Energy Conversion and Management 144 (2017) 132–142

Contents lists available at ScienceDirect

Energy Conversion and Management

journal homepage: www.elsevier.com/locate/enconman

Optimal stochastic short-term thermal and electrical operation of fuel

cell/photovoltaic/battery/grid hybrid energy system in the presence of
demand response program
Majid Majidi, Sayyad Nojavan ⇑, Kazem Zare
Faculty of Electrical and Computer Engineering, University of Tabriz, P.O. Box: 51666-15813, Tabriz, Iran

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

Article history: In this paper, cost-efficient operation problem of photovoltaic/battery/fuel cell hybrid energy system has
Received 2 February 2017 been evaluated in the presence of demand response program. Each load curve has off-peak, mid and peak
Received in revised form 8 April 2017 time periods in which the energy prices are different. Demand response program transfers some amount
Accepted 14 April 2017
of load from peak periods to other periods to flatten the load curve and minimize total cost. So, the main
goal is to meet the energy demand and propose a cost-efficient approach to minimize system’s total cost
including system’s electrical cost and thermal cost and the revenue from exporting power to the
Keywords:
upstream grid. A battery has been utilized as an electrical energy storage system and a heat storage tank
Fuel cell
Battery bank
is used as a thermal energy storage system to save energy in off-peak and mid-peak hours and then sup-
Heat storage tank ply load in peak hours which leads to reduction of cost. The proposed cost-efficient operation problem of
Hybrid energy system photovoltaic/battery/fuel cell hybrid energy system is modeled by a mixed-integer linear program and
Demand response program solved by General algebraic modeling system optimization software under CPLEX solver. Two case stud-
ies are investigated to show the effects of demand response program on reduction of total cost.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction In [5], cost-efficient sizing of a hybrid energy system containing

By taking a simple look to the recent works in power system
technologies, it can be understood that the upstream grid is not
the only power supply resource in residential section anymore
and various types of distributed energy sources are getting more
and more utilized by system operators. Photovoltaic system
receives solar irradiation and then generates clean energy [1]. Con-
suming natural gas or hydrogen, fuel cell generates heat and elec-
tric power [2]. Electrical storage systems can be also employed in
power systems to store energy in peak periods and supply load
in other periods [3]. So, integrated energy systems system like
hybrid systems including such resources can have significant por-
tion in handling energy demand in residential section [4].
1.1. Literature review
Due to fast-going improvement of hybrid energy systems, many
researches and studies have been done by researchers. Here we
have some of the important ones done recently as follows:
⇑ Corresponding author.
E-mail addresses: majidmajidi95@ms.tabrizu.ac.ir (M. Majidi), sayyad.nojavan@
tabrizu.ac.ir (S. Nojavan), kazem.zare@tabrizu.ac.ir (K. Zare).
PV system, battery, wind turbine and fuel cell has been studied. A
battery/fuel cell hybrid system generating energy without any
interruption has been utilized to supply an electrical vehicle in
[6]. In [7], optimal performance of a wind/PV/fuel cell/grid hybrid
system supplying an electrical load and a heat pump water heater
has been investigated. Optimal size of a battery/wind turbine/PV
hybrid system has been found utilizing Big Bang–Big Crunch algo-
rithm in [8]. Optimal size of a fuel cell/battery/PV hybrid system
has been found with different approaches and considering several
power management strategies in [9]. Economic operation of a
hydrogen/battery/PV hybrid system has been evaluated and com-
pared with diesel-PV generator and battery-PV in [10]. In [11],
genetic algorithm (GA) has been employed to optimize and model
a diesel/wind/PV hybrid system. In [12], different operating strate-
gies have been utilized to evaluate performance of a hybrid energy
system in residential section. In [13], with the aim of minimizing
cost and CO2, best possible size of an ESS/PV/diesel hybrid system
has been found in a ship power system. In [14], several solar
thermal/PV/micro-CHP hybrid systems have been designed to
determine optimal economic and environmental performance of
a building. Optimal performance of an off-grid wind/battery/diesel
hybrid system has been evaluated in [15] using genetic algorithm
(GA). Genetic algorithm and weighted Ah ageing model have been

http://dx.doi.org/10.1016/j.enconman.2017.04.051
0196-8904/Ó 2017 Elsevier Ltd. All rights reserved.
 M. Majidi et al. / Energy Conversion and Management 144 (2017) 132–142 133

