Professional Documents
Culture Documents
Battery-Assisted Low-Cost H2 Production From Solar Energy - Cost Analysis
Battery-Assisted Low-Cost H2 Production From Solar Energy - Cost Analysis
ScienceDirect
Article history: The massive implementation of renewable energy requires sophisticated assessments
Received 16 July 2018 considering the combination of feasible technology options. In this study, a techno-
Received in revised form economic analysis was conducted for hydrogen production from photovoltaic power gen-
13 November 2018 eration (PV) utilizing a battery-assisted electrolyzer. The installed capacity of each compo-
Accepted 16 November 2018 nent technology was optimized for the wide range of unit costs of electricity from the PV,
Available online 13 December 2018 battery, and proton-exchange membrane electrolyzer. Leveling of PV output by battery, the
necessary capacity of electrolyzer is suppressed and the operating ratio of electrolyzer in-
Keywords: creases. The battery-assist will result in a lower hydrogen production cost when the benefit
Techno-economic analysis associated with the smaller capacity and higher operation ratio of the electrolyzer exceeds
Technology roadmap the necessary investment for battery installation. The results from this study indicated the
Off-grid production cost of hydrogen as low as 17 to 27 JPY/Nm3 using a combination of technologies and the
Proton-exchange membrane achievement of ambitious individual cost targets for batteries, PV, and electrolyzers.
electrolyzer © 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
published technology roadmaps with research targets for PV, time t, 4ðtÞ, is converted into solar power at time t, PowPV ðtÞ,
batteries, and electrolyzers to meet the international goals. through a PV. The generated PV power is transmitted to a PEM
Ideally, the realization of such national targets should lead to electrolyzer through a DC/DC converter to adjust its voltage. If
changing the implementation of renewable resources in the PowPV ðtÞ is larger than the capacity of the electrolyzer,
Japan. However, the target settings for individual technologies Capely , the generated power is stored in a battery by trans-
have met with difficulties in advancing the implementation of mitting Powbat ðtÞ. When the operating ratio of the PEM electro-
renewable resources. For example, the frequency fluctuation lyzer and the charging ratio of battery storage are both full, the
associated with the grid connection of PV has necessitated the excess power is cut off by an energy management system (EMS)
power output suppression of PV in some regions in Japan, and lost as Powloss1 ðtÞ. A part of the charged electricity in the
which is recognized as a significant social issue. The Japanese battery is lost due to the internal resistance during the roundtrip
roadmap for PV projected the targets based on the levelized of charging and discharging, and this power loss is denoted as
cost of electricity (LCOE) only and did not address the grid Powloss2 ðtÞ. This operation is managed by the EMS connected to
connection issue [44]. Therefore, the implementation of PV is the DC/DC converter. The battery assists the stabilization of the
ambiguous even though the LCOE target is reached. Batteries DC flow to PEM electrolyzer, Powely ðtÞ, by charging the excess
can be a solution for such issues; however, the technology PowPV ðtÞ during daytime and discharging it during nighttime as
targets set for batteries for stationary purposes are relatively Powbat ðtÞ. It resulted in the temporally stabilized production of
obscure [45]. Some technology targets should be examined in hydrogen, ProdH . The time integral for generated power is
the context of systems combining multiple technologies balanced with those of power consumed by the PEM electro-
because the focused technology may not function as the lyzer and power losses at the EMS and battery as follows:
system component without adequate development and Z Z Z Z
incorporation of other technologies. Rational targets should PowPV ðtÞdt ¼ Powely ðtÞdt þ Powloss1 ðtÞdt þ Powloss2 ðtÞdt
be set based on such system design and analysis.
In this study, the techno-economic performance of a (1)
battery-assisted hydrogen production system is analyzed
Settings for techno-economic performance
considering technology development based on governmental
incentivization. The case study of off-grid hydrogen produc-
The techno-economic parameters for technology options are
tion is explored for Japan, where almost all fossil resources are
selected by referring to the target values found in the tech-
imported. The renewable resource is set as the solar energy in
nology roadmaps or in the literature. This enables us to
Nagano, one of the best locations for solar irradiation in Japan.
examine if those targets separately set for individual tech-
A proton-exchange membrane (PEM) electrolyzer is suggested
nologies will lead to the feasibility of the battery-assisted
as the hydrogen production technology. An advanced battery
hydrogen production system as a whole. Table 1 summa-
is assumed to be connected for managing the balance of PV
rizes the parameter settings in this study. At first, the existing
and PEM electrolysis hydrogen production. Future techno-
technology roadmaps for electricity from PV, batteries, and
economic performance in their individual technology road-
PEM electrolyzers are investigated. Based on the target values
maps are systematically analyzed based on the minimization
in such roadmaps, the settings for evaluating the levelized
of the unit cost of hydrogen.
unit cost of hydrogen production, UCostH2 , are extracted or
assumed in this study. Note that the time of the target values
Materials and methods is defined as 2030.
Table 1 e Settings for techno-economic performance. The efficiencies of PV, batteries, and electrolyzers are assumed to be
constant in this study.
Technology Parameters Values
Photovoltaic solar panel UCostelec 7 [44], 5, and 3 JPY/kWh assumed from USDOE [46] and IRENA [47]
CapPV Variables for the minimization of UCostH2 . s, where 4base ¼ 1 kW/m2 and
hPV was assumed as 0.15.
Battery storage UCostbat More than 5000 JPY/kWh [45]
Capbat Variables for the minimization of UCostH2
Lbat 20 years [45]
hbat Assumed as 0.9
PEM electrolyzer UCostely CAPEXely
, where CAPEXely is based on Eq. (6) [55], the cases of which are
Capely
30 000 JPY/kW (2.47 105 kW), 50 000 JPY/kW (5.01 104 kW), and 100 000
JPY/kW (5.74 103 kW)
Capely 3:544
Scaleely , where hely was assumed as 0.826 [55] and 3.544 kWh/Nm3 is
hely
the electrolysis energy requirement calculated from the standard enthalpy
change at 298K.
