Applied Energy 342 (2023) 121198

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

Techno-economic analysis of wind-powered green hydrogen production to

facilitate the decarbonization of hard-to-abate sectors: A case study
on steelmaking
Francesco Superchi , Alessandro Mati , Carlo Carcasci , Alessandro Bianchini *
Università degli Studi di Firenze, Department of Industrial Engineering, Via di Santa Marta 3, 50139 Firenze, Italy

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• Techno economic analysis on green

hydrogen from an industrial
perspective.
• Levelized cost of hydrogen and green
index evaluation.
• Hybrid production system using wind
farms, H2 storage tanks and batteries.
• Sensitivity analysis on component and
electricity prices useful for policy
makers.
• Evaluation of emission reduction po­
tential in the context of EU-27 countries.

A B S T R A C T

Green hydrogen is among the most promising energy vectors that may enable the decarbonization of our society. The present study addresses the decarbonization of
hard-to-abate sectors via the deployment of sustainable alternatives to current technologies and processes where the complete replacement of fossil fuels is deemed
not nearly immediate. In particular, the investigated case study tackles the emission reduction potential of steelmaking in the Italian industrial framework via the
implementation of dedicated green hydrogen production systems to feed Hydrogen Direct Reduction process, the main alternative to the traditional polluting routes
towards emissions abatement. Green hydrogen is produced via the coupling of an onshore wind farm with lithium-ion batteries, alkaline type electrolyzers and the
interaction with the electricity grid. Building on a power generation dataset from a real utility-scale wind farm, techno-economic analyses are carried out for a large
number of system configurations, varying components size and layout to assess its performance on the basis of two main key parameters, the levelized cost of
hydrogen (LCOH) and the Green Index (GI), the latter presented for the first time in this study. The optimal system design and operation logics are investigated
accounting for the necessity of providing a constant mass flow rate of H2 and thus considering the interaction with the electricity network instead of relying solely on
RES surplus. In-house-developed models that account for performances degradation over time of different technologies are adapted and used for the case study. The
effect of different storage technologies is evaluated via a sensitivity analysis on different components and electricity pricing strategy to understand how to favour
green hydrogen penetration in the heavy industry. Furthermore, for a better comprehension and contextualization of the proposed solutions, their emission-reduction
potential is quantified and presented in comparison with the current scenario of EU-27 countries. In the optimal case, the emission intensity related to the steel­
making process can be lowered to 235 kg of CO2 per ton of output steel, 88 % less than the traditional route. A higher cost of the process must be accounted, resulting
in an LCOH of such solutions around 6.5 €/kg.

* Corresponding author.
E-mail address: alessandro.bianchini@unifi.it (A. Bianchini).

https://doi.org/10.1016/j.apenergy.2023.121198
Received 20 December 2022; Received in revised form 31 March 2023; Accepted 21 April 2023
Available online 29 April 2023
0306-2619/© 2023 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
F. Superchi et al. Applied Energy 342 (2023) 121198

Nomenclature Acronyms
AEL Alkaline Electrolyzer
Symbols BESS Battery Energy Storage System
C Cost [€] BF-BOF Blast Furnace-Basic Oxygen Furnace
E Energy [kWh] CAPEX Capital expenditures
i Interest rate [%] CCUS Carbon Capture Utilization and Storage
I Current [A] CF Capacity Factor
P Power [kW] DR Direct Reduction
T Temperature [◦ C] DRI Direct Reduced Iron
V Voltage [V] EAF Electric Arc Furnace
η Efficiency [%] G Grid
φ Conversion factor [kg/MWh] GI Green Index
H-DR Hydrogen-Direct Reduction
Subscripts KPI Key Performance Index
bess battery LCOE Levelized Cost of Electricity
el electrolyzer LCOH Levelized Cost of Hydrogen
exc excess NG Natural Gas
grid electrical grid NPV Net Present Value
grid,hc high cost from electrical grid O&M Operation and Maintenance
grid,lc low cost from electrical grid OPEX Operating expenditures
id ideal PEM Proton Exchange Membrane Electrolysis
min minimum PPA Power Purchase Agreement
prod produced PtG Power to Gas
op operating PUN Prezzo Unificato Nazionale (Unified National Price)
purch purchase SCADA Supervisory Control And Data Acquisition
time,deg time related degradation SF Shaft Furnace
Thermal,deg temperature cooling related degradation SOC State Of Charge
rated rated value SOH State Of Health
req requested SOEC Solid Oxide Electrolysis Cell
res renewable energy sources tls Tons of liquid steel
sell sold WF Wind Farm
tot total
work actual working hours

1.1. Decarbonizing the steel industry

Accounting for about 8% of the global final energy demand, the iron
1. Introduction and steel industry is responsible for 5% of CO2 emissions in the EU and
7% globally [3], and thus constitutes a critical sector in the challenge of
The path towards a net zero emissions economy is characterized by industry decarbonization. With an annual global production of
different challenges, among which the decarbonization of the so-called
approximately 1950 Mt of crude steel in 2021 [4] and an average output
“hard-to-abate” sectors, as these industries constitute about 30% of growth rate of 3.8% driven by increasing demand, the steel
global CO2 emissions from all sectors [1]. Conventionally speaking, this
manufacturing industry is characterized by high energy intensity, huge
label indicates those sectors for which the transition is not straightfor­ production capacities and strong dependence on coal. Regarding this
wardly connected to the adoption of renewables for energy production,
last point, steel production required around 15% of global coal demand
because of either the technical characteristics of their production pro­ in 2019 accounting for an average emission factor of 1.8 tCO2/tls [5],
cesses or the large costs associated to their reconversion. Heavy industry
with the main production technology represented by the blast furna­
falls into this category due to the lack of readily deployable solutions in
ce–basic oxygen furnace (BF-BOF) process. In such process, iron ores are
fields like cement, iron and steel, and chemicals production. These
reduced to pig iron in a blast furnace at temperatures above 2000 ◦ C
sectors can be hard to abate for many reasons, mainly due to the highly
through a high carbon-intense reduction employing coke referred as
integrated and complex nature of the production processes, which often
ironmaking before the conversion to crude steel in the basic oxygen
demand for extremely high temperatures (steel and aluminum) or pro­
furnace [6]. The process-related carbon emissions are estimated around
ducing emissions from non-energy sources (ammonia production). The
90% of the entire production chain [7], therefore technological efforts to
heavy reliance on high-temperature heat for many of the processes
mitigate steelmaking environmental footprint have been attempted in
involved in these industries constitutes a major technological limitation,
recent years. Their progress is measured by the development of key
as it cannot currently be sustained without generating significant
projects meant to close the gap between speed of innovation and the
greenhouse gas emissions derived from the direct use of fossil fuels.
milestones of 2030 Net Zero Scenario [8]. The two main categories of
Similarly, economic constraints refer to the high cost associated with the
mitigation routes are broadly represented by carbon capture utilization
deployment of low-carbon alternatives, which, although promising, may
and storage (CCUS) [9] and carbon direct avoidance technologies, with
be prohibitively expensive or not mature enough to have a significant
the latter encompassing options like hydrogen [10], bioenergy [11–13],
impact in reducing emissions in the short term as required by current
direct electrification [14] and energy efficiency measures [15].
policies [2]. Achieving economies of scale and reducing development
Because of its energy-intensive nature, tackling efficiency improve­
costs if thus the only way to make abatement technologies become
ment and energy-saving has long been the main priority of the industry.
commercially viable.

2
F. Superchi et al. Applied Energy 342 (2023) 121198

Efforts have been put on efficiency measures to increase the productivity accounting for dedicated RES), the specific emissions of steel production
of the companies and their competitiveness, and most of waste energy could be slashed by more than 35% by means of this technology. The
streams are nowadays valorized. Over the past decades, the energy in­ appropriate design and operation of dedicated green hydrogen pro­
tensity of steel production has been reduced by roughly 80%, from over duction systems must be addressed to identify innovative strategies that
110 GJ consumed per ton of crude steel produced in 1970 s to the cur­ can enable their uptake, hitherto considered too expensive [28]. It is
rent levels of about 20 GJ/t [16]. Nevertheless, consensus exists on the well known that one of the main issues regarding green hydrogen pro­
fact that the technology has now reached its maturity [17] and the room duction is the intermittent power input from renewables and its coupling
for further improvement of process efficiency is small (15–20%) [18]. with the dedicated generation system. Hydrogen request by heavy in­
The primary alternative to the BF-BOF technology is represented by the dustry applications is instead usually constant over time, thus leading to
shaft-type Direct Reduction-Electric Arc Furnace (DR-EAF), which relies possible mismatches between demand and production. One of the
on natural gas (NG) to convert iron ore to direct reduced iron (DRI), possible solutions to cope with this issue is the introduction of storage
subsequently processing it in an EAF. The use of NG for the direct technologies.
reduction operation results in a CO2 emission profile of ~ 0.9 tons of CO2 When considering green hydrogen hubs, current literature generally
per ton of crude steel (tCO2/t) which, although almost halved compared refers to the coupling of power-to-gas (PtG) installations with fluctu­
to the traditional BF-BOF route [19], makes it a process that still ating energy supply from wind and solar power stations. As for wind
struggles with achieving the climate goals defined for the steelmaking energy, both offshore [29]–[32] and onshore configurations [33–35] are
industry by IEA Sustainable Development Scenario [3]. In fact, accord­ addressed, exploring the interplay between different combinations of
ing to this study, by 2050 the average direct CO2 emission intensity in electrolysis technologies, storage systems and end uses both for stand-
the iron and steel sector must decline to the value of 0.6 tCO2/t. Within alone and grid-connected applications.
this context, the global search for more sustainable pathways in steel Focusing on electrolysis, this study considers alkaline electrolyzers as
manufacturing has been focusing lately in replacing CO2-intensive the reference technology, since to date they are recognized as the most
processes with a direct reduction technology based on green hydrogen technologically mature and reliable technology, since it has been widely
[20–22]. The basis of this approach is represented by the implementa­ deployed globally in the last decades, resulting to be the one with the
tion of Hydrogen Direct Reduction (H-DR) in conjunction with the EAF largest share of installed capacity for large-scale industrial applications
(H2-DRI-EAF process), where hydrogen is meant to replace NG as a worldwide [36–38]. Alkaline electrolyzers present some advantages like
reducing agent in the production of DRI. Considering the European ready market availability, non-reliance on noble metals as constitutive
scenario, several projects have been recently kicked-off across EU to materials, higher longevity, and lower investment costs in comparison
explore the technical and commercial feasibility of hydrogen-based with the other considered typologies of electrolysis [39]. However, they
steelmaking. For example, the HYBRIT project [23] launched in Swe­ also have to deal with some technical limitations like low operating
den is aimed at investigating hydrogen-based sponge iron production by pressure levels and limited values for the operational current densities
entirely relying on fossil-free electricity. The pilot plant has been (below 400 mA/cm2) associated with the formation of potentially
commissioned in August 2020 producing the first world’s sponge iron flammable mixtures of hydrogen and oxygen diffusing through mem­
reduced via fossil-free hydrogen gas in June 2021 [24]. The steel branes [40,41]. More importantly, they must operate in a range between
manufacturing corporation ArcelorMittal S.A. is developing an innova­ 20% and 100% of the declared rated power. This feature, to some extent,
tive project in Germany, aiming at the first industrial scale production negatively affects their coupling potential with unpredictable produc­
and use of DRI from 100% H2 reduction to reach the annual output of tion from RES, and, as the level of detail of the analysis grows, the effects
100,000 tons of steel [25]. These examples are just a few among the of an intermittent functioning must be considered both in the prediction
noteworthy applications of hydrogen in the steel industry, and many of the system performance and in the modelling of the resulting wear
other important industrial firms are developing similar projects, namely and tear effects over the lifetime [42].
Tata steel, Baowu steel, Thyssenkrupp, Voestalpine etc. To better track Another cornerstone for wind-fed hydrogen production is the
recent developments in the sector, the reader is referred to the “Green deployment of large scale and low-cost storage, i.e., the key component
Steel Tracker”, a public database that tracks low-carbon investments in able to convert the intermittent production of renewable sources into a
the steel industry by screening among projects associated with the constant hydrogen flow rate as required by steelworks applications.
pursuit of ambitious climate goals in line with the Paris Agreement Currently, hydrogen storage is addressed via several technologies with
targets [26]. main solutions being represented by physical storage, both as com­
pressed gas or liquid, and material-based technologies [43]. While the
1.2. Technologies, obstacles and prospects of green hydrogen for heavy latter is still in its development phase, and liquid storage is better suited
industry decarbonization for long haul transportation, physical storage of hydrogen in tanks via
compression emerges as the optimal solution when considering large-
Based on the outlined scenario, it is important to focus the attention scale production hubs like the one investigated in this study. Such an
on those hydrogen production technologies that could make it approach is not only well-developed because of the strong similarities
competitive on a commercial scale. Among these, green hydrogen pro­ with natural gas industry [44], but also allows for a better dynamic
duced via water electrolysis powered by RES is regarded as the cleanest operation of the resource in terms of filling and releasing procedures,
and most appealing enabler for the development of H-DR systems, and thus better adapting itself to the needs of complex dedicated hydrogen
thus the basis for energy transition of heavy industry, and particularly of hubs in hard-to-abate scenarios [45].
the steel one. It is then apparent that the techno-economic feasibility of utility-
A recent study analyzed the implementation of this process, scale applications of green hydrogen in hard-to-abate industries must
assuming that the electricity consumption is entirely covered by the be addressed in depth to assess prospects and constraints, enabling the
grid. This means that in addition to the operation of the EAF and definition of policy-driven strategies for a successful market uptake.
ancillary processes, water electrolysis is also supplied without ac­ Lucas et al. [30] analyzed the feasibility of offshore wind-generated
counting for the presence of dedicated renewable plants [27]. Results hydrogen in Portugal’s electricity market for the WindFloat Atlantic
show that, considering the current average EU grid emission factor of case study, considering the variability of electricity price correlated with
295 kgCO2/MWh, the emission intensity of the H2-DRI-EAF process to­ the respective national wind production. McDonagh et al. [32] looked at
tals 1101 kgCO2/tls. inland H2 production fed by offshore wind power to define the optimal
Given that the traditional BF-BOF route reaches the value of 1688 economic outcomes, reporting a Levelized Cost Of Hydrogen (LCOH) of
kgCO2/tls, it is worth noting that, also in this baseline case (i.e., not 3.77 €/kg in correspondence of an LCOE of 38.1 €/MWh. Meier [46]