Nomenclature

Abbreviations DRPemax maximum amount of electrical DRP

PV photovoltaic system f a small value used for considering loss of heat in heat
DRP demand response program storage tank
MIP mixed-integer linear program gstin heat storage’s charging efficiency
GAMS general algebraic modeling system gstout heat storage’s discharging efficiency
Hst
capacity rated capacity of heat storage tank
Indices
h; s Hour and scenario indices HEfu
ratio
heat to electricity ratio of fuel cell
Hbb
capacity rated capacity of backup burner
Parameters
H; S number of hours and scenarios Variables
qgrid price of procured power from upstream grid Cos t total total cost
h;s
Cos t grid cost of power procurement from upstream grid
gfce fuel cell electrical efficiency Cos t gas cost of gas procurement
pv
gbb efficiency of backup burner Sale revenue from exporting power to the upstream grid by
qgas
h
price of purchased gas PV
qph;sv ;sale price of sold power to upstream grid by PV Pgrid;ch
h;s
power supplied by upstream grid to charge battery
M a constant number Pgrid;l
h;s
power supplied by upstream grid to supply load
ugrid;pur
h;s
binary variable; 1 if upstream grid attempts to supply Pfc;l power supplied by fuel cell to supply load
h;s
load and charge battery; otherwise 0
Pfc;ch power supplied by fuel cell to charge battery
a; b efficiency of PV system and inverter h;s

ProbðsÞ probability of each scenario Hab

h;s generated heat by backup burner
pv ;s
A area for PV installation Ph;s power supplied by PV to be exported to the upstream
Rh;s amount of solar irradiation grid
pv ;l
ugrid;s binary variable; 1 if battery PV system attempts to sell Ph;s power supplied by PV to supply load
h;s
power to the upstream grid; otherwise 0 Pb;dis
h;s
discharged power by battery bank to supply load
Pfc
capacity rated power of fuel cell pv ;ch
Ph;s generated power by PV to charge battery
d a small value used for considering self-discharge of
battery bank Ebh;s available energy of battery storage
gbch battery’s charging efficiency Pl;DRP
h;s
electrical load with considering DRP
gbdis battery’s discharging efficiency PTOU
h;s new load with considering time-of-use (TOU) program
Ebat
min minimum capacity of battery Hlh;s thermal load
st;out
Ebat
max maximum capacity of battery Hh;s output heat of heat storage tank
N a constant number Hst available heat of heat storage tank
h;s
ub;ch
h;s
binary variable; 1 if battery storage attempts to charge; st;in
Hh;s input heat of heat storage tank
otherwise 0
ub;dis binary variable; 1 if battery storage attempts to dis- Hfu
h;s
produced heat by fuel cell
h;s
charge; otherwise 0 Hbb produced heat by backup burner
h;s
Plh;s electrical load without considering DRP

used to investigate the effect of lead-acid batteries on efficient
operation of hybrid systems in [16]. In [17], performance of a fuel
cell/battery/PV hybrid energy system connected to an electrolyser
and H2 tank has been evaluated. Operation of a hybrid fuel
cell/PV/electrolyser system for micro co-generation applications
has been studied in [18]. In [19], particle swarm optimization
(PSO) algorithm has been utilized to ideally size a hybrid
battery/PV/H2 energy system. A fuzzy-logic based Energy
management of a grid-connected fuel cell/PV hybrid system has
been investigated in [20]. A complete review of techniques and
methods that can be employed for optimal management of hybrid
PV/battery/fuel cell system is presented in [21]. Finally, ideal sizing
problem of a wind/PV/battery/diesel hybrid energy system along
with minimizing CO2 emission, dump energy and life cycle cost,
has been investigated in [22]. Using energy conversion system
for wind turbine and maximum power point tracking method for
PV system, efficient operation of PV-wind-fuel cell hybrid energy
system has been evaluated in [23]. Employing Hammersley
Sequence Sampling, multi-objective model including three
objective functions namely total cost, electricity efficiency and
energy supply reliability is solved in [24]. Different hydrogen
technologies have been used in [25] to facilitate linkage of heat
and electricity for integration of renewable energy resources in
various scales. Using thermo-economic based model, a hybrid fuel
cell/gas turbine energy system is optimally configured in [26].
1.2. Demand response program (DRP)
Electrical loads can participate in DRP to reduce their operation
costs. Participating in DRP, customers are responsible to change
their energy consumption pattern to reduce their expenses. As a
reward for reduction of consumption, consumers get incentives
or they pay less to the utility [27]. Totally, demand response pro-
grams are divided into two groups: Incentive-Based Programs
(IBPs) and Price-Based Programs (PBPs). In PBPs, dynamic pricing
rates including Time of Use (TOU) rate, Critical Peak Pricing
(CPP), Extreme Day Pricing (EDP), Extreme Day CPP (ED-CPP) and
Real Time Pricing (RTP) are used [28,29].
IBPs are divided into market-based programs and classical pro-
grams. In these programs, customers get incentives as much as
they reduce their consumption.
134 M. Majidi et al. / Energy Conversion and Management 144 (2017) 132–142

