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 8 ( 2 0 2 3 ) 3 7 5 7 9 e3 7 5 9 3

Available online at www.sciencedirect.com

ScienceDirect

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

Feasibility analysis of green hydrogen production

from wind

Angelica Liponi, Andrea Baccioli, Lorenzo Ferrari*

Department of Energy, Systems, Territory and Construction Engineering, University of Pisa, Pisa, Italy

highlights

Effects of overall size and number of separated electrolysis groups are investigated.

Stack degradation included in the electrolyzer model.

Higher production in 2-group configurations compared to equivalent single-group ones.

Minimum hydrogen selling price between 4.5 and 6.5 V/kg required for profitability.

Dividing the electrolysis plant into more than 2 groups didn't result profitable.

article info abstract

Article history: Renewable hydrogen production has an important role in global decarbonization. However,
Received 23 January 2023 when coupled with intermittent and variable sources, such as wind or PV, electrolyzers are
Received in revised form subjected to part-load and dynamic operation. This can lead to low utilization factors and
8 April 2023 faster degradation of the electrolyzers and affect the specific hydrogen cost. The design
Accepted 4 May 2023 and sizing of such electrolysis systems are fundamental to minimize costs. In this study,
Available online 25 May 2023 several configurations of an electrolysis system producing green hydrogen from a 39 MW-
wind farm are compared. The effects of both the size of the plant and the number of
Keywords: separated groups into which it is divided are investigated. Dividing the plant into two
Wind energy separated groups resulted to be enough to increase hydrogen production; a further increase
Alkaline electrolysis in the number of groups didn't produce significant differences. The most profitable con-
Degradation model figurations resulted that with one or two groups, depending on the hydrogen selling price.
Techno-economic analysis © 2023 The Authors. Published by Elsevier Ltd on behalf of Hydrogen Energy Publications
Hydrogen price LLC. This is an open access article under the CC BY license (http://creativecommons.org/
licenses/by/4.0/).

energy systems. Hydrogen production through water elec-

Introduction trolysis allows the conversion of electrical energy into chem-
ical energy and can contribute to making RE penetration in the
Hydrogen is recognized to be one of the key elements that will grid more reliable and reducing RE curtailment [2]. Further-
help in the global decarbonization efforts [1]. In order to more, hydrogen allows the storage of significant amounts of
reduce carbon dioxide (CO2 ) emissions, the deployment of energy even for extended periods [3].
renewable energies (RE), in particular wind and photovoltaic Hydrogen is also required by many industrial processes,
(PV), for electricity production is continuously increasing. The such as ammonia production, steel making and oil refining,
high share of these sources is causing an increasing need for where it is currently produced for the greatest part with fossil
energy storage systems and the transition to multi-vector fuel-based methods. Its replacement with electrolytic

* Corresponding author.
E-mail addresses: [email protected] (A. Liponi), [email protected] (A. Baccioli), [email protected] (L. Ferrari).
https://doi.org/10.1016/j.ijhydene.2023.05.054
0360-3199/© 2023 The Authors. Published by Elsevier Ltd on behalf of Hydrogen Energy Publications LLC. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).
37580 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 8 ( 2 0 2 3 ) 3 7 5 7 9 e3 7 5 9 3

hydrogen produced using RE will therefore give a great Zheng et al. [16] proposed an alkaline electrolyzer model
contribution to decarbonize these sectors [4], together with taking into account temperature variations and electrolyzer
the adoption of carbon capture systems. Water electrolysis state transitions and employed it in an optimal day-ahead
technologies can be divided into low-temperature (alkaline operation problem of a wind-electrolyzer system. Varela
and polymer electrolyte membrane electrolyzers) and high- et al. [17] proposed an optimal scheduling model for an alka-
temperature electrolyzers (solid-oxide electrolyzers) [5]. line electrolysis system for the efficient conversion of RE into
Among these technologies, the alkaline one is the most hydrogen. Their model takes into account different opera-
mature and widespread [6]. However, alkaline electrolyzers tional states and transitions of the electrolyzers. They applied
have some limitations when operating with fluctuating REs. the model to a case study of a 50 MW wind plant. They found
Indeed, they must operate in the range of 20e100% of the that the optimal number of electrolyzers corresponded to an
nominal power [7] for safety reasons since, at lower loads, the overall nominal power equal to 54% of the wind power peak
ratio of hydrogen to oxygen increases and there is the risk that and was capable of using 89.7% of the annual available energy.
it exceeds the explosion limit [8]. When coupled to fluctuating They did not consider any grid electricity contribution and
renewable sources, electrolyzers are subjected to dynamic found 764 start-up/shutdown cycles evenly distributed among
and part-load operation and to frequent shutdowns that can the units.
cause faster stack degradation [9]. In the case of large-scale electrolysis plants, multi-stack
Many studies deal with coupling of water electrolysis with systems are adopted. The optimal configurations and power
RE, in particular wind and solar PV, and with the sizing of allocation strategies of such systems are worthy of investi-
these systems. The design and the choice of the sizes of gation in order to both maximize the overall efficiency and/or
electrolysis systems coupled to REs is a trade-off between the minimize stacks degradation. Lu et al. [18] proposed an opti-
maximization of RE utilization and the minimization of mization of a power allocation strategy for a wind-hydrogen
hydrogen production costs. On one side, a higher nominal system with a multi-stack PEM water electrolyzer by consid-
power of the electrolysis system increases the RE utilization. ering the degradation status of each stack. Five different
On the other side, it leads to a decrease in the electrolyzers’ operating conditions were considered to determine the stack
utilization factor and, consequently, to an increase in the degradation rate during operation: maintaining, low power
specific hydrogen production cost, as highlighted also by Khan fluctuation, constant turning power, high power fluctuation,
et al. [10]. The effects of the source variability related to the and constant rated power. The proposed strategy was
location and climate conditions on techno-economic perfor- compared to other strategies (average allocation, daisy chain
mances have been investigated both for wind [11] and PV allocation, and optimal efficiency allocation) and showed an
sources [12]. energy efficiency comparable to that of the optimal efficiency
In order to mitigate production fluctuations due to RE allocation strategy (the highest one). Meanwhile, the single-
power input variability, electrical storage can be added to the stack degradation was almost the same for each stack and it
system. The choice of the storage size should be a compro- was lower than the stack degradation (averaged among the
mise between the improvement in the operation of the elec- stack) obtained with the other power allocation strategies.
trolysis system and the limitation of costs. Liponi et al. [13] Hong et al. [19] proposed a control strategy based on the
investigated the effects of the presence of a battery on the segmented-fuzzy control for improving the overall efficiency
mitigation of fluctuations of hydrogen production by an of multi-electrolyzer systems for hydrogen production from
electrolyzer from wind energy. The adopted management wind. Luxa et al. [20] developed a multi-stack PEM modeling
strategy assigned the electrolyzer power input as a function of and simulation framework by using a multi-linear model
both the battery state of charge and a moving average of the class. The model included a linear stack degradation model
wind turbine power. Battery capacities of at least about 6 h and an electrolyzer controller for allocating the operating
were necessary to achieve significant effects but led to non- power to each stack. Rizwan et al. [21] proposed a shared
competitive hydrogen costs. Fang and Liang [14] adopted a balance of plant (BoP) and power supply design for an indus-
modular adaptive control strategy to optimize the electrolyzer trial-scale alkaline electrolyzer plant that reduces capital
operation to overcome wide power fluctuations in a wind- costs with minimum negative effects on operational flexibility
hydrogen system. Supercapacitors were integrated to absorb and performance. The electrolyzer stacks were assumed at
instantaneous power fluctuations. They found a significant different stages of their lifetime and consequently had
reduction in the number of electrolyzer start-stop switching different performances because of stack degradation. They
and an increase in hydrogen production in comparison with found that the layout with separate BoP and power systems
other control strategies. has higher capital investments but offers greater operational
When an electrolysis system is coupled with variable and flexibility when the electrolyzer stacks are degraded. Results
intermittent power input such as RE one, another option to showed that individual BoP per stack (with higher capital
limit frequent shutdowns and consequent fast degradation is costs) is optimal when the plant is expected to operate at high
to keep the electrolyzers in warm stand-by state for short capacity. For shared BoP, the hydrogen production was
periods. In this way, by consuming a small amount of elec- increased by 8e12% when operated with variable lye flowrate
tricity, the stack lifetime can be increased. Matute et al. [15] compared to fixed lye flowrate [21].
developed a multi-state techno-economic model for the In this study, a techno-economic comparison of several
optimal operation of a power-to-hydrogen system. By testing configurations of an alkaline electrolysis system for green
it with real data and comparing it with a model with only two hydrogen production from a 39 MW grid-connected wind farm
states, they found the multi-state model more profitable. has been carried out. Hydrogen is produced with wind
 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 8 ( 2 0 2 3 ) 3 7 5 7 9 e3 7 5 9 3 37581

