1 s2.0 S0959652621014104 Mainext

Download as pdf or txt
Download as pdf or txt
You are on page 1of 24

i An update to this article is included at the end

Journal of Cleaner Production 306 (2021) 127191

Contents lists available at ScienceDirect

Journal of Cleaner Production


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

Decarbonizing copper production by power-to-hydrogen: A techno-


economic analysis
€ ben a, b, Nikolas Scho
Fritz T.C. Ro € ne a, Uwe Bau a, Markus A. Reuter c, d, Manuel Dahmen a,
Andre  Bardow e, a, f, g, *
a
Institute of Energy and Climate Research - Energy Systems Engineering (IEK-10), Forschungszentrum Jülich GmbH, 52425, Jülich, Germany
b
RWTH Aachen University, 52062, Aachen, Germany
c
SMS Group GmbH, 40237, Düsseldorf, Germany
d
Helmholtz Institute Freiberg for Resource Technology (HIF), 09599, Freiberg, Germany
e
JARA-ENERGY, 52425, Jülich, Germany
f
Institute of Technical Thermodynamics, RWTH Aachen University, 52062, Aachen, Germany
g
Energy & Process Systems Engineering, ETH Zürich, 8092, Zürich, Switzerland

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

Article history: Electrifying energy-intensive processes is currently intensively explored to cut greenhouse gas (GHG)
Received 20 December 2020 emissions through renewable electricity. Electrification is particularly challenging if fossil resources are
Received in revised form not only used for energy supply but also as feedstock. Copper production is such an energy-intensive
18 March 2021
process consuming large quantities of fossil fuels both as reducing agent and as energy supply.
Accepted 18 April 2021
Available online 25 April 2021
Here, we explore the techno-economic potential of Power-to-Hydrogen to decarbonize copper pro-
duction. To determine the minimal cost of an on-site retrofit with Power-to-Hydrogen technology, we
^as de
Handling editor: Cecilia Maria Villas Bo formulate and solve a mixed-integer linear program for the integrated system. Under current techno-
Almeida economic parameters for Germany, the resulting direct CO2 abatement cost is 201 EUR/t CO2-eq for
Power-to-Hydrogen in copper production. On-site utilization of the electrolysis by-product oxygen has a
Keywords: substantial economic benefit. While the abatement cost vastly exceeds current European emission cer-
Decarbonization tificate prices, a sensitivity analysis shows that projected future developments in Power-to-Hydrogen
Power-to-Hydrogen technologies can greatly reduce the direct CO2 abatement cost to 54 EUR/t CO2-eq. An analysis of the
Copper production
total GHG emissions shows that decarbonization through Power-to-Hydrogen reduces the global GHG
Greenhouse gas emissions (GHG)
emissions only if the emission factor of the electricity supply lies below 160 g CO2-eq/kWhel.
The results suggest that decarbonization of copper production by Power-to-Hydrogen could become
economically and environmentally beneficial over the next decades due to cheaper and more efficient
Power-to-Hydrogen technology, rising GHG emission certificate prices, and further decarbonization of
the electricity supply.
© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).

1. Introduction supply (Luderer et al., 2018).


However, replacing fossil fuels with electricity is not always
The reduction of CO2 emissions in the industrial sector will be directly possible since fossil fuels serve two use cases in energy-
crucial to meet the goals for climate change mitigation (Rogelj et al., intensive industries: 1) the supply of process heat by combustion
2018; de Coninck et al., 2018). Besides increased energy efficiency, and 2) the use as chemical feedstock. Process heat demand is
additional options are necessary to reduce emissions (Fischedick typically large and only few alternatives to fossil fuels can supply
et al., 2014b). An increasingly powerful tool for reducing emis- heat at the required consistency, fluxes, and in particular high
sions is electrification due to the rapid decarbonization of power temperatures (Friedmann et al., 2019). A promising alternative to
fossil fuels is the combustion of renewable hydrogen (H2) to pro-
vide high-temperature process heat (Friedmann et al., 2019).
* Corresponding author. Energy & Process Systems Engineering, ETH Zürich, CLA Renewable H2 can often also be used as an alternative chemical
F19.1, Tannenstrasse 3, 8092, Zürich, Switzerland. feedstock.
E-mail address: [email protected] (A. Bardow).