Lely 10 years
Note: The power conversion efficiency of PV depends on conditions, such as the installed angle and direction and temperature. Battery
charging/discharging efficiency will degrade during its cycle operation. The part load operation will change the efficiency of the PEM electro-
lyzer. Although these conditional changes in performance can occur hourly in the actual installation, they were not considered in this study for
simplicity.
by Eq. (2), where 4base is the base irradiation and hPV is the Battery
power conversion efficiency. The generated power, PowPV ðtÞ, Various materials have been adopted and examined for the
is be obtained in Eq.(3), where 4ðtÞ is the actual hourly irra- anode and cathode materials in batteries, which have mainly
diation in the target region. As shown in Eq. (3), a larger CapPV been considered for application in electric vehicles [48]. The
induces wider gaps of power generation between daytime and costs for battery packs have fallen rapidly [49]. The milestone
nighttime, which may affect the required capacity for battery cost for battery packs, UCostbat , has been recognized as 125
storage. USD/kWhuse, i.e., 13 750 JPY/kWhuse, and 100 USD/kWhuse, i.e.,
11 000 JPY/kWhuse, for Lithium-ion and Lithium-metal batte-
CapPV 4base ries, respectively [50]. Considering the estimation of UCostbat
AreaPV ¼ (2)
hPV and applying advanced materials in the anode and cathode, a
milestone for the selling price was set as 50 USD/kWhcharge, i.e.,
POWPV ðtÞ ¼ AreaPV $hPV $4ðtÞ (3) 5500 JPY/kWhcharge in 2030 [48]. The latest major technology
The cost of electricity from a PV is problematic because roadmap for batteries in Japan gives the milestones as 20 000
diversified analyses have been conducted on the LCOE derived JPY/kWh, 10 000 JPY/kWh, and 5000 JPY/kWh in 2020, 2030, and
from PV, UCostelec . The latest major roadmap for solar cells in after 2030, respectively [45]. Therefore, 5000 JPY/kWh can be
Japan is the NEDO PV Challenges [44]. According to their sys- regarded as an ambitious target value for UCostbat .
tematic review of solar cells, the LCOE derived from PV is Capbat is defined as a variable for minimizing UCostH2 . The
targeted at 14 JPY/kWh after the increase of module conver- fluctuation of irradiation, 4ðtÞ, can be stabilized by a battery
sion efficiency and the decrease in production cost. With for increasing the operation ratio of the electrolyzer. Although
material and structural optimization, 7 JPY/kWh in 2030 is the the generated power could be utilized, the height of peaks in
target. The US Department of Energy (USDOE) has launched an irradiation sharply changes daily and seasonally and requires
initiative for the reduction in the LCOE, with targets of significant battery capacity, Capbat . This results in the
0.03e0.05 USD/kWh, about 3e5 JPY/kWh, in 2030 [46], which decrease in the operating ratio of the battery. The optimal
has become an international target [47]. Although UCostelec balance of the capacities of the electrolyzer, Capely , battery,
can be broken down and correlated with CapPV , the actual cost Capbat , and PV, CapPV , should be scrutinized and can be
composition is region-specific and based on irradiation con- changed by the balance of their unit costs, UCostely , UCostbat ,
dition, land, construction, wheeling charge, and trans- and UCostelec . Depending on cell chemistry, a high C-rate will
portation costs. Therefore, UCostelec was set as 7, 5, and 3 JPY/ increase battery ageing and decrease its lifetime [51,52]. If
kWh as discrete variables in this study without explicitly there is a constraint on the operable C-rate in battery
considering its dependence on CapPV . Therefore, the optimi- charging/discharging, the Capbat increases and reduces the C-
zation results indicate the minimized CostH2 produced from rate; otherwise, its lifetime can be shortened compared to the
opt
the electricity with the UCostelec generated from CapPV . The expected life. Note that the lifetime of that battery is a con-
cost of the power converter is also included in the LCOE in the stant in this study at 20 years for simplicity, while the lifetime
extracted UCostelec . will depend on the charging/discharging profiles. The scale
factor of the battery is not explicitly considered in this study.
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 5 1 e1 4 6 5 1455
Z
UCostelec $ PowPV ðtÞdt þ CAPEXely Lely þ OPEXely þ ðCAPEXbat =Lbat þ OPEXbat Þ
UCostH2 ¼ (4)
ProdH
cludes the peripheral devices [52], assuming 130 JPY/EUR. The The LCOE assumed from the PV, UCostelec , is set as shown
lifetime has also been elongated from 20 000e60 000 h in Table 1, which is defined by the first part of the algorithm in
(1500e3800 USD/kW) in 2015 to 80 000 h (830 USD/kW) in 2030 Fig. 3. The capital expenditure, CAPEXely , and the annual
[53]. According to the latest major roadmap on hydrogen operating expense, OPEXely , of the PEM electrolyzer are
strategy in Japan, the unit plant cost of the PEM electrolyzer, defined next. In this study, the existing cost estimation func-
UCostely , is targeted at 50 000 JPY/kW, i.e., about 385 EUR/kW tions [55] are adopted to represent the scaling factors of the
and 455 USD/kW [54]. Here, the unit cost of an electrolyzer is electrolyzer as shown below:
considered under its scale factor [55].