1.3. Novelty, contributions of this research pv

X
S X
H
Sale ¼ ProbðsÞ  ðPph;sv ;s  qph;sv ;sale Þ ð4Þ
s h
The purpose of proposed paper is to optimize performance of a
PV/battery/fuel cell system with considering DRP. In this paper, Eq. (2) details the cost of purchased power. The cost of procured
time-of-use (TOU) rate of demand response program (DRP) has power from upstream grid includes the cost of imported power for
been proposed to improve the hybrid system’s operation and min- charging battery and supplying load. Eq. (2) shows that the gas
imize its total cost. In fact, DRP flattens the load curve by transfer- procurement cost is equal to the cost of purchased gas for fuel cell
ring some percentage of load from peak time (expensive) periods and backup burner operation. Finally, the benefit from selling
to the off-peak time (cheaper) periods which leads to cost- power to the upstream grid is expressed by Eq. (4). According to
efficient operation of hybrid system and decreases system total (4), benefit from selling power is equal to excess produced power
cost. By employing DRP in the proposed paper, it has been tried of PV multiplying an agreed price.
to operate distributed generation units (PV, battery and fuel cell)
with their maximum capacity to supply load and reduce procured 2.2. Electrical load balance
power from upstream grid as much as possible to minimize oper-
ation cost of hybrid system. According to the given explanation, the Each generation system including different power resources
novelty and contributions of proposed paper are presented as should be able to supply energy demand. The mentioned sentence
follows: is formulated as power balance equation in (5). It should be noted
that demand response program (DRP) has been implemented and
1. Economic performance of grid-connected PV/battery/fuel cell all the equations are formulated with DRP consideration.
hybrid system. pv ;l
2. Implementation of demand response program (DRP) to improve
Pl;DRP
h;s ¼ P grid;l fc;l b;dis
h;s þ P h;s þ P h;s þ P h;s ð5Þ
economic performance of hybrid PV/battery/fuel cell system It can be understood from Eq. (5) that the procured power from
and to minimize system’s total cost. upstream grid and the produced power by fuel cell and also the
3. Optimal operation of distributed energy sources (PV, battery generated power by PV system besides the discharged power of
and fuel cell) to reduce dependency on the upstream grid. battery storage should be equal to the load demand with consider-
1.4. Paper organization ing DRP.

The rest of proposed work is organized as follows: formulation 2.3. Upstream grid constraints
of optimal performance problem of PV/battery/ fuel cell hybrid
energy system is presented in Section 2. In Section 3, a sample sys- In the proposed paper, upstream grid has been considered as
tem is studied with two different study cases (with and without one of power resources which is able to sell (purchase) power to
DRP) and the results are compared to show the effects of proposed (from) the hybrid system but the amount of exchanged power
approach. Finally, conclusions of proposed paper are presented in between these two sides is limited. Eq. (6) constrains purchased
Section 4. power from upstream grid in which the purchased power from
upstream grid should not exceed the grid capability.
2. Mathematical formulation Pgrid;ch þ Pgrid;l 6 M  ugrid;pur ð6Þ
h;s h;s h;s

In this paper, besides upstream grid, a hybrid PV/fuel cell/bat-

tery energy system has been utilized to economically supply resi- 2.4. PV constraints
dential section energy demand (electrical and thermal). The
optimal operation problem of hybrid PV/FC/battery energy system Produced power by PV system can be used to supply load and
is formulated in this section. also it can be utilized to charge battery storage and finally the
excess power can be sold to the upstream grid with an agreed price
2.1. Objective function to gain profit. Sum of the produced power by PV system for differ-
ent applications should not exceed the rated power that PV system
The objective function of presented paper is economic opti- can produce. The sentence above in mathematical form is
mization of a hybrid energy system’s performance to minimize expressed by Eq. (7)
total cost of hybrid system. The mathematical form of the sentence
Pph;sv ;l þ Pph;sv ;s þ Pph;sv ;ch ¼ a  b  A  Rh;s ð7Þ
above is described by (1) [13].
pv Sold power by PV system is limited by Eq. (8). To avoid simul-
Min Cos ttotal ¼ Cos t grid þ Cos tgas  Sale ð1Þ
taneous power exchange between upstream grid and hybrid sys-
Total cost of hybrid system includes cost of procured power tem, Eq. (9) is utilized.
from upstream grid, cost of procured gas for fuel cell and backup
burner operation and finally the revenue from exporting power Eph;sv ;s 6 M  ugrid;s
h;s ð8Þ
to the upstream grid. Since the obtained revenue is benefit for
the system, it has a negative value in the formulation. To more clar- ugrid;pur
h;s þ ugrid;s
h;s 61 ð9Þ
ification, the mentioned costs in the objective function have been
detailed in Eqs. (2)–(4) [13].
2.5. Fuel cell electrical constraints
X
S X
H
Cos t grid
¼ ProbðsÞ  ððPgrid;ch
h;s þ Pgrid;l
h;s Þ q grid
h;s Þ ð2Þ
s
In the considered hybrid system, fuel cell is able to produce heat
h
and electrical power. Produced electric power by fuel cell to supply
! load and charge battery storage is restricted by Eq. (10).
X
S X
H Pfc;l fc;ch
h;s þ P h;s Hab
h;s
Cos t gas ¼ ProbðsÞ  þ  qgas ð3Þ
s h gfce gbb h Pfc;l fc;ch fc
h;s þ P h;s 6 P capacity ð10Þ
 M. Majidi et al. / Energy Conversion and Management 144 (2017) 132–142 135