https://doi.org/10.1016/j.jclepro.2021.127191
0959-6526/© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
€ben, N. Scho
F.T.C. Ro €ne, U. Bau et al. Journal of Cleaner Production 306 (2021) 127191

Nomenclature opxfix
c Fixed annual operation and maintenance cost factor
[%/a]
Pcmax Maximal installation size [kW]
Abbreviations Pcmin Minimal installation size [kW]
AEL Alkaline electrolysis ps;t Commodity prices [EUR/kWh, EUR/kg]
AF Anode furnace LHVs Lower heating value [kWh/mol]
ASU Air separation unit
C Carbon Variables
CH4 Methane E_ e;c;t Energy/material flow e2S ∪ I ∪ D [kW, kg/h]
CO2-eq Carbon dioxide equivalent S_s;t Supply of energy/material [kW, kg/h]
EL Electrolyzer system
COdir2t Time-specific direct CO2 emissions [g/h]
GHG Greenhouse gas
COdir2 Annual direct CO2 emissions [g/a]
H2 Hydrogen
CPXc Capital expenditures [EUR]
H2C Hydrogen compressor
GHGind Annual indirect GHG emissions [g/a]
H2st Hydrogen storage
GHGtot Annual total GHG emissions [g/a]
MILP Mixed-integer linear program
OPXcfix Fixed annual operation and maintenance costs [EUR/
O2 Oxygen
a]
PEMEL Polymer electrolyte membrane electrolysis var
OPXs;t Variable operating costs [EUR/h]
PES Process electricity supply
PcN Capacity of component [kW]
PHS Process heat supply O
Pc;t Operation rate of component [kW]
SCF Slag-cleaning furnace
SOC0 Initial state of charge of the H2 storage [kWh]
SOEL Solid oxide electrolysis
SOCt State of charge of the H2 storage [kWh]
TAC Total annualized cost
xc;t Decision to operate component [-]
VL Valve
yc Decision to install component [-]
WACC Weighted average cost of capital
Subscripts
Sets
Dphd; H2 Process heat demand savings when using H2 as
c 2C Components ¼ {EL, H2st, H2C, VL, PHS, AF, SCF, PES}
reducing agent
d 2D Demands ¼ {phd, O2d, rad, rsd, phsd, eld}
Dphsd; H2 Process heat demand savings in SCF when using H2
i 2I Intermediates ¼ {H2, pH2, O2e}
as reducing agent
s 2S Sources ¼ {el, CH4, C, O2s}
ei Material/energy input
t 2T Time steps
eo Material/energy output
stk Electrolyzer stack
Parameters
el electricity
a CO2 cap in relation to conventional operation [-]
eld electricity demand
DhR Enthalpy of reaction [kJ/mol]
O2d Oxygen demand
D_ d;t Time-specific process demands [kW, kg/h] O2e Oxygen from water electrolysis
εdir
s;c Direct CO2 emission factors [g/kWh] O2s Oxygen source
εind
s Indirect GHG emission factors [g/kWh] pH2 Pressurized hydrogen
hc;eo ;ei Efficiency of component [-, kgO2/kWhH2] phd Process heat demand
nc;Deo ;H2 Correction factor for H2 use [-] phsd Process heat demand SCF
COAL2 Annual limit of direct CO2 emissions [g/a] rad Reducing agent demand AF
COconv
2 Conventional CO2 emissions [g/a] rsd Reducing agent demand SCF
cpxfix
c Fixed investment costs [EUR]
cpxvar
c Size-specific investment costs [EUR/kW] Superscripts
CRF Capital recovery factor [1/a] dir Direct
int Interest rate of investment [%] fix Fixed
n Lifetime of the project [a] ind Indirect
nstk Lifetime electrolyzer stack [a] tot Total
omin
c Minimal part-load factor [-] var Variable