3
F. Superchi et al. Applied Energy 342 (2023) 121198

considered electrolysis from SOEC and PEM technologies fed with sea f. It assesses the emission reduction potential of the proposed solutions
water on an offshore platform in Norway, analyzing techno-economic and its contextualization in the EU-27 steel manufacturing scenario.
implications of hydrogen compression and transportation. Franco g. Ultimately, it provides a techno-economic insight relevant for short-
et al. [31] carried out energy and economic analysis of offshore pro­ and long-term planning of investments and policy planning. In fact,
ductions studying the viability of different pathways for transporting since the support of the national electricity grid appears most of the
hydrogen to land. Correa et al. [47] instead considered delocalized time to be unavoidable to match the industrial demand of hydrogen,
hydrogen production based on the wind resource availability in different efforts are aimed at quantifying and comparing the effect that in­
countries, performing LCOH evaluations under various comparative centives on fundamental components or grid electricity (both sell
scenarios that take also transportation into account. However, all these and purchase price) would have on the techno-economic outcomes of
studies do not consider a specific user or the need of a constant hydrogen the system.
output, but only aim to produce H2 in the most convenient way. When
referring to hard-to-abate sectors, Nascimento da Silva et al. [48] eval­ To address these objectives, the study is organized as follows. First,
uated the use of wind energy to produce hydrogen for oil refineries, the reference case study is illustrated in Section 2, where a detailed
defining a potential GHGs emission reduction of 22.1% for the best case description of the wind power generation dataset is given, followed by a
scenario. Nevertheless, a gap still exists in the literature about the study discussion on the models adopted in the thermodynamic simulations.
of green hydrogen systems dedicated to address decarbonization in the Main economic assumptions and parameters are also presented in this
heavy industry. section. Section 3 outlines the main findings of the sensitivity analysis on
the system layout. Results are presented and discussed under the light of
1.3. Aims of the study and novelty both economic and environmental viability. Moreover, a sensitivity
analysis is reported based on the projections of components prices and
The present study aims at studying the techno-economic perfor­ different market scenarios. Then, Section 4 presents and contextualizes
mance of a customized hydrogen plant to convert the intermittent power the effective decarbonization potential of the proposed solutions in the
production of a wind farm into a constant output of green hydrogen to be broader European scenario. Finally, the main conclusions of the study
delivered to a steel industry located in Italy. Being the second largest and recommendations are outlined in the Section 5.
producer in the EU-27 scenario [49] and the eleventh worldwide [50],
Italy is among the nations where the adoption of hydrogen in the 2. Materials and methods
steelmaking sector could have a significant positive impact in terms of
emissions abatement and serve as a benchmark for large economies of 2.1. Reference case study
similar size and energy mix.
Differently from other studies which use average aggregate data for A steel mini-mill has been considered as the reference case study to
wind (namely wind distributions), here experimental data from a real represent the final user for hydrogen production. This steelworks ty­
utility-scale wind farm with a temporal resolution of 10 min are used. pology is a recently introduced kind of industrial plant implementing the
Building on such dataset, a series of 84,240 different plant configuration EAF technology to produce continuous casting steel mainly from scrap
layouts and component sizes is simulated. Annual simulations are per­ material. More specifically, this study refers to an integrated process
formed for the entire set of case studies, evaluating the influence of facility comprising the H-DR stage, as it is already commercially avail­
multiple key parameters on two major metrics, i.e. the LCOH and a able [51,52]. If compared to traditional plants, mini-mills are charac­
newly proposed metric represented by the percentage of renewable terized by higher operation flexibility, shorter start-up and stop times
energy used for the produced unit of hydrogen, referred to as Green and lower production volumes, with the latter feature being key for
Index (GI). Since the employment of electricity from both the wind farm deriving the exact amount of hydrogen to be supplied to decarbonize the
and the power grid is considered for all cases, the GI is meant to assess process. In the present analysis, an annual yield of 100,000 tons of steel
the real environmental impact of the system by presenting the per­ has been considered, since this order of magnitude represents a bench­
centage of “green” electricity that is turned into hydrogen. To the best of mark for various green hydrogen-related flagship projects currently
authors knowledge, such an indicator has not yet been defined in liter­ underway at European level [53,54].
ature and may represent a fundamental metric for hybrid hydrogen- As discussed, hydrogen is necessary in this technology to substitute
generation systems. the coke as reducing agent in the furnace for the production process of
The study goes beyond existing literature in many respects. More direct reduced iron (DRI). Subsequently, the DRI is fed to the EAF
specifically: together with steel scarp to be recycled, in equal shares. According to
Vogl et al. [21], this process requires around 25 kg of hydrogen input per
a. It sheds light on the optimal system design and operation for each ton of output steel, thus totaling an amount equal to 2500 ton of
different sizes of the wind farm and the related downstream chain in hydrogen per year. This corresponds approximately to a constant flow
a real application scenario based on calculations from the input time- rate of 285 kg/h that must be continuously fed to the plant and thus
varying wind energy production. produced via the dedicated electrolysis facility.
b. It assesses how the necessity to ensure a fixed H2 mass flow rate for To satisfy this demand, an industrial-scale stack of commercial
the H-DR process needs affects the control logic and the management alkaline electrolyzers has been considered in modelling the system.
of the different energy streams instead of the reliance on RES pro­ Given that a single module can produce around 17.8 kg/h of hydrogen
duction surplus. per MW of electric power at rated conditions, an installation of at least
c. It accounts for the implementation of realistic technologies models, 16 MW of capacity is required to operate nonstop. The investigated
also considering how the operation and degradation of components general plant layout is presented in Fig. 1, albeit many configurations
affect the system performance over time. have been tested in the study also excluding some of the components in
d. It provides a techno-economic evaluation of the entire set of tested some cases. A dedicated wind farm facility provides electric energy to
configurations based on LCOH and GI parameters, with a focus on the electrolyzers stack when wind is available while storing over­
the effects of different storage solutions on the decarbonization rate. production in a BESS or selling it to the grid when exceeding electro­
e. It carries out an in-depth sensitivity analysis on different energy lyzers rated capacity. In case wind resources are not sufficient to meet
prices and types of incentives to investigate how explored scenarios the minimum required hydrogen amount, the missing energy share is
can help new policies to foster the penetration of green hydrogen in withdrawn from the national grid, thus allowing for a certain percentage
the heavy industry. of yellow hydrogen [55] to be fed to the mini-mill. Downstream the

4
F. Superchi et al. Applied Energy 342 (2023) 121198

Fig. 1. Schematic diagram of the hydrogen production system. Electric power flows are represented by yellow lines, hydrogen streams by blue lines.

electrolyzers, a storage tank system can be installed to absorb any sur­ present, it is reasonable to consider the installation of an electrolyzer
plus of green hydrogen production. The motivation behind the presence stack with a nominal power higher than 16 MW to produce more
of the two different storage solutions is to investigate and assess how to hydrogen when wind production peaks occur. Different sizes for the
maximize the exploitation of the renewable source, allowing for higher portrayed system solutions are modelled, simulated and compared
shares of decarbonization of the process. It follows that, if tanks are considering economic and environmental aspects.

Fig. 2. Original wind farm power production scaled by 1 (a), 2 (b), 3 (c) and 4 times (d). Deficit (light blue area) and surplus (yellow area) energy quantification with
respect to the constant power request form the steel mill of 16 MW (red line).

5
F. Superchi et al. Applied Energy 342 (2023) 121198

2.1.1. Wind farm The energy surplus becomes higher than the deficit only in the 4-time
To estimate as realistically as possible the green hydrogen produc­ scaled scenario, corresponding to 74 GWh of surplus and 65 GWh of
ibility in one year of operation, real production data of a utility scale deficit. However, the power trend in Fig. 2 (d) shows that the surplus is
wind farm (WF) have been fed to the simulation. Original data have not homogeneously distributed during the year. Because of this, the
been harvested from the supervisory control and data acquisition 100% self-sufficiency might not be reached by means of storage systems
(SCADA) system of a WF located in Greece. The real production history even in this scenario.
of one year of operation with a 10-minute resolution has been kindly Histograms in Fig. 3 quantify the time frame in which a certain
provided by Eunice Energy Group, the owner of the system, which is power level is maintained by the WF, again for four plant different
acknowledged for this. The plant is composed by six 2.3 MW onshore scales: 1 (a), 2 (b), 3 (c) and 4 times (d) the original production. Power
wind turbines; thus, the nominal power of the farm is 13.8 MW. The levels have been discretized into 2 MW intervals, except for the first step
dataset has been analyzed and cleaned: corrupted data, measuring errors that ranges from 0.2 to 2 MW. The reason behind this exception is the
and values related to periods of maintenance, lightning and icing were minimum power required by the smallest considered alkaline electro­
removed. The data analysis showed that the capacity factor of the farm is lyzer module, which corresponds to 20% of its nominal power (0.2 MW
about 30%; therefore, the number of equivalent working hours of the for the selected technology). Since the hours in which the wind farm
offshore farm corresponds to approximately 2660 h/year. Being the produces less than 0.2 MW cannot be exploited by the electrolyzer stack,
farm nominal power production lower than the constant power request those have been excluded from the counting. Hours in which the power
from the electrolyzer to meet the hydrogen demand, the original dataset production was lower than the power request of electrolyzers (16 MW)
has been scaled up to analyze how different plant sizes would behave in have been highlighted in orange. These correspond to times when
this kind of application. external support for the operation is required (i.e., battery or grid
Fig. 2 shows the power production trend during the considered year activation).
of operation. The original power trend is scaled by 1 (a), 2 (b), 3 (c) and
4 times (d) and compared to the power request from the electrolyzer (red
line). For each multiplier, the amount of deficit (blue area) and surplus 2.2. System modelling
energy (yellow area) that is produced with respect to the constant
request from the user (red line) are shown. The original power pro­ A dedicated simulation framework has been developed to estimate,
duction reported in Fig. 2 (a) lays entirely below the required threshold, as realistically as possible, the capabilities of the hydrogen production
meaning that, in the non-scaled scenario, a constant grid support would system for several combinations of different electrolyzers, storage sys­
be required to satisfy the demand. The 2-time scaled production in Fig. 2 tems, and wind farm scales. The 10-minute time resolution for the wind
(b) presents several production peaks emerging above the required production data enables a step-by-step assessment of the behavior that
power line, especially during months of high production in the second the electrolyzer, the grid, the battery and the tank would follow when
half of the year. Nevertheless, the quantification of the missing energy subjected to a control algorithm that aims to satisfy the hydrogen de­
(81 GWh) is still considerably lower than the surplus energy (16 GWh), mand from the steel mill.
meaning that a massive grid support would be still required. In a sce­
nario that considers a 3-time scaled production (Fig. 2 (c)) the deficit 2.2.1. Electrolyzer
energy drops to 71 GWh and the surplus rises to 43 GWh. The energy To assess the hydrogen production capabilities, an original electro­
that must be provided by the grid decreases and the introduction of a lyzer model developed by the same authors has been used. For a detailed
storage system may increase the self-sufficiency degree of the system. explanation of the model, see Superchi et al. [56]. Based on commercial
devices produced by McPhy Energy, a leading company in the alkaline

Fig. 3. Histograms of power production from wind farm scaled by 1 (a), 2 (b), 3 (c) and 4 times (d). Hours of power production lower than the required one (16 MW)
are represented in orange.