0:68
CAPEXely ¼ 4:3 106 $ Scaleely (5)
Evaluation algorithm of hydrogen production cost
Z
hely
3:544
$ Powely ðtÞdt
Major assumptions in evaluation OPEXely ¼ 0:075$CAPEXely þ 0:075$CAPEXely $
The ancillary facilities of for PV, batteries, and electrolyzers, ADays$Scaleely
such as wiring cables, are implicitly considered in the cost (6)
settings above. Independently, a PV power converter, battery
where the coefficients have the physical meaning as param-
management system, and AC/DC converter are equipped in
eters of the PEM electrolyzer. The coefficient, 4:3 106 , and
the PV, battery, and electrolyzer, respectively. The set costs
power, 0.68, are obtained from the curve fitting of techno-
above, UCostely , UCostbat , and UCostelec , are expected to be for
economic performance for four different scales of hydrogen
standalone installations and include such ancillary facilities,
production, i.e., 300, 3000, 10 000, and 32 000 Nm3/h [55,56].
some of which are not necessary in the battery-assisted
The coefficient, 0.075, used in (both terms in right hand side
hydrogen production investigated in this study, in their cost
of) Eq. (6) indicates the ratio of fixed and variable costs in
targets found in various roadmaps [44,45,56]. Although
OPEXely . ADays is the total number of hours per year. The
battery-assisted hydrogen production requires a novel EMS for
second term in the right side of Eq. (6) is the OPEXely related
optimizing the system operation, it is assumed that the cost for
with operating ratio. The costs of battery were set as shown in
the ancillary facilities, which is included in the independent
Eqs. (7) And (8). The unit cost of battery storage, UCostbat , is
system but is not necessary in the battery-assisted hydrogen
assumed as more than 1000 JPY/kWh to include the target
production, can compensate for the additionally required cost
value shown in Table 1. In this study, the OPEXbat is not
for the hybridization of the PV, battery storages, and PEM
considered.
electrolyzer. More specifically, the PV power converter, battery
management system, and AC/DC converter included in the CAPEXbat ¼ Capbat $UCostbat (7)
unit costs are assumed compensate for those of the DC/DC
converter and EMS shown in Fig. 1. The irradiation patterns,
OPEXbat ¼ 0 (8)
4ðtÞ, can change the optimal points, the location of which is set
as Nagano-city in Japan in this study as shown in Fig. 2. Based on these equations, the hydrogen production cost,
The quality and property of produced hydrogen were set for CapH2 , is minimized by adjusting CapPV and Capbat . The profile
conventional use as fuel gas. The purity of hydrogen produced of generated power, PPV ðtÞ, is obtained from Eqs. (2) And (3),
from a PEM electrolyzer is 99.9e99.9999% [52]. The pressure of according to the hourly irradiation pattern, 4ðtÞ. If the elec-
hydrogen can range from 3 MPa to 10 MPa [57]. It can be utilized trolyzer cannot consume all generated power from the PV
as the heat quantity adjuster for city gas and the pressure in due to the limitation shown in Eq. (10), the remaining power
pipelines, which is from 0.3 to 1 MPa for medium pressure or is stored in the battery (Eq. (9)). The transmission of power
more than 1 MPa for high pressure city gas [58]. to the battery is constrained by Eq. (11). The power that
cannot be consumed by the electrolyzer nor stored by the
Flow chart battery, Powloss1 ðtÞ, is shown as Eq. (12). A part of the stored
In the system shown in Fig. 1, the major design parameters of power in battery is lost during the charging/discharging
facilities related with hydrogen production cost, UCostH2 , are cycle due to internal losses as shown in Eq. (13). The golden-
1456 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 5 1 e1 4 6 5
Fig. 2 e Irradiation in Nagano-city for solar photovoltaic power generation. The time begins at 0:00 1 January 2016, which
includes 8784 h due to the bissextile. The detailed graphs are shown in Fig. S1.
Results
expenditure for the battery. Increased battery capacity con- loss associated with charging/discharging of the battery,
tributes to the improvement in the operation ratio of the increase.
electrolyzer, reducing its capital expenditure. When the bat-
tery becomes sufficiently cheap, the installed capacity does Capacities of the battery and PV at the optimal point of
not increase further, and the capital expenditure of the bat- hydrogen production cost
tery decreases in proportion to its unit cost. The costs for
R
Powloss1 ðtÞdt are reduced by the hybridizing battery in all opt
Fig. 5 and Fig. 6 show the profiles of Capbat =Capely and
R
cases in Fig. 4, although the costs for Powloss2 ðtÞdt, i.e., the opt
CapPV =Capely , respectively, for UCostbat for three cases of
1458 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 5 1 e1 4 6 5
CAPEX
Fig. 5 e Profiles of Capopt 4 5 4
bat =Capely for Costbat for the three cases of Capely : 3.00 £ 10 JPY/kW (2.47 £ 10 kW), 5.00 £ 10 JPY/
ely
4 5 3 CAPEXely
kW (5.01 £ 10 kW), and 1.00 £ 10 JPY/kW (5.74 £ 10 kW). The calculation for the three cases of Capely has the range
limitation: 2.00 £ 104 JPY/kWh for the 3.00 £ 104 JPY/kW and 5.00 £ 104 JPY/kW cases and 4.00 £ 104 JPY/kWh for 1.00 £ 105
JPY/kW.