Water electrolysis systems can produce H2 from water and A prominent example for H2 as a decarbonization option for
renewable electricity and thus are often summarized under the energy-intensive industries is steel production (Fischedick et al.,
term Power-to-H2. These systems are already commercially avail- 2014a; Weigel et al., 2016; Otto et al., 2017; Thyssenkrupp, 2019).
able at large system sizes of several MW power rating (Siemens AG, Studies show that steel production with direct hydrogen reduction
2018). Additionally, further improvements are expected with re- is economically and environmentally the most promising process-
gard to technical and economic performance (Buttler and ing route (Fischedick et al., 2014a; Weigel et al., 2016). In practice,
Spliethoff, 2018). Due to these attractive features, Power-to-H2 Thyssenkrupp, a large steel producer, recently started injecting
systems are intensively investigated in literature (Parra et al., 2019) hydrogen as reducing agent into a blast furnace (Thyssenkrupp,
and in numerous demonstration projects around the world (Thema 2019). In addition to their use as reducing agents, fossil fuels are
et al., 2019). currently required for the provision of process heat. Here, the

2
€ben, N. Scho
F.T.C. Ro €ne, U. Bau et al. Journal of Cleaner Production 306 (2021) 127191

combustion of synthetic natural gas produced by methanation of To evaluate the environmental and economic potential of ret-
renewable H2 and carbon-rich off-gases, e.g., from steel production, rofitting copper production with a Power-to-H2 system, a detailed
represents a further decarbonization option in integrated steel model is necessary for the integrated supply system for energy and
plants (Rosenfeld et al., 2020). materials. Although the potential of Power-to-H2 systems has been
Hydrogen could take a similar role in the energy-intensive analyzed for many applications (Bertuccioli et al., 2014; Brunner
production of copper. Reducing copper production's environ- et al., 2015; Parra et al., 2017; van Leeuwen and Mulder, 2018;
mental impact is important as the global copper demand is ex- Nguyen et al., 2019; Morgenthaler et al., 2020), the integration with
pected to triple from 2010 to 2050 (Elshkaki et al., 2016). This strong co-utilization of O2 and H2 has not yet been investigated for copper
increase in global copper demand is mostly attributed to socio- production.
economic development (Elshkaki et al., 2016; Deetman et al., In this contribution, our main objective is to determine the
2018; Henckens and Worrell, 2020). A significant driver for techno-economic potential of decarbonizing copper production by
growth is due to the fact that copper is a key element for low- Power-to-H2 technology. The techno-economic potential is quan-
carbon technologies like renewable electricity and electric vehi- tified by the minimal abatement cost of direct CO2 emissions. To
cles and thus a central material for a sustainable energy transition also evaluate the overall climate change impact, we assess the total
(Deetman et al., 2018; Kuipers et al., 2018). The specific climate greenhouse gas emissions, including upstream processes, e.g.,
change impact per copper produced is expected to decrease due to electricity supply. For this purpose, an optimal system design is
a cleaner electricity supply. However, this decrease can only proposed for a representative copper production site. A model is
partially compensate for the additional emissions due to higher developed for the energy supply system for copper production
copper demand (Kuipers et al., 2018). Consequently, several studies integrating a Power-to-H2 system. Through numerical optimiza-
consistently forecast that copper production's climate change tion, this model is used to analyze the techno-economic potential of
impact will significantly increase if no additional measures are decarbonizing copper production. Specifically, we formulate and
taken (Van der Voet et al., 2018; Kuipers et al., 2018; Dong et al., solve a mixed-integer linear program (MILP) that minimizes the
2020). total annualized cost (TAC), varying GHG emissions reductions. As
Copper's mass-specific climate impact is three times higher than several parameters such as investment cost and efficiency of the
for iron (Van der Voet et al., 2018). Accordingly, several recent water electrolyzer in Power-to-H2 systems are expected to change
studies analyze the life-cycle of copper with a focus on the impact in the future, we conduct a broad sensitivity analysis to give an
of mines (Song et al., 2017; Segura-Salazar et al., 2019), primary outlook on the future economic and environmental impacts.