6
F. Superchi et al. Applied Energy 342 (2023) 121198

electrolyzers field, the electrolyzer model has been developed with the milliseconds, its dynamic behavior has been neglected in this study
aim to simulate a realistic operation of such a complex system. The because of the considered time step of 10 min.
hydrogen production capabilities of an electrolyzer change considerably Limits are imposed to the state of charge (SOC) of the battery: to
when an intermittent utilization, as the one simulated in this work, avoid harmful cycles, minimum SOC is set to 15% and maximum SOC to
undergoes. With respect to a conventional operation in which the 95%. The maximum power that the battery can absorb and release (C-
component processes a constant power equal to its nominal value, here rate) is limited as well, considering at least 1 h for a full charge (1C) and
it works at a power level that goes from the bare minimum of 16 MW to 30 min for a full discharge (2C). Considering those limitations and
power production peaks that may occur any time. The alkaline elec­ depending on the instantaneous power production from the wind farm
trolyzer modelled here is characterized by an electric efficiency around and the power demand from the electrolyzer, the SOC of the battery is
60%. In line with the manufacturer’s datasheet, this parameter has been updated at each step of the simulation. Based on effects reported in the
quantified by a power-to-gas conversion factor φ, that expresses how component datasheet, the battery model also considers the charging and
much hydrogen the electrolyzer it is able to produce per each MWh of discharging efficiency dependency on the SOC (Fig. 5). A focus on the
input energy, equal to 18 kg/MWh. In the actual high current stack management of the component is reported in Section 2.2.4.
technology, time degradation causes an increase in the required voltage
per working hour, while thermal degradation causes the same effect per 2.2.3. Tank
each degree of cool down with respect to rated conditions. Considering A tank storage system has the function to store the hydrogen excess
this, Eq. (1) adjusts the operating voltage Vop from its ideal value that may be produced by high power electrolyzers to exploit moments of
considering the time effect (second term) and cooling effect (third term). power production peaks of the wind farm. For the sake of the analysis, a
simplified model that tracks the quantity of H2 that flows inside a hy­
Vop = Vid + ΔVtime,deg ⋅hwork + ΔVThermal,deg ⋅(Trated − Tel ) (1)
pothetical vessel has been introduced in the simulation. A storage
The operating voltage Vop has, in turn, an impact on the power to H2 pressure of 30 bar is considered, which is the same pressure level that
conversion factor φ, computed according to Eq. (2): the gas has at the output of the alkaline electrolyzer. Therefore, a further
compression is not required. The model updates the SOC of this
H2,id
φ=( ) (2) component at each time-step. In this case, for the sake of simplicity, this
Iid ⋅Vop ⋅time parameter is not limited and is allowed to go from 0 to 100%, as assumed
Fig. 4 shows the trend of φ. With the same 10-minute time resolution by Mah et al. [59]. In further analyses, the tank storage sizes are
of wind power history, the conversion efficiency of each module is quantified in terms hydrogen mass (in tons) that can be contained inside
updated to reach an accurate estimation of the hydrogen production: the the tank at 30 bars. Fig. 6 shows the trend of the gas volume that could
step-by step hydrogen production is computed according to the adjusted be stored in a configuration characterized by 37 MW electrolyzers
conversion factor φ. The algorithm considers the instant availability of combined with tanks able to store 117 tons of gas. Large scale modules
wind-generated electrical power, as well as the battery or grid support, can be used to perform a long-term storage and exploit the higher pro­
to estimate the electrical energy Eel that can be converted to hydrogen in ducibility of windy seasons in moments of low production or during
the given time frame. Hydrogen production is then defined by Eq. (3). periods of maintenance of the farm. As for the BESS, a focus on the
management of the tank SOC is reported in section 2.2.4.
H2,prod = φ⋅Eel (3)

At the end of the year-long simulation, the sum of the step-by-step H2 2.2.4. Control strategy
output gives an accurate estimate of the yearly hydrogen production For each considered wind farm scale and storage capacity, the target
capability of the system. of the hydrogen production system remains to feed the steel mill with a
flow rate of 285 kg/h. To achieve this, it would be required to feed 16
2.2.2. Battery MW of electrical power to the electrolyzers continuously. Due to the
Similarly to the electrolyzer model and based on a previous work by inherent intermittent nature of a wind farm production, it is possible to
some of the authors [57], the BESS model simulates the real behavior of install electrolyzers with a higher nominal power to produce a hydrogen
a lithium-ion battery. Among the other BESS technologies, Li-ion has excess when an electricity production peak occurs. This hydrogen excess
been chosen for their high efficiency and resilience to cyclic operations must be stored to be subsequently fed to the user. RES and BESS together
[58]. Since this technology shows a response time in the order of feed the electrolyzer to produce hydrogen and, if the production exceeds
the instantaneous request from the steel mill, the excess is stored inside
the tanks. As previously mentioned, the SOC of the two different storage
means is updated at each step of the simulation.
A parametric control strategy was applied to simulate the behavior of
the system when integrating all the components. Fig. 7 reports the flow
chart that schematize the control strategy to manage power and
hydrogen fluxes at each considered timestep (i). The two inputs of the
iteration are reported at the bottom, the current wind farm production
(Pwind) and the hydrogen request form the steel mill (H2req).
At the end the flow chart reports instead the two main outputs, the
hydrogen produced by modules (H2prod) and either the grid support
(Gsupport) would be required at the beginning of the subsequent timestep
or not. The battery activation always comes first: this component is the
first to be charged when an excess of RES production occurs and the first
to be discharged when the electrolyzer starts requiring more electricity
(Pel) than the instantaneous production. At each timestep, the BESS
control algorithm sets a target power (Pgoal) according to the current
RES production (Pwind) and the state of charge of the tank.
Two operational modes are considered: grid supported (Gsupport =
Fig. 4. Conversion factor (φ) variation in time during a year of operation of the True) and islanded (Gsupport = False). Grid supported mode is activated
electrolyzer. when the hydrogen contained in the tank (H2tank) is lower than the

7
F. Superchi et al. Applied Energy 342 (2023) 121198

necessary to sustain the operation. Moments of power production

peaks are mainly absorbed by the BESS. The last third of the month
was characterized by a power production sufficient to maintain the
autonomous operation but the tank never reaches its full capacity
during this month.
c. November is an example of a highly self-sufficient operation. For
almost half of the month, the tank remains full and is able, together
with the BESS, to cover periods of low-RES production. Only the 1st
and 25th day of the month required a grid support.

2.3. Economic parameters

When referring to the economic assessment of multi-energy systems

considering the production of green hydrogen, existing literature usually
Fig. 5. Dependency of the battery charging and discharging efficiency on the relies on two main parameters, namely the Net Present Value (NPV) and
state of charge. the Levelized Cost of Hydrogen (LCOH). These indicators can be consid­
ered in conjunction [60–62] or as alternative assessment methods
constant request from the user (H2req). In those moments, energy is [63–65].
taken from the grid (Pgrid) to reach the target of 16 MW power to feed the If the wind farm was not included in the considered system, the
electrolyzer, and the BESS is discharged only to reduce the amount of Levelized Cost Of Energy (LCOE) [66] of the renewable source should
electricity that must be bought. It charges in instants of power produc­ have been considered as an additional operating expense to drive the
tion excess and immediately discharges when power is required. Instead, electrolysis process. This study considers a system comprising the entire
islanded mode is activated when the tank is filled and contains more farm; hence the cost of its energy is included in the calculation of the
hydrogen than the constant request. In these moments, the electrical green hydrogen.
grid support is not activated and the electrolyzer is powered only by the For a system that sells the hydrogen to a general market, the Payback
intermittent power that comes from the wind farm. Period (PBT) [67] would have been another important parameter for the
As reported before, alkaline electrolyzers requires at least 20% of evaluation of the techno-economic performance of the plant. PBT re­
their nominal power to produce hydrogen (Pmin), hence in this mode the flects the time needed to recover the investment, given the revenue
BESS is dedicated to keep this minimum required power level and pre­ coming from the sale of produced hydrogen. Instead, focus of this work
vent the stand-by of the modules. is to assess how to satisfy the H2 request of a specific user, thus the LCOH
Power fed to the electrolyzer (Pel) is converted to the equivalent that the steel mill must face remains the most important parameter to
amount of hydrogen that modules are able to produce, thanks to the CF evaluate.
update realized by the electrolyzer mode. The difference between the In this paper, LCOH has been adopted as the main index for the
hydrogen produced and required will be stored inside the tanks. economic analysis, such parameter is used to assess the cost of producing
a unit of hydrogen for a certain time and using a certain production
2.2.5. Components SOC tracking system as presented in several other studies concerning techno-
Fig. 8 displays the variation trend of power from the grid, battery and economic analyses of hydrogen production systems (i.e.,
tank SOC during three months of the year-long simulation, for a [30–32,42,47,48,68–70]). Since this methodology accounts for all the
configuration composed by 37 MW of electrolyzers, a tank capable to capital and operating costs of producing hydrogen, it enables the com­
store 18 tons of hydrogen and a 20 MWh battery. parison of different production methods and the performance of
The red line shows that the BESS SOC, due the relatively small ca­ different plant configurations. The value of this indicator is given by the
pacity of the component (20 MWh), shifts from the minimum to the point in which the actualized sum of revenues equals the actualized sum
maximum allowed values. The tank SOC line (light blue) varies in a of costs: if the market price of hydrogen equals its levelized cost, the
smoother way because of the relatively high dimensions of the compo­ investor will recover the expenditure in the predetermined time [71].
nent (18 tons of H2 equals to 594 MWh of energy). When both storage To assess the economic feasibility of different system configurations,
systems are out of charge, the grid support becomes necessary, and the LCOH is calculated as the ratio between the discounted cash flow
power (yellow line) is purchased to reach the 16 MW power level at the and the discounted hydrogen output (Eq. (4)), considering a time span
electrolyzer. The selected three months exemplify three possible (t) of 20 years and an interest rate (i) of 6 %.
scenarios: ∑20 (CAPEXt +OPEXt )
(1+i)t
(4)
t=0
LCOH = ∑20
a. January is an example of medium self-sufficiency. An initial SOC of Hprod
t=0 (1+i)t
40% is hypothesized for both the BESS and the tank at the beginning
of the simulation. Due to the low initial RES production, the residual CAPEX = Cel + CBESS + Ctank + CWF (5)
energy on the battery is immediately sent to the electrolyzer and the
tank is discharged in a few days. After that, the grid support is OPEX = O&Mel + O&MBESS + O&Mtank + O&MWF + Ppurch ⋅Egrid − Psell ⋅Eexc
required, and power is absorbed from outside. Since the 13th day of (6)
the month, when a new period of high wind speed begun, the
The Capital Expenditure (CAPEX) at time zero equals the initial in­
operation of the system was completely autonomous: BESS and tank
vestment for all the system components (Eq. (5)), it is then considered to
were able to store the power and hydrogen excess and feed the steel
be zero in the remaining period, except for the technologies replacement
mill without absorbing any electricity from the grid. During the last
interventions.
days of the month, the tank and BESS SOC shows that, during a
Another important cost factor is the Operational Expenditure (OPEX)
period of very low wind power production, the storage systems
presented in Eq. (6), which considers the Operational and Maintenance
combined can support the operation for around 3 days.
costs (O&M) of each component, the cost of electricity purchased by the
b. June is as an example of highly grid-dependent operation. For two
national grid and earnings from electricity sold to the grid.
thirds of the month, a massive power absorption from the grid was
To drive the electrolysis process, fresh water is needed as input for

8
F. Superchi et al. Applied Energy 342 (2023) 121198

Fig. 6. Trend of the H2 volume stored in tank storage in a configuration equipped with a 117 tons of hydrogen storage capacity.

the electrolyzer. The hydrogen production cost breakdown by Corera 150 €/MWh for purchase, respectively. Assumptions on the sale price for
et al. [47] shows that water contributes to the 0.2% of the total. In this the excess wind energy are derived from both the global weighted
study, water cost was not directly included in the calculation but average Levelized Cost Of Electricity (LCOE) of new onshore wind
considered included in the OPEX of the electrolyzer. In addition, one projects [77] and the most recent wholesale energy prices under the
valuable subproduct of the electrolysis process is oxygen [30], that may power purchase agreement (PPA) schemes adopted in Europe [78]. On
represent another earning source for the plant. This option was the other hand, the selected purchase price for electricity is related to the
neglected because the selling of large quantities of O2 is favorable only average values of the Italian unified national price (PUN) for the two-
in specific markets. year period 2021–2022 [76].
Storage at high pressure, transmission and distribution costs of Table 2 reports the results of a configuration without any storage for
hydrogen were neglected as well since the analysis considers a scenario the four wind farm sizes introduced before. Since there is no possibility
in which the H2 is produced in close proximity to the user. to store any excess of power, an electrolyzer with a rated power of 16
All the major economic figures considered for the calculations are MW is installed, sufficient to produce the amount of hydrogen required
reported in Table 1. by the steel mill. When the power production from the wind farm is
below the required 16 MW, energy is purchased from the grid. Instead,
2.4. Green index power excess during wind production peaks is entirely sold to the grid.
As expected, due to the significant difference between the selling and
According to the act on green hydrogen by the European Commis­ purchase price, scenarios with a larger wind farm that increases the self-
sion, the rules for counting electricity from directly connected in­ sufficiency of the plant lead to an inevitable drop in the LCOH.
stallations as fully renewables are several. Among these, hydrogen can The lowest LCOH that can be obtained in absence of a storage system
be labelled as green if produced “during a one-hour period where the is 5.89 €/kg, when the electrolyzer is connected to a 4-time scaled wind
clearing price of electricity […] is lower or equal to 20 €/MWh” [79]. farm. This configuration leads to a GI of 70.29%, meaning that only 30%
According to these definitions, the hydrogen produced by the system of the hydrogen that is fed into the steel mill is “yellow hydrogen”. This
considered in this study cannot be considered fully “green” (as often value is still considerably higher than the market price of “grey”
erroneously claimed by similar studies), but might be partially “yellow”, hydrogen, that currently ranges between 1 and 2 €/kg [81]. An energy
because the electricity sources are both the wind farm and the national storage system allows rising the GI of the hydrogen produced by the
grid. Bearing this in mind, a Green Index (GI) is defined to assess the plant thank to the increasing in the degree of exploitation of wind en­
different environmental impact of hydrogen produced with different ergy. To assess economically reasonable capacity ranges for the BESS
system configurations. The GI is calculated with Eq. (7), where Ewf is the and the hydrogen tank, a preliminary and wide range analysis was
electricity that the wind farm produces and the electrolyzers uses, which performed. The two storage means were considered separately, and the
is considered 100% green, while Egrid,lc and Egrid,hc are the electricity aim was to evaluate the cost of reaching 100% of green hydrogen by
purchased by the national grid at times of low and high cost, respec­ using one technology or the other. The analysis was carried out
tively. According to the current energy mix of a country like Italy, the considering a wind farm scaled by a factor of 4. Then, the LCOH figures
first is considered 100% green, while the second is yellow but can be resulting from such solutions have been computed to compare the eco­
considered 36% green [80]. Etot is the sum of the three terms. nomic performance of BESS-based solutions with tank-based solutions.
Results are plotted in Fig. 9.
GI =
EWF + Egrid,lc + 0.36⋅Egrid,hc
(7) All BESS-based solutions consider an electrolyzer stack of 16 MW
Etot nominal power. The storage is located between the power source (wind
farm) and the electrolyzer, thus there is no possibility to store an excess
3. Results and discussion of hydrogen that could be produced by a larger electrolyzer. On the
other hand, if one wills to exploit a tank storage system, it is necessary to
This section presents the results of an in-depth parametric analysis increase the electrolyzer power at higher levels than the bare minimum
aimed at assessing the sensitivity of techno-economic parameters, required to meet the steel mill demand. For this reason, tank-based so­
namely LCOH and GI on the variation of system configuration in terms of lutions consider electrolyzer capacities proportional to the increasing
wind farm size, battery, and tank capacity. Market values for electricity tank size.
price in the baseline scenario have been set to 70 €/MWh for sale and

9
F. Superchi et al. Applied Energy 342 (2023) 121198

Fig. 7. Flow chart of the control strategy that manages energy and hydrogen fluxes in each considered timestep.

