i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 4 2 1 7 e1 4 2 2 7

Available online at www.sciencedirect.com

ScienceDirect

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

Stochastic security-constrained operation of wind

and hydrogen energy storage systems integrated
with price-based demand response

Mohammad Amin Mirzaei a, Ahmad Sadeghi Yazdankhah a,*,

Behnam Mohammadi-Ivatloo b
a
Department of Electrical Engineering, Sahand University of Technology, Tabriz, Iran
b
Faculty of Electrical and Computer Engineering, University of Tabriz, Tabriz, Iran

article info abstract

Article history: Because of highly increasing energy consumption, environmental issues and lack of
Received 28 August 2018 common energy sources, the use of renewable energy sources especially wind power
Received in revised form generation technology is increasing with significant growth in the world. But due to the
3 December 2018 variable nature of these sources, new challenges have been created in the balance between
Accepted 6 December 2018 production and consumption of power system. The hydrogen energy storage (HES) system
Available online 28 December 2018 by storing excess wind power through the technology of power to hydrogen (P2H) and
delivering it to the electricity network through hydrogen-based gas turbine at the required
Keywords: hours reduces not only wind alternation but can play an important role in balancing power
Stochastic security-constrained production and consumption. On the other hand, power consumers by participating in
scheduling demand response (DR) programs can reduce their consumption at peak load or wind power
Hydrogen energy storage system shortage hours, and increase their consumption at low-load or excess wind power hours to
Price-based demand response reduce wind power spillage and system energy cost. This paper proposes a stochastic se-
Wind energy curity constrained unit commitment (SCUC) with wind energy considering coordinated
operation of price-based DR and HES system. Price-based DR has been formulated as a
price responsive shiftable demand bidding mechanism. The proposed model has been
tested on modified 6-bus and 24-bus systems. The numerical results show the effect of
simultaneous consideration of HES system and price-based DR integrated with wind en-
ergy on hourly generation scheduling of thermal units. As a result there is some reduction
in wind generation power spillage and daily operation cost.
© 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.

International Energy Agency (IEA), the annual wind produc-

Introduction tion will be increased to 2182 TWh by year 2030 which is seven
times more compared to year 2009 [1]. Therefore, increasing
In recent decades, public concern about climate change and the penetration of renewable energy sources has created new
the reduction of fossil fuels has led to significant investment challenges in the balance between power production and
in renewable energy production. According to the

* Corresponding author.
E-mail address: [email protected] (A. Sadeghi Yazdankhah).
https://doi.org/10.1016/j.ijhydene.2018.12.054
0360-3199/© 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
14218 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 4 2 1 7 e1 4 2 2 7

Nomenclature dmax
n;j;t;s Upper limit of the nth block of PRSL
hP2H
h ; h H2P
h Generation/Storage efficiency of HES system
t Index of time periods
Amax
h ; Amin
h Min/Max capacity of HES system
s Index of scenarios
PFr;t;s Available forecasted wind power at time t and
i Index of thermal units
scenario s
r Index of wind power plant
XL Reactance of line L
h Index of HES system
PFmax Maximum capacity of line L
b; b' Index of buses L
Pi;t;s Generation power of unit i
j Index of loads
Ii;t;s Binary on/off status indicator of unit i
L Index of transmission lines
RO S
i;t;s ; Ri;t;s Operation/Spinning reserve of unit i
NT Number of time periods
RRU RD
i;t;s ; Ri;t;s Up/Down regulation reserve of unit i
NS Number of scenarios on off
Xi;t
1;s ; Xi;t
1;s On/Off time of unit i
NU Number of thermal units
bP2H
h ;c P2H
h Cost function coefficient of HES system in storage
NH Number of HES system
mode
NR Number of wind farms H2P H2P
aH2P
h ; b h ; ch Cost function coefficient of HES system in
NB Number of buses
generation mode
ai ; bi ; ci Cost function coefficient of units
IH2P P2H
h;t;s ; Ih;t;s Binary generation/storage status indicator of HES
ROt System operation reserve requirement
system
RSt System spinning reserve requirement
PP2H H2P
h;t;s ; Ph;t;s Generated/Stored hydrogen of HES system
RRU
t Up-regulation reserve requirement
Ah;t;s Stored hydrogen level of HES system
RRD
t Down -regulation reserve requirement
dn;j;t;s Amount of PRSL in the nth block
SUi Start-up cost of thermal unit i
DRj;t;s Amount of adjustable load
Pi max ; Pi min Min/Max generation capacity of thermal unit i
Pr;t;s Dispatched wind power
RUi ; RDi Ramp up/down thermal unit i
Off Dj;t Expected hourly load j
TOn
i ; Ti Minimum up/downtime of unit i
PFL;t;s Line flow at line L
Ph;max ; PP2H
H2P
h;max Upper limit of the corresponding modes
db;t;s Voltage angle of network buses
PH2P
h;min ; P P2H
h;min Lower limit of the corresponding modes
ps The probability of scenario
DRj;t;max Maximum adjustable load j
CBLj;t Fix load j