electricity while grid electricity is used only to keep electro- MW-scale wind farm. Indeed, actual power data already
lyzers in stand-by for short periods of low wind power and, include turbine interaction effects on power production and
thus, limit the number of electrolyzers’ shutdowns. The power fluctuations that are typical of a wind farm and can not
excess of wind electricity that is not used for the electrolysis be easily taken into account by starting from wind speed data.
process is sold to the grid. Each configuration has a different The wind power is sent to the electrolysis system as a
overall electrolysis capacity and consists of a different num- priority while the possible exceeding power is released to the
ber of electrolysis groups that can operate separately. Each grid.
group consists of several 1 MW-electrolyzers that operate The electrolysis system is divided into several electrolysis
together at the same operating conditions and have shared groups that can operate separately. Each group consists of a
power supply and BoP. The shared BoP allows a reduction of number (nelz per group ) of 1 MW alkaline electrolyzers that oper-
BoP capital costs because of the greater size of the compo- ate simultaneously. Electrolyzers in the same groups have a
nents and consequently lower specific capital costs. The shared balance of plant (BoP), including the separators, buffer
purpose is to investigate the effects of both the overall ca- tank, pump, and cooler in the lye circulation loop, and also
pacity of the electrolysis plant and the number of groups on power electronics and gas conditioning. This allows to reduce
the system operation and economic feasibility and to find the the number of BoP components and increase their size. In this
most profitable configuration. way, their capital cost can be reduced. The same number of
Annual simulations of the system operation were per- electrolyzers per group is assumed for each group of the same
formed for each configuration. An analysis of the operational electrolysis system.
states of the electrolysis groups and a comparison in terms of Several combinations of the number of electrolysis groups
annual performance indicators have been carried out. The (ngroups ), from 1 up to 39, and number of electrolyzers per group
profitability of the configurations analyzed is evaluated by (from 1 up to 13) have been analyzed up to overall electrolysis
assuming several hydrogen selling prices with respect to the capacities capable to use the whole nominal power of the
case without any electrolysis plant in which all wind elec- wind plant (39 MW). Combinations with an overall electrolysis
tricity is sold to the grid. capacity under 7 MW were not considered since they could
use just a very small fraction (less than 1/6) of the wind
nominal power.
System description and modeling The power feeding the electrolysis system is equally
divided among all the electrolysis groups in operation. Before
The system under investigation consists of an alkaline elec- being sent to the electrolyzer stacks, wind power is adjusted to
trolysis system for green hydrogen production from a MW- the desired voltage by a transformer and converted to direct
scale grid-connected wind power plant. The electrolysis sys- current by a rectifier. The power feeding the electrolyzers of
tem is divided into several units (groups) of the same size that the same group is equally divided among them. Consequently,
can operate separately. The effects of both the overall nominal electrolyzers of the same group operate all the time at the
power of the electrolysis system and the number of groups same conditions as one single bigger electrolysis unit and are
into which it is divided have been investigated. Several con- subject to similar stack degradation throughout their lifespan.
figurations of the electrolysis system characterized by The model of each single 1 MW electrolyzer takes into ac-
different overall capacities up to a nominal power equal to the count both electrochemical and thermal aspects. The 3 MW
wind plant one and divided into a different number of groups alkaline electrolyzer adopted in Ref. [24] is taken as a refer-
have been compared through annual simulations of the sys- ence. In addition, degradation effects on the polarization
tem operation in MATLAB [22]. Power production data of an curve and consequent losses in the efficiency have been added
actual wind farm, located in South Australia, were taken as to the model. An operating power range between 20% and
reference [23]. They consist in 5 min-averaged power data of 100% of the nominal power is imposed as a typical range for
one year (Fig. 1). The wind farm has a nominal power of 39 MW alkaline electrolyzers [7] to avoid safety problems at low cur-
and consists of 13 turbines, each with a nominal power of rent densities.
3 MW. Its capacity factor is 30.4%. The reason for choosing Given the electrical power input to each electrolyzer (Pelz ),
these data is to use power data that are representative of a real assigned at each time step by the management strategy, the
power feeding the stack (Pstack ) is determined by subtracting
the power required by the BoP, which is assumed to be 8% of
the power input. This percentage was derived from Ref. [24] at
the nominal electrolyzer power. Then, the cell current (Icell )
and voltage (Ucell ) are determined as a function of the stack
temperature (Telz ) by solving the system of Eqs. (1) and (2) by
means of the Newton iteration method.

Pstack ¼ Ucell $Icell $ncells (1)

Ucell ðTelz Þ ¼ Urev ðTelz Þ þ ða1 þ a2 $Telz Þ$i

" ! #
b b (2)
þ s$log b1 þ 2 þ 23 $i þ 1
Telz Telz
Fig. 1 e Annual distribution of the wind farm power output.
37582 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 8 ( 2 0 2 3 ) 3 7 5 7 9 e3 7 5 9 3