Results show that, when the target is to reach GIs slightly higher than different sizes of the electrolyzer for systems powered by the wind farms
that obtained by a configuration without storage systems, the optimal with increasing scale factor, i.e., 2x (Fig. 10 (a)), 3x (Fig. 10 (b)) and 4x
solution is to install a small battery. In this GI range (70–72%), a BESS (Fig. 10 (c)). For the purposes of this analysis, only discretized intervals
allows to increase the GI without increasing the size of the electrolyzer. of 18 tons for the tank capacity have been considered in order to portray
Indeed, when the target is to reach GIs higher than 74%, the combina­ the qualitative trend of performance. An electrolytic power ranging from
tion of a hydrogen tank and a higher electrolyzer installed power be­ 16 MW up to the nominal power of the wind farm (i.e.: 28 MW for the 2x,
comes more effective. To reach high GIs, it is considerably more 42 MW for the 3x and 56 MW for the 4x) was considered.
convenient to install high-capacity tanks with respect to high-capacity It is important to emphasize that the black line shown in the graphs
batteries. relates to a configuration with no tank installed and confirms that in
Based on these results, the capacity range for the BESS was set to absence of a storage medium, an increase in electrolyzer power only
0–20 MWh and the tank sizes goes from 0 to 117 tons of hydrogen. The produces a linear increase in the LCOH and no improvements in the GI
main technical assumptions and the capacity range of each component and should therefore be avoided. On the other hand, when a tank is
considered in the parametric analysis have been summarized in Table 3. present, the installation of a high electrolytic capacity not only improves
the GI but, in configurations with a high amount of available renewable
3.1. Hydrogen tank power, it may also decrease the LCOH. A higher level of self-sufficiency,
given by the possibility of exploiting power production peaks, increases
Fig. 10 shows the LCOH and GI trend varying the tank capacity for the use of the renewable resource and decreases the amount of

10
F. Superchi et al. Applied Energy 342 (2023) 121198

Fig. 8. Tank and BESS SOC and electrical grid power request trend during a month of operation. January (a), June (b), November (c).

electricity that must be purchased from the grid.

Table 1
In configurations paired with a wind farm scaled by 2 times (Fig. 10
Main economic parameters considered in the study.
(a)) compared to the original dimension, the GI starts from 63% (16 MW,
no tank) and reaches values around 70% (28 MW with tank). The dash Component CAPEX OPEX Lifetime Reference
dotted blue line represents the trend of the LCOH resulting from systems Electrolyzer 650 €/kW 2.75% I0 €/y 10 years [72]
that utilize tanks able to store 18 tons of H2. This line presents a mini­ BESS 117 €/kWh 2.5% I0 €/y 10 years [73]
mum when paired with a 24 MW electrolyzer: a good trade-off between H2 Tank (30 bar) 460 €/kg 1% I0 €/y 20 years [72,74]
Wind Farm 1279 €/kW 42 €/kW/y 25 years [75]
GI (68.75%) and LOCH (6.96 €/kg). The effect on the GI of tank sizes Ppurch grid – 150 €/MWh – [76]
larger than the latter is negligible, and it only produces an increase in the Psell grid – 70 €/MWh – [77,78]
LCOH: for each additional 18 tons of H2 storage capacity, the GI gain is
around 0.1%, while the LOCH increment is around 32c€/kg.
When a wind farm scaled 3 times the original size (Fig. 10 (b)) the tank storage system. This scenario brings to the lowest LCOH
powers the system, GI starts around 68% (16 MW, no tank) and can be reachable for the discretized analysis: a 30 MW electrolyzer equipped
increased up to 87.5% (42 MW with tank). In this case, due to the high with a tank capable to store 18 tons of H2 results in a LOCH of 5.71 €/kg,
availability of cheap renewable energy, the installation of a tank can and a GI of 83.5%. This is also the scenario that allows one to reach the
reduce the LCOH to values lower than the standard configuration with highest possible GI: a 56 MW electrolyzer equipped with a 117 tons
no storage means and low electrolyzer power. A tank capable to store 18 capacity H2-tank system reaches a GI of 95.53%, but the resulting LCOH
tons of H2, paired with a 25 MW electrolyzer, results in a LOCH reduc­ is 7.36 €/kg. Generally, presented analysis shows that, when a storage
tion of 7c€/kg with respect to the hydrogen cost resulting from the system is present, the LCOH trend varying the electrolyzer power always
configuration without storage. This effect is even more pronounced in shows a minimum. For each tank size, there is an electrolyzer rated
configurations powered by a wind farm 4 times the original size (Fig. 10 power that optimizes the trade-off between the higher initial investment
(c)), where the GI can be increased from 70% to almost 96% thanks to for the storage system and the consequent electricity purchase savings. A

11
F. Superchi et al. Applied Energy 342 (2023) 121198

better description of this phenomenon is reported in section 3.4. When a 3-time scaled wind farm is considered (Fig. 12(b)), also
components size ranges are broadened. For this case, the influence of the
3.2. Battery energy storage system (BESS) tank storage system begins to make its impact, allowing a minimum
LCOH value of 6.17 €/kg to be reached when 24 MW of electrolyzers are
Fig. 11 shows the effect that a battery storage system can have on the connected to a 9 tons capacity tank system and no BESS is installed.
levelized cost of hydrogen and green index in the configurations with 16 Finally, values of Fig. 12(c) are representative of the largest system ca­
MW electrolyzers. As pointed before, when only this kind of storage is pacities simulated and coupled with a 4-time scaled wind farm. For the
present, electrolyzers must produce exactly the constant request from best case, higher storage capacities are paired with high power elec­
the user, and thus only 16 MW of installed power is considered. The trolyzers, higher investment costs may be better balanced by the cost
capacity range of the BESS goes from 0 to 20 MWh. As expected, the reduction in hydrogen production. In fact, the LCOH index significantly
introduction of a BESS increases the GI and, until a certain dimension, decreases to a minimum value of 5.69 €/kg in the presence of a 13.5 tons
reduces the LCOH. As for the tank, the augmented self-sufficiency makes capacity for the tanks and an electrolysis rated power of 28 MW without
the produced gas greener and reduces the amount of electricity that the support of any electrical storage mean.
must be acquired from the grid. The minimum is again given by the As important as the economic assessment, the environmental score of
trade-off between purchase savings and higher initial investment. The all the combinations tested is presented in Fig. 13. Similar to the image
positive effect that the storage has on the GI scales up with the size of the discussed above, GI results are portrayed for the different wind farm
wind farm. The increase in GI by a 20 MWh capacity BESS is 1.92% for dimensions and plotted over the same interval to display the evolution of
the smallest considered plant (2x), 2.33% in the middle case (3x) and performance as considered size ranges increase. Starting from Fig. 13(a),
2.54% for the biggest one (4x). Noteworthy, the GI trend seems to flatten it is apparent that the system is under-dimensioned to achieve an
out with increasing BESS capacity, thus indicating that a further gain in acceptable GI value for the constant grid support required to meet
electrical storage size would result in minor environmental returns. The hydrogen demand. The best outcome is obtained for 28 MW electro­
increment in wind farm scale makes the minimum of the LCOH shifting lyzers, 20 MWh of BESS capacity and 72 tons of tank size, accounting for
towards bigger BESS sizes: 10 MWh in the 2x case, 13 MWh for the 2x a GI of 70.5%.
and 18 MWh for the 4x. Fig. 13 (b) is instead useful to understand the trend for higher
These results prove that a relatively small BESS (compared to installed storage capacities. Configurations with high storage volumes
considered capacities for the tank) can reduce the LCOH of configura­ allow a higher penetration of renewable energy generated by the wind
tions with lower electrolytic capacities. farm, leading the 3D graph to a distinct transition to green in corre­
spondence with the latter. For the wind farm scaled by a factor three, the
3.3. Combined system highest GI is 86.35%, generated by a process layout with 42 MW of
electrolyzers and 20 MWh and 72 tons capacity for BESS and tanks
The effect of the combination of the two storage systems is analyzed system, respectively. Among the many cases studied, the best possible
in this section. The three-dimensional plots presented in Fig. 12 and in outcome is achieved by the highest storage capacities coupled with a 56
Fig. 13 are meant to give a global overview on how the variation in size MW electrolyzers stack, accounting for a remarkable 94.28% GI score.
and the interplay of the three main components affect the tech­ These results are depicted in Fig. 13(c) and show that for a wind farm
noeconomic outcomes of the system for the three different wind farm four times the original size, it is possible to decarbonize the hydrogen
dimensions considered. Defined ranges are presented in Table 3; greater production process nearly completely.
wind farm dimensions imply the simulation of a wider capacity spec­ Fig. 14 helps to better grasp the results presented above by repre­
trum also for the considered production and storage technologies to senting in the same graph the evolution of the two KPIs for different
expand the coverage of the present analysis. The total number of scales of the wind farm. While the effects of a storage capacity increase
different configurations tested is to 84,240 and, for each of them, a are detrimental both from an economic and an environmental point of
yearlong simulation has been performed leading to specific LCOH and GI view for the smallest wind farm, as shown in Fig. 14(a), the same does
results that help in the visualization and interpretation of the com­ not apply for larger systems. It is interesting to observe that the family of
plexities of this analysis (given the number of variables). curves tends to move progressively towards the lower right-hand side of
Fig. 12 displays the results of the LCOH for all the tested different the graphs, implying that both the parameters improve as the system
layouts, presented in the same scale of magnitude for a better under­ size increases, LCOH declines while GI grows. Minimum values for the
standing of the effects of the overall system size variation. Fig. 12 (a) is cost of production are found in solutions that consider the installation of
characterized by the highest LCOH figures and refers to a configuration storage systems: the variation in BESS size is not relevant with respect to
of the wind farm scaled by 2 times. The simultaneous installation of both the effects of increasing the tank system capacity, also due to the small
large tank and BESS systems when little electrolysis capacity is available range considered for the first technology. Fig. 14(c) allows to visualize
has inevitably a negative impact on the metric since all the storage po­ that for the 4-time scaled wind farm, the presence of high-capacity
tential is not fully exploited. On the other hand, the lowest LCOH value storage tanks strongly affects the minimum value of LCOH, which,
of 6.75 €/kg is obtained when an electrolyzer stack of 16 MW is coupled despite the higher initial investment compared to the case without
with a BESS of 10 MWh and no other storage mean is present. storage, enables improved economic performance due to the electricity
Given the poor matching between the renewable source availability cost difference between purchase and sale prices. As previously
and the facility needs, the grid support is frequently activated; the mentioned, minimum LCOH value is 5.69 €/kg, and it occurs for a tank
presence of electrical storage for the configuration with the lowest LCOH capacity of 13.5 tons. Larger tank sizes initially cause a significant
is prime indicator of the influence of the price delta between the sale and translation towards higher GI figures, progressively decreasing their
purchase of electricity, which leads to a preference for storing energy at contribution thereafter when the shift becomes vertical, yielding only a
zero cost when available. significant increase in LCOH compared to a negligible gain in GI.