consumption due to the variable nature of these sources. To investigated on power system operation cost. Ref. [8] corre-
respond to these challenges, several approaches have been sponds to the impact of participating DR in energy and reserve
proposed in the literature. Operation improvement through market with the integration of wind energy on system oper-
the design of new markets such as flexible ramping market ation cost. The effect of DR in the day-ahead market on
[2,3], coordination of demand response (DR) and wind energy locational marginal prices (LMP) and system operation cost
[4,5], coordinated operation of energy storage system with with high penetration wind energy has been studied in [9]. In
wind farm [6] are some of these approaches. [10], the effect of DR program has been studied on optimal
Demand response (DR) programs play an important role in operation of power and heat micro-grids considering the un-
reliable and economic operation of future power systems and certainties of load demand and market price. The effect of
electricity markets. By performing demand response pro- price-based demand response on the system operation cost
grams, end users reduce their electricity consumption at peak has been investigated in [11] by solving a problem of sto-
load or shortage wind power hours and increase it at low load chastic security constrained unit commitment (SCUC)
or surplus wind power hours. As a result, DR can influence the considering natural gas network constraints. Authors in [12]
electricity prices, peak load reduction, wind power curtail- have proposed a two-stage stochastic SCUC problem for co-
ment, and system operation cost. DR programs are classified ordinated electricity and natural gas networks with the inte-
as incentive-based DR and price-based DR. In incentive-based gration of hourly DR and wind energy considering flexible
DR, the power consumers bid their power consumption ramping products. This study has focused mainly on the
reduction to the independent system operator (ISO), and if participation of DR in energy and flexible ramping markets
their bid is accepted after the market clearing process, the considering natural gas supplies limits. In [13], a full DR model
power consumer has to perform the contract, otherwise will has been introduced to participation in energy and spinning
be fined. On the other hand, in price-based DR which is used in reserve markets simultaneously.
restructured power systems, consumers are able to respond to Hydrogen energy storage (HES) system like the other en-
changes in the price of electricity causing network's load ergy storage systems such as pumped storage unit [14,15],
profile improvement and load reduction in peak periods. In compressed air energy storage (CAES) unit, batteries and
[4,7], the impact of DR coordinated with wind energy has been electric vehicles [16] play important role in enhancing
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 4 2 1 7 e1 4 2 2 7 14219

balance between generation and consumption. As shown in the integration of wind energy. The main features of this
Fig. 1, in HES system, the excess power generated by paper are as follows:
renewable energy sources is converted to hydrogen through
the electrolysis and stored in hydrogen storage. Then, during  Simultaneous consideration of price-based demand
periods of high power demand and low wind power genera- response and HES system in stochastic SCUC with wind
tion, the stored energy in hydrogen storage will be converted energy, which increases system flexibility compared to the
to electricity through the hydrogen-based gas turbine. The condition where these resources are considered alone.
main feature of HES with respect to other similar storage  Price-based demand response has been considered as the
systems is that the stored hydrogen could be used in price responsive shiftable demand bidding mechanism
hydrogen dependent industries and/or injected into the nat- that shifts electricity load from high prices to low prices
ural gas network for gas consumers. Several studies have and reduces wind power spillage and daily operation cost.
been done about the coordinate operation of HES system and  Both power to hydrogen (P2H) technology and hydrogen-
renewable energy sources [17]. For the first time in the decay based gas turbine (H2P) have been considered that con-
of 1990, coordination of HES unit and PV was performed in verts wind power to hydrogen when wind power genera-
[18]. During 2003 and 2004, the Norwegian energy company tion is extra and then converts stored hydrogen to
Norsk Hydro together with the German wind turbine manu- electricity when the network requires power.
facturer Enercon installed the first and largest wind-
hydrogen plant in Utsira, Norway, in 2004, which operates In this paper, SCUC is modeled as a mixed integer
as an isolated power system with 90% availability [19]. In nonlinear programming (MINLP) problem which is solved
Nakskov Denmark, a project of integration HES into a wind using GAMS software with DICOPT solver.
power plant has been successfully producing, storing, and The rest of the paper is organized as follows. The problem
utilizing hydrogen since 2007 [20]. formulation is presented in Section Problem formulation. The
A lot of researches have focused on the effect of energy numerical results of case studies are presented in Section
storage systems on the cost of operation and the wind spillage Numerical simulations. Finally, the conclusions are provided
reduction. The impact of pumped hydro storage unit on wind in Section Conclusion.
power dispatch has been determined to solve stochastic unit
commitment in [21]. In [14], the demand response and pum-
ped hydro storage scheduling considering wind power un- Problem formulation
certainties has been performed to reduce energy and reserve
costs. In [22], flexible interaction of plug-in electric vehicle The formulation of the problem involves the objective func-
parking lots for efficient wind integration is studied in a joint tion, thermal unit constraints, compressed air energy storage
market clearing problem of energy and spinning reserve. A and also natural gas network constraints in the first and sec-
problem of stochastic network constrained energy and ond steps.
reserve market clearing with bulk energy storages, DR, plug-in
electric vehicle parking lots and wind energy has been dis- Objective function
cussed in [23]. In [24], the effect of power to gas storage (P2G)
has been investigated on daily operation cost and wind power The objective of the proposed SCUC is to determine the day-
spillage in a problem of robust scheduling of integrated elec- ahead scheduling of generation power plants, hydrogen en-
tricity and natural gas systems. Finally, the impact of inte- ergy storage system, and price responsive shiftable loads
gration of HES system in a day-ahead stochastic SCUC on the (PRSL) to maximize system social welfare with the integration
reduction of wind power curtailment and system operation of wind energy. The objective function given in Eq. (1) is
cost has been investigated in [25]. maximizing the social welfare which is defined as the con-
In a number of previous mentioned works, the effect of sumer consumption benefits minus the operation cost, and
each emerging technologies such as demand response pro- consists of four parts. The first part represents the revenue of
grams and energy storage systems have been investigated in PRSL. Second part corresponds to the fuel and start-up cost of
the power system operation, individually. Also, in some other, thermal plants. The third and fourth parts are the HES system
the effect of consideration of demand response programs with cost in storage and generation mode. It is assumed that the
storage technologies except HES have been investigated. In HES system in generation mode has cost function the same as
this paper, we have focused on the interaction of price-based gas-fired power plants given in Eq. (4), and the profit of DR
demand response and HES system in day-ahead market with providers only depends on PRSL, not base loads.