Eq. (1) states that the overall stack power is given by the
Table 1 e Parameters of the electrolyzer model.
product of the cell current (Icell ), cell voltage (Ucell ), and the
number of cells (ncells ) in the stack. Eq. (2) is the cell polarization Parameter Value
curve in which the cell voltage is expressed as the sum of the ncells 109
reversible voltage, ohmic overpotential, and activation over- Acell 1.3295 m2
potential. The polarization curve is a function of the operating pressure 16 bar
%wt KOH 25%
conditions and, in particular, of the temperature. Some of the
a1;0 8$105 U m2
coefficients in Eq. (2) are expressed as a function of the stack
a2  7:63$107 U m2 =C
degradation level as described later. s0 0.1795 V
The stack temperature (Telz ) is calculated through the b1 2$103 m2 =A
thermal model proposed by Ulleberg [25], in which a lumped- b2 1$105 m2 K=A
capacitance model is applied and it is determined at each time b3 35 m2 K2 =A
step k as: hF 0.86
Cth 21:7$106 J=K
Dt h i Rth 3$102 K=W
Telz ðkÞ ¼ Telz ðk  1Þ þ Qgen ðk  1Þ  Qloss ðk  1Þ  Qcool ðk  1Þ Ccw 2:5$104 J=K
Cth
UAhx 5$103 W=K
(3)

where Dt ¼ 300 s is the time step of the simulations, Cth is the

overall thermal capacitance, and Qgen , Qloss , and Qcool are the
internal heat generation, the total heat loss to the ambient, proper temperature rise of the cooling water at nominal con-
and the cooling heat, respectively. ditions of about 10  C. The values adopted for the thermal
The internal heat generation is calculated as: model in this study are consistent with the corresponding
values of electrolyzers with similar sizes having the same
Qgen ðkÞ ¼ ðU  Utn Þ$I$ncells (4)
orders of magnitudes and can be considered reasonable for
where Utn is the thermo-neutral voltage. the aim of this study.
The total heat loss is determined by assuming an overall As degradation occurs, the cell voltage-current relation-
thermal resistance (Rth ) and a constant ambient temperature ship changes. In particular, at the same current density,
(Ta ) of 25  C as: overvoltage increases and, as a consequence, the electrolyzer
efficiency decreases. There are no generic phenomenological
Qloss ðkÞ ¼ ðTelz ðkÞ  Ta Þ = Rth (5) degradation models for alkaline electrolyzers published, that
In order to keep the stack temperature within the desired can represent the influence on partial or dynamic load, as well
operating range, a cooling system is required to absorb the as the number of on/offs. However, for the aim of this study, it
excess of heat produced. The cooling system consists of a is not essential to apply a degradation model that describes
cooling water heat exchanger with a simple on/off regulation. the electrochemical processes in detail. The effect of stack
The cooling system is turned on when the stack temperature degradation has been included in the model by expressing the
exceeds the upper threshold value of 73  C and turned off polarization curve also as a function of the stack degradation
when it drops below the lower threshold value of 67  C. The level. A constant degradation rate of 1 mV per equivalent
cooling heat, Qcool , is calculated as: operating hour at nominal load was assumed as typical value
 for alkaline electrolyzers. The maximum number of equiva-
Ccw ðTcwo ðkÞ  Tcwi Þ when the cooling system is on lent operating hours after which a stack replacement is
Qcool ðkÞ ¼
0 when the cooling system is off needed (heq;max ) is fixed equal to 100,000 h (projected
(6) 2030e2050 lifetime of alkaline electrolyzers [1]). As a conse-
where Ccw is the cooling water thermal capacitance, expressed quence, the maximum degradation leads to a degradation
in [J/K], and Tcwi ¼ 15  C and Tcwo are the cooling water inlet and overvoltage (DVnom;max degr ) of 0.1 V (¼ 1 mV/heq $ 100,000 heq ) at
outlet temperatures, respectively. The outlet cooling water the nominal current density.
temperature is calculated as: At each time step, the degradation level is defined as the
ratio between the equivalent operating hours (heq;degr ) at that
  
UAhx time step and the maximum equivalent operating hours:
Tcwo ðkÞ ¼ Tcwi þ ðTelz ðtÞ  Tcwi Þ$ 1  exp  (7)
Ccw 
Ldegr ðkÞ ¼ heq;degr ðkÞ heq;max (8)
where UAhx is the product of the global heat transfer coeffi-
cient and the heat transfer area of the cooling water heat The additional overvoltage (DVnom;degr ) due to degradation at
exchanger. nominal power is assumed to be proportional to the degra-
The values of the parameters for the thermal model (Table dation level and calculated at each time step as:
1) are based on the reference model and appropriately scaled
DVnom;degr ðkÞ ¼ Ldegr ðkÞ$DVnom;max degr (9)
to be applied to the nominal power of the single electrolyzer
that is 1/3 of that of the reference electrolyzer. Indeed, as for The additional overvoltage changes with the current den-
the thermal model, it should be considered that the thermal sity. To take into account this, the polarization curve has been
capacitance and resistance changes with the size. The UAhx expressed as a function of the degradation level. In particular,
coefficient of the heat exchanger is chosen in order to have a the parameters s and a1 are expressed as:
 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 8 ( 2 0 2 3 ) 3 7 5 7 9 e3 7 5 9 3 37583

Therefore, the number of stack replacements (nrepl )

s ¼ s0 $kdegr ; a1 ¼ a1;0 $kdegr (10)
required during the entire plant lifetime is determined as the
where kdegr is determined, by imposing the degradation over- maximum between the ratio between the plant lifetime (Ltpl )
voltage (DVnom;degr ) at nominal current, as: and the stack lifetime, rounded down to the nearest whole
 number:
I
 Vnom;degr Ldegr e Urev $ ða2 $ Telz Þ$ cell;max
Acell  
kdegr Ldegr ¼ I Icell;max
(11) Ltpl  0:01
a1;0 $ cell;max
Acell
þ s0 $ log b$ Acell
þ 1 nrepl ¼ max (19)
Ltelz
where b¼b1þb2/Telzþb3/TelTelz. In Eq. (19), 0.01 is subtracted to the plant lifetime just to
The equivalent operating hours of each electrolyzer used avoid counting a replacement that may occur just at the end of
for determinate the degradation level (heq;degr ) at each time the plant lifetime. It is simply a mathematical expedient that
step k are calculated as: do not alter the results.
Pk
j¼1 Pelz ðjÞ Dt
heq;degr ðkÞ ¼ $ þ nOFF/ON ðkÞ (12) Management strategy
Pelz;nom 3; 600 s=h