production technologies (Kulczycka et al., 2016), and the recycling The remainder of the paper is structured as follows: Section 2
of copper-rich waste (Zhang et al., 2021) for a circular economy gives a brief overview of copper production, Power-to-H2 tech-
(Soulier et al., 2018; Bonnin et al., 2019; Reuter et al., 2015, 2019). nology and states the assumptions and parameters used in our
The reported climate change impact of copper production ranges analysis. Section 3 presents the mixed-integer linear program. In
from 1.1 to 8.9 kg CO2-eq per kg of refined copper (Dong et al., Section 4, the results of the optimization and sensitivity analysis are
2020). The wide variation of environmental impacts is due to presented. Finally, conclusions are drawn in Section 5.
different production technologies, ore grades, and local energy
mixes (Dong et al., 2020), especially at the location of the mine 2. Material and data: Retrofitting copper production with
(Abadías Llamas et al., 2019). The climate change impact, in Power-to-H2
particular, is strongly linked to energy consumption (Kuipers et al.,
2018). Thus, a clean energy supply is important to reduce copper In this section, we give an overview of the reference copper
production's impact (Dong et al., 2020), particularly to mitigate the production plant and a Power-to-H2 system. Section 2.1 provides a
rising energy demand due to the general decline of copper ore brief overview of the conventional reference process. In Section 2.2,
grades at the mining sites. The energy-focused analysis by Moreno- we discuss the assumptions and parameters for the use of H2 and
Leiva et al. (2019) concluded that detailed assessment and optimal O2 in the process. Technical and economic parameters of the Power-
design of energy systems in copper production can enable eco- to-H2 system are presented in Section 2.3. Indirect GHG emissions
nomic and environmental benefits, with the integration of from upstream processes are examined in Section 2.4. Section 2.5
renewable energy supply offering a large potential. However, the details the assumptions with respect to commodity prices and
direct use of low-carbon electricity only reduces greenhouse gas energy demand profiles.
(GHG) emissions related to electricity consumption, while emis-
sions remain from process-related use of fossil fuels. To further 2.1. Conventional copper production
reduce the climate impact, production systems need to employ
clean fuels across the production chain of copper (Dong et al., The production of copper consists of several process steps to
2020). Power-to-H2 provides a potential pathway to also elimi- produce highly pure copper from copper concentrate. Our refer-
nate process-related emissions. So far, the integration of Power-to- ence production process is based on Schlesinger et al. (2011) and its
H2 both in the energy supply system and the production process retrofit with Power-to-H2 technology is shown in Fig. 1.
has not been analyzed, despite the importance of copper and the In the flash furnace, copper concentrate is smelted into a molten
promising application due to the co-utilization of oxygen. mass. The required energy originates from oxidizing iron and sul-
First tests on the use of H2 in copper production were recently fides in the concentrate. The oxidation takes place in an oxygen-
carried out, identifying promising applications in hydrogen enriched blast of 50e70% O2, thus requiring large quantities of
burners, anode refining furnaces, and melting units (Degel et al., O2. In the furnace bath, the molten mass separates into two layers
2019). H2 has also successfully been employed experimentally as with copper-rich matte (50e70 wt-% Cu) at the bottom and lighter
an alternative reducing agent (Goyal et al., 1982; Parra De Lazzari slag, containing mostly iron oxides, at the top. Matte is sent to the
and Capocchi, 1997; Marin and Utigard, 2010; Qu et al., 2020). As converters and slag to the slag-cleaning furnace. The slag-cleaning
copper production also requires oxygen, on-site integration of a furnace recovers matte droplets entrapped in the slag. To enhance
Power-to-H2 system becomes particularly attractive, as the elec- recovery and maintain a low-viscosity slag, the slag is kept at
trolysis by-product O2 can partially replace its current supply by operating temperature and a layer of coke acts as reducing agent.
energy-intensive cryogenic air separation. Recovered matte is added to the converters along with matte from
3
€ben, N. Scho
F.T.C. Ro €ne, U. Bau et al. Journal of Cleaner Production 306 (2021) 127191