Fig. 1 e Schematic diagram of HES system.

14220 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 4 2 1 7 e1 4 2 2 7

2 3
XNP X
NT X NJ XNT X NU
    Any increase or decrease in the power generation of ther-
6 B n;j;t d n;j;t;s
Fi P i;t;s þ SUCi;t;s 7
XNS 6 n¼1 t¼1 j¼1 7 mal unit i at consecutive periods is bounded by RUi and RDi as
ps 6 7
t¼1 i¼1
max 6 X   X   7
s¼1 4 NT X NH NT X NH
5 shown in Eqs. (15) and (16).

FPH
h P P2H
h;t;s
FHP
h P H2P
h;t;s
t¼1 h¼1 t¼1 h¼1 Pi;t;s
Pi;t
1;s  RUi (15)
(1)
Pi;t
1;s
Pi;t;s  RDi (16)
where:
  Minimum on/off times for each unit transition are given in
Fi Pi;t;s ¼ ai P2i;t;s þ bi Pi;t;s þ ci (2)
Eqs. (17) and (18), respectively.
    
P2H P2H
FPH h;t;s ¼ bh Ph;t;s þ ch
PP2H P2H
i;t
1;s
Ti
Xon Ii;t
1;s
Ii;t;s  0
(3) on
h (17)

   2   
H2P off off
FHP
h h;t;s ¼ ah
PH2P H2P
PH2P
h;t;s þ bh PH2P
h;t;s þ ch
H2P
(4) Xi;t
1;s
Ti Ii;t;s
Ii;t
1;s  0 (18)

Maximization problem in this study is subjected to the Finally, start-up cost of each thermal unit is dictated by Eq.
thermal unit and system constraints, wind power plant and HES (19).
constraints, and DR constraints which are described below.  
SUCi;t;s  SUi Ii;t;s
Ii;t
1;s (19)
Thermal unit and system constraints
HES system constraints
Generated power of thermal unit ⅰ is limited by its upper and
Hydrogen energy storage system like the other energy storage
lower values as shown in Eq. (5).
systems has three operating states: power to hydrogen state,
Pmin
i Ii; t; s  Pi; t; s  Pmax
i Ii; t; s (5) hydrogen to power state, and no load state in which Eq. (20)
should be satisfied.
The values of operation reserve, spinning reserve and up/
h;t;s þ Ih;t;s  1
IH2P P2H
down regulation reserve enhanced by thermal generation are (20)
specified as Eqs. (6) and (7). The corresponding reserve con-
Eqs. (21) and (22) show the hydrogen and power generated
straints are given in Eqs. 8e11.
by HES system which are limited by some upper and lower
  limit values.
ROi;t;s þ RSi;t;s þ RRU
i;t;s  Ii;t;s min RUi ; Pi
max