where nOFF/ON ðkÞ is the number of electrolyzer start-ups from A management strategy is applied in order to decide the
the OFF state from the beginning of the plant lifetime up to the operational state of each electrolysis group and the elec-
time step k. Each transition from OFF to ON state is assumed tricity flows. Wind power is supposed to feed the electrolysis
equivalent to one operating hour. system as a priority, i.e., whenever possible by respecting the
The imposed minimum stack input power slightly increases imposed constraints. The excess power that can not be used
with the degradation level to comply with the minimum by the electrolysis system is released to the external grid.
allowed current density. Instead, the imposed maximum Three operational states are defined for the electrolysis
power input to the electrolyzer stack does not change with groups: ON, stand-by (SB), and OFF states. If the state of a
degradation and is set equal to the nominal stack power. group is ON, the electrolyzers of the group are operating at
The hydrogen production rate, expressed in mols per sec- the same power that must be within the allowed operating
ond, and the electrolysis efficiency based on the lower heating power range. If the state of a group is OFF, all the electro-
value (LHV ¼ 241 kJ=mol) are calculated at each time step as, lyzers of the group are OFF, no power is consumed, and no
respectively: hydrogen is produced by the group. If the state of a group is
nc $I SB, all the electrolyzers of the groups are kept in temperature
n_H2 ¼ hF (13) and pressure, and no hydrogen is produced. The power
2$F
consumption during SB is assumed to be 5% of the nominal
n_H2 $LHV power and to be provided by the grid. No power is provided by
hLHV ¼ (14)
Pelz the grid for the electrolysis operation apart from that
required for SB states.
where hF is the Faraday efficiency assumed to be 0.86, as re-
The following further constraints are imposed:
ported in Ref. [24].
The annual hydrogen production (MH2 ;ann ), expressed in
- An electrolysis group can be turned ON from the OFF state
kilograms, is calculated as:
only after a minimum time of 2 h.
X
2:016 105120 - If the wind power is not sufficient to keep in the ON state
MH2 ;ann ¼ $ n_H2 ðkÞ$Dt (15) all the groups operating in the previous time step, the
1000 k¼1
maximum possible number of groups already operating
The annual equivalent operating hours of each electrolyzer are kept on, while the rest is put in SB state. First, the
(heq;degr;ann ) are calculated as: groups with a higher number of equivalent operating
P105120 hours are switched SB.
k¼1 Pelz ðkÞ h - If the wind power exceeds the maximum allowed power
heq;degr;ann ¼ $8760 þ nOFF/ON (16)
105120$Pelz;nom y
of the groups operating in the previous time step, further
The annual utilization factor (UF) of each electrolyzer is groups are turned ON, if possible. The number of groups
defined as the ratio between average input power, including turned ON is the minimum number that allows to use the
OFF periods, and the nominal electrolyzer power: exceeding power as much as possible. First, the groups in
P105120 SB for more time are turned ON. Then, if there are no more
k¼1 Pelz ðkÞ groups in SB, the groups that were OFF in the previous
UF ½% ¼ $100 (17)
105120$Pelz;nom time step are turned on if possible and if the minimum
The stack of each electrolyzer is supposed to be replaced imposed off time has been already reached. When
when the degradation level reaches 100% or after a maximum switching groups from OFF to ON, the priority is given to
of 10 years of operation. Consequently, the electrolyzer stack the groups with a higher number of equivalent operating
lifetime (Ltelz ), expressed in years, is determined as: hours.
- If the electrolyzers of a group are not turned ON after a
 
heq;max maximum duration of continuous SB state of half an hour,
Ltelz ¼ min ; 10 (18)
heq;degr;ann they are turned OFF.
37584 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 8 ( 2 0 2 3 ) 3 7 5 7 9 e3 7 5 9 3

The strategy adopted aims to limit as much as possible 1 MW-electrolyzers; and the third one has only one electrol-
state changes. Furthermore, it aims to keep the equivalent ysis group made of twenty-four 1 MW-electrolyzers operating
operating hours of the groups as much similar as possible. The always together.
minimum OFF duration is imposed to avoid too frequent By increasing the number of groups of a smaller size,
electrolyzers on/offs. changes in the operational state are more frequent. On the
other side, the electrolysis system is in operation (with at least
one group) for a greater percentage of time. It can be notice
Analysis of the operational states of the that there are short frequent SB periods between two suc-
electrolysis groups cessive ON states when wind power has strong fluctuations.
This is more marked in the configurations with more and
In general, at the same overall electrolysis capacity, several smaller groups. SB state allows to avoid too frequent shut-
configurations with a different number of electrolysis groups downs and consequent losses in production.
are possible. The lower is the number of groups, the greater is For each configuration, each electrolysis group operates
the number of 1 MW-electrolyzers per group. For example, for separately and, consequently, has its own power input and
an overall electrolysis capacity of 24 MW, there are eight production trend over time. However, because of the adopted
possible configurations (ngroups  nelzs per group ): 1  24, 2  12, management strategy, all the groups of the same configura-
3  8, 4  6, 6  4, 8  3, 12  2, and 24  1. Given an overall tion have similar annual operating conditions. This was
electrolysis capacity, the plant operation is influenced by the confirmed by the results obtained. For this reason and for
configuration adopted. Indeed, smaller and more numerous making an easier comparison, the results obtained for each
groups allow the system to operate even at lower wind power. configuration that are shown in the following are averaged
On the other side, smaller groups have a narrower power among all the groups of each configuration.
operating range and, therefore, they may be more frequently At the same overall nominal electrolysis power, by
subject to changes in the operational state. As an example, increasing the size of each group (i.e., by reducing the number
Fig. 2 shows the operation of all the groups in three different of groups), the average annual percentage of time during
configurations with the same overall nominal electrolysis which a group is in operation (ON state) increases (Fig. 3). In
power of 24 MW during a 7-day period. The first configuration the configurations with only one group, the operating power
has six electrolysis groups, each with four 1 MW-electrolyzers; range of the single larger group is wider than in the configu-
the second one has two electrolysis groups, each with twelve rations with a higher number of smaller groups. Therefore,

Fig. 2 e Example of operational states of the electrolysis groups for a 7-day period in three cases with a different number of
groups (6, 2, and 1 groups) at the same overall electrolysis capacity of 24 MW.
 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 8 ( 2 0 2 3 ) 3 7 5 7 9 e3 7 5 9 3 37585

Fig. 3 e Annual percentage of time during which each electrolysis group is in the ON, SB, and OFF states (averaged among all
the groups for each configuration) for configurations with a different number of electrolysis groups at the same overall
nominal electrolysis power of (a) 12 MW, (b) 24 MW, and (c) 36 MW.

the single group can be in operation for more time. It must be The distributions of SB durations (Fig. 4) give information
pointed out that this does not mean that the overall plant on how much the SB state helps in limiting the number of
produces more hydrogen or is in operation for more time in shutdowns and on the appropriateness of the maximum
the configuration with only one group. Indeed, in the case of duration imposed. A maximum SB duration of 30 min was
two or more groups, each group can operate in a different imposed by the management strategy, after which, if the
period and globally cover a greater percentage of the year in group was not turned on, it was turned off. In all the config-
comparison to the configuration with one group. In confir- urations, the two more frequent SB durations are the smallest
mation of this, as shown later (Fig. 7b), the utilization factor of and the greatest one. Slightly more than 1/3 of the SB periods
the electrolyzers is about the same or even slightly higher in lasts the maximum time of 30 min for all the configurations.
the configurations with more than one group. This high frequency is due to imposed maximum SB duration:
The overall SB duration, averaged among all the groups of an electrolyzer that would have been in the SB state for more
each configuration, accounts for between about 4 and 6% of than 30 min is forced to be turned off. Instead, the frequency
the year in all the configurations analyzed. It decreases with of the shortest (5 min-) SB periods is more variable depending
the decrease in the number of groups when the overall nom- on the configuration. Generally, SBs that last only 5 min are
inal power is under 25 MW (Fig. 3aeb). This is consistent with less frequent (in percentage terms) in the configurations with
the more frequent changes in the operational states in the a higher number of groups. The rest of SB durations
configurations with more electrolysis groups. For overall (10e25 min) accounts for between about 25 and 35% of the
electrolysis power greater than 25 MW, differences in the total number of SBs.
overall group-averaged SB duration decrease (as occurs in The distribution of OFF durations has a decreasing trend
Fig. 3c). with the OFF duration as can be seen from the decreasing