Table 2 3.4. LCOH contributors

LCOH and GI for a system without any storage for different WF scales.
To understand the weight of different contributors on the final price
WF Scale 1x 2x 3x 4x
of green hydrogen, and how this contribution varies when the storage of
LCOH [€/kg] 7.66 6.78 6.29 5.89 the electrolyzer size is increased, Fig. 15 shows the LCOH variation as a
GI [%] 53.17 62.89 67.31 70.29
function of different cost contributors. This analysis considers a fixed

12
F. Superchi et al. Applied Energy 342 (2023) 121198

Fig. 9. Comparison of LCOH trends between BESS (orange line) and tank (blue line) supported operation to parity of Green Index (GI). Lines interpolate discrete
simulations (represented by marks) and labels indicates the size of the considered storage system: MWh for BESS and tons of hydrogen for tanks.

performed under different market scenarios that may present significant

Table 3
variations of the electricity and components prices. An increment in the
Technical assumptions of the main components of the system: electrolyzer,
market price of electricity can be produced by market fluctuations, as it
battery, H2 tank and wind farm.
is discussed below. A reduction in components prices may be given by
Component Technical assumptions Capacity range the advancement and diffusion of such devices. Additionally, a price
Electrolyzer Variable Temperature (71 ◦ C nominal) 16–28 MW (2x) reduction in those two areas may be given by public incentives that aim
Variable Voltage (1.91 V initial) 16–42 MW (3x) to boost the penetration of green technologies.
18 kg/MWh 16–56 MW (4x)
The prediction of future outcomes is intrinsically characterized by
Battery 15–95% SOC 0–20 MWh
SOC dependent efficiency high levels of uncertainty, and this influences all the factors and quan­
H2 Tank 0–100% SOC 0–117 tons of H2 tities that this sensitivity analysis considers. Hence, all displayed results
30 bar are meant to give an insight into different potential evolutions of the
Wind Farm Enercon E-82 (2.3 MW) 14 MW (1x) market scenario. To address this uncertainty, the analyses consider a
10 min time resolution data from SCADA system 28 MW (2x)
42 MW (3x)
wide range of assumptions with the aim to cover as many probable
56 MW (4x) techno-economic developments as possible and provide a comprehen­
sive picture of their potential impact. In this context, this analysis may
help the reader to understand how such subsidies would affect the
tank size of 9 tons of hydrogen and a wind farm scaled 4 times. The optimal solution. All analyses in this section consider a wind farm scaled
increment of the electrolyzer power produces a linear increase in the 4 times with respect to the original plant.
initial investment cost (dotted line) and, consequently, a linear increase
in the operation and maintenance of the plant (dash-dotted line). On the 3.5.1. Price of system components
other hand, this increment results in a drop of the electricity purchase As pointed in Section 3.4, the production cost of hydrogen is influ­
expenditure (dashed line). This drop is not linear and flattens around a enced by several technical and economic contributions. One of the most
44 MW electrolyzer: since the tank size is limited, a further increment in important is the CAPEX requirement for the various components that
the electrolyzer power does not result in a correspondent increment in make up the plant, i.e., wind turbines, electrolyzers, batteries and tanks.
the system self-sufficiency. Technological improvements, new materials, new production methods
The minimum present in the global LOCH (continuous line) corre­ and economy of scales will reduce the current costs of these components.
sponds to the optimal trade-off between the electricity purchase Regarding electrolyzers, the development of new materials for elec­
expenditure drop and the initial investment increment. If the electro­ trodes and membranes and a future economy of scale in the
lyzer power is additionally increased, the higher initial investment is not manufacturing process could lead to a drastic drop in the production
balanced by the electricity savings. The introduction of a BESS system cost of the stacks. IEA foresees a cost reduction of 20% in alkaline
(different line colors) reduces the LCOH when the electrolyzer power is electrolyzers because of the advent of scaled-up electrolyzers and
low, but only translates in an increase when the power of this compo­ automated production processes [82]. The Norwegian electrolyzer
nents becomes higher than 23 MW: the electricity expenditure savings maker Nel declared a plan to cut the cost of their devices by 75% in a
become smaller and smaller, as evidenced by the reciprocal position of new Gigafactory that they are building [83]. According to Irena [84],
dashed lines. the cost of hydrogen electrolyzers could fall during the next decade at
similar rates to those seen in solar panels and wind turbines: minus 82%
and 39%, respectively, between 2010 and 2019. The cost of lithium-ion
3.5. Sensitivity analysis
batteries has declined by 97% between 1991 and 2018 [85]. According
to IRENA [86], Li-ion batteries are still a relatively new technology, and
This section presents the results of an economic optimization

13
F. Superchi et al. Applied Energy 342 (2023) 121198

Fig. 10. LCOH (dash-dotted line) and Green Index (continuous line) trend varying tank capacity in tank supported operation for different scales (2x, 3x and 4x).
Different tank sizes are represented by different colors.

Fig. 11. LCOH (dash-dotted line) and Green Index (continuous line) trend varying battery capacity in BESS supported operation for different scales. Different wind
farm scales are represented by different colors.

their potential cost reduction is large: increase in the scale of production materials and the growing increase on the hydrogen tanks demand could
capacity, new materials, more competitive supply chains and perfor­ produce similar price drops even in these components. To estimate the
mance improvements are the most promising factors for a further cost future economic implications of market evolution on hydrogen pro­
reduction. BNEF forecasts a LI-ion batteries prices fall to 73 $/kWh in duction plants, the following sections present the effect that the com­
2030 [87]. In line with electrolyzers and batteries, research on new ponents price drop could have on the final LCOH.

14
F. Superchi et al. Applied Energy 342 (2023) 121198

Fig. 12. LCOH indicators varying BESS capacity, tank capacity and electrolyzer power. Comparison among different wind farm size scales. All graphs use the same
color scale.

Fig. 13. GI indicators varying BESS capacity, tank capacity and electrolyzer power. Comparison among different wind farm scales. All graphs use the same
color scale.

3.5.2. Price of electricity 3.5.3. Market and components price variation

Fig. 16 reports the trend of the Italian unified national price (PUN) Fig. 17 shows the effect of a variation in the electricity purchase and
during the past 3 years and a half: 2019 (a), 2020 (b), 2021 (c) and selling price on the techno-economic outcome of the optimal configu­
January to August 2022 (d). PUN history was obtained from the Italian ration under different market circumstances. Lines in the graph show the
electric market manager, “Gestore del Mercato Elettrico (GME)”. Mean trend of LCOH (dotted line) and GI (dashed line), as well as the tank size
price recorded during the year is represented by the orange flat line, (continuous line) and electrolyzer power (dash-dotted line). For what
while the green flat line represents the price threshold of 20 €/MWh that concerns the electricity purchase price variation, a price range from 50
is required to classify the hydrogen produced by means of that electricity to 300 €/MWh was considered, to assess how the optimal solution varies
as “green”. Fig. 16 (a) shows that, during 2019, the price of electricity if the scenario shifts from a normal situation (2019 with an average price
fluctuated around its mean value of 52 €/MWh, going below the “green of 52 €/MWh) to the current market situation (311 €/MWh on average
threshold” only a few times. In 2020 (Fig. 16 (b)) the price dropped and rising). Four possible selling prices were considered (60, 80, 100
during the months of April and May (COVID-19 lockdown months) and and 120 €/MWh) to understand how the difference between the two
raised again during the last months of the year. The situation changed price points may modify the optimal solution. Analyses consider a fixed
radically in 2021 (Fig. 16 (c)) and in fist months of 2022 (Fig. 16 (d)), electrolyzer price of 600 €/MW and a tank price of 400 €/kg. Fig. 17
when the electricity price has seen a sharp rise starting from September shows that an increase in the grid electricity expenditure of the system
2021. Average electricity price rose considerably: 311 €/MWh in 2022, produces a massive drop in the final LCOH: from around 2 €/kg when the
six times the average price of 2019. Due to this strong uncertainty when the market presents prices in the 2019 range of 50 €/MWh, it rises
behind the electricity expenditure that the system must face during its to 7 €/kg when the electricity reaches current values of 300 €/kg.
lifetime to support the constant hydrogen production, it is important to This clearly reflects the strong dependency of LCOH on the electricity
analyze how techno-economic metrics as the levelized cost of hydrogen price fluctuations, due to the required energy from the grid. However,
and the green index would vary under different market conditions. the optimal storage size gradually increases as the electricity purchase

15
F. Superchi et al. Applied Energy 342 (2023) 121198

Fig. 14. LCOH value over GI varying BESS and tank size for three different farm scales (2x, 3x and 4x).

from the grid becomes more expensive, and this consequently rises the effect of the increase in the electricity selling price proves that the most
GI of the optimal solution: from less than 70% to more than 90%. The important incentive for an investment on a storage system is given by the
trade-off between the higher initial investment for a storage system and difference between the sale and purchase price: to parity of purchase
electricity savings shifts to bigger tank capacities. In the cheap elec­ price, a selling price of (120 €/MWh, red line) brings the GI of the
tricity market scenario, the configuration without any storage means, i. optimal solution to significantly lower values with respect to solutions
e., the most gird-dependent one, is indeed the most convenient. The that considers a lower selling price (60 €/MWh, yellow line). If it is

Fig. 15. LCOH contributors for a configuration provided with a tank capacity of or 9 tons of hydrogen and a BESS size of 0 MWh.

16
F. Superchi et al. Applied Energy 342 (2023) 121198

convenient to sell the excess energy, there is no advantage in investing in €/kW. To make a fair comparison between the impact of an incentive on
a storage system. Buying electricity from the grid during deficit hours components and an incentive on electricity purchase, the amount of
would be more cost-effective. money that is required to produce a certain drop in the electrolyzer price
Fig. 18 shows the effect of a reduction in the electrolyzer and tank was first quantified. This was then converted in the equivalent achiev­
price on the techno-economic outcome. Lines in two graphs show again able drop in electricity purchase. To reduce the component price of 450
the trend of the LCOH (dotted line) and GI (dashed line), as well as the €/kW when installing electrolyzer size of 37 MW, an incentive of 16.65
tank size (continuous line) and electrolyzer power (dash-dotted line). An M€ is required. Considering the 20 years of lifetime for the analysis, this
electrolyzer price drop from the actual value of 650 €/kW down to 200 translates into an equivalent subsidy of 832.5 k€ per year. Since the
€/kW was considered together with three tank prices: 400, 300 and 200 initial optimal configuration requires 39.14 GWh per year, if this subsity
€/kg. A purchase price of electricity of 150 €/MWh and a selling price of is directed to reduce the electricity expenditure, it would result in a
70 €/MWh were considered. This analysis shows that the effect of a 21.27 €/MWh discount. Based on these assumptions, Fig. 20 (a) con­
reduction in the electrolyzer price also produces a considerable LCOH siders an electricity purchase price ranging from 128 to 150 €/MWh. The
drop: a decrease on the investment price for the electrolyzer of almost electrolyzer price reduction reduces the LCOH down to 4.93 €/kg and
70% (650 to 200 €/kW) reduces the LCOH of around 1 €/kg. This time, it increases the GI up to 86%. On the other hand, the electricity purchase
must be noticed that the price drop is followed by an increase in the GI. price reduction also decreases the LCOH, but to a lesser extent, down to
Cheaper components promote the installation of high-capacity storage 5.27 €/kg. In addition, this effect combines with a reduction of the GI of
systems that, in turn, enhance the self-sufficiency of the system. In the 6.5% due to the inconvenience of installing a large capacity tank and a
same way, a tank price drop reduces the LCOH and, since it pushes the high electrolyzer power. Not only the same subsidy produces a lower
installation of bigger tanks, makes the final product greener: to parity of LCOH decrement when directed towards the electricity market, but also
electrolyzer cost, a tank price of 200 €/kg (light blue line) increases the produces negative environmental impacts.
GI of the optimal solution of around 3 percentage points with respect to
a tank price of 400 €/kg. 3.5.5. LCOH resilience to market fluctuations
Fig. 19 shows the effect of a BESS price reduction on the same Due to the uncertain trend of the grid electricity purchase price, it is
quantities analyzed above. From a BESS starting price of 120 €/kWh, the worth assessing how the final LCOH varies when subjected to market
analysis considers a price reduction down to 50 €/kWh, which corre­ fluctuations. For each configuration, this analysis focuses on the effect of
sponds approximately to a 60% reduction. For BESS prices higher than the electricity purchase price variation on the resulting LCOH in con­
100 €/kWh, results show that is not economically convenient to invest in figurations with different storages installed. Three different electrolyzer
this kind of storage. When the price falls below this threshold, the bat­ installed power rates are considered, namely 26, 36 and 46 MW.
tery becomes a viable candidate. The optimal battery size increases Fig. 21 shows the relationship between the LCOH and the electricity
steeply at price levels lower than 90 €/kWh: optimum of 1 MWh at 90 purchase price variation (x axis). Lines of different colors (same for each
€/kWh, 5 MW at 80 €/kWh. In the latter point, the optimal tank capacity of the three graphs) show the LOCH trend of configurations character­
(continuous line) decreases from 13.5 to 9 tons of hydrogen, producing a ized by five different tank sizes: 0, 18, 36, 54 and 72 tons of hydrogen.
drop in the GI (dashed line). This trend is determined by the discretized The blue line represents the behavior of a configuration with no storage
nature of the parametric analysis that may produce discontinuities on installed (16 MW electrolyzer, no tank). This can be considered as the
results. After this abrupt change, the optimal tank size remains constant, reference for the most grid-dependent configuration, i.e., that providing
while a further drop in BESS price makes the optimal BESS size increase the lowest LCOH when the grid electricity is cheap, although very sen­
up to 20 MWh when the price is 50 €/kWh. Due to the small dimensions sitive to market fluctuations. Lines relative to a large installed tank re­
of this component with respect to the rest of the plant, this variation sults in a higher LCOH when the electricity purchase price is low (less
produces negligible changes on the LCOH (dotted line, 2c€/kg drop). than75 €/MWh) but starts becoming convenient when the price rises.
Table 4 summarizes the main results obtained by the sensitivity Fig. 22, similarly to Fig. 10, shows the LCOH variation according to
analyses on a) the electricity purchase and selling price variation, b) the the installed tank size, for increasing electricity purchase prices (rep­
electrolyzer and tank price variation and c) the battery price variation. resented in in different colors). A capacity range for the tank storage
Those results highlight the LCOH and the GI obtained at the maximum varying from 0 to 70 tons of hydrogen is considered. Fig. 22 (b) and
and minimum cost values considered for electricity and components. Fig. 21 (c) show the LCOH minimums that the installation of a storage
Additionally, it reports the optimal size of devices that best perform in system produces when the electricity purchase price is sufficiently high
each hypothetical market scenario. and a consistent difference between the selling and purchase price is
created. Fig. 22 (a) considers a minor electrolyzer power of 26 MW. The
3.5.4. potential impact of incentives span between LCOH lines of different electricity purchase prices remains
In a future scenario with rising electricity prices (accordingly to rather constant, even with increasing tank capacities. On the other hand,
latest trends) and a drop in the cost of technologies, configurations that Fig. 22 (c) shows that, when large capacity tanks are coupled with
involve large storage systems able reach high levels of self-sufficiency higher electrolyzer power (46 MW), the distance between the LCOH
seem promising. From the standpoint of a policy maker that aims to lines can be reduced.
facilitate the decarbonization of hard-to-abate sectors as the steel Fig. 21 and Fig. 22 also help in visualizing a key concept: a large
manufacturing, those results help to understand where incentives could installed tank, resulting in higher self-sufficiency of the system, makes
represent a catalyst of energy transition. Generally speaking, incentives the price of hydrogen less sensitive to the electric market fluctuation. If
targeted at lowering the expenditure in electricity purchase make the the electricity market sees a variation as the one that characterized the
system less “clean” with respect to those on the capital cost of technol­ recent years (2020–2021), the average electricity price may vary from
ogies. Fig. 20 reports the effect of the same amount of subsidy applied to 50 to more than 300 €/MWh. For those two extremes, the LCOH
electricity purchase price (Fig. 20 (a)) and electrolyzer price (Fig. 20 generated by a grid-dependent configuration like the one adopting a16
(b)). Both analyses consider, as a starting point, the configuration that MW of electrolyzer and no tank (blue line in Fig. 20) varies from less
brings to the lowest LCOH in a market condition in which the electricity than 4 €/kg to almost 9 €/kg (a 125% increase). The LCOH derived from
purchase price is 150 €/MWh and the selling price is 70 €/MWh; the a configuration with a higher degree of self-sufficiency the one imple­
system comprises a 37 MW electrolyzer paired with a tank able to store menting a 46 MW of electrolyzer and 72 tons of hydrogen tank (pink line
13.5 tons of hydrogen and no BESS. in graph (c)) sees a notably smaller variation, starting from a price of
This results in a LCOH of 5.7 €/kg and a GI of 82%. Fig. 20 (b) reports almost 6 €/kg and reaching a maximum price slightly higher than 7
again the effect of a reduction in the electrolyzer price from 650 to 200 €/kg, for a total variation around 1.5 €/kg. The resilience of self-