Pi;t;s (6)

h;min Ih;t;s  Ph;t;s  Ph;max Ih;t;s

PP2H P2H P2H P2H P2H
n o (21)
RRD
i;t;s  Ii;t;s min RDi ; Pi;t;s
Pmin
i (7)

h;min Ih;t;s  Ph;t;s  Ph;max Ih;t;s

PH2P H2P H2P H2P H2P
(22)
X
NU
ROi;t;s  ROt (8) Hourly stored hydrogen in HES system given by Eq. (23) which
i¼1 is limited to the maximum and minimum capacity of the unit as
Eq. (24). Eq. (25) defines the initial value of HES capacity, and Eq.
X
NU
(26) specifies that the initial value at t¼0 and final value at t¼24 of
RSi;t;s  RSt (9)
i¼1
the stored hydrogen are the same. Mh;t;s is denoted as the amount
of hydrogen converted to some other types of energy which is
X
NU limited to zero andMh;max as given by Eq. (27).
i;t;s  Rt
RRU RU
(10)
PH2P
h;t;s
i¼1 Ah;t;s ¼ Ah;t
1;s þ hP2H
h Ph;t;s

P2H

Mh;t;s (23)
hH2P
h

X
NU

i;t;s  Rt
RRD RD
(11) Amin  Ah;t;s  Amax (24)
h h
i¼1

Power balance equation in each bus is given by Eq. (12). Ah;0;s ¼ Ah;in;s (25)
Power transmission from bus b to bus b' is determined in Eq.
(13). The transmission line capacity limit is dictated by Eq. (14). Ah;0;s ¼ Ah;24;s (26)
X
NUb X
NRb X
NHb X
NHb X
NJb
 
Pi;t;s þ Pr;t;s þ h;t;s

PH2P h;t;s

PP2H Dj;t;s
DRj;t;s 0  Mh;t;s  Mh;max (27)
i¼1 r¼1 h¼1 h¼1 j¼1

X
NLb Demand response constraints
¼ PFL;t;s (12)
L¼1
Demand response model has been considered as a price
db;t;s
db' ;t;s responsive shiftable demand bidding mechanism in which
PFL;t;s ¼ (13) the loads are classified as the consumer base load (CBL) and
xL
price responsive shiftable load (PRSL). CBLs are not sensitive

PFmax
L  PFL;t;s  PFmax
L (14)
to price variation and predicted by historical data. But PRSLs
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 4 2 1 7 e1 4 2 2 7 14221

are dependable to electricity market price variation and can bus IEEE-RTS system. The proposed mixed integer nonlinear
shift their demand to the other operation times because of programming (MINLP) model is implemented in the general-
economical reasons. The correlation between PRSLs blocks ized algebraic modeling systems (GAMS) software and solved
and total load at each hour and scenario, and also its limits are using DICOPT solver.
defined in Eqs. (28) and (29), respectively [26].
Modified six-bus system
X
NP
dn;j;t;s ¼ Dj;t
DRj;t;s
CBLj;t (28)
n¼1 In order to evaluate the proposed model, a modified 6 bus
system has been studied which depicted in Fig. 2. The modi-
max
0  dn;j;t;s  dn;j;t;s (29) fied 6-bus system consists of three thermal units, three elec-
tric loads, seven transmission lines, a HES system, and a wind
Amount of adjustable load at each hour and each scenario
power plant. The specifications of the thermal units, network
is limited by Eq. (30). Finally, Eq. (31) denotes that the total
load, transmission lines, and wind power generation are given
shifted load for the whole study time is zero.
in Appendix A. The required minimum operating, spinning
  and regulation reserves are assumed 10, 5 and 2% of the
DRj;t;s   DRmax (30)
j;t
network load, respectively.
The cases described below show the simultaneous
X
NT
DRj;t;s ¼ 0 (31) consideration of the hourly demand response and HES to
t¼1 reduce the cost of daily operation and also increase wind
power dispatch.
Wind power plant constraint
Case 1: Solving deterministic SCUC.
The limit of dispatched wind power is the only wind power Case 2: Solving stochastic SCUC.
plant constraint given by Eq. (32). Case 3: Integration of price-based DR in stochastic SCUC.
Case 4: Integration of HES in stochastic SCUC.
0  Pr;t;s  PFr;t;s (32)
Case 5: Integration of price-based DR and HES system in
The wind power generation forecast error can be repre- stochastic SCUC.
sented by a normal distribution function. The probability Case 1: In this case, a wind power plant is located at bus 4.
distribution function (PDF) of the normal distribution is given The hourly generation dispatch of units G1, G2 and G3 has
as been shown in Fig. 3, where unit G1 is considered to be the
ε
ε 2 cheapest unit that is committed over the entire time
1 1
PDFðεÞ ¼ pffiffiffiffiffiffiffiffiffiffiffie
2
0
s ;
∞  ε  ∞ (33) period. Also, units G2 (most expensive unit) and G3 are
2ps2
committed for 2 and 12 h, respectively. The curtailed wind
power and daily operation cost, in this case, are
142.58 MWh and $73919.17, respectively.
Numerical simulations
Case 2: In this case, the Monte Carlo simulation is used to
model the uncertainty of wind power generation. The wind
The proposed stochastic SCUC model considering price-based
power generation forecasting error follows a truncated
DR and HES system is evaluated on a six bus system and the 24