CAPEXely The threshold point must be the point where the hybridi-
Capely : 3.00 104 JPY/kW (2.47 105 kW), 5.00 104 JPY/kW
zation of the battery effectively reduces UCostH2 as shown in
(5.01 104 kW), and 1.00 105 JPY/kW (5.74 103 kW). The
Figs. 5 and 6 The higher UCostely induces higher incentives for
calculated C-rate for 54 patterns is shown in Fig. S4. By tracing
installing batteries to increase the operation ratio of the
the profiles shown in Figs. 5 and 6 from the highest to lowest
electrolyzer shown in Fig. S3, that is, the amount of produced
UCostbat , the system conditions change as observed by major
hydrogen, ProdH and decrease the costs for the electrolyzer in
inflection points. At the UCostbat of 1.00 105 JPY/kWh, no
opt UCostH2 . A small battery installation needs a relatively higher
battery storage was installed in all cases. The CapPV relative to
C-rate to adjust the intermittent power from the PV. This
the capacity of the electrolyzer is obtained to increase the
provides the threshold point for the cases with C-rate con-
amount of produced hydrogen, ProdH , by decreasing the
R straints needing a lower UCostbat than those without the C-
Powloss1 ðtÞdt. When the cost of the battery is too expensive to rate constraint. The inflection points of UCostbat for the
be installed, the capacity ratio of PV and electrolyzer does not UCostely cases without the C-rate constraint were about
change.
2.50 104, 3.30 104, and 6.50 104 JPY/kWh for 3.00 104,
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 5 1 e1 4 6 5 1459
CAPEX
Fig. 6 e Profiles of Capopt 4 5 4
PV =Capely for Costbat for the three cases of Capely : 3.00 £ 10 JPY/kW (2.47 £ 10 kW), 5.00 £ 10 JPY/
ely
4 5 3 CAPEXely
kW (5.01 £ 10 kW), and 1.00 £ 10 JPY/kW (5.74 £ 10 kW). The calculation for the three cases of Capely has the range
limitation: 2.00 £ 104 JPY/kWh for the 3.00 £ 104 JPY/kW and 5.00 £ 104 JPY/kW cases and 4.00 £ 104 JPY/kWh for 1.00 £ 105
JPY/kW.
5.00 104, and 1.00 105 JPY/kW, respectively, while they absolute value of discharging C-rate reached a constant for
were about 1.00 104, 1.60 104, and 3.20 104 JPY/kWh for opt
the entire year according to Fig. S4. This indicates that CapPV =
the cases with the C-rate limit. At the same time, the relative Capely is sufficient for supplying power to the electrolyzer
capacity of the PV also increases as shown in Fig. 6. This is
most of the day through the battery installed. The inflection
because the excess PowPV ðtÞ during the daytime is provided to
zones correspond to the extremal points of battery capital
the battery to use during the nighttime. Through this mech- opt
expenditure in Fig. 4. However, for Capbat =Capely , the values
anism, the optimized UCostH2 is reduced as shown in Fig. 4.
opt
At the inflection zones circled in Figs. 5 and 6, the slope increase. Fig. 5(a) shows an exponential increase in Capbat
opt opt along with the reduction in Costbat under the inflection zones,
factors of Capbat and CapPV were considerably changed to
reach pseudo-plateau regions. In these zones, the maximum while Fig. 5(b) shows mild increase along with the reduction of
1460 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 5 1 e1 4 6 5
CAPEXely
Table 2 e Results of base settings and advanced technology settings for. .
Capely
Parameters w/o C-rate limitation jC-ratej 0.1
Base setting Advanced electrolyzer Base setting Advanced electrolyzer
setting setting
UCostH2 [JPY/Nm3] 25.27 17.42 26.39 18.43
Electricity 12.89 8.58 12.89 8.58
Lost electricity (loss 1) 0.62 0.42 0.64 0.41
Storage 2.64 2.53 3.74 3.29
Lost electricity (loss 2) 0.80 0.63 1.00 0.89
Electrolysis plant 3.71 2.34 3.59 2.34
Electrolysis operation 4.62 2.91 4.53 2.91
Operating ratio of electrolysis 0.66 0.66 0.68 0.66
opt
Capbat [kWh] 7.12 105 6.84 105 1.04 106 8.91 105
opt
CapPV [kW] 2.13 105 2.16 105 2.24 105 2.22 105
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 5 1 e1 4 6 5 1461
savings in the costs for the electrolyzer and its operation. The The inflection zones clarified by the optimization in this
change in the balances of UCostelec , UCostbat , and UCostely study should be referred to as a rationale for the target in
moved the optimal point of Capbat and CapPV , which results in technology roadmaps. Battery-assist has been recognized as
opt opt
the reduction in Capbat . Regarding CapPV , it is increased by the the factor for increasing hydrogen production cost according
reduction in UCostelec , which means that lost electricity can be to previous research on the hybridization of power-to-gas
opt with batteries [40]. In that work, the UCostely and UCostbat
offset by the reduced UCostbat and Capbat . In both cases of C-
were set as 550 to 949 EUR/kW and 240 to 1000 EUR/kWh,
rate limitation, the optimized UCostH2 could be reduced by ca.
respectively. The lower values were set referring to future
7 JPY/Nm3.
development targets in 2030 [66]. The battery is set for the
The results in Table 2 were compared with the previous
minimum load requirement of the electrolyzer without
studies in literature. Because various system components
connection to the power grid, resulting in an increase in
have been considered in the reported results, the comparison
hydrogen production cost for the case of 71 500 and 117 000
should be carefully conducted to adjust the premises of set-
JPY/kW for UCostely [40]. This is because the cost of the battery
tings. The results of systems with the same components, i.e.,
is not low enough to be installed. Here, we optimize the sys-
PV, battery, and electrolyzer, were extracted from a literature
tem assuming the same unit cost of 550 and 900 EUR/kW and
[40], where UCostH2 are 13.2 JPY/Nm3 for no-battery case, and
determine that ca. 60 and 120 EUR/kWh, respectively, are the
ranged from 49.0 to 59.2 JPY/Nm3 in 2030 assuming 10.78 MJ/
threshold unit costs of the battery to be installed. Apparently,
Nm3 as heat quantity. First of all, the battery assistance has no
the UCostbat target for 2030 used in the preceding study are not
effect to decrease UCostH2 in Ref. [40]. This is partly because
sufficiently low to support the economic feasibility of the
the relative ratios of capacities of component systems were
battery-assist hydrogen production; therefore, the battery
not optimized. The optimization of installation capacity ratios
cost target of 60e120 EUR/kWh should be shared among stake-
enhances the value of battery assistance. The lowest value in
holders and one should note that such targets depend on the
the reference is less than that in this work. This is because
cost target of the PV and electrolyzer.