2.6. Battery storage constraints which means that the amount of base load in a day is fixed and
it is just transferred from peak time periods to off-peak time peri-
In addition to other resources utilized to supply load, a battery ods which leads to cost reduction.
has been used as an electrical storage system to manage and con-
trol demand in different time periods. Utilization of battery storage 2.8. Thermal load
along with implementation of DRP results in optimal performance
of hybrid energy system. Utilized battery storage includes some In the proposed paper, in addition to electrical load, we also
technical constraints related to charge, discharge and etc. The fol- have a thermal load which should be supplied. Constraints of ther-
lowing constraints are described by Eqs. (11)–(15). mal load are expressed by Eqs. (19)–(26).
pv ;ch
Ebh;s ¼ ð1  dÞ  Ebh1;s þ ðPgrid;ch
h;s þ Pfc;ch b;dis
h;s þ P h;s Þ  gch  ðP h;s =gdis Þ
b b
Hlh;s ¼ Hst;out
h;s ð19Þ
ð11Þ
Hsth;s ¼ ð1  fÞ  Hsth1;s þ Hst;in st st;out
h;s  gin  ðH h;s =gout Þ
st
ð20Þ
Ebmin 6 Ebh;s 6 Ebmax ð12Þ
Hsth;s 6 Hstcapacity ð21Þ
pv ;ch
ðPgrid;ch
h;s þ Pfc;ch
h;s þ P h;s Þ 6N ub;ch
h;s ð13Þ
Hst;in st
h;s 6 Hcapacity ð22Þ

Pb;dis 6N ub;dis ð14Þ

h;s h;s
Hst;out
h;s 6 Hstcapacity ð23Þ

ub;ch b;dis
h;s þ uh;s 6 1
Eq. (11) states that the amount of stored energy in battery stor-
age is equal to the amount of energy at former hour plus the
amount of energy at present time, deducting the discharged power
by battery storage. It should be noted that the stored energy in
each hour should obey its maximum and minimum limitations
(12).
Eq. (13) expresses that the charging power of battery storage is
not allowed to exceed its nominal value. Also, discharging power
limitation of battery storage is expressed by Eq. (14). In order to
avoid battery storage from simultaneous charging and discharging,
Eq. (15) is utilized.
2.7. Time-of-use rates of demand response program
One of the popular and well-known programs used extensively
to economically optimize energy systems with different perfor-
mances, is demand response program (DRP). DRP includes different
series of optimization programs inside itself and the one used in
the proposed work is time-of-use (TOU) of demand response pro-
gram [30–32]. TOU flattens the load curve by shifting some amount
of load from peak time periods to the off-peak time periods which
leads to minimization of total cost. Mathematical formulation of
explanation above is presented in Eq. (16).
Pl;DRP
h;s ¼ Plh;s þ PTOU
h;s
According to Eq. (16), the new electrical load with considering
DRP is equal to the base electrical load plus the variable power,
P TOU
h . This variable can be either positive or negative. If it is positive,
it means that the load decreases and if it is negative, it means that
the load increases and this is the base of time-of-use (TOU) pro-
gram. Eqs. (17) and (18) express some technical constraints related
to DRP.
ð15Þ
Hst;in fu bb
h;s ¼ H h;s þ Hh;s ð24Þ
Hfu fc;l fc;ch fu
h;s 6 ðEh;s þ Eh;s Þ  HEratio ð25Þ
Hbb bb
h;s 6 H capacity ð26Þ
Heat storage tank, which is charged by fuel cell and backup bur-
ner, supplies thermal load (19). Eq. (20) states that the amount of
stored heat in heat storage tank is equal to the amount of heat at
former hour plus the amount of heat at present time, deducting
the discharged heat by heat storage tank. Eq. (21) expresses that
the heat inside the heat storage tank at hour h should not exceed
its nominal value. Eqs. (22) and (23) are utilized to limit charging
and discharging heat of heat storage tank, respectively. Fuel cell
and backup burner use city gas to generate heat and then, the pro-
duced heat charges heat storage tank (24). Eqs. (25) and (26) limit
produced heat by fuel cell and backup burner, respectively.
2.9. Stochastic
Uncertain behavior of parameters like electrical load, thermal
load, market price and solar irradiation can be modeled using
stochastic programming. In this approach, the forecast error distri-
bution curves are divided into five intervals with the width of one
standard deviation [33] and the mean values are the parameters
ð16Þ used in deterministic programming. The percentage of increase
or decrease for uncertain variables is considered to be 10%. Fig. 1
shows a sample discrete form of the predication error probability
distribution function.
It is essential for every available scenario that 2 values be
computed:

jPTOU e l
h;s j 6 DRP max  P h;s ð17Þ

X
24
PTOU
h;s ¼ 0 ð18Þ
h¼1

In TOU program, the amount of increase or decrease that can be

achieved in the load should be less than the base load (17). It
should be noted that the allowable amount of increase or decrease
the load can have is considered to be 20%. Also, according to Eq.
(18), the increased or decreased load at one day should be equal Fig. 1. Probability distribution function for uncertainty parameters.
136 M. Majidi et al. / Energy Conversion and Management 144 (2017) 132–142

Table 1
Probability of scenarios approximated normal distribution function.