Fig. 2 e Illustration of 6-bus system.

14222 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 4 2 1 7 e1 4 2 2 7

240 G1 G2 G3
220
200
180

Generation power (MW)

160
140
120
100
80
60
40
20
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (hour)

Fig. 3 e Hourly generation dispatch of units G1,G2, and G3 for case 1.

normal distribution with a mean value of zero and the Fig. 6 shows the power generated and stored by HES in
standard deviation of 5%. The 100 generated scenarios scenario 4. HES system, stores excess wind power at 2, 3, 4,
have been reduced to the most probable 5 scenarios using and 7 h, and increases the dispatched wind power in this
the SCENRED tool in GAMS software. The cost of daily scenario. In addition, at hours 13, 14, 15 and 21, HES system
operation and the curtailed wind power in different sce- works at generation mode which by injection of power to
narios are shown in Table 1. In Scenarios 4 and 5, by the network, the hourly commitment and the dispatched
increasing predicted wind power, the cost of daily opera- power of units G2 and G3 are reduced in this scenario. The
tion is reduced in comparison with the case 1 and also, the cost of daily operation and the curtailed wind power, in
curtailed wind power is increased compared to case 1. The this case, are $71866.24 and 39.953 MWh, which are lower
cost of expected daily operation and the curtailed wind than case 2.
power, in this case, are $73946.80 and 128.35 MWh, Case 4: In this case, the demand response program has
respectively. been implemented in all load buses. The PRSL participation
Case 3: In this case, a HES with 180 MWh maximum factor, which is defined as the ratio of available PRSL to the
hydrogen storage capacity and 40 MWh minimum reserve expected load, is fixed at 0.03. The PRSL bid consists of a
capacity is located at bus 4. The corresponding parameters single energy block with a marginal benefit of 45 $/MWh.
of HES are given in Table 2. The initial value of HES is 40% of The total cost of operation, in this case, is $70,432.98, which
its maximal capacity. It is also assumed that the stored is lower than cases 2 and 3. Meanwhile, the total curtailed
hydrogen can only be converted to electricity through a wind power is 103.31 MWh, which is high compared to case
hydrogen-based gas turbine (Mh;t;s ¼ 0). In addition, it is 3. In order to cases fully assess the effect of hourly demand
assumed that HES system in generation mode has cost response on the power system operation cost and the
function the same as gas-fired units [25]. The comparison curtailed wind power, we consider the PRSL participation
of the hourly generation dispatch of units G2 and G3 and factor from 1 to 9%. As shown in Table 3, by increasing PRSL
the dispatched wind power for cases 2 and 3 in scenario 4 participation factor, the cost of daily operation and the
has been shown in Figs. 4 and 5, respectively. In addition, curtailed wind power have been decreased.
Case 5: In this case, the hourly demand response and HES
system are considered simultaneously. The PRSL partici-
Table 1 e Operation cost and curtailed wind power in pation factor is fixed at 0.03, HES system has the same
different scenarios. specifications as for the case 3. The effect of simultaneous
Scenario Probability Operation Curtailed wind consideration of the hourly demand response and HES on
cost ($) power (MW) the reduction of the operation cost and the curtailed wind
1 0.265 74322.20 114.16 power compared with the previous cases has been shown
2 0.128 75288.09 113.26 in Figs. 7 and 8, respectively. In this case, the total curtailed
3 0.284 74028.89 117.89 wind power and the daily operation cost are equal to
4 0.155 73021.77 155.50 27.997 MWh, $69,143.42 respectively, which are signifi-
5 0.168 73047.45 154.86
cantly decreased, compared to the results of previous
Total scenarios 1 73946.80 128.35
cases.
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 4 2 1 7 e1 4 2 2 7 14223

Table 2 e HES system parameters.

Mode c ($/h) b ($/MWh) a ($/MW2h) Pmax Pmin ƞ
Generation 153.4 15.60 0.0027 30 10 0.80
Storage 0 1.41 0.125 30 10 0.6

Fig. 4 e Hourly generation dispatch of units G2 and G3 for cases 2 and 3 in scenario 4.