Fig. 1. Flowsheet of copper production process retrofitted with Power-to-H2 technology to supply hydrogen as reducing agent in the anode furnace and slag-cleaning furnace, and
to supply high-temperature process heat. The electrolysis by-product oxygen is utilized in the flash furnace and converters. The structure of the conventional copper production
process is based on Schlesinger et al. (2011) and represented by solid lines, while the new material streams for the Power-to-H2 system are represented by dashed lines. The other
process heat (aggregated) represents the further process heat demands in the plant besides those explicitly mentioned.

flash smelting. In the converters, oxygen-enriched air (22e28 vol-% (Abadías Llamas et al., 2019), but also indicates rather high effi-
O2) is blown through the matte to oxidize iron and sulfides and ciency of the analyzed process. Thus, the considered plant rather
produce blister copper of 99 wt-% copper. The blister copper is sent corresponds to a challenging case for decarbonization of copper.
to the anode furnaces where remaining sulfur and oxygen are
removed by blowing first air and then a reducing agent through the 2.2. Copper production with Power-to-H2 and eO2
melt. Typically, natural gas is used as reducing agent, but other
fossil fuels are also in use (Ramachandran et al., 2003). The The considered integration of a Power-to-H2 system into copper
resulting melt has a purity of 99.5 wt-% copper and is cast into production is shown in Fig. 1. In the following, we discuss our as-
copper anodes. The anodes are electrolytically refined to copper sumptions for the use of H2 and O2 in copper production.
cathodes of 99.99 wt-% purity in the final process step. The corre-
sponding energy demands of an exemplary copper production 2.2.1. H2 as reducing agent in anode furnaces
plant are shown in Table 1 (Aurubis, 2019). Different copper production plants use different reducing agents
Copper production processes are very heterogeneous due to to deoxidize the molten blister copper (Ramachandran et al., 2003).
different geological, geographical, and technological conditions In our reference process, natural gas (CH4) is used as reducing agent
(Kuipers et al., 2018). However, the chosen reference process ap- (Schlesinger et al., 2011). While there is little public information
plies the most commonly used technologies for smelting and about the industrial practice of using H2 instead of CH4 in anode
converting, i.e., flash smelting and Peirce-Smith converting (Wang furnaces, experimental investigations have shown that H2 is an
et al., 2019). Further, virtually all molten copper is fire refined and effective reducing agent for refining molten copper (Goyal et al.,
electrolytically refined (Schlesinger et al., 2011), as performed in 1982; Parra De Lazzari and Capocchi, 1997; Marin and Utigard,
our reference process. Thus, our reference process reflects the most 2010).
common setup of copper production and should therefore be Due to a lack of public data, we assume that the reducing agents
representative to establish the techno-economic potential of Po- CH4 and H2 react completely in the melt. Note that utilization in
wer-to-H2. The carbon footprint of copper produced by the refer- industrial practice is typically not complete and will likely differ
ence plant corresponds to a comparatively low value in relation to between CH4 and H2 due to differing transport phenomena at the
other reported copper productions (Dong et al., 2020), which interface between gas and liquid metal. To estimate the theoretical
strongly depends on the specific impacts of mining operations potential of replacing CH4 by H2, we consider complete conversion
in the two reactions:

1 1 1 
Table 1 Cu2 O þ CH4 /2 Cu þ CO2 þ H2 O DhR1 ¼ 56:02 kJ mol
Annual energy demands of a conventional copper plant (Aurubis, 2019). 4 4 2
(R1)
Energy form Unit Annual demand

Natural gasa GWh 481


Metallurgical cokea GWh 48
Cu2 O þ H2 /2 Cu þ H2 O DhR2 ¼ 88:73 kJ=mol (R2)
Electricity (direct use) GWh 552
The enthalpies of reaction are stated in relation to 1 mol Cu2O
Electricity (oxygen production) GWh 127
Other energy sources GWh 16
and assume that the reducing agents enter at 25  C while the other
Annual production of copper kt 473 reactants are at 1200  C (Schlesinger et al., 2011). The stoichiometry
Specific energy demand MWh/t 2.6 shows that reducing 1 mol Cu2O requires 1 mol H2 instead of 14 mol
a
Energy amount stated as lower heating value (LHV). CH4. Since 1 mol H2 has an energy content of 241.8 kJ (LHV)
4
€ben, N. Scho
F.T.C. Ro €ne, U. Bau et al. Journal of Cleaner Production 306 (2021) 127191