UCostelec in the reference was set as zero. If UCostelec was set
Systems optimizations for hydrogen production should be
as zero in Table 2, the modified lowest UCostH2 could become
conducted for process design or operation. The optimization in
8.84 JPY/Nm3. Due to the optimization of installed capacity
this study revealed the inflection in the process to be designed
ratios and battery assistance, the modified lowest UCostH2 in
based on the values of battery assistance for hydrogen produc-
this work is less than that in the literature.
tion with the variables of capacities of PV, battery, and elec-
trolysis considering changes in their CAPEXs by technology
Characteristics in this evaluation
development. We can find previous studies on power-to-gas
technologies analyzing the performances of hybridization of
Optimization for system design
different technologies. For example, the maximum power
The combination of PV and electrolysis with batteries has
points (MPP) were simulated for hydrogen production by
numerous opportunities for optimization of system design.
different number of series and parallel cells of PEM electrolyzer
Their balance is significantly sensitive to hydrogen production
[62], and power generation by battery-assisted PV considering
cost due to the changes in operation ratio of the electrolysis
dynamic changes of PV output by every second [63] or applying
plant. Optimization can generate edging alternatives, which
fuzzy-logic modeling [64]. These optimizations in process op-
are the alternatives related to the Pareto solutions of
opt opt erations are based on fixed capacities and CAPEX of devices,
Capbat =Capely and CapPV =Capely ; by screening the possibilities
rather than the dynamic simulation of power management [65],
of the combinations. The optimization of UCostH2 by adjusting although the power management in this study was defined as
Capbat and CapPV clarifies the optimized installation capacities the algorithm represented in the equations in Section Evaluation
of battery-assisted electrolyzers utilizing the electricity from algorithm of hydrogen production cost. Optimizations of both
the PV at the set of UCostelec , UCostbat , and UCostely . UCostelec the design and operation of systems should be addressed for
has a relationship with the utilization ratio of generated obtaining optimal hydrogen production under the conditions on
electricity by the PV. When UCostelec and UCostbat are suffi- the renewable resources as primary energy sources, which de-
ciently high and low, respectively, the utilization ratio of pends on the technology levels of renewable resource utiliza-
generated electricity approaches 100% to avoid discarding tion. Such optimization, however, may require heavy
electricity. However, the results shown in Fig. 4 demonstrate computational costs if one intends to obtain rigorous solutions.
that the optimal settings are not always for 100% use of The results in this study mean the preliminary sets of capacities
generated electricity. In Fig. 5, the values around the lowest and CAPEXs near to optimal, which can become the initial values
UCostbat sharply increase, which means that the required for integrated optimization of process design and operation.
relative capacity of the battery is considerably large to recover
all of the generated electricity. This is because the vertical axis
Limitation of modeling
gives the capacity of the battery at the optimal point. The
The evaluation results are limited in their numerical accuracy
relative battery capacity in Fig. 5 increases as UCostbat de-
originating from the assumption and premises in the adopted
creases. Through optimization of the adjustable parameters,
data and models. One of the major premises in the data is that
i.e., Capbat and CapPV and with changes in UCostelec , UCostbat ,
hourly data for irradiation are defined as the minimum unit of
and UCostely , the rationale for the targets in technology
data shown in Fig. 2. To represent the actual behavior for bat-
development can be examined. tery storage and the electrolyzer, the data should be obtained
1462 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 5 1 e1 4 6 5
with much shorter time interval, such as a second. Such methods. Although the results are different due to the dif-
detailed data are not available for irradiation and PV. The use of ferences in the settings and scopes of impact assessment
such data will be useful in identifying the actual battery cycle methods [75], not only climate change caused by global
characteristics and may influence the C-rate requirement of warming, but also other environmental impacts should be
the battery. The required C-rate will change the electrode taken into account for battery and PV cells as a part of
design; i.e. dense packing of the electrode for high capacity but chemical risk management [76].
low C-rate and loose packing for low-capacity and high C-rate.
Note that the latter may result in higher capital expenditure for Rational target settings for future technology systems
the battery. At the same time, Pely ðtÞ can be stabilized via bat-
tery storage for such actual fluctuations because battery stor- A technology roadmap is a map illustrating the future direc-
age can adjust the voltage profiles of Pely ðtÞ. These conditions tion of technological development for advancing technologies
are not included in the simulations evaluated in this study. and fostering understanding among stakeholders. By articu-
The lifetimes of the battery and electrolyzer are defined lating technological development and plans, research and
since their operating ratios have no relationship with the development can be accelerated by enabling a concentration
lifetimes. However, the actual lifetimes should take into ac- of effort toward technological options selected in the roadmap
count their usage. For batteries, the C-rate should be adjusted [77]. In this study, the target values for techno-economic
to the design range to avoid electrode degradation [67]. performance of electricity from PV, battery storage, and elec-
Although the actual lifetime is often defined as cycle times, trolysis were acquired from their roadmaps articulated by
the full charging/discharging is not always found during NEDO of Japan [44,45,56] and the US DOE [46]. NEDO and DOE
actual use. As for the electrolyzer, the operational conditions, are some of the largest organizations promoting research and
such as operating ratio and start-up/shut-down frequency, development and deployment of industrial, energy, and
become the factors for lifetime [68]. The estimation of lifetime environmental technologies into society. The roadmaps from
under certain conditions of use should be conducted for such organizations have the potential to direct technology
analyzing the practical expenditures, otherwise the analysis development toward their target settings and change the
cannot represent the actual situation even if detailed data technology readiness levels. The roadmaps for the three
with short time intervals could become available. technologies are articulated separately, in addition to road-
The optimization of UCostH2 by adjusting Capbat and CapPV maps for hydrogen production. While these roadmaps
in this study is classified as a nonlinear and multiple solutions implicitly refer to the other roadmaps, they separately set
problem. If rigorous solutions should be obtained, applied targets that do not guarantee the feasibility of the systems
optimization methods could be changed to another algorithm. integrating each technology. The results from this study give
In this study, the golden-section search [59] was employed for an example of determining research targets rationally.