150 Dispatched wind power in case 3 and scenario 4

140 Dispatched wind power in case 2 and scenario 4
130
Dispatched wind power (MW)

120
110
100
90
80
70
60
50
40
30
20
10
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (hour)

Fig. 5 e Hourly dispatched wind power for cases 2 and 3 in scenario 4.

Modified 24-bus RTS system system are the same as specification of previous case study.
Electric load profile of the network, technical data of trans-
To evaluate the proposed method on medium systems, a mission lines and thermal units are expected from [28]. The
modified 24-bus RTS system has been considered. This system wind power prediction error tracks a normal distribution
consists of 24 thermal units, 34 transmission lines, and 17 function with zero mean and standard deviation of 5%. 100
electric loads. A wind power plant with the capacity of 500 MW scenarios generated by Monte Carlo was reduced to the most
[27] and a HES system with maximum storage capacity of probable 4 scenarios using the SCENRED tool in GAMS soft-
300 MW and maximum generation power and storage of ware. The PRLS participation factor has been assigned as 10%
50 MW were included at bus 6. Other specifications of the HES including an energy block with marginal benefit of 45 $/MWh.
14224 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 4 2 1 7 e1 4 2 2 7

32 HES in generation mode HES in storage mode

28
24
20

Power (MW) 16
12

4
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (hour)

Fig. 6 e Power generated and stored by HES system for case 3 in scenario 4.

Table 3 e Compression of the results with different values of PRSL participation factor.
PRSL participation factor (%) 1 3 5 7 9
Cost of operation ($) 72952.75 70432.98 69998.71 68711.37 67923.64
Total curtailed wind power (MWh) 127.40 103.31 90.10 79.24 67.69

75000

74000

73000
Operation cost ($)

72000

71000

70000

69000

68000

67000

66000
Case 2 Case 3 Case 4 Case 5

Fig. 7 e The effect of simultaneous consideration of hourly demand response and HES on the reduction of the operation cost.

To evaluate the impact of coordinated scheduling of HES operation cost without considering wind power uncertainty
system and price-based DR on both wind power spillage and is $ 614578.93. Table 4 demonstrates the daily operation cost
system operation cost, the following four cases have been and the curtailed wind power for different scenarios in case
considered: 1. It is clear that the operation cost in scenarios 2 and 3 has
been increased in compared to the deterministic cost, and
Case 1: Solving stochastic SCUC. the curtailed wind power has been decreased because of the
Case 2: Integration of price-based DR in stochastic SCUC. reduction in wind power generation prediction. The pre-
Case 3: Integration of HES in stochastic SCUC. dicted operation cost and the total curtailed wind power in
Case 4: Integration of price-based DR and HES system in case 1 are $614629.05 and 167.67 MWh, respectively. In case 2
stochastic SCUC. the impact of price-based DR is considered on the curtailed
wind power and system operation cost. The operation cost
In case 1, the stochastic SCUC problem is solved where a and curtailed wind power have been decreased to $589246.72
wind power plant is located at bus 6. Knowing that the daily and 137.75 MWh. Fig. 9 presents the effect of price-based DR
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 4 2 1 7 e1 4 2 2 7 14225

140
130
120

Curtailed wind power (MWh)

110
100
90
80
70
60
50
40
30
20
10
0
Case 2 Case 3 Case 4 Case 5

Fig. 8 e The effect of simultaneous consideration of hourly demand response and HES on the reduction of the curtailed wind
power.

power. Case 3 corresponds to the impact of integrated HES

Table 4 e Operation cost and curtailed wind power in
system on both operation cost and wind power curtailment.
different scenarios.
The results show that the operation cost in case 3
Scenario probability Operation Curtailed wind
($613794.88) is less than to case 1. Also, the curtailed wind
cost ($) power (MWh)
power has been decreased to 20.66 MWh which is lower than
1 0.09 610354.52 270.465 the results of cases 1 and 2. In case 4, the price-based DR and
2 0.412 615118.83 152.139
HES system are considered simultaneously. Table 5 shows
3 0.236 617986.08 95.848
4 0.262 612303.32 221.486
the results of simultaneous impact of DR and HES system on
system operation cost and total curtailed wind power (last
column of the Table). Comparing to the results of cases 1, 2,
on system load profile in scenario 2. Price based DR is
and 3, the coordinated scheduling of price-based DR and HES
considered at all load buses. Load shifting to off-peak hours
system has positive impact on system operation.
results in smoother network load profile and consequently
reduces the operation cost and the total curtailed wind

Network load profile before implementation of DR

Network load profile after implementation of DR

3000
2800
2600
2400
Load (MW)

2200
2000
1800
1600
1400
1200
1000
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)

Fig. 9 e The effect of DR on network load profile in scenario 2.