17
F. Superchi et al. Applied Energy 342 (2023) 121198

Fig. 16. Variation on Italian unified national price of electricity (PUN) for: a) 2019b) 2020c) 2021 d) January to August 2022. Average price recorded during the
year is represented by the orange line. Green line represents the price threshold required to classify the hydrogen as “green”.

sufficient systems to market variations in electricity price is key to un­ of equal share of HBI and scrap, as it is in the assumptions of this study,
derstand the potential of storage systems in this kind of applications. the required volumes turn into 738 kg of iron ore, 536 kg of scrap ma­
terial and 25 kg of hydrogen per ton of liquid steel. The definition of the
4. Emissions reduction material flows involved in the process is preliminary to the calculation of
both specific consumption and emissions and to make it comparable to
Based on the comprehensive analysis presented in the above para­ other case studies.
graphs, this section aims to assess and quantify the effective decarbon­ In terms of energy requirements, the H2-SF-EAF route is reported to
ization potential of the proposed solutions in the broader EU-27 context account for 4.25 MWh/tls by Bhaskar et al. [10], considering an elec­
of steel production. A comparison is made with both the traditional BF- trolyzer efficiency of 53 kWh/kgH2, for the techno-economic assessment
BOF route and with new generation H2-SF-EAF plants that rely entirely of a grid connected plant located in Norway using 100% HBI. By
on the electricity grid for energy supply. adopting the method derived from the above-mentioned study, the
On the basis of data available in the literature, an estimation of corresponding figure for this work turns out to be 2.201 MWh/tls,
material flows, electricity consumption and related emissions has been considering a nominal efficiency of 56.2 kWh/kgH2 for the electrolyzer
carried out for the case study under consideration to align with the most technology adopted in this study. The great reduction in consumption is
widely established indicators. In this framework, it is important to largely due to the use of a 50% share of scrap material. It is also worth
consider that hydrogen-based steel production can be divided into three noting that electrolysis in this case contributes about 65% of the total
different sub-processes, namely the production of Hot-Briquetted Iron energy consumption for steel production.
(HBI) in the shaft furnace, the iron-to-steel conversion in the EAF and Given these necessary assumptions, the specific emissions for the
the production and storage of the hydrogen required for the reduction case study have been calculated accounting for the impact of wind-
process. All the calculations for material and energy flows, as well as for produced hydrogen, in order to accurately define the decarbonization
the emission factor, are ultimately referred to the production of one ton potential of the proposed solutions. Total emissions can be classified into
of liquid steel. direct and indirect as made by Bhaskar et al. [10]. The only direct
Vogl et al. [21] report that, for the process under consideration, contribution is represented by emissions from EAF operations and ac­
1504 kg of iron ore pellets are required per each ton of liquid steel counts for 73 kgCO2/tls, due to lime production, carbon oxidation and
produced, together with 51 kg of hydrogen as reducing agent for the FeO reduction. Indirect emissions figures are also reported and re-
same output. They also specify that, if the feedstock for the EAF is made adapted from [21] and [10], presenting values of 53 and 55.90

18
F. Superchi et al. Applied Energy 342 (2023) 121198

Fig. 17. Electricity purchase and selling price effect on the variation on optimal configuration parameters: LCOH (dotted line), GI (dashed line), tank size
(continuous line) and electrolyzer size (dash-dotted line).

Fig. 18. Tank and electrolyzer price effect on the variation on optimal configuration parameters: LCOH (dotted line), GI (dashed line), tank size (continuous line) and
electrolyzer size (dash-dotted line).

kgCO2/tls respectively for carbon, lime and graphite electrodes con­ intensity of the grid to which the plant is connected.
sumption and for iron ore pellet. The parameter of main interest for the Fig. 23 shows the GI and the correspondent process emission
analysis is represented by the indirect emissions from electricity con­ reduction that can be achieved at different LCOH values. Starting from
sumption, which is a function of both material flows and the emission the price of the optimal configuration (5.7 €/kg), the graph shows how

19
F. Superchi et al. Applied Energy 342 (2023) 121198

Fig. 19. BESS price effect on the variation on optimal configuration parameters as: LCOH (dotted line), GI (dashed line), tank size (continuous line) and electrolyzer
size (dash-dotted line).

Table 4
Main outcomes of the sensitivity analysis on a) electricity purchase and selling price variation, b) electrolyzer and tank price variation and c) battery price variation.
El purchase price El EC cost Tank cost BESS cost EC power [MW] BESS size [MWh] Tank GI LCOH
selling [€/kW] [€/kg] [€/kWh] size [tons of H2] [%] [€/kg]
[€/MWh]
price
[€/MWh]

a) Electricity price sensitivity analysis

50* 120** 600 400 117 16 0 0 70.3 1.75
50* 60* 600 400 117 16 0 0 70.3 3.55
300** 120** 600 400 117 42 0 40 90.8 6.66
300** 60* 600 400 117 44 0 58 92.75 7.36
b) Electrolyzer and tank prices sensitivity analysis
150 70 650** 400** 117 28 0 13 82 5.7
150 70 650** 200* 117 30 0 22 84.5 5.55
150 70 200* 400** 117 37 0 17 86 4.93
150 70 200* 200 117 44 0 40 91.2 4.67
c) Battery price sensitivity analysis
150 70 650 400 120** 28 0 13 82 5.7
150 70 650 400 50* 24 20 9 80 5.67
*
minimum and ** maximum price value considered in the sensitivity analyses.

the GI can increase if a higher cost is accepted, thanks to higher sizes of can be achieved by configuration C, which brings to a GI of almost 96%
the storage systems. Direct emissions from reactions occurring in the but a considerably higher LCOH of 7.6 €/kg. In this case, the emission
electric arc furnace (EAF) are displayed in orange and are constant for intensity can be lowered to 218 kgCO2/tls. Between those two extremes
all configurations. Global emissions from the entire process, in grey, there is a trade-off between price and emission savings: configuration B.
vary according to the GI (green line) of the hydrogen fed to the plant. In this case, for a hydrogen price of 6.5 €/kg, the GI can reach almost
The optimal system size in economic terms does not match the most 94% and emissions can be reduced to 235 kgCO2/tls.
environmentally friendly solution and to reach higher green shares, the For the sake of comparison, Fig. 24 presents the re-adaptation of data
cost of the hydrogen must increase. presented in the work of Bhaskar et al. [10] that compare the carbon
Three configurations, summarized in Table 5, have been selected as a intensity the H2-DRI-EAF route in different countries, producing
reference and for comparison. hydrogen from electrolysis that entirely relies on electricity from the
Configuration A leads to the lowest LCOH, thus the optimal solution national grid (yellow hydrogen). The graph also reports the country-
in economic terms, and results in a LCOH of 5.7 €/kg. In this case the GI related emissions of the natural gas driven manufacturing process NG-
of the final product is around 82% - a remarkable result - which leads to DRI- EAF (red dots) and the emission band of the traditional
a carbon intensity of the steel manufacturing process of 334 kg of CO2 manufacturing pathway based on blast furnace (grey bands).
per ton of liquid steel. On the other hand, the most significant reduction The three case studies illustrated above are included in the graph

20
F. Superchi et al. Applied Energy 342 (2023) 121198

Fig. 20. Incentive comparison of the same magnitude on a) electricity purchase price and b) electrolyzer price.

Fig. 21. LCOH trends by varying electricity price and tank size for three different electrolyzer sizes: 26 (a), 36 (b) and 46 MW (c).

next to the bar representing the Italian scenario. In Italy, due to the high The carbon intensity of a process that utilizes hydrogen produced by
reliance on fossil fuels for electricity production, a steel manufacturing configuration A would be comparable to what can be achieved in France
process that fully relies on yellow hydrogen can reach emissions of or Finland, i.e., countries in which the electricity generation highly re­
almost 1 tCO2/tls, even higher than NG-EAF (844 kgCO2/tls). The three lies on low carbon sources. Using configurations B and C the result starts
proposed configurations show considerably lower specific emissions. to approach even less carbon intensive countries as Norway. This

21
F. Superchi et al. Applied Energy 342 (2023) 121198

Fig. 22. LCOH varying electricity price and tank size for three different electrolyzer sizes: 26 (a), 36 (b) and 46 MW (c).

comparison shows that, in a country in which electricity generation is new small-scale plant concept that processes half scrap and half raw
still relying on non-renewable fuel sources, hydrogen production sys­ materials. The decarbonization of the steel production sector is part of
tems that directly exploits the energy produced by renewable power the path towards a cleaner manufacturing industry. In that context,
stations are the only way to decarbonize hard-to-abate processes as the green hydrogen can play a key role in achieving the greenhouse emis­
steel manufacturing. sions reduction goals of our society. Real data from an existing wind
farm were used to estimate the producibility of such systems. The rise of
5. Conclusions the intermittent renewable energy generation opens new possibilities for
producing hydrogen in a sustainable way, but also brings new chal­
The study provides a techno-economic analysis on the potential lenges. Real data from an existing wind farm were used to estimate the
production of a constant flow rate of green hydrogen for an industrial producibility of such systems. Due to the intermittent nature of wind
user fed by a dedicated wind farm. A steel mill was selected as the hy­ power production, two storage means were also considered and
pothetical final user of the produced hydrogen. The industrial demand compared to match wind fluctuations with the constant request of the
was modelled on a H2-DRI-EAF steel making process, considering the steel mill, namely batteries and hydrogen tanks. In addition, alkaline

Fig. 23. Emission intensity and GI varying the LCOH of the system.

22
F. Superchi et al. Applied Energy 342 (2023) 121198

Table 5 a large capacity battery to store electricity upstream the electrolyzer.