simplification. Although multiple solutions can be found, the Specifically, the target values of UCostH2 are found to be 30
optimized UCostH2 is not changed much because the effects of JPY/Nm3 [78,79] or 20 JPY/Nm3 [80]. The results in this study
changes in Capbat and CapPV are physically understood as show the possibility of achieving the target values of UCostH2
explained in Section Results and are monotonic. However, the through the combination of technologies along with the in-
optimized UCostH2 can become less than the current values by dividual technology development from the roadmaps for
applying rigorous optimization methods for nonlinear and batteries, PV, and electrolyzers. Batteries have the potential
multiple solution problems. to support the reduction in UCostH2 by increasing the oper-
ating ratio of electrolyzers. The thresholds of UCostbat to
Limitation of evaluation indicators reduce UCostH2 are 1.00 104 to 6.50 104 JPY/kWh associ-
In addition to economic aspects of hydrogen production, so- ated with UCostely . This indicates that greater development
cial and environmental aspects should be also examined. The of electrolyzer technologies leads to lower Capbat to assist in
social acceptance of hydrogen and other energy technologies hydrogen production. In other words, batteries can assist the
is one of the most important concerns for social imple- reduction in UCostH2 even if electrolyzers cannot develop
mentation and has been assessed through various ap- according to its technology roadmap. As shown in Fig. 4, the
proaches [69] including analyses on, for example, dominant component of UCostH2 is the electricity for elec-
transportation in Spain [70] and fuel cell vehicles in London trolysis, which means that the sensitivity of an electrolyzer
[71]. The environmental impacts are also one of most impor- to battery assistance is low. The C-rate required for battery-
tant concerns in massive implementation of hydrogen pro- assisted hydrogen production is ranges widely depending
duction. As for the key impact categories induced by battery, on Capbat , which are obtained through the optimization of
abiotic depletion, acidification, and human toxicity have UCostH2 with CapPV . We note that the C-rate for assisting
higher potential environmental impacts than global warming hydrogen production in this study is much less than that for
[72], which were confirmed for various types of batteries [73]. electric vehicles, i.e., 2.14 calculated from the battery road-
PV cell production is also associated with various environ- map of Japan [45]. Due to the significant demand for batteries
mental impacts as analyzed in and retrieved from ecoinvent for vehicles, the technology development of batteries has
[74], the largest life-cycle inventory database in the world. been directed for such usage. If this is shifted to assistance
Fig. S5 and Fig. S6 show the results of cradle-to-gate life cycle for hydrogen production, the required C-rate should be
assessment on Li-ion battery in ecoinvent, and two types of addressed by investigating irradiation patterns and the
PV cell productions represented by other impact assessment optimal CapPV .
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 5 1 e1 4 6 5 1463
[21] Ghaib K, Ben-Fares FZ. Power-to-Methane: A state-of-the-art [39] Shaner MR, Atwater HA, Lewis NS, McFarland EW. A
review. Renew Sustain Energy Rev 2018;81:433e46. comparative technoeconomic analysis of renewable
[22] Zoss T, Dace E, Blumberga D. Modeling a power-to- hydrogen production using solar energy. Energy Environ Sci
renewable methane system for an assessment of power grid 2016;9:2354e71.
balancing options in the Baltic States' region. Appl Energy [40] Gillessen B, Heinrichs HU, Stenzel P, Linssen J. Hybridization
2016;170:278e85. strategies of power-to-gas systems and battery storage using
[23] Parra D, Zhang X, Bauer C, Patel MK. An integrated techno- renewable energy. Int J Hydrogen Energy 2017;42:13555e67.
economic and life cycle environmental assessment of [41] Belmonte N, Girgenti V, Florian P, Peano C, Luetto C, Rizzi P,
power-to-gas systems. Appl Energy 2017;193:440e54. Baricco M. A comparison of energy storage from renewable
[24] Rivera-Tinoco R, Farran M, Bouallou C, Aupretre F, sources through batteries and fuel cells: A case study in
Valentin S, Millet P, Ngameni JR. Investigation of power-to- Turin, Italy. Int J Hydrogen Energy 2016;41:21427e38.
methanol processes coupling electrolytic hydrogen [42] Shabani B, Andrews J, Watkins S. Energy and cost analysis of
production and catalytic CO2 reduction. Int J Hydrogen a solar-hydrogen combined heat and power system for
Energy 2016;41(8):4546e59. remote power supply using a computer simulation. Sol
[25] Tsupari E, Karki J, Vakkilainen E. Economic feasibility of Energy 2010;84:144e55.
power-to-gas integrated with biomass fired CHP plant. J [43] Lagorse J, Simo ~ es MG, Miraoui A, Costerg P. Energy cost
Energ Stor 2016;5:62e9. analysis of a solar-hydrogen hybrid energy system for stand-
[26] Otto A, Robinius M, Grube T, Schiebahn S, Praktiknjo A, alone applications. Int J Hydrogen Energy 2008;33:2871e9.