compared to 200.6 kJ for 14 mol CH4, more energy is consumed Table 3


Direct emission factors of the processes when using fossil fuels.
when using H2. Thus, switching from CH4 to H2 increases the en-
ergy demand for reduction by 20.5%. The efficiency is calculated Process Factor Unit Value
relative to conventional operation using methane by Anode furnace reduction (AF) εdir g CO2/kWhCH4 197.4
AF;CH4
Slag-cleaning furnace reduction (SCF) εdir g CO2/kWhc 402.6
1LHV SCF;C
CH4
hAF;rad;H2 ¼ 4
¼ 0:830; (1) Process heat supply (PHS) εdir
PHS;CH4
g CO2/kWhCH4 197.4
LHVH2

where hAF;rad;H2 is the efficiency for using H2 to meet the reducing waste cooking oil has also been tested as reducing agent in an in-
agent demand in the anode furnace instead of CH4, LHVCH4 and dustrial slag-cleaning furnace (Li et al., 2018). The corresponding
LHVH2 are the lower heating values of CH4 and H2, respectively. The analysis showed that the bio-diesel is primarily pyrolyzed to C and
efficiency hAF;rad;CH4 for using CH4 as a reducing agent is per defi- H2, which then react with Fe3O4 in the slag. Just recently, Qu et al.
nition equal to 1. The hydrogen-based reduction (Reaction (R2)), (2020) have conducted experiments to analyze the reduction with
however, is 58% more exothermic than the methane-based reduc- H2 by injecting a gas stream of 70% H2 and 30% N2 into the slag.
tion (Reaction (R1)). The additional reaction enthalpy thus reduces Their results show good reduction of Fe3O4 and settling conditions.
the process heat demand (phd). Setting the additional enthalpy of Based on these applications and experiments, we assume that
reaction in relation to the LHV of the H2 input gives the correction the switch from coke (C) to H2 is feasible. Our calculations further
factor nAF;Dphd;H2 for the process heat demand as assume that coke consists entirely of carbon and that utilization of
both C and H2 is 100%. Note that utilization is again likely not
DhR1  DhR2 complete in industrial practice and will differ between reducing
nAF;Dphd;H2 ¼ ¼ 0:135: (2)
LHVH2 agents due to the differing transport phenomena, e.g., C particles
gasify and create a gas film around themselves (Warczok and
An overview of the efficiencies and correction factors is given in
Riveros, 2007).
Table 2.
To estimate the theoretical potential of H2 in slag reduction, the
According to Reaction (R1), 1 mol CO2 is released per 1 mol CH4.
reaction
Therefore, we calculate the direct CO2 emissions of the methane-
based operation with respect to the energy demand as 
1 1
  Fe3 O4 þ C/3 FeO þ CO2 DhR3 ¼ 57:8 kJ mol (R3)
MCO2 gCO2 2 2
εdir
AF;CH4 ¼ ¼ 197:4 ; (3)
LHVCH4 kWhCH4 is replaced by the reaction

where εdir
AF;CH4 are the CO2 emissions per kWh input of the anode Fe3 O4 þ H2 /3 FeO þ H2 O DhR4 ¼ 30:2 kJ=mol: (R4)
furnace AF when using CH4, MCO2 is the molar mass of CO2, and
The enthalpies of reaction are stated in relation to 1 mol Fe3O4
LHVCH4 is the molar lower heating value of CH4 (see Table 3 for the
and assume that the reducing agents enter at 25  C while the other
emission factors).
reactants are at 1230  C (Schlesinger et al., 2011). The stoichiome-
tries in Reactions (R3) and (R4) show that 1 mol H2 is necessary
2.2.2. H2 as reducing agent in slag-cleaning furnace instead of 12 mol C to reduce the same amount of Fe3 O4 , assuming
In the slag-cleaning furnace, slag is kept at the operating tem- complete reaction. The energy content of 12 mol C and 1 mol H2 are
perature by electric resistance heating while suspended matte 196.8 kJ and 241.8 kJ, respectively. Therefore, switching the
droplets finish settling (Schlesinger et al., 2011). To maintain a low- reducing agent from C to H2 increases the energy demand by 22.9%
viscosity slag, magnetite (Fe3O4) is reduced to FeO with coke as in terms of lower heating value. The corresponding efficiencies in
reducing agent because magnetite forms a solid in slag (Schlesinger relation to conventional operation using C are calculated as
et al., 2011).
As reducing agents, pulverized coal, diesel, and natural gas have 1LHV
C
been used (Sallee and Ushakov, 1999; Li et al., 2017). Bio-diesel from hSCF;rsd;H2 ¼ 2
¼ 0:814; (4)
LHVH2