Fig. 4 e Group-averaged annual distribution of SB duration for several configurations with a different number of electrolysis
groups at the same overall nominal electrolysis power of (a) 12 MW, (b) 24 MW, and (c) 36 MW.
37586 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 8 ( 2 0 2 3 ) 3 7 5 7 9 e3 7 5 9 3

slope of the cumulative frequency in Fig. 5. The ordinate of the indicated by the marker “x” in Fig. 5). Most of the OFF periods
points on the curves of the graphs (Fig. 5) represents the (>50%) have a duration of less than 6 h (as can be seen from
group-averaged annual number of OFF periods that last less the marker “þ”, indicating the cumulative frequency of the
than the OFF duration indicated by the abscissa. The abscissa 50% of the annual OFF events). According to these results, by
of the point marked by the “x” symbol is the maximum OFF allowing grid electricity support for a maximum of 6 contin-
duration occurred (the averaged value among all the groups) uous hours when wind power is not enough to continue to
while the y-coordinate is the annual (group-averaged) number operate an electrolysis group, the number of OFF periods
of OFF periods. The most frequent OFF lengths are the shortest would be at least halved.
ones, as can be observed from the steeper slope at low average Most of the annual overall OFF duration (>50%) is consti-
OFF durations. Reducing the number of short OFF periods, tuted by less than one-fourth of the total number of OFF pe-
which can be seen as an index of the frequency of shutdowns, riods with the highest OFF durations as can be observed by the
has a positive effect on the electrolyzers lifetime. The per- increased slope of the curves between the 50%- and the 100%-
centage of short OFF periods with a length under 2.5 h de- annual cumulative OFF duration (marked by the symbols "þ”,
pends on the configuration and varies between 14 and 57%. It and “x”, respectively, in Fig. 6). This is more accentuated in the
decreases with the increase in the overall nominal power and configurations with only one electrolysis group. To eliminate
with the increase in the number of groups. The annual this part of the cumulative OFF duration, consisting of the
maximum OFF duration can reach about 4e5 days (as longest OFF periods, would require a large amount of external

Fig. 5 e Group-averaged cumulative frequency of OFF durations for several configurations with a different number of
electrolysis groups at the same overall nominal electrolysis power of (a) 12 MW, (b) 24 MW, and (c) 36 MW “þ” indicates 50%
of the annual OFF events. “x” indicates 100% of the annual OFF periods.

Fig. 6 e Group-averaged OFF-duration cumulative sum in ascending order for several configurations with a different number
of electrolysis groups at the same overall nominal electrolysis power of (a) 12 MW, (b) 24 MW, and (c) 36 MW “þ” indicates
50% of the annual cumulative OFF duration. “x” indicates the annual cumulative OFF duration.
 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 8 ( 2 0 2 3 ) 3 7 5 7 9 e3 7 5 9 3 37587

electricity or large electric storage and would not be environ- obtained at the expense of a reduction in the utilization factor
mentally (in the case of non-RE electricity support) nor of the electrolyzers (Fig. 7b) and an increase in the number of
economically convenient. Furthermore, the presence of long operational state changes. In turn, this has a negative impact
OFF periods affects to a lesser extent the annual number of on the specific hydrogen production costs.
electrolyzer shutdowns in comparison with the more frequent At the same number of electrolysis groups, the number of
shorter OFF periods. start-ups increases up to a certain overall electrolysis capacity
and then decreases for higher capacities (Fig. 8). The overall
electrolysis capacity at which the maximum number of start-
Annual performance indicators ups is reached depends on the number of groups and is
generally greater in the case with fewer groups. At low overall
All the configurations analyzed are compared in terms of electrolysis capacities, the configurations with fewer elec-
several annual performance indicators. The first indicator is trolysis groups show a lower number of start-ups. The oppo-
the annual overall hydrogen production (Fig. 7a). Annual site happens at higher capacities (greater than about 25 MW).
hydrogen production generally increases with the overall The annual number of start-ups from the OFF state,
electrolysis capacity since more wind electricity can be con- together with the utilization factor of the electrolyzers, in-
verted into hydrogen at higher electrolysis capacities. In the fluences stack degradation. In turn, stack degradation de-
configurations with only one group, the increase in hydrogen creases the electrolyzers’ efficiency and, consequently, the
production with the increase in the overall nominal electrol- hydrogen production. Furthermore, it may cause the need of
ysis power is lower, especially when the electrolysis power is an early stack replacement. Stack degradation decreases
above about half the wind nominal power. This is accentuated with the increase in the overall nominal electrolysis power
for increasing overall electrolysis capacities. (Fig. 9) because of the decrease in the utilization factor of the
Given the same number of groups, when the electrolysis electrolyzers (Fig. 7b). At greater overall nominal electrolysis
capacity increases, the minimum operating power of the power, stack degradation is lower in the configurations with
system increases in all the configurations since the size of only one group compared to the other configurations at the
each group increases. However, in the case with only one
group, the minimum operating power increases more. This
causes the impossibility of using wind power lower than that
minimum value and, consequently, a lower hydrogen pro-
duction. Even if the percentage of the annual time during
which each electrolysis group is in operation is higher in the
case with only one group (Fig. 3), the annual time during
which at least one group of the plant is in operation is lower in
comparison to the configurations with more than one group.
The presence of two groups is already enough to avoid a
marked reduction of the overall hydrogen production at the
greatest overall electrolysis capacities.
Furthermore, a small reduction in the annual overall
hydrogen production with the increase in the overall elec-
trolysis capacity is obtained at the highest overall nominal
power in the configurations with only one group.
In general, when increasing the overall electrolysis capac- Fig. 8 e Annual average number of start-ups from the OFF
ity, the increase in the hydrogen production (Fig. 7a) is state of each electrolyzer vs overall nominal electrolysis
power for a different number of electrolysis groups.

Fig. 7 e (a) Annual overall hydrogen production and (b) group-averaged utilization factor of the electrolyzers vs overall
nominal electrolysis power for a different number of electrolysis groups.
37588 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 8 ( 2 0 2 3 ) 3 7 5 7 9 e3 7 5 9 3

The annualized cost of the electrolysis system, including

operation and maintenance costs and the possible stack
replacement cost, is determined as:
 
Celz;ann ¼ CRF$ Ccap þ Crepl þ CO&M (21)

where Ccap is the capital cost of the electrolysis system, Crepl is

the cost of stack replacement, and CO&M is the annual opera-
tion and maintenance cost, assumed to be 2% of the capital
cost [26].
The capital recovery factor, CRF, is given by:
.h i
CRF ¼ IR$ð1 þ IRÞLtpl ð1 þ IRÞLtpl  1 (22)

in which an interest rate (IR) of 5% [27] is assumed.