Configurations for comparison. The best option to reach a high share of green hydrogen, while guar­
Conf. Electrolyzer Battery Tank GI Emission LCOH anteeing the constant flow rate needed to meet steel mill demand, is to
power [MW] Capacity Capacity [%] intensity [€/kg] enlarge the downstream storage capacity of the system, allowing the
[MWh] [tons of [kgCO2/ electrolyzer stack to follow the power fluctuations of the wind farm.
H2] tls]
Furthermore, four different upscaling of the original wind farm were
A 28 1 13.5 82 334 5.7 analyzed to adapt the power production potential to the constant
B 48 4 68 94 235 6.5 request of the electrolyzers (16 MW). Electrolyzers power levels from 16
C 56 17 117 96 218 7.6
to 56 MW were considered, coupled with battery capacities ranging
from 0 to 20 MWh and tanks able to store from 0 to 117 tons of
electrolyzers and lithium-ion batteries models that account for aging hydrogen. Global results show that the configuration enabling to reach
and degradation effects were employed to simulate a realistic behavior the lowest levelized cost of hydrogen consists in a large-scale wind farm
under the fluctuating operation regime that they will face. (4 times the original one), coupled with 28 MW electrolyzers, a 1 MWh
Several system configurations were considered, and their techno- battery and tanks able to contain 13.5 tons of hydrogen; with this
economic outcome was evaluated by means of two parameters, configuration, the resulting LCOH is 5.7 €/kg, with a GI of 82%. If the
namely the green index (GI), which considers the self-consumption of electrolyzer cost is decreased to 200 €/kW, due to the future market
the system and the share of renewables in the national electric grid, and development or thanks to incentives, the optimal configuration could
the Levelized Cost of Hydrogen (LCOH). possibly reach a LCOH of 4.93 €/kg, still considerably higher than cur­
Results show that it is not possible to reach a 100% green hydrogen rent grey hydrogen price (1–2 €/kg). Inevitably, a hydrogen production
flow through the whole year, unless the wind farm is significantly system that must provide a constant flow rate of gas is still dependent on
oversized, which makes anyhow the investment not convenient. It must grid electricity, whose high cost still hinders the potential cost
be stressed out that, due to the intermittent nature of wind power pro­ reduction.
duction, the national grid support still plays a key role in meeting the The sensitivity analysis conducted over the price of electricity and
constant hydrogen demand. Nevertheless, storage means are key to in­ the capital investment of components shows that it is key to reduce the
crease the self-consumption of the system, the resulting green index, and individual cost technologies (electrolyzers, batteries and tanks) to both
the global emission reduction potential of the process. This latter point reduce the LCOH and increase the green index. Among the various
was thoroughly addressed by accounting for different configurations of components, the electrolyzer price has shown the most relevant influ­
the system. In particular, it is shown that it is not convenient to invest in ence on the LCOH trend and is the most likely to decline in the near

Fig. 24. Emission intensity for steel manufacturing in EU-27 countries, re.
adapted from [10]

23
F. Superchi et al. Applied Energy 342 (2023) 121198

future. A price drop of 70% in electrolysis technologies would produce a Acknowledgements

1 €/kg reduction on the final LCOH. Regarding storage means, a 50%
cost reduction in tanks would have a smaller effect on the final hydrogen This research did not receive any specific grant from funding
price but could increase the green index of the optimal solution up to agencies in the public, commercial, or not-for-profit sectors.
7%. On the other hand, a sensitivity analysis has shown that a large-scale The authors would like to sincerely thank Eng. Mattia Pasqui for the
battery storage would be economically unsustainable to reach high intellectual assistance and support received during the conceptualiza­
levels of self-sufficiency with respect to tanks. Due to the relatively small tion of this work.
size range considered in the size optimization, the contribution of this
component to the final LCOH is negligible (i.e., 2c€/kg for a 60% price References
drop of the component). A decline in electricity purchase prices would
be crucial to reduce the cost of hydrogen, but it is very difficult to state [1] Energy Transitions Commission, “Mission Possible: Reaching Net-Zero Carbon
Emissions - ETC,” 2018. https://www.energy-transitions.org/publication
how the electricity market might evolve even in the nearest future. s/mission-possible/ (accessed Dec. 02, 2022).
These results are of particular interest for policy makers aiming to push [2] Schleussner C-F, et al. Science and policy characteristics of the Paris Agreement
forward the penetration of green hydrogen in industrial processes, since temperature goal. Nat Clim Chang 2016;6(9):827–35. https://doi.org/10.1038/
nclimate3096.
it is apparent that incentives should be focused on the reduction of [3] Iea. Iron and Steel Technology Roadmap. Paris 2020;2020. https://doi.org/
components prices rather than on electricity purchase prices. Configu­ 10.1787/3dcc2a1b-en.
rations that involve high-capacity storage means have been also proved [4] “World Steel in Figures 2022 - worldsteel.org.” https://worldsteel.org/stee
l-topics/statistics/world-steel-in-figures-2022/ (accessed Oct. 17, 2022).
to be more resilient to the electricity market fluctuations. One [5] IEA, “IEA (2021),” Paris, 2021. doi: 10.1787/90c8c125-en.
conceivable way to reduce both the cost of producing hydrogen and its [6] M. D. Fenton and C. A. Tuck, “Iron and Steel. 2016 Minerals Yearbook,” US Geol.
carbon footprint is to decarbonize the electricity grid itself. Surv., 2019.
[7] Zhao J, Zuo H, Wang Y, Wang J, Xue Q. Review of green and low-carbon
Results show that the emission reduction potential of hydrogen
ironmaking technology. Ironmak Steelmak 2020;47(3):296–306. https://doi.org/
streams characterized by a high green index is remarkable. The emission 10.1080/03019233.2019.1639029.
intensity of a H2-DRI-EAF steelmaking process located in Italy that takes [8] “Iron and Steel – Analysis - IEA.” https://www.iea.org/reports/iron-and-steel
as input the hydrogen mix of the most cost-effective configuration stands (accessed Mar. 29, 2023).
[9] Lee H, Lee J, Koo Y. Economic impacts of carbon capture and storage on the steel
at around 334 kgCO2/tls, a reduction of almost 84% if compared to the industry–A hybrid energy system model incorporating technological change. Appl
traditional BF-BOF process. This quantity can be further reduced if Energy Jul. 2022;317:119208. https://doi.org/10.1016/J.
higher costs of hydrogen are accepted: the emission intensity of con­ APENERGY.2022.119208.
[10] Bhaskar A, Abhishek R, Assadi M, Somehesaraei HN. Decarbonizing primary steel
figurations presenting green indexes of 94% corresponds to 235 kgCO2/ production: Techno-economic assessment of a hydrogen based green steel
tls, but a LOCH of 6.5 €/kg must be taken into account. As for the Italian production plant in Norway. J Clean Prod 2022;vol. 350, no. March:131339.
energy mix, the emission intensity of the same process powered by grid https://doi.org/10.1016/j.jclepro.2022.131339.
[11] Nwachukwu CM, Olofsson E, Lundmark R, Wetterlund E. Evaluating fuel switching
electricity would be more than four times higher (1000 kgCO2/tls). In options in the Swedish iron and steel industry under increased competition for
countries where the electrical grid is still heavily dependent on fossil forest biomass. Appl Energy Oct. 2022;324:119878. https://doi.org/10.1016/J.
fuels, plants similar to the one analyzed in this work could enable a APENERGY.2022.119878.
[12] H. Suopajärvi et al., “Use of biomass in integrated steelmaking – Status quo, future
carbon reduction of the steel making process similar to what can be needs and comparison to other low-CO2 steel production technologies,” Appl.
achieved in countries whose electrical grid is characterized by a really Energy, vol. 213, no. November 2017, pp. 384–407, 2018, doi: 10.1016/j.
small carbon footprint. apenergy.2018.01.060.
[13] Jahanshahi S, et al. Development of Low-Emission Integrated Steelmaking Process.
Finally, upon examination of the potential emission reduction of the
J Sustain Metall 2015;1(1):94–114. https://doi.org/10.1007/s40831-015-0008-6.
steelmaking process, it is shown that, in a country in which electricity [14] Toktarova A, Walter V, Göransson L, Johnsson F. Interaction between electrified
generation is still relying on non-renewable fuel sources, hydrogen steel production and the north European electricity system. Appl Energy 2022;310
production systems that directly exploits the energy produced by (January). https://doi.org/10.1016/j.apenergy.2022.118584.
[15] “PROCESS INTEGRATION IN THE IRON AND STEEL INDUSTRY: IEA IETS ANNEX
renewable power stations are the only way to decarbonize hard-to-abate XIV TECHNICAL REPORT,” Accessed: Mar. 29, 2023. [Online]. Available: http://
processes as the steel manufacturing. www.iea-industry.org.
[16] World Steel Association, “STEEL ’ S CONTRIBUTION TO A LOW CARBON FUTURE
AND CLIMATE RESILIENT SOCIETIES worldsteel position paper,” World Steel
CRediT authorship contribution statement Assoc., pp. 1–6, 2015, [Online]. Available: http://www.worldsteel.
org/dms/internetDocumentList/booksh
Francesco Superchi: Methodology, Software, Validation, Formal op/Steel-s-Contribution-to-a-Low-Carbon-Future-/document/Steel’s Contribution
to a Low Carbon Future .pdf.
analysis, Investigation, Data curation, Writing – original draft, Visuali­ [17] Nwachukwu CM, Toffolo A, Wetterlund E. Biomass-based gas use in Swedish iron
zation. Alessandro Mati: Software, Validation, Formal analysis, Inves­ and steel industry – Supply chain and process integration considerations. Renew
tigation, Data curation, Writing – original draft. Carlo Carcasci: Energy Feb. 2020;146:2797–811. https://doi.org/10.1016/J.
RENENE.2019.08.100.
Resources, Supervision, Project administration, Funding acquisition. [18] S. Tian, J. Jiang, Z. Zhang, and V. Manovic, “Inherent potential of steelmaking to
Alessandro Bianchini: Conceptualization, Methodology, Investigation, contribute to decarbonisation targets via industrial carbon capture and storage,”
Resources, Data curation, Writing – review & editing, Supervision, doi: 10.1038/s41467-018-06886-8.
[19] “Impact of Hydrogen DRI on EAF Steelmaking - Midrex Technologies, Inc.” http
Project administration, Funding acquisition.
s://www.midrex.com/tech-article/impact-of-hydrogen-dri-on-eaf-steelmaking/
(accessed Mar. 29, 2023).
Declaration of Competing Interest [20] Fan Z, Friedmann SJ. Low-carbon production of iron and steel: Technology options,
economic assessment, and policy. Joule Apr. 2021;5(4):829–62. https://doi.org/
10.1016/J.JOULE.2021.02.018.
The authors declare that they have no known competing financial [21] Vogl V, Åhman M, Nilsson LJ. Assessment of hydrogen direct reduction for fossil-
interests or personal relationships that could have appeared to influence free steelmaking. J Clean Prod 2018;203:736–45. https://doi.org/10.1016/j.
the work reported in this paper. jclepro.2018.08.279.
[22] Fischedick M, Marzinkowski J, Winzer P, Weigel M. Techno-economic evaluation
of innovative steel production technologies. J Clean Prod 2014;84(1):563–80.
Data availability https://doi.org/10.1016/J.JCLEPRO.2014.05.063.
[23] M. Pei, M. Petäjäniemi, A. Regnell, and O. Wijk, “metals Toward a Fossil Free
Future with HYBRIT: Development of Iron and Steelmaking Technology in Sweden
Data will be made available on request. and Finland,” doi: 10.3390/met10070972.
[24] “HYBRIT pilot plant produces first sponge iron using fossil-free hydrogen gas | S&P
Global Commodity Insights.” https://www.spglobal.com/commodityinsights/en/

24
F. Superchi et al. Applied Energy 342 (2023) 121198

market-insights/latest-news/electric-power/062121-hybrit-pilot-plant-produces- [52] “MINIMILLS and INTEGRATED PLANTS | Metech STG.” http://www.stggroup.it/