Scenario Value of each relevant Probability of each relevant

number scenario scenario
S1 l  2:5r 0.0123
S2 l  1:5r 0.136
S3 l 0.682
S4 l þ 1:5r 0.136
S5 l þ 2:5r 0.023

i. By integrating the area below the probability distribution

curve in every period, we can acquire each scenario’s
probability.
ii. The realized prediction error in each relevant scenario is
considered to be the average amount of period. Table 1
shows the amount and its probability in each relevant
scenario.

Fig. 3. Energy demand.

3. Numerical study

In this section, a sample hybrid system like the one shown in
Fig. 2 is studied in two different case studies without and with con-
sidering DRP and by comparing the obtained results, the effective-
ness of proposed approach is concluded.
3.1. Input data
As shown in Fig. 3, the sample system includes two different
types of loads: electrical load and thermal load. To supply electrical
demand, upstream grid and a hybrid energy system containing fuel
cell, battery storage and PV system have been utilized.
Also, generated heat by backup burner and recovered heat from
fuel cell are used to supply thermal load. In the proposed paper, it
has been considered that a part of electrical load is supplied by the
purchased power from upstream grid. Also, some of purchased
power from upstream grid is used to charge battery storage. The
price of purchased power is various in different seasons and hours
as shown in Table 2. PV system is one of the distributed energy
resources used in this paper. The output power of PV system is
used to supply load, charge battery and at the time of excess, it
is sold to the upstream grid at an agreed price which is shown in
Table 2. The considered area in which the PV system is installed
is 50 m2 and the solar irradiation which has straight effect on the
output of PV system is illustrated for various times in Fig. 4.
Also, technical parameters of utilized PV system have been pre-
sented in Table 3. Some technical data about fuel cell is presented
in Table 3. A battery has been utilized as an electrical energy stor-
age system to save energy at off-peak hours and supply load in
peak hours which leads to cost reduction. Utilized battery storage
is charged by upstream grid, PV system and fuel cell. The data
related to battery storage is presented in Table 3. The only avail-
able resource to supply thermal load is the heat storage tank. Fuel
cell and backup burner charge the heat storage and then, the saved
heat is used to supply thermal load. It should be noted that heat
storage tank has the ability to charge and discharge simultane-
ously. Technical information about the heat storage is presented
in Table 3. As another energy source used in this paper, fuel cell
consumes gas and produces heat and electrical power which is
used to supply load and charge battery storage. The price of pur-
chased gas is presented in Table 4. The optimal operation problem
of PV/battery/fuel cell hybrid energy system is solved by GAMS
optimization package utilizing CPLEX 11.0 [34].
3.2. Comparison results of deterministic and stochastic approaches in
different cases
In this section, a sample hybrid energy system has been studied
using deterministic and stochastic approaches in two different

Fig. 2. Sample hybrid system schematic diagram.

 M. Majidi et al. / Energy Conversion and Management 144 (2017) 132–142 137

Table 2 cases and the results are compared to show the effectiveness of
Price of exchanged power between upstream grid and hybrid system. proposed model. The investigated cases are:
Item Unit Value
Energy charge Peak period Summer $/kw h 0.30 Case 1: operation of hybrid energy system without DRP.
Mid-peak period $/kw h 0.19 Case 2: operation of hybrid energy system with DRP.
Off-peak period $/kw h 0.08
PV sell price $/kw h 0.29 In case 1, no economic strategy is considered and the studied
system is investigated without DRP consideration. The obtained
results show that the components of system do not have optimal
operation and therefore, there is no definite strategy for power
exchange between system and upstream grid. As presented in
Table 5, total cost of hybrid system in deterministic approach is
equal to 42.872$ which includes 24.587$ for electrical power pro-
curement from upstream grid, 22.600$ for gas procurement for fuel
cell and backup burner operation and 4.315$ benefit from selling
power to the upstream grid. Also, in stochastic approach, the elec-
trical power and gas procurement costs are 23.624$ and 22.210$,
respectively. Furthermore, the revenue from selling power is
3.819$. Therefore, the total cost of system is 42.016$ which is less
than deterministic approach due to uncertainty modeling. Evalua-
tion of obtained results indicates that optimal operation of dis-
tributed energy resources and decreasing dependency to the
upstream grid can lead to cost reduction and this problem is inves-
tigated and solved in the next case.
In case 2, optimal operation of system is investigated with con-
sidering DRP. DRP transfers some part of load from expensive
(peak) periods to the cheaper (off-peak) periods which leads to
load flattening and cost reduction. As shown in Table 5, by employ-
ing DRP in deterministic approach in case 2, total cost of system is
Fig. 4. Solar irradiation.
equal to 40.521$ which includes 21.440$ for electrical power pro-
curement from upstream grid, 22.268$ for gas procurement for fuel
cell and backup burner operation and 3.187$ benefit from selling
power to the upstream grid. Also, in stochastic approach, the elec-
Table 3
Technical data of utilized components in the hybrid system.
trical power and gas procurement costs are 21.211$ and 22.080$,
respectively. Furthermore, the revenue from selling power is
Parameters Value (%) 3.600$. Therefore, the total cost of system is 39.690$ which is less
Efficiency of PV system 0.19 than deterministic approach due to uncertainty modeling.
Efficiency of inverter 0.90 It can be concluded from Table 5 that the total cost in determin-
Electrical efficiency of fuel cell 0.39
istic and stochastic approaches in case 2 has been reduced 5.48 %
Heat to electricity ratio of fuel cell 0.56
Battery’s charging efficiency 0.95 and 5.53 % in comparison with case 1 which is due to DPP
Battery’s discharging efficiency 0.95 implementation.
Self-discharge value of battery bank 0.05 To more clarification and to show the effect of DRP implemen-
Heat storage’s charging efficiency 0.95
tation, operation of system’s components in deterministic and
Heat storage’s discharging efficiency 0.95
Loss of heat value in heat storage tank 0.05
stochastic programming is illustrated by Figs. 5–22. New load
Efficiency of backup burner 0.95 profile (with DRP consideration) in deterministic and stochastic
programming is illustrated in Figs. 5 and 6, respectively. According
to these Figures, load has been shifted from peak times to the off-
peak and mid-peak times which has led to minimization of total
Table 4
cost.
Price of gas procurement. Procured power from upstream grid in deterministic and
stochastic programming is illustrated in Figs. 7 and 8, respectively.
Item Unit Value
According to these Figures, purchased power in expensive periods
Energy charge $/m3 0.12 is reduced and thus increased in mid-peak and off-peak periods