Fig. 9 e Annual increment of the stack degradation level vs The specific capital cost of a single electrolyzer (celz;ref ),
overall nominal electrolysis power for a different number including BoP, is assumed to be 1100 V/kW [28]. Half of this
of electrolysis groups. cost is assumed to be the cost of BoP (including power elec-
tronics and gas conditioning) [28]. When the electrolysis
same overall electrolysis capacity. Instead, there are no groups have more than one electrolyzer, the BoP is shared by
relevant differences in the annual stack degradation among all the electrolyzers of the group. The specific cost of the
the configurations with more than one group. This result is electrolyzers without including the shared BoP is supposed
consistent with the lower utilization factor of the electro- to remain the same in all the cases. Instead, the specific cost
lyzers in the case with only one group that is enough to of BoP is supposed to decrease with the increase in the size of
compensate the missed reduction of start-ups at the highest each group. A specific cost (cBoP;ref ) of 550 V/kW is assumed for
overall electrolysis capacities (Fig. 8). The annual average the reference size of 1 MW (the size of the single electrolyzer)
stack degradation level varies between 2.5 and 6.5%, that lead to a reference BoP capital cost (Ccap;BoP;ref ) of 5:50$105
depending on the configuration. As a consequence, stacks V. The generally adopted exponential law was used to eval-
are always replaced after the maximum imposed period of 10 uate the BoP cost at different sizes of the groups. The BoP cost
years, and at that moment, the maximum degradation is not is calculated for each group as:
yet reached.
nelz per group fscale
Ccap;BoP;group ¼ Ccap;BoP;ref $ (23)
1
Economic and feasibility analysis A scale factor, fscale , of 0.677 was assumed as in Ref. [28].
The capital cost of each group is determined as the sum of
In order to compare all the different sizes and configurations the cost of the stacks and the cost of the BoP. The capital cost
of the electrolysis system in economic terms, the levelized of the overall electrolysis system (Ccap ) is the sum of the capital
cost of hydrogen and the equivalent hydrogen price (EHP), costs of the electrolysis groups.
defined below, are determined. Since all the groups are operated to have a similar degra-
The levelized cost of hydrogen (LCOH) is calculated as dation level, all the stacks are supposed to be replaced in the
the sum of the contribution given by the capital and oper- same period. Therefore, the cost of stack replacements is
ation and maintenance costs of the electrolysis plant, determined as:

8 X
nrepl
<
ngroups $nelzper group $ 0:3$celz;ref $Pelz;nom ð1 þ IRÞr $ Ltelz if nrepl > 0
Crepl ¼ (24)
: r¼1
0 if nrepl ¼ 0

including stack replacements (LCOHelz sys ); the contribution In Eq. (24), the replacement cost of each stack is assumed to
of the wind electricity cost (LCOHw el ); and the contribution be 30% of the overall capital cost [29] of a single electrolyzer.
of the grid electricity cost purchased for keeping the elec- The annual overall operation & maintenance cost of the
trolyzers in the SB state (LCOHSB el ). The first contribution is electrolysis system, CO&M , is assumed to be 2% of the capital
given by the ratio between the annualized cost of the cost [26].
electrolysis system (Celz;ann ) and the annual hydrogen pro- The contribution of wind electricity cost to the LCOH is
duction (MH2 ;ann ). given by:

LCOHelz sys ¼ Celz;ann MH2 ;ann (20) Ew/elz sys;ann $cw el
LCOHw el ¼ (25)
MH2 ;ann
 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 8 ( 2 0 2 3 ) 3 7 5 7 9 e3 7 5 9 3 37589

The contribution of grid electricity cost for SBs to the LCOH two groups. At low capacities, the configurations with only
is given by: one group have a lower capital cost of BoP due to the greater
BoP size in comparison to configurations with more groups
Egrid/elz sys ðSBÞ;ann $cSB el
LCOHSB el ¼ (26) and, at the same time, hydrogen production is almost the
MH2 ;ann
same. At higher capacities, even if they have lower capital
An average specific cost of wind electricity of 40 V/MWhel is costs of BoP, their hydrogen production is lower than that of
assumed by considering the projected decrease of wind lev- the configurations with at least two groups. This is the reason
elized cost of electricity and the 2019 global weighted-average why the lowest LCOH is obtained with configurations with two
levelized cost of electricity for onshore wind of 53 $/MWh, re- groups. It results not convenient to have more than two
ported by IRENA [30]. A slightly higher average electricity cost of groups for any of the considered overall electrolysis capacities.
60 V/MWhel is assumed for grid electricity used during SB states. The equivalent hydrogen price (EHP), defined by Eq. (27),
The second economic parameter defined and evaluated to increases with the overall electrolysis capacity (Fig. 11) and
make a comparison of the configurations of the electrolysis varies between 4.5 and 6.5 V/kg, depending on the configura-
plant is the equivalent hydrogen price (EHP). It has been tion. The reason is that at low electrolysis capacities, the uti-
defined as the price at which the hydrogen produced should lization factor is higher. This lowers the specific costs related
be sold to have the same net income that would be obtained by to the electrolysis plant. As a consequence, lower hydrogen
selling all the wind electricity directly to the grid. Accordingly, selling prices are required to obtain the same net income as
it was calculated as: that in the case of directly selling all the wind electricity to the


Ew $psell;w el þ Celz;ann þ Egrid/elz sys ðSBÞ;ann $cSB el e psell;w el $ Ew/grid;ann
EHP ¼ (27)
MH2 ;ann

In Eq. (27), the first addend in the numerator represents the

annual income in the case without electrolytic hydrogen
production, in which all the wind electricity produced (Ew ) is
sold at the price psell;w el . The second addend is the annualized
cost of the electrolysis system. The third addend is the annual
cost for purchasing grid electricity for SBs at the price cSB el .
Finally, the subtrahend is the annual revenue from selling the
remaining wind electricity to the grid (Ew/grid;ann ) at the same
selling price of the case without hydrogen production (psell;w el ).
For this evaluation, a fixed selling price of wind electricity to
the grid of 50 V/MWh and a slightly greater purchase price of
grid electricity for SBs of 60 V/MWh were imposed.
Finally, for all the configurations analyzed, the difference
in the annual gain (DG) between the cases with and without
the electrolysis system has been evaluated by assuming a Fig. 10 e Levelized cost of hydrogen (LCOH) vs overall
hydrogen selling price. Three values of hydrogen selling price nominal electrolysis power for a different number of
(pH2 ;sell ) have been considered: 5, 7.5, and 10 V/kg. electrolysis groups.

DG ¼ pH2 ;sell $MH2 ;ann þ psell;w el $ Ew;ann  Ew/elz sys;ann
 cSB el $Egrid/elz sys ðSBÞ;ann  Celz sys;ann  psell;w el $Ew;ann (28)

Following this definition, when the hydrogen selling price

is greater than EHP, DG is positive and, therefore, it is profit-
able to add the electrolysis plant to the wind farm.
The levelized cost of hydrogen (Fig. 10) increases with the
overall electrolysis capacity and varies between 3.9 and 5.9
V/kg, depending on the configuration. Indeed, only wind
electricity is used for hydrogen production. This causes a
decrease in the utilization factor of the electrolyzers (and,
consequently, in the hydrogen production of each electro-
lyzer) when the overall electrolysis capacity increases while
the capital cost increases. At the same electrolysis capacity,
the LCOH of the configuration with only one group is the Fig. 11 e Equivalent hydrogen price (EHP) vs overall
lowest for nominal capacities of up to about 28 MW. At higher nominal electrolysis power for a different number of
electrolysis capacities, it becomes more convenient to have electrolysis groups.
37590 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 8 ( 2 0 2 3 ) 3 7 5 7 9 e3 7 5 9 3

grid. At the same electrolysis capacity, the EHP of the config- not an electrolysis system, and all the wind electricity is sold
uration with only one group is the lowest for nominal capac- to the grid has been evaluated (Fig. 12, graphs on the left). It
ities of up to about 28 MW. At higher electrolysis capacities, shows a maximum for a certain value of the overall electrol-
the lowest EHP is found with two groups because the higher ysis capacity which depends on the hydrogen price. Indeed, at
hydrogen production overcomes the higher specific BoP cost low overall electrolysis capacities, an increase in the overall
in comparison to the configuration with only one group. capacity produces an increase in hydrogen production that is
By fixing several hydrogen selling prices, the annual gain sufficient to overcome the increase in the capital costs.
difference (Eq. (28)) with respect to the case in which there is Instead, at high overall capacities, the increase in the capital