where hSCF;rsd;H2 is the efficiency for using H2 as a reducing agent,


Table 2
Efficiencies and correction factors of different fuels to supply the demands LHVC and LHVH2 are the lower heating values of C and H2,
normalized to conventional operation: reducing agent anode furnace demand (rad), respectively. The efficiency hSCF;rsd;C for using C as a reducing agent
process heat demand (phd), reducing agent slag-cleaning furnace demand (rsd), is per definition equal to 1. However, hydrogen-based slag reduc-
process heat slag-cleaning furnace demand (phsd), and process electricity demand
tion (Reaction (R4)) is less endothermic than coke-based reduction
(eld). The energy inputs are natural gas (CH4), hydrogen (H2), coke (C), and elec-
tricity (el). (Reaction (R3)) and thus requires 48% less heat. Therefore, the
electrical energy input to heat the slag can be reduced accordingly.
Process Parameter Value
The process heat savings in relation to the H2 input are given by the
Anode Furnace (AF) hAF;rad;CH4 1 correction factor nSCF;Dphsd;H2 as
hAF;rad;H2 0.830
nAF;Dphd;H2 0.135
DhR3  DhR4
Slag-cleaning furnace (SCF) hSCF;rsd;C 1 nSCF;Dphsd;H2 ¼ ¼ 0:114: (5)
hSCF;rsd;H2 0.814
LHVH2
nSCF;Dphsd;H2 0.114 The efficiency for the resistive-electric heating is assumed to be
hSCF;phsd;el 1
100%, with the corresponding efficiency hSCF;phd;el being equal to 1
Process heat supply (PHS) hPHS;phd;CH4 1
(Table 2).
hPHS;phd;H2 1
Process electricity supply (PES) hPES;eld;el 1
According to Reaction (R3), 1 mol CO2 is released per mol C.
Therefore, we calculate the direct CO2 emissions of the coke-based
5
€ben, N. Scho
F.T.C. Ro €ne, U. Bau et al. Journal of Cleaner Production 306 (2021) 127191