Table 5
Comparison results of deterministic and stochastic approaches in cases 1 and 2.

Approaches Deterministic approach Stochastic approach

Different parameters Case 1 Case 2 Case 1 Case 2
Cost of electrical power procurement ($) 24.587 21.440 23.624 21.211
Cost of gas consumption ($) 22.600 22.268 22.210 22.080
Revenue from sold power ($) 4.315 3.187 3.819 3.600
Total cost ($) 42.872 40.521 42.016 39.690
Total cost reduction (%) 0 5.48 0 5.53
138 M. Majidi et al. / Energy Conversion and Management 144 (2017) 132–142
Fig. 5. New load profile with DRP in deterministic programming.
Fig. 6. New load profile with DRP in stochastic programming.
which have less expensive price in comparison with peak time
periods and this has led to reduction of total cost.
Electrical power generated by fuel cell in deterministic and
stochastic programming is illustrated in Figs. 9 and 10, respec-
tively. According to these Figs, fuel cell electrical generation is
reduced due to implementation of DRP which increases lifespan
of fuel cell.
Figs. 11 and 12 depict the sold power by PV system in determin-
istic and stochastic programming. As shown in these figures, due to
implementation of DRP, sold power by PV system is reduced
and therefore, PV system portion in supplying load is increased
which means that the purchased power from upstream grid is
decreased.
Figs. 13 and 14 express that by employing DRP in case 2, the
number of charge and discharge and also and the amount charging
and discharging power of battery storage has been reduced which
increases the storage’s lifespan.
Fig. 7. Purchased power from upstream grid in deterministic programming.
Fig. 8. Purchased power from upstream grid in stochastic programming.
Gas procurement in deterministic and stochastic programming
is reduced in expensive periods and then increased in cheaper peri-
ods which is depicted by Figs. 15 and 16, respectively.
Optimal operation of fuel cell in case 2 has changed the opera-
tion of backup burner and therefore, as shown in Figs. 17 and 18,
backup burner has attempted to produce heat with an optimal pre-
determined plan instead of normal production. As mentioned
before, due to utilization of DRP, electrical generation of fuel cell
is reduced. Since the heat production of fuel cell is proportional
with its electrical power production, then according to Figs. 19
and 20, fuel cell heat generation is reduced in case 2 which
increases fuel cell’s lifespan.
Finally, charge and discharge of heat storage tank in determin-
istic and stochastic programming is illustrated in Figs. 21 and 22,
respectively. According to these Figs, heat storage has been mostly
charged in mid and off-peak periods which has led to reduction of
total cost.
 M. Majidi et al. / Energy Conversion and Management 144 (2017) 132–142 139

Fig. 12. Sold power by PV system in stochastic programming.

Fig. 9. Electrical production of fuel cell in deterministic programming.

Fig. 13. Charge and discharge of battery storage in deterministic programming.

Fig. 10. Electrical production of fuel cell in stochastic programming.

Fig. 11. Sold power by PV system in deterministic programming. Fig. 14. Charge and discharge of battery storage in stochastic programming.
140 M. Majidi et al. / Energy Conversion and Management 144 (2017) 132–142

Fig. 15. Purchased gas volume in deterministic programming. Fig. 18. Produced heat by backup burner in stochastic programming.

Fig. 16. Purchased gas volume in stochastic programming. Fig. 19. Produced heat by fuel cell in deterministic programming.