14226 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 4 2 1 7 e1 4 2 2 7

Table 5 e Comparison of the operation cost and curtailed wind power in different cases.
Case studies Case 1 Case 2 Case 3 Case 4
Operation cost ($) 614629.05 589246.72 613794.88 588320.02
Curtailed wind power (MWh) 167.67 137.75 20.66 16.43

consideration of these resources has a significant impact on

Conclusion reducing wind power spillage and daily operation cost
compared with the condition when these resources are taken
This paper proposed a problem of stochastic day-ahead se- into account alone.
curity-constrained scheduling with the integrated wind en-
ergy and HES systems considering hourly demand response Appendix A
for managing the uncertainty of wind power generation. The
proposed model has been formulated as a mix-integer The required data of case studies are provided in Tables
nonlinear programming (MINLP) problem. The modeled HES A1.1eA1.3.
system converts the excess wind power to hydrogen using an

Table A1.1 e Thermal units' data for a six-bus system.

Unit c ($/h) b ($/MWh) a ($/MW2h) Pmax (MW) Pmin (MW) Initial Status (h) Min Down (h) Min Up (h) Ramp (MW/h)
G1 176.95 13.51 0.0004 220 100 4 4 4 55
G2 129.97 32.63 0.001 100 10
3 3 2 50
G3 137.41 17.7 0.005 20 10
1 1 1 20

Table A1.2 e Electricity load and wind power data of 24 h.

Hour Demand Wind Hour Demand Wind Hour Demand Wind Hour Demand Wind
(h) (MW) (MW) (h) (MW) (MW) (h) (MW) (MW) (h) (MW) (MW)
1 187.3 69.6 7 292 120.8 13 263.5 130.6 19 282.4 139.0
2 181.9 103.2 8 271.3 95.5 14 258.3 102.4 20 241.6 44.5
3 179.2 128.4 9 255.1 70.2 15 261.1 110.2 21 219 5.4
4 204.1 139.8 10 262.1 67.6 16 259 118.8 22 211.86 10.4
5 220.1 115.6 11 265.7 79.2 17 269 102.5 23 181.8 13.8
6 255 95.6 12 277.3 122.8 18 241.1 137.2 24 163.9 30.8

references
Table A1.3 e Parameters of power transmission branch.
Branch From bus To bus X (p.u.) Flow limit (MW)
[1] He C, Zhang X, Liu T, Wu L, Shahidehpour M. Coordination of
Line 1 1 2 0.17 200
interdependent electricity grid and natural gas networkda
Line 2 1 4 0.258 100
review. Current Sustainable/Renewable Energy Reports.
Line 3 2 3 0.197 100
2018;5:23e36.
Line 4 2 4 0.140 100
[2] Wang B, Hobbs BF. Flexiramp market design for real-time
Line 5 3 6 0.037 100
operations: can it approach the stochastic optimization
Line 6 4 5 0.037 100
ideal?. Power and energy society general meeting (PES) IEEE;
Line 7 5 6 0.018 100
2013. p. 1e5.
[3] Navid N, Rosenwald G. Market solutions for managing ramp
flexibility with high penetration of renewable resource. IEEE
electrolyzer, which then stores hydrogen using a gas storage Transactions on Sustainable Energy 2012;3:784e90.
tank, afterward converts hydrogen to electricity using a [4] De Jonghe C, Hobbs BF, Belmans R. Value of price responsive
hydrogen-based gas turbine. In addition, the price-based de- load for wind integration in unit commitment. IEEE Trans
mand response has been considered as a price responsive Power Syst 2014;29:675e85.
[5] Hajibandeh N, Shafie-khah M, Talari S, Dehghan S,
shiftable demand bidding mechanism which shifts the load
Amjady N, Mariano S, et al. Demand response based
from peak hours to off-peak hours and maximizes social
operation model in electricity markets with high wind power
welfare. In this paper, the effect of simultaneous consider- penetration. IEEE Transactions on Sustainable Energy
ation of the price-based demand response and HES system 2018:1e10. https://doi.org/10.1109/TSTE.2018.2854868.
has been investigated on the dispatched wind power and daily [6] Daneshi H, Srivastava A. Security-constrained unit
operation cost of the power system in a problem of stochastic commitment with wind generation and compressed air
day-ahead scheduling. The results show that simultaneous energy storage. IET Gener, Transm Distrib 2012;6:167e75.
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 4 2 1 7 e1 4 2 2 7 14227