first-sponge-iron-using-fossil-free-hydrogen-gas (accessed Oct. 18, 2022). en/plants/minimills (accessed Nov. 08, 2022).
[25] “Hydrogen-based steelmaking to begin in Hamburg | ArcelorMittal.” https://corp [53] “Hydrogen steel plant: voestalpine X Mitsubishi Heavy Industries | EU-Japan.”
orate.arcelormittal.com/media/case-studies/hydrogen-based-steelmaking-to-begi https://www.eu-japan.eu/publications/hydrogen-steel-plant-voestalpine-x-mits
n-in-hamburg (accessed Oct. 18, 2022). ubishi-heavy-industries (accessed Nov. 08, 2022).
[26] V. Vogl et al., “Green Steel Tracker.” Stockholm, 2021, [Online]. Available: www. [54] “Hydrogen-based steelmaking to begin in Hamburg | ArcelorMittal.” https://corp
industrytransition.org/green-steel-tracker. orate.arcelormittal.com/media/case-studies/hydrogen-based-steelmaking-to-begi
[27] Bhaskar A, Assadi M, Somehsaraei HN. Decarbonization of the iron and steel n-in-hamburg (accessed Nov. 08, 2022).
industry with direct reduction of iron ore with green hydrogen. Energies 2020;13 [55] Ajanovic A, Sayer M, Haas R. The economics and the environmental benignity of
(3):1–23. https://doi.org/10.3390/en13030758. different colors of hydrogen. Int J Hydrogen Energy 2022;no. xxxx. https://doi.
[28] Davis SJ, et al. Net-zero emissions energy systems. Science (80-.) Jun. 2018;360 org/10.1016/j.ijhydene.2022.02.094.
(6396). https://doi.org/10.1126/SCIENCE.AAS9793. [56] Superchi F, Papi F, Mannelli A, Balduzzi F, Ferro FM, Bianchini A. Development of
[29] Dinh VN, Leahy P, McKeogh E, Murphy J, Cummins V. Development of a viability a reliable simulation framework for techno-economic analyses on green hydrogen
assessment model for hydrogen production from dedicated offshore wind farms. Int production from wind farms using alkaline electrolyzers. Renew Energy May 2023;
J Hydrogen Energy Jul. 2021;46(48):24620–31. https://doi.org/10.1016/J. 207:731–42. https://doi.org/10.1016/J.RENENE.2023.03.077.
IJHYDENE.2020.04.232. [57] A. Mannelli, F. Papi, G. Pechlivanoglou, G. Ferrara, and A. Bianchini, “Discrete
[30] Lucas TR, Ferreira AF, Santos Pereira RB, Alves M. Hydrogen production from the Wavelet Transform for the Real-Time Smoothing of Wind Turbine Power Using Li-
WindFloat Atlantic offshore wind farm: A techno-economic analysis. Appl Energy Ion Batteries,” Energies 2021, Vol. 14, Page 2184, vol. 14, no. 8, p. 2184, Apr. 2021,
Mar. 2022;310. https://doi.org/10.1016/J.APENERGY.2021.118481. doi: 10.3390/EN14082184.
[31] Franco BA, Baptista P, Neto RC, Ganilha S. Assessment of offloading pathways for [58] Divya KC, Østergaard J. Battery energy storage technology for power systems-An
wind-powered offshore hydrogen production: Energy and economic analysis. Appl overview. Electr Power Syst Res Apr. 2009;79(4):511–20. https://doi.org/
Energy Mar. 2021;286:116553. https://doi.org/10.1016/J. 10.1016/J.EPSR.2008.09.017.
APENERGY.2021.116553. [59] Mah AXY, et al. Optimization of a standalone photovoltaic-based microgrid with
[32] McDonagh S, Ahmed S, Desmond C, Murphy JD. Hydrogen from offshore wind: electrical and hydrogen loads. Energy Nov. 2021;235. https://doi.org/10.1016/J.
Investor perspective on the profitability of a hybrid system including for ENERGY.2021.121218.
curtailment. Appl Energy 2020;vol. 265, no. February:114732. https://doi.org/ [60] Nicita A, Maggio G, Andaloro APF, Squadrito G. Green hydrogen as feedstock:
10.1016/j.apenergy.2020.114732. Financial analysis of a photovoltaic-powered electrolysis plant. Int J Hydrogen
[33] Olateju B, Kumar A, Secanell M. A techno-economic assessment of large scale wind- Energy Apr. 2020;45(20):11395–408. https://doi.org/10.1016/j.
hydrogen production with energy storage in Western Canada. Int J Hydrogen ijhydene.2020.02.062.
Energy 2016;41(21):8755–76. https://doi.org/10.1016/j.ijhydene.2016.03.177. [61] Jang D, Kim K, Kim KH, Kang S. Techno-economic analysis and Monte Carlo
[34] Weimann L, Gabrielli P, Boldrini A, Kramer GJ, Gazzani M. Optimal hydrogen simulation for green hydrogen production using offshore wind power plant. Energy
production in a wind-dominated zero-emission energy system. Adv Appl Energy Convers Manag Jul. 2022;263. https://doi.org/10.1016/J.
Aug. 2021;3. https://doi.org/10.1016/J.ADAPEN.2021.100032. ENCONMAN.2022.115695.
[35] Glenk G, Reichelstein S. Economics of converting renewable power to hydrogen. [62] Superchi F, Mati A, Pasqui M, Carcasci C, Bianchini A. Techno-economic study on
Nat Energy 2019;4(3):216–22. https://doi.org/10.1038/s41560-019-0326-1. green hydrogen production and use in hard-to-abate industrial sectors. J Phys Conf
[36] Saba SM, Müller M, Robinius M, Stolten D. The investment costs of electrolysis – A Ser 2022.
comparison of cost studies from the past 30 years. Int J Hydrogen Energy Jan. [63] Lubello P, Pasqui M, Mati A, Carcasci C. “Assessment of hydrogen based long term
2018;43(3):1209–23. https://doi.org/10.1016/J.IJHYDENE.2017.11.115. electrical energy storage in residential energy systems”, Smart. Energy 2022;vol. 8,
[37] Shiva Kumar S, Himabindu V. Hydrogen production by PEM water electrolysis – A no. August:100088. https://doi.org/10.1016/j.segy.2022.100088.
review. Mater Sci Energy Technol Dec. 2019;2(3):442–54. https://doi.org/ [64] Singlitico A, Østergaard J, Chatzivasileiadis S. Onshore, offshore or in-turbine
10.1016/J.MSET.2019.03.002. electrolysis? Techno-economic overview of alternative integration designs for
[38] “Nel ASA: Receives 4.5 MW electrolyzer purchase order for fossil free steel green hydrogen production into Offshore Wind Power Hubs. Renew Sustain Energy
production | Nel Hydrogen.” https://nelhydrogen.com/press-release/nel-asa-rece Transit Aug. 2021;1:100005. https://doi.org/10.1016/J.RSET.2021.100005.
ives-4-5-mw-electrolyzer-purchase-order-for-fossil-free-steel-production/ [65] Kazi MK, Eljack F, El-Halwagi MM, Haouari M. Green hydrogen for industrial
(accessed Oct. 26, 2022). sector decarbonization: Costs and impacts on hydrogen economy in qatar. Comput
[39] Schmidt O, Gambhir A, Staffell I, Hawkes A, Nelson J, Few S. Future cost and Chem Eng Feb. 2021;145. https://doi.org/10.1016/J.
performance of water electrolysis: An expert elicitation study. Int J Hydrogen COMPCHEMENG.2020.107144.
Energy Dec. 2017;42(52):30470–92. https://doi.org/10.1016/J. [66] Duffy A, et al. Land-based wind energy cost trends in Germany, Denmark, Ireland,
IJHYDENE.2017.10.045. Norway, Sweden and the United States. Appl Energy Nov. 2020;277. https://doi.
[40] Marini S, et al. Advanced alkaline water electrolysis. Electrochim Acta Nov. 2012; org/10.1016/J.APENERGY.2020.114777.
82:384–91. https://doi.org/10.1016/J.ELECTACTA.2012.05.011. [67] Wang R, Lam CM, Hsu SC, Chen JH. Life cycle assessment and energy payback time
[41] Ursúa A, Barrios EL, Pascual J, San Martín I, Sanchis P. Integration of commercial of a standalone hybrid renewable energy commercial microgrid: A case study of
alkaline water electrolysers with renewable energies: Limitations and Town Island in Hong Kong. Appl Energy Sep. 2019;250:760–75. https://doi.org/
improvements. Int J Hydrogen Energy Aug. 2016;41(30):12852–61. https://doi. 10.1016/J.APENERGY.2019.04.183.
org/10.1016/J.IJHYDENE.2016.06.071. [68] S. Furfari and A. Clerici, “Green hydrogen: the crucial performance of electrolysers
[42] Liponi A, Baccioli A, Ferrari L, Desideri U. Techno-economic analysis of hydrogen fed by variable and intermittent renewable electricity,” Eur. Phys. J. Plus, vol. 136,
production from PV plants. E3S Web Conf 2022;334:01001. https://doi.org/ no. 5, May 2021, doi: 10.1140/EPJP/S13360-021-01445-5.
10.1051/e3sconf/202233401001. [69] D. Pivetta, C. Dall’armi, and R. Taccani, “Multi-Objective Optimization of a
[43] Usman MR. Hydrogen storage methods: Review and current status. Renew Sustain Hydrogen Hub for the Decarbonization of a Port Industrial Area,” J. Mar. Sci. Eng.,
Energy Rev Oct. 2022;167:112743. https://doi.org/10.1016/J. vol. 10, no. 2, 2022, doi: 10.3390/jmse10020231.
RSER.2022.112743. [70] Yang Y, et al. The scheduling of alkaline water electrolysis for hydrogen production
[44] Abe JO, Popoola API, Ajenifuja E, Popoola OM. Hydrogen energy, economy and using hybrid energy sources. Energy Convers Manag Apr. 2022;257. https://doi.
storage: Review and recommendation. Int J Hydrogen Energy Jun. 2019;44(29): org/10.1016/J.ENCONMAN.2022.115408.
15072–86. https://doi.org/10.1016/J.IJHYDENE.2019.04.068. [71] Fan JL, Yu P, Li K, Xu M, Zhang X. A levelized cost of hydrogen (LCOH) comparison
[45] Zhang F, Zhao P, Niu M, Maddy J. The survey of key technologies in hydrogen of coal-to-hydrogen with CCS and water electrolysis powered by renewable energy
energy storage. Int J Hydrogen Energy Sep. 2016;41(33):14535–52. https://doi. in China. Energy Mar. 2022;242:123003. https://doi.org/10.1016/J.
org/10.1016/J.IJHYDENE.2016.05.293. ENERGY.2021.123003.
[46] Meier K. Hydrogen production with sea water electrolysis using Norwegian [72] Gorre J, Ruoss F, Karjunen H, Schaffert J, Tynjälä T. Cost benefits of optimizing
offshore wind energy potentials: Techno-economic assessment for an offshore- hydrogen storage and methanation capacities for Power-to-Gas plants in dynamic
based hydrogen production approach with state-of-the-art technology. Int J Energy operation. Appl Energy Jan. 2020;257:113967. https://doi.org/10.1016/J.
Environ Eng Jul. 2014;5(2–3):1–12. https://doi.org/10.1007/S40095-014-0104- APENERGY.2019.113967.
6/TABLES/6. [73] W. Cole, A. W. Frazier, and C. Augustine, “Cost Projections for Utility-Scale Battery
[47] Correa G, Volpe F, Marocco P, Muñoz P, Falagüerra T, Santarelli M. Evaluation of Storage: 2021 Update,” 2030, Accessed: Nov. 10, 2022. [Online]. Available: www.
levelized cost of hydrogen produced by wind electrolysis: Argentine and Italian nrel.gov/publications.
production scenarios. J Energy Storage Aug. 2022;52. https://doi.org/10.1016/J. [74] S. Kharel and B. Shabani, “Hydrogen as a Long-Term Large-Scale Energy Storage
EST.2022.105014. Solution to Support Renewables,” Energies 2018, Vol. 11, Page 2825, vol. 11, no. 10,
[48] Nascimento da Silva G, Rochedo PRR, Szklo A. Renewable hydrogen production to p. 2825, Oct. 2018, doi: 10.3390/EN11102825.
deal with wind power surpluses and mitigate carbon dioxide emissions from oil [75] “Future renewable energy costs: onshore wind Renewable Energies.”.
refineries. Appl Energy Apr. 2022;311. https://doi.org/10.1016/J. [76] “GME - Statistiche - dati di sintesi MPE-MGP.” https://www.mercatoelettrico.org/
APENERGY.2022.118631. it/Statistiche/ME/DatiSintesi.aspx (accessed Nov. 10, 2022).
[49] Eurofer, European steel in figures 2022. 2022. [77] International Renewable Energy Agency, Renewable Power Generation Costs in 2021.
[50] “World Steel in Figures 2022 - worldsteel.org.” https://worldsteel.org/stee 2021.
l-topics/statistics/world-steel-in-figures-2022/ (accessed Mar. 30, 2023). [78] LevelTen, “PPA Price Index Executive Summary,” 2022.
[51] “Danieli.” https://www.danieli.com/it/about-us/storia/integrated-minimills_18_ [79] European Commission, “COMMISSION DELEGATED REGULATION (EU)
14.htm (accessed Nov. 08, 2022). supplementing Directive (EU) 2018/2001 of the European Parliament and of the

25
F. Superchi et al. Applied Energy 342 (2023) 121198

Council by establishing a Union methodology setting out detailed rules for the [84] IRENA, “World energy transitions outlook,” Irena, pp. 1–54, 2022, [Online].
production of renewable liquid and gaseous transport fuels of,” pp. 9–25, 2022. Available: https://irena.org/publications/2021/March/World-Energy-Transitions-
[80] “Terna: nel 2021 deciso recupero dei consumi elettrici +5,6% rispetto al 2020, Outlook.
tornati sui valori del 2019 - Terna spa.” https://www.terna.it/it/media/comunicat [85] Ziegler MS, Trancik JE. Re-examining rates of lithium-ion battery technology
i-stampa/dettaglio/consumi-elettrici-2021 (accessed Nov. 10, 2022). improvement and cost decline. Cite this Energy Environ Sci 2021;14:1635. https://
[81] I. - International Energy Agency, “Global Hydrogen Review 2022,” 2022, Accessed: doi.org/10.1039/d0ee02681f.
Nov. 10, 2022. [Online]. Available: www.iea.org/t&c/. [86] International Renewable Energy Agency, “Electricity storage and renewables: Costs
[82] IEA, “The Future of Hydrogen,” Paris, 2019. Accessed: Nov. 14, 2022. [Online]. and markets to 2030,” Int. Renew. Energy Agency, no. October, p. 132, 2017,
Available: https://www.iea.org/reports/the-future-of-hydrogen. [Online]. Available: http://irena.org/publications/2017/Oct/Electricity-storage
[83] “Nel to slash cost of electrolysers by 75%, with green hydrogen at same price as -and-renewables-costs-and-markets%0Ahttps://www.irena.org/-/media/Files
fossil H2 by 2025 | Recharge.” https://www.rechargenews.com/transition/nel-to-s /IRENA/Agency/Publication/2017/Oct/IRENA_Electricity_Storage_Costs_2017.pd
lash-cost-of-electrolysers-by-75-with-green-hydrogen-at-same-price-as-fossil-h2- f.
by-2025/2-1-949219 (accessed Nov. 14, 2022). [87] C. Curry, “Lithium-ion Battery Costs and Market Squeezed margins seek technology
improvements & new business models,” 2017.

26