Fig. 17. Produced heat by backup burner in deterministic programming. Fig. 20. Produced heat by fuel cell in stochastic programming.
 M. Majidi et al. / Energy Conversion and Management 144 (2017) 132–142
Fig. 21. Charge and discharge of heat storage in deterministic programming.
Fig. 22. Charge and discharge of heat storage in stochastic programming.
3.3. Sensitivity analysis
Sensitivity analysis on the impact of the energy cost on the DRP
benefit is done and the results are presented in Table 6.
It can be seen from the obtained results that with the
increase of load participation in DRP, total operation cost of
hybrid energy system decreases consequently. It is due to that
by more participation of customer in DRP, more load will be
transferred and then more reduction of operation cost will be
obtained.
4. Conclusion
Optimal operation problem of hybrid battery/PV/fuel cell
energy system has been investigated in the proposed paper.
Demand response program (DRP) has been employed improve per-
formance of mentioned system. Therefore, to see the effective of
proposed approach on economic performance of hybrid
141
Table 6
Sensitivity analysis results.
DRP (%) Operation cost ($) Reduction of total cost (%)
0 42.872 0
5 42.498 0.87
10 42.122 1.74
15 41.730 2.66
20 41.327 3.6
25 40.924 4.54
30 40.521 5.48
PV/battery/fuel cell system, 2 study cases have been simulated.
To simulate the proposed problem, an MIP model was utilized
and the results are compared. According to the results, total cost
of hybrid system in case 2 has been decreased 5.48% in comparison
to case 1 and this is all due to DRP implementation. As expressed in
former sections, DRP flattened load curve by shifting some part of
load from peak periods (expensive) to off-peak (cheaper) periods
which led to cost reduction. Also, owing to DRP utilization, optimal
operation of distributed energy resources (PV-battery-fuel cell) is
established and the purchased power from upstream grid to supply
load is decreased. Also, by employing DRP, lifespan of utilized dis-
tributed generation units (PV-fuel cell-battery) is increased. So, it
can be concluded that from point of economic view, DRP can be
a suitable choice for system operators to minimize their operation
cost as much as possible.
References
[1] Ghalelou AN, Fakhri AP, Nojavan S, Majidi M, Hatami H. A stochastic self-
scheduling program for compressed air energy storage (CAES) of renewable
energy sources (RESs) based on a demand response mechanism. Energy
Convers Manage 2016;120(Jul 15):388–96.
[2] Nojavan S, Allah Aalami H. Stochastic energy procurement of large electricity
consumer considering photovoltaic, wind-turbine, micro-turbines, energy
storage system in the presence of demand response program. Energy
Convers Manage 2015;103(Oct 31):1008–18.
[3] Nojavan S, Zare K, Mohammadi-Ivatloo B. Selling price determination by
electricity retailer in the smart grid under demand side management in the
presence of the electrolyser and fuel cell as hydrogen storage system. Int J
Hydrogen Energy 2017;42(5):3294–308.
[4] Song J, Yang W, Higano Y, Wang XE. Introducing renewable energy and
industrial restructuring to reduce GHG emission: application of a dynamic
simulation model. Energy Convers Manage 2015;96(May 15):625–36.
[5] Hosseinalizadeh R, Shakouri H, Amalnick MS, Taghipour P. Economic sizing of a
hybrid (PV–WT–FC) renewable energy system (HRES) for stand-alone usages
by an optimization-simulation model: case study of Iran. Renew Sustain
Energy Rev 2016;54(Feb 29):139–50.
[6] Mebarki N, Rekioua T, Mokrani Z, Rekioua D, Bacha S. PEM fuel cell/battery
storage system supplying electric vehicle. Int J Hydrogen Energy 2016;41
(45):20993–1005.
[7] Sichilalu S, Tazvinga H, Xia X. Optimal control of a fuel cell/wind/PV/grid
hybrid system with thermal heat pump load. Sol Energy 2016;135(Oct
31):59–69.
[8] Ahmadi S, Abdi S. Application of the Hybrid Big Bang-Big Crunch algorithm for
optimal sizing of a stand-alone hybrid PV/wind/battery system. Sol Energy
2016;134(Sep 30):366–74.
[9] Behzadi MS, Niasati M. Comparative performance analysis of a hybrid PV/
FC/battery stand-alone system using different power management strategies
and sizing approaches. Int J Hydrogen Energy 2015;40(1):538–48.
[10] Guinot B, Champel B, Montignac F, Lemaire E, Vannucci D, Sailler S, et al.
Techno-economic study of a PV-hydrogen-battery hybrid system for off-grid
power supply: impact of performances’ ageing on optimal system sizing and
competitiveness. Int J Hydrogen Energy 2015;40(1):623–32.
[11] Merei G, Berger C, Sauer DU. Optimization of an off-grid hybrid PV–Wind–