Stolten D. Power-to-Steel: Reducing CO2 through the [44] NEDO (New Energy and Industrial Technology Development
Integration of Renewable Energy and Hydrogen into the Organization). NEDO PV Challenges: Strategy in solar cell
German Steel Industry. Energies 2017;10(4):451. development. 2014. Available from, http://www.nedo.go.jp/
[27] McKenna RC, Bchini Q, Weinand JM, Michaelis J, Konig S, content/100573590.pdf [in Japanese only].
Koppel W, Fichtner W. The future role of Power-to-Gas in the [45] NEDO (New Energy and Industrial Technology Development
energy transition: Regional and local techno-economic Organization). Battery RM2013: Roadmap of technology
analyses in Baden-Wurttemberg. Appl Energy development on secondary battery. 2013. Available from:
2018;212:386e400. http://www.nedo.go.jp/content/100535728.pdf [in Japanese
[28] Posso F, Sa nchez J, Espinoza JL, Siguencia J. Preliminary only].
estimation of electrolytic hydrogen production potential [46] USDOE (US Department of Energy). The SunShot Initiative's
from renewable energies in Ecuador. Int J Hydrogen Energy 2030 Goal. 2016. Available from: https://www.energy.gov/
2016;41:2326e44. sites/prod/files/2016/12/f34/SunShot%202030%20Fact%
[29] Joshi AS, Dincer I, Reddy BV. Solar hydrogen production: A 20Sheet-12_16.pdf.
comparative performance assessment. Int J Hydrogen Energy [47] IRENA (International Renewable Energy Agency). Renewable
2011;36(17):11246e57. Power Generation Costs in 2017. 2018. Available from:
[30] Dincer I, Acar C. Review and evaluation of hydrogen https://www.irena.org/-/media/Files/IRENA/Agency/
production methods for better sustainability. Int J Hydrogen Publication/2018/Jan/IRENA_2017_Power_Costs_2018.pdf.
Energy 2015;40:11094e111. [48] Berckmans G, Messagie M, Smekens J, Omar N,
[31] Joshi AS, Dincer I, Reddy BV. Exergetic assessment of solar Vanhaverbeke L, Mierlo JV. Cost Projection of State of the Art
hydrogen production methods. Int J Hydrogen Energy Lithium-Ion Batteries for Electric Vehicles Up to 2030.
2010;35(10):4901e6. Energies 2017;10:1314e33.
[32] Kalogirou SA, Karellas S, Braimakis K, Stanciu C, Badescu V. [49] Nykvist B, Nilsson M. Rapidly falling costs of battery packs
Exergy analysis of solar thermal collectors and processes. for electric vehicles. Nat Clim Change 2015;5:329e32.
Prog Energy Combust Sci 2016;56:106e37. [50] Vehicle Technologies Office, USDOE (US Department of
[33] Joshi AS, Dincer I, Reddy BV. Effects of various parameters on Energy). Overview of the DOE VTO Advanced Battery R&D
energy and exergy efficiencies of a solar thermal hydrogen Program. Available from: https://www.energy.gov/sites/
production system. Int J Hydrogen Energy prod/files/2016/06/f32/es000_howell_2016_o_web.pdf.
2016;41(19):7997e8007. [51] Samsung SDI. Smart battery systems for energy storage. 2016.
[34] Gutierrez-Martı́n F, Ochoa-Mendoza A, Rodrı́guez-Anto n LM. 20.09.2016]; Available from: http://www.samsungsdi.com/
Pre-investigation of water electrolysis for flexible energy upload/ess_brochure/Samsung%20SDI%20brochure_EN.pdf.
storage at large scales: the case of the Spanish power system. [52] Leclanche. Graphite/NMC industrial storage. Fuel Cells and
Int J Hydrogen Energy 2015;40(15):5544e51. Hydrogen Joint Undertaking, Study on development of water
[35] Van Dael M, Kreps S, Virag A, Kessels K, Remans K, electrolysis in the EU; 2016. 20.09.2016]. Available from:
Thomas D, De Wilde F. Techno-economic assessment of a http://www.leclanche.com/technology-products/products/
microbial power-to-gas plant - Case study in Belgium. Appl graphitenmc-racks/. http://www.fch.europa.eu/sites/
Energy 2018;215:416e25. default/files/FCHJUElectrolysisStudy_FullReport%20(ID%
[36] Lee B, Chae H, Choi NH, Moon C, Moon S, Lim H. Economic 20199214).pdf. 2014. Available from.
evaluation with sensitivity and profitability analysis for [53] International Energy Agency. Technology Roadmap:
hydrogen production from water electrolysis in Korea. Int J Hydrogen and fuel cells. 2015. Available from: https://www.
Hydrogen Energy 2017;42:6462e71. iea.org/publications/freepublications/publication/
[37] Yao J, Kraussler M, Benedikt F, Hofbauer H. Techno- TechnologyRoadmapHydrogenandFuelCells.pdf.
economic assessment of hydrogen production based on dual [54] CO2 free hydrogen working group. 2017. Available from:
fluidized bed biomass steam gasification, biogas steam http://www.meti.go.jp/press/2016/03/20170307003/
reforming, and alkaline water electrolysis processes. Energy 20170307003.html [in Japanese only].