[7] Wu H, Shahidehpour M, Al-Abdulwahab A. Hourly demand [18] Kauranen P, Lund P, Vanhanen J. Development of a self-
response in day-ahead scheduling for managing the sufficient solar-hydrogen energy system. Int J Hydrogen
variability of renewable energy. IET Generation. Transm Energy 1994;19:99e106.
Distrib 2013;7:226e34. [19] Ulleberg Ø, Nakken T, Ete A. The wind/hydrogen
[8] Heydarian-Forushani E, Golshan M, Shafie-khah M, Catala ~ o J. demonstration system at Utsira in Norway: evaluation of
Impacts of stochastic demand response resource scheduling system performance using operational data and updated
on large scale wind power integration. Power Engineering hydrogen energy system modeling tools. Int J Hydrogen
Conference (AUPEC), 2015. Australasian Universities: IEEE; Energy 2010;35:1841e52.
2015. p. 1e6. [20] Alto P, Holeby. First Danish hydrogen energy plant is
[9] Zhao Z, Wu L. Impacts of high penetration wind generation operational. Renewable Energy Access; 2007. http://www.
and demand response on LMPs in day-ahead market. IEEE renewable.energyworld.com/rea/news/article/2007/06/
Transactions on Smart Grid 2014;5:220e9. firstdanish-hydrogen-energy-plant-is-operational-48873.
[10] Nazari-Heris M, Abapour S, Mohammadi-Ivatloo B. Optimal [21] Khodayar ME, Shahidehpour M, Wu L. Enhancing the
economic dispatch of FC-CHP based heat and power micro- dispatchability of variable wind generation by coordination
grids. Appl Therm Eng 2017;114:756e69. with pumped-storage hydro units in stochastic power
[11] Zhang X, Shahidehpour M, Alabdulwahab A, Abusorrah A. systems. IEEE Trans Power Syst 2013;28:2808e18.
Hourly electricity demand response in the stochastic day- [22] Heydarian-Forushani E, Golshan M, Shafie-khah M. Flexible
ahead scheduling of coordinated electricity and natural gas interaction of plug-in electric vehicle parking lots for
networks. IEEE Trans Power Syst 2016;31:592e601. efficient wind integration. Appl Energy 2016;179:338e49.
[12] Zhang X, Che L, Shahidehpour M, Alabdulwahab A, [23] Heydarian-Forushani E, Golshan M, Siano P. Evaluating the
Abusorrah A. Electricity-natural gas operation planning with benefits of coordinated emerging flexible resources in
hourly demand response for deployment of flexible ramp. electricity markets. Appl Energy 2017;199:142e54.
IEEE Transactions on Sustainable Energy 2016;7:996e1004. [24] Chuan H, Tianqi L, Lei W, Shahidehpour M. Robust
[13] Liu G, Tomsovic K. A full demand response model in co- coordination of interdependent electricity and natural gas
optimized energy and reserve market. Electr Power Syst Res systems in day-ahead scheduling for facilitating volatile
2014;111:62e70. renewable generations via power-to-gas technology. Journal
[14] Kiran BDH, Kumari MS. Demand response and pumped of Modern Power Systems and Clean Energy 2017;5:375e88.
hydro storage scheduling for balancing wind power [25] Mingfei B, Jilai Y, Shahidehpour M, Yiyun Y. Integration of
uncertainties: a probabilistic unit commitment approach. Int power-to-hydrogen in day-ahead security-constrained unit
J Electr Power Energy Syst 2016;81:114e22. commitment with high wind penetration. Journal of Modern
[15] Nazari-Heris M, Madadi S, Mohammadi-Ivatloo B. Optimal Power Systems and Clean Energy 2017;5:337e49.
management of hydrothermal-based micro-grids employing [26] Nojavan S, Zare K, Mohammadi-Ivatloo B. Selling price
robust optimization method. Classical and recent aspects of determination by electricity retailer in the smart grid under
power system optimization. Elsevier; 2018. p. 407e20. demand side management in the presence of the electrolyser
[16] Heydarian-Forushani E, Golshan MEH, Shafie-khah M, and fuel cell as hydrogen storage system. Int J Hydrogen
Siano P. Optimal operation of emerging flexible resources Energy 2017;42:3294e308.
considering sub-hourly flexible ramp product. IEEE [27] Sahin C, Shahidehpour M, Erkmen I. Allocation of hourly
Transactions on Sustainable Energy 2018;9:916e29. reserve versus demand response for security-constrained
[17] Gutierrez-Martı́n F, Confente D, Guerra I. Management of scheduling of stochastic wind energy. IEEE Transactions on
variable electricity loads in windeHydrogen systems: the Sustainable Energy 2013;4:219e28.
case of a Spanish wind farm. Int J Hydrogen Energy [28] Force RT. The IEEE reliability test system-1996. IEEE Trans
2010;35:7329e36. Power Syst 1999;14:1010e20.