Fig. 12 e Annual gain difference (graphs on the left) and corresponding annual specific gain difference (graphs on the right)
vs overall nominal electrolysis power for a different number of electrolysis groups with a hydrogen selling price (pH2 ;sell ) of (a)
5 V/kg, (b) 7.5 V/kg, and (c) 10 V/kg.
 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 8 ( 2 0 2 3 ) 3 7 5 7 9 e3 7 5 9 3 37591

costs is not compensate by an adequate income from the electrolysis capacities, the lowest LCOH is obtained with
increased hydrogen production. The electrolysis plant capac- two groups.
ity giving the maximum gain difference increases for
The equivalent hydrogen price (EHP), defined as the selling
increasing hydrogen selling prices because of the increased price at which the net income obtained is the same as that
specific income from selling hydrogen. in the case without an electrolysis plant in which all the
At lower overall electrolysis capacities, at the same ca- wind electricity is sold to the grid, increases with the
pacity, the gain difference is the highest for the configuration overall electrolysis capacity, and varies between 4.5 and 6.5
with only one group. At a hydrogen selling price of 7.5 V/kg or V/kg, depending on the configuration. Configurations with
higher, the maximum gain difference is obtained with two only one or two groups, depending on the hydrogen selling
groups and at an overall capacity that increases with the price, give the maximum net income at an overall nominal
hydrogen selling price. The maximum gain is obtained with a power increasing with the hydrogen selling price and
single group with a nominal capacity of 10 MW when pH2 ;sell ¼ 5 resulted the most profitable.
V/kg, with two groups and an overall nominal electrolysis
capacity of 24 MW when pH2 ;sell ¼ 7.5 V/kg, and with two groups
and an overall nominal electrolysis capacity of 28 MW when Declaration of competing interest
pH2 ;sell ¼ 10 V/kg.
The specific gain difference per kilogram of hydrogen de- The authors declare that they have no known competing
creases with the overall electrolysis capacity (Fig. 12, graphs financial interests or personal relationships that could have
on the right), coherently with the opposite trend of EHP. At the appeared to influence the work reported in this paper.
same overall electrolysis capacity, the highest specific gain
differences are obtained with one group up to a nominal
electrolysis power of 28 MW. By further increasing the overall Nomenclature
electrolysis capacity, slightly higher maximum specific gain
differences are obtained with two groups.
Symbols
C cost [V]
c specific cost [V/MWh]
Conclusions
Ccw cooling water thermal capacitance [J=K]
Cth electrolyzer thermal capacitance [J=K]
Several configurations of an alkaline electrolysis system for
E annual energy [MWh]
green hydrogen production from a 39 MW-wind farm were
heq max maximum equivalent operating hours before stack
compared by analyzing techno-economic indicators. The ef-
replacement is needed [h]
fect of both the size of the electrolysis plant and the number of
heq;degr equivalent operating hours for degradation [h]
separated groups into which it is divided was investigated.
F Faraday constant (F ¼ 96485 C=mol)
Three operational states were considered for each group: on,
DG annual gain difference [V]
stand-by, and off states. The model adopted took into account
I current [A]
stack degradation as a function of the equivalent operating
i current density [A=m2 ]
hours and the number of start-ups from the off state.
k index for the time step
The main findings are:
Ldegr degradation level
Lt lifetime [y]

Most of the OFF periods have a duration of less than 6 h.
MH2 ;ann annual hydrogen production [kg]
This result suggests investigating the possibility of both
n_H2 hydrogen production rate [mol=s]
adding electrical storage or allowing grid support during
ncells number of stack cells
short periods of low wind power in order to reduce the
ngroups number of electrolysis groups
number of OFF periods. The effect on CO2 emissions per
nelz per group number of electrolyzers in each group
kilogram of produced hydrogen should be evaluated since
nOFF/ON number of state transitions from OFF to ON
a small increase in specific emissions could allow to obtain
nrepl number of stack replacements
both a more continuous production and a significant
P power [W]
reduction in specific costs by increasing the electrolyzer
Q heat [W]
utilization factor and reducing stack degradation.
p price [V=MWh]

When the overall nominal electrolysis power is greater
pH2 sell hydrogen selling price [V=kg]
than half the wind one, higher hydrogen production is
Rth overall electrolyzer thermal resistance [K=W]
obtained with configurations with two groups instead of
T temperature [ C]
those with only an equivalent bigger group. The difference
U voltage [V]
becomes more marked for increasing overall electrolysis
UAhx product of the global heat transfer coefficient and
capacities. A further increase in the number of groups does
the heat transfer area of the exchanger [W/K]
not influence much hydrogen production.
Dt time step [s]

The lower is the electrolysis capacity, the lower is LCOH. At
hF Faraday efficiency
the same electrolysis capacity, the lowest LCOH is obtained
in the configurations with only one group for a nominal Subscripts/superscripts
electrolysis power of up to about 28 MW. At higher a ambient
37592 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 8 ( 2 0 2 3 ) 3 7 5 7 9 e3 7 5 9 3

