.
many, Springer; Softcover reprint of the original 2nd ed. [47] P.K. Tripathi, S. Bandyopadhyay and S.K. Pal, Multi-
1992 edition, 2013. Objective Particle Swarm Optimization with time variant
[30] A. Mawardi and R. Ptchumani, Effects of parameter uncer- inertia and acceleration coefficients, Information Sciences
tainty on the performance variability of proton exchange 177 (2007), 5033–5049.
membrane (PEM) fuel cells, J Power Sources 160 (2006), [48] P. Zhang, H. Chen, X. Liu and Z. Zhang, An iterative multi-
232–245. objective particle swarm optimization-based control vector
[31] H. Ren, W. Zhou, K. Nakagami, W. Gao and Q. Wu, parameterization for state constrained chemical and bio-
Multi-objective optimization for the operation of distributed chemical engineering problems, Biochemical Engineering
energy systems considering economic and environmental Journal 103 (2015), 138–151.
aspects, Applied Energy 87(12) (2010), 3642–3651. [49] Y. Wang and Y. Yang, Particle swarm with equilibrium strat-
[32] R. Billinton and R.N. Allan, Reliability evaluation of power egy of selection for multi-objective optimization, European
systems. 2nd ed. Germany, Springer; Softcover reprint of Journal of Operational Research 200(1) (2010), 187–197.
the original 1st ed. 1996 edition, 2013. [50] K.E. Parasopoulos and M.N. Vrahatis, On the computation
[33] R.S. Garcia and D. Weisser, A wind–diesel system with of all global minimizers through particle swarm optimiza-
hydrogen storage: Joint optimization of design and dispatch, tion, IEEE Trans on Evolutionary Computation 8(3) (2004).
Renewable Energy 31 (2006), 2296–2320. [51] W. Fang, J. Sun, H. Chen and X. Wu, A decentralized
[34] A.T.D. Perera, R.A. Attalage, K.K.C.K. Perera and V.P.C. quantum-inspired particle swarm optimization algorithm
Dassanayake, A hybrid tool to combine multi-objective with cellular structured population, Information Sciences
optimization and multi-criterion decision making in design- 330 (2016), 19–48.
ing standalone hybrid energy systems, Applied Energy 107 [52] B. Jarboui, N. Damak, P. Siarry and A. Rebai, A combina-
(2013), 412–425. torial particle swarm optimization for solving multi-mode
[35] D.B. Nelson, M.H. Nehrir and C. Wang, Unit sizing resource-constrained project scheduling problems, Applied
and cost analysis of stand-alone hybrid wind/PV/fuel cell Mathematics and Computation 205(2) (2008), 509–510.
H.R. Baghaee et al. / Multi-objective optimal power management and sizing of a reliable wind 1773
[53] R.R. Tan, Hybrid evolutionary computation for the develop- on Web Technologies and Internet Commerce, 2005, Vol. 1,
ment of pollution prevention and control strategies, Journal pp. 1115–1121.
of Cleaner Production 15 (2007), 902–906. [55] R. Saborido, A.B. Ruiz, J.D. Bermúdez, E. Vercher
[54] Z. Jinghui, H. Xiaomin, Z. Jun and G. Min, Comparison and M. Luque, Evolutionary multi-objective optimization
of Performance between Different Selection Strategies on algorithms for fuzzy portfolio selection, Applied Soft Com-
Simple Genetic Algorithms, International Conference on puting, Accepted Manuscript In Press, Available online 12
Intelligent Agents, Computational Intelligence for Model- November 2015.
ing, Control and Automation, and International Conference