Diesel system with different battery technologies using genetic algorithm. Sol
Energy 2013;97(Nov 30):460–73.
[12] Ren H, Wu Q, Gao W, Zhou W. Optimal operation of a grid-connected hybrid
PV/fuel cell/battery energy system for residential applications. Energy
2016;113(Oct 15):702–12.
[13] Lan H, Wen S, Hong YY, David CY, Zhang L. Optimal sizing of hybrid
PV/diesel/battery in ship power system. Appl Energy 2015;158(Nov 15):
26–34.
[14] Rodríguez LR, Lissén JM, Ramos JS, Jara EÁ, Domínguez SÁ. Analysis of the
economic feasibility and reduction of a building’s energy consumption and
emissions when integrating hybrid solar thermal/PV/micro-CHP systems. Appl
Energy 2016;165(Mar 1):828–38.
142 M. Majidi et al. / Energy Conversion and Management 144 (2017) 132–142
[15] Gan LK, Shek JK, Mueller MA. Optimised operation of an off-grid hybrid wind-
diesel-battery system using genetic algorithm. Energy Convers Manage
2016;126(Oct 15):446–62.
[16] Lujano-Rojas JM, Dufo-López R, Atencio-Guerra JL, Rodrigues EM, Bernal-
Agustín JL, Catalão JP. Operating conditions of lead-acid batteries in the
optimization of hybrid energy systems and microgrids. Appl Energy 2016;179
(Oct 1):590–600.
[17] Dash V, Bajpai P. Power management control strategy for a stand-alone solar
photovoltaic-fuel cell–battery hybrid system. Sustain Energy Technol Assess
2015;9(Mar 31):68–80.
[18] Özgirgin E, Devrim Y, Albostan A. Modeling and simulation of a hybrid
photovoltaic (PV) module-electrolyzer-PEM fuel cell system for micro-
cogeneration applications. Int J Hydrogen Energy 2015;40(44):15336–42.
[19] Paulitschke M, Bocklisch T, Böttiger M. Sizing algorithm for a PV-battery-H 2-
hybrid system employing particle swarm optimization. Energy Procedia
2015;73(Jun 30):154–62.
[20] Tiar M, Betka A, Drid S, Abdeddaim S, Becherif M, Tabandjat A. Optimal energy
control of a PV-fuel cell hybrid system. Int J Hydrogen Energy 2016(Jul 4).
[21] Bigdeli N. Optimal management of hybrid PV/fuel cell/battery power system: a
comparison of optimal hybrid approaches. Renew Sustain Energy Rev 2015;42
(Feb 28):377–93.
[22] Ogunjuyigbe AS, Ayodele TR, Akinola OA. Optimal allocation and sizing of PV/
Wind/Split-diesel/Battery hybrid energy system for minimizing life cycle cost,
carbon emission and dump energy of remote residential building. Appl Energy
2016;171(Jun 1):153–71.
[23] Fathabadi H. Novel standalone hybrid solar/wind/fuel cell power generation
system for remote areas. Sol Energy 2017;146(Apr 30):30–43.
[24] Chen H, Yang C, Deng K, Zhou N, Wu H. Multi-objective optimization of the
hybrid wind/solar/fuel cell distributed generation system using Hammersley
Sequence Sampling. Int J Hydrogen Energy 2017(Feb 24).
[25] Nastasi B, Basso GL. Hydrogen to link heat and electricity in the transition
towards future Smart Energy Systems. Energy 2016;110(Sep 1):5–22.
[26] Pirkandi J, Mahmoodi M, Ommian M. An optimal configuration for a solid
oxide fuel cell-gas turbine (SOFC-GT) hybrid system based on thermo-
economic modelling. J Cleaner Prod 2017(Jan 9).
[27] U.S. Department of Energy, Energy policy act of 2005, section 1252, February,
2006. <http://energy.gov/sites/prod/files/oeprod/DocumentsandMedia/DOE_
Benefits_of_Demand_Response_in_Electricity_Markets_and_
Recommendations_for_Achieving_Them_Report_to_Congress.pdf>.
[28] Staff Report, Assessment of demand response and advanced metering, 2006.
<http://www.FERC.gov>.
[29] Aalami H, Parsa Moghaddam M, Yousefi GR. Modeling and prioritizing demand
response programs in power markets. Electric Power Syst Res
2010;80:426–35.
[30] Nojavan S, Mohammadi-Ivatloo B, Zare K. Optimal bidding strategy of
electricity retailers using robust optimization approach considering time-of
use rate demand response programs under market price uncertainties. IET
Gener Transm Distrib 2015;9(4):328–38.
[31] Nojavan S, Majidi M, Najafi-Ghalelou A, Ghahramani M, Zare K. A cost-
emission model for fuel cell/PV/battery hybrid energy system in the presence
of demand response program: e-constraint method and fuzzy satisfying
approach. Energy Convers Manage 2017;138(Apr 15):383–92.
[32] Dehnavi E, Abdi H, Mohammadi F. Optimal emergency demand response
program integrated with multi-objective dynamic economic emission
dispatch problem. J Oper Automat Power Eng 2016;4(1):29–41.
[33] Nojavan S, Qesmati H, Zare K, Seyyedi H. Large consumer electricity
acquisition considering time-of-use rates demand response programs. Arab J
Sci Eng 2014;39(12):8913–23.
[34] The GAMS Software Website, 2017. Available: <http://www.gams.com/
dd/docs/solvers/cplex.pdf>.