Convers Manag 2017;145:278e92. [55] Asaoka Y, Uotani M. Feasibility study on hydrogen
[38] Han X, Qu Y, Dong Y, Zhao J, Jia L, Yu Y, Zhang P, Li D, Ren N, production with off-peak electricity. -Evaluation of the
Feng Y. Microbial electrolysis cell powered by an aluminum- effects of availability and electric power transmission. CRIEPI
air battery for hydrogen generation, in-situ coagulant report No. T02039. 2003.
production and wastewater treatment. Int J Hydrogen Energy [56] NEDO (New Energy and Industrial Technology Development
2018;43:7764e72. Organization). Reports of World Energy Network (WE-NET)
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 5 1 e1 4 6 5 1465
projects. 2003. Available from: https://www.enaa.or.jp/WE- [69] Gaede J, Rowlands IH. Visualizing social acceptance research
NET/report/report_j.html [in Japanese only]. A bibliometric review of the social acceptance literature for
[57] Parra D, Patel MK. Techno-economic implications of the energy technology and fuels. Energ Res Soc Sci
electrolyser technology and size for power-to-gas systems. 2018;40:142e58.
Int J Hydrogen Energy 2016;41(6):3748e61. [70] Iribarren D, Martı́n-Gamboa M, Manzano J, Dufour J.
[58] The Japan Gas Association. Supply of city gas in Japan Assessing the social acceptance of hydrogen for
(original title in Japanese). 2017. http://www.gas.or.jp/ transportation in Spain: An unintentional focus on target
gastodokumade/kyokyu/. population for a potential hydrogen economy. Int J Hydrogen
[59] Kiefer J. Sequential minimax search for a maximum. Proc Energy 2016;41:5203e8.
Am Math Soc 1953;4(3):502e6. https://doi.org/10.2307/ [71] Mourato S, Saynor B, Hart D. Greening London's black cabs:
2032161, JSTOR 2032161, MR 0055639. a study of driver's preferences for fuel cell taxis. Energy Pol
[60] USDOE. DOE hydrogen and fuel cells program record. 2016. 2004;32:685e95. https://doi.org/10.1016/S0301-4215(02)
https://www.hydrogen.energy.gov/pdfs/16020_fuel_cell_ 00335-X.
system_cost_2016.pdf. [72] Peters JF, Baumann M, Zimmermann B, Braun J, Weil M. The
[61] Renewable Energy Project Development Office. Saudi Arabia environmental impact of Li-Ion batteries and the role of key
Opens Bids for First Utility e Scale 300 MW Solar PV Project. parameters e A review. Renew Sustain Energy Rev
2017. https://www.powersaudiarabia.com.sa/web/attach/ 2017;67:491e506.
news/PRESS-RELEASE-Sakaka-Bid-Opening.pdf. [73] Peters JF, Weil M. Providing a common base for life cycle
[62] Sayedin F, Maroufmashat A, Sattari S, Elkamel A, assessments of Li-Ion batteries. J Clean Prod 2018;171:704e13.
Fowler Michael. Optimization of Photovoltaic Electrolyzer [74] ecoinvent, ecoinvent v3.3. 2016. https://www.ecoinvent.org/.
Hybrid systems; taking into account the effect of climate [75] Bare JC, Gloria TP. Critical Analysis of the Mathematical
conditions. Energy Convers Manag 2016;118:438e49. Relationships and Comprehensiveness of Life Cycle Impact
[63] Meskani A, Haddi A, Becherif M. Modeling and simulation of Assessment Approaches. Environ Sci Technol
a hybrid energy source based on solar energy and battery. Int 2006;40:1104e13.
J Hydrogen Energy 2015;40:13702e7. [76] Kikuchi Y, Hirao M. Risk Classification and Identification for
[64] Lekhchine S, Bahi T, Abadlia I, Bouzeria H. PV-battery energy Chemicals Management in Process Design. J Chem Eng Jpn
storage system operating of asynchronous motor driven by 2013;46(7):488e500.
using fuzzy sliding mode control. Int J Hydrogen Energy [77] Kato Y, Koyama M, Fukushima Y, Nakagaki T. Energy
2017;42:8756e64. Technology Roadmaps of Japan; Future Energy Systems
[65] Cano MH, Kelouwani S, Agbossou K, Dube Y. Power Based on Feasible Technologies Beyond 2030. Tokyo:
management system for off-grid hydrogen production based Springer; 2016. ISBN13: 9784431559498.
on uncertainty. Int J Hydrogen Energy 2015;40:7260e72. [78] Center for Research and Development Strategy, Japan
[66] Bertuccioli L, Chan A, Hart D, Lehner F, Madden B, Standen E. Science and Technology Agency. Strategic Proposal:
Development ofwater electrolysis intheEuropeanUnion. Fundamental Technology of Energy Carriers for
2014. https://www.fch.europa.eu/sites/default/files/study% Transportation, Storage and Utilization of Renewable Energy.
20electrolyser_0-Logos_0_0.pdf. 2012. CRDS-FY2012-SP-08, https://www.jst.go.jp/crds/pdf/
[67] Kim S, Wee J, Peters K, Shadow Huang HY. Multiphysics 2012/SP/CRDS-FY2012-SP-08.pdf.
coupling in lithium-ion batteries with reconstructed porous [79] The Institute of Applied Energy, Japan. A report on research
microstructures. J Phys Chem C 2018;122(10):5280e90. of scenario to implement hydrogen energy wihout carbon
https://doi.org/10.1021/acs.jpcc.7b12388. dioxide emission. 2017. http://www.iae.or.jp/wp/wp-
[68] Feng Q, Yuan XZ, Liu G, Wei B, Zhang Z, Li H, Wang H. A content/uploads/2018/05/ap-fy2017_r3.pdf.
review of proton exchange membrane water electrolysis on [80] Council of Ministers on renewable energy and hydrogen in
degradation mechanisms and mitigation strategies. J Power the Cabinet Office. Japan. Basic strategy on hydrogen energy.
Sources 2017;366(31):33e55. https://doi.org/10.1016/j. 2017. https://www.cas.go.jp/jp/seisaku/saisei_energy/pdf/
jpowsour.2017.09.006. hydrogen_basic_strategy.pdf.