ann annual, annualized specific energy consumption. Int J Hydrogen Energy

cap capital 2020;45:14625e39. https://doi.org/10.1016/
cw cooling water j.ijhydene.2020.03.195.
[7] Go € tz M, Lefebvre J, Mo € rs F, McDaniel Koch A, Graf F, Bajohr S,
cwi cooling water inlet
Reimert R, Kolb T. Renewable Power-to-Gas: a technological
cwo cooling water outlet and economic review. Renew Energy 2016;85:1371e90.
cool cooling https://doi.org/10.1016/j.renene.2015.07.066.
degr degradation [8] Brauns J, Turek T. Alkaline water electrolysis powered by
el electrical energy renewable energy: a review. Processes 2020;8:248. https://
elz electrolyzer doi.org/10.3390/pr8020248.
gen generated [9] Abdel Haleem A, Huyan J, Nagasawa K, Kuroda Y, Nishiki Y,
Kato A, Nakai T, Araki T, Mitsushima S. Effects of operation
nom nominal
and shutdown parameters and electrode materials on the
purch purchase reverse current phenomenon in alkaline water analyzers. J
O&M operation and maintenance Power Sources 2022;535:231454. https://doi.org/10.1016/
pl plant j.jpowsour.2022.231454.
repl replacement [10] Ali Khan MH, Daiyan R, Han Z, Hablutzel M, Haque N,
rev reversible Amal R, MacGill I. Designing optimal integrated electricity
supply configurations for renewable hydrogen generation in
SB el grid electricity for SBs
Australia. iScience 2021;24:102539. https://doi.org/10.1016/
sell selling
j.isci.2021.102539.
tn thermo-neutral [11] Liponi A, Frate GF, Baccioli A, Ferrari L, Desideri U. Impact of
w wind wind speed distribution and management strategy on
w el wind electricity hydrogen production from wind energy. Energy
grid/elz sys ðSBÞ from the grid to the electrolysis system for 2022;256:124636. https://doi.org/10.1016/
SBs j.energy.2022.124636.
[12] Abbas MK, Hassan Q, Tabar VS, Tohidi S, Jaszczur M,
w/elz sys from the wind farm to the electrolysis system
Abdulrahman IS, Salman HM. Techno-economic analysis for
w/grid from the wind farm to the grid clean hydrogen production using solar energy under varied
Acronyms climate conditions. Int J Hydrogen Energy 2023;48:2929e48.
https://doi.org/10.1016/j.ijhydene.2022.10.073.
BoP balance of plant
[13] Liponi A, Frate GF, Baccioli A, Ferrari L, Desideri U. Green
CRF capital recovery factor hydrogen from wind energy: mitigation of operating point
EHP equivalent hydrogen price [V=kg H2] fluctuations. In: Proceedings of ECOS - the 34th international
IR interest rate conference on efficiency, cost, optimization, simulation and
LHV lower heating value environmental impact of energy systems; 2021. Taormina,
LCOH levelized cost of hydrogen Italy, June 27-July 2.
[14] Fang R, Liang Y. Control strategy of electrolyzer in a wind-
PV photovoltaic
hydrogen system considering the constraints of switching
RE renewable energy
times. Int J Hydrogen Energy 2019;44:25104e11. https://
doi.org/10.1016/j.ijhydene.2019.03.033.
[15] Matute G, Yusta JM, Beyza J, Correas LC. Multi-state techno-
references economic model for optimal dispatch of grid connected
hydrogen electrolysis systems operating under dynamic
conditions. Int J Hydrogen Energy 2021;46:1449e60. https://
[1] IEA. The Future of Hydrogen: seizing today's opportunities. doi.org/10.1016/j.ijhydene.2020.10.019.
2019. Available online: https://www.iea.org/reports/the- [16] Zheng Y, You S, Bindner HW, Münster M. Optimal day-ahead
future-of-hydrogen. dispatch of an alkaline electrolyser system concerning
[2] Liponi A, Pasini G, Baccioli A, Ferrari L. Hydrogen from thermaleelectric properties and state-transitional dynamics.
renewables: is it always green? The Italian scenario. Energy Appl Energy 2022;307:118091. https://doi.org/10.1016/
Convers Manag 2023;276:116525. https://doi.org/10.1016/ j.apenergy.2021.118091.
j.enconman.2022.116525. [17] Varela C, Mostafa M, Zondervan E. Modeling alkaline water
[3] Egeland-Eriksen T, Hajizadeh A, Sartori S. Hydrogen-based electrolysis for power-to-x applications: a scheduling
systems for integration of renewable energy in power approach. Int J Hydrogen Energy 2021;46:9303e13. https://
systems: achievements and perspectives. Int J Hydrogen doi.org/10.1016/j.ijhydene.2020.12.111.
Energy 2021;46:31963e83. https://doi.org/10.1016/ [18] Lu X, Du B, Zhou S, Zhu W, Li Y, Yang Y, Xie C, Zhao B,
j.ijhydene.2021.06.218. Zhang L, Song J, et al. Optimization of power allocation for
[4] Dolci F. Green hydrogen opportunities in selected industrial wind-hydrogen system multi-stack PEM water electrolyzer
processes: workshop summary report (26th of June 2018, considering degradation conditions. Int J Hydrogen Energy
Centre Albert Borschette, Brussels, Belgium). Luxembourg: 2022;48:5850e72. https://doi.org/10.1016/
Publications Office of the European Union; 2018. 978-92-79- j.ijhydene.2022.11.092.
99135-6. [19] Hong Z, Wei Z, Han X. Optimization scheduling control
[5] Schmidt O, Gambhir A, Staffell I, Hawkes A, Nelson J, Few S. strategy of wind-hydrogen system considering hydrogen
Future cost and performance of water electrolysis: an expert production efficiency. J Energy Storage 2022;47:103609.
elicitation study. Int J Hydrogen Energy 2017;42:30470e92. https://doi.org/10.1016/j.est.2021.103609.
https://doi.org/10.1016/j.ijhydene.2017.10.045. [20] Luxa A, Jo € res N, Ya
n~ ez C, Souza M, Pangalos G, Schnelle L,
[6] Herna  ndez-Go  Ramirez V, Guilbert D. Investigation of
 mez A, Lichtenberg G. Multilinear modeling and simulation of a
PEM electrolyzer modeling: electrical domain, efficiency, and multi-stack PEM electrolyzer with degradation for control
 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 8 ( 2 0 2 3 ) 3 7 5 7 9 e3 7 5 9 3 37593

concept comparison. 14/07/2022 - 16/07/2022. In: Proceedings [26] Armijo J, Philibert C. Flexible production of green hydrogen
of the 12th international Conference on Simulation and and ammonia from variable solar and wind energy: case
modeling methodologies, Technologies and applications. study of Chile and Argentina. Int J Hydrogen Energy
Lisbon, Portugal: SCITEPRESS - Science and Technology 2020;45:1541e58. https://doi.org/10.1016/
Publications; 2022. p. 52e62. 978-989-758-578-4. j.ijhydene.2019.11.028.
[21] Rizwan M, Alstad V, Ja € schke J. Design considerations for [27] Schnuelle C, Wassermann T, Fuhrlaender D, Zondervan E.
industrial water electrolyzer plants. Int J Hydrogen Energy Dynamic hydrogen production from PV & wind direct
2021;46:37120e36. https://doi.org/10.1016/ electricity supply e modeling and techno-economic
j.ijhydene.2021.09.018. assessment. Int J Hydrogen Energy 2020;45:29938e52. https://
[22] MATLAB and statistics toolbox release. The MathWorks, Inc; doi.org/10.1016/j.ijhydene.2020.08.044.
2019a. [28] Bo€ hm H, Zauner A, Rosenfeld DC, Tichler R. Projecting cost
[23] Power technology website. Available online: https://www. development for future large-scale power-to-gas
power-technology.com/marketdata/lake-bonney-australia/. implementations by scaling effects. Appl Energy
[24] Sakas G, Iban
~ ez-Rioja A, Ruuskanen V, Kosonen A, Ahola J, 2020;264:114780. https://doi.org/10.1016/
Bergmann O. Dynamic energy and mass balance model for j.apenergy.2020.114780.
an industrial alkaline water electrolyzer plant process. Int J [29] Olateju B, Kumar A, Secanell M. A techno-economic
Hydrogen Energy 2022;47:4328e45. https://doi.org/10.1016/ assessment of large scale wind-hydrogen production with
j.ijhydene.2021.11.126. energy storage in Western Canada. Int J Hydrogen Energy
[25] Ulleberg O. Modeling of advanced alkaline electrolyzers: a 2016;41:8755e76. https://doi.org/10.1016/
system simulation approach. Int J Hydrogen Energy j.ijhydene.2016.03.177.
2003;28:21e33. https://doi.org/10.1016/S0360-3199(02) [30] IRENA. Renewable power generation costs in 2019. 2020. Abu
00033-2. Dhabi.