operation with respect to the energy demand as Today, O2 required for copper production is usually obtained
  from external production in a dedicated cryogenic air separation
MCO2 gCO2 unit (ASU). In the investigated copper process, the annual elec-
εdir
SCF;C ¼ ¼ 402:6 ; (6)
LHVC kWhC tricity demand for the required O2 is reported as 127 GWh (Aurubis,
2019). With a typical ASU efficiency of 200e300 kWhel/tO2
where εdir
SCF;C are the CO2 emissions per kWh of the slag reduction (Banaszkiewicz et al., 2014), this energy demand corresponds to an
when using carbon C, MCO2 is the molar mass of CO2, and LHVC is annual O2 demand of around 423e635 ktO2 of the copper pro-
the heating value of C (see Table 3 for a summary of the emission duction. Replacing the annual fuel demand (Table 1) with H2 based
factors). In contrast, Reaction (R4) with H2 causes no direct CO2 on LHV requires around 16 kt of H2, corresponding to around 128 kt
emissions. The indirect greenhouse gas emissions from upstream of O2. Even if the lower estimate for the O2 demand is taken as
electricity supply are introduced in Section 2.4 along with other reference, the electrolysis by-product O2 can only fulfill around 30%
indirect upstream emissions. of the O2 demand in copper production. Thus, all by-product O2 can
be utilized and we assume that every ton of O2 produced by the
2.2.3. H2 as substitute for CH4 in process heat supply water electrolyzer substitutes a ton of O2 otherwise produced
We assume that hydrogen burners can be used to provide pro- externally, following the approach taken for O2 utilization in other
cess heat (Degel et al., 2019). Further, we assume that the efficiency Power-to-Gas studies (Parra et al., 2017; Guilera et al., 2018; Breyer
of heat supply is 100% for both CH4 and H2 with regard to their et al., 2015; Bailera et al., 2015; Kuparinen and Vakkilainen, 2017;
lower-heating values. The respective direct CO2 emissions for Morgenthaler et al., 2020; Rosenfeld et al., 2020).
methane-based heat supply are calculated as
  2.3. Power-to-H2 system
MCO2 gCO2
εdir
PHS;CH4 ¼ ¼ 197:4 ; (7)
LHVCH4 kWhCH4 To evaluate the techno-economic potential of a retrofit, we
analyze the Power-to-H2 system components in the following. The
assuming complete combustion. In Eqn. (7), εdir
PHS;CH4 are the CO2 discussed parameters are summarized in Table 4.
emissions of natural gas use (CH4) to fulfill the process heat supply Besides the Power-to-H2 system itself, retrofitting an existing
(PHS), MCO2 is the molar mass of CO2, and LHVCH4 is the lower copper production plant would incur additional costs, e.g., for
heating value of CH4. replacing burners or furnace equipment. However, reliably defining
such cost data is difficult as the data depends on site-specific factors
2.2.4. Utilization of electrolysis by-product O2 and non-disclosed data (Cusano et al., 2017). Therefore, we exclude
One advantage of retrofitting a Power-to-H2 system into copper costs that arise from the substitution of process equipment and
production is the utilization of the electrolysis by-product O2. point out that such costs must be added when applying the results
Typically, the by-product O2 is emitted to the atmosphere. When to an existing plant.
producing 1 kg of H2 from water electrolysis, around 8 kg of O2 are
produced based on the stoichiometric reaction of water 2.3.1. Water electrolysis system
electrolysis: The central component of a Power-to-H2 system is the water
electrolyzer with the three main technology options: alkaline
1 electrolysis (AEL), polymer electrolyte membrane electrolysis
H2 O / H2 þ O2 (R5)
2 (PEMEL), and solid oxide electrolysis (SOEL). Generally, AEL is

Table 4
Parameters for the Power-to-H2 system.

Component Parameter Unit Value Reference

Electrolyzer CAPEX cpxvar


EL EUR=kWel 1000 assumption in range of [1]
Electrolyzer stack reinvest cpxvar
stk EUR=kWel 271 45% of investment cost [2]
discounted by 10 years
Electrolyzer fix. OPEX opxfix %-Invest/a 3 [1]
EL
Electrolyzer lifetime system n a 20 assumption in range of [1]
Electrolyzer lifetime stack nstk a 10 assumption in range of [1]
Discount rate int % 5.2 based on WACC from [3]
Electrolyzer efficiency hEL;H2;el e 0.6 [1][2]
hEL;O2e;H2 kgO2/kWhH2 0.238 based on stoichiometric
reaction, see Section 2.3
Electrolyzer min. part-load omin;EL e 0.2 [1]
H2 storage CAPEX cpxvar
H2st EUR/kgH2 470 [4]
H2 storage fix. OPEX opxfix %-Invest/a 2.0 [4]
H2st
H2 compressor CAPEX cpxfix EUR 500,000 fit to model in [4]
H2C
cpxvar
H2C EUR/(kgH2/h) 2440 fit to model in [4]
H2 compressor fix. OPEX opxfix %-Invest/a 1.5 [5]
H2C
H2 compressor efficiency hH2C;pH2;el kWhel/kWhH2 0.051 based on [4]
hH2C;pH2;H2 e 1
Pressure reduction hVL;H2;pH2 e 1 assuming no H2 losses
[1]: (Buttler and Spliethoff, 2018) [2]: (Smolinka et al., 2018)
[3]: (KPMG, 2019) [4]: (Chardonnet et al., 2017)
[5]: (Parra et al., 2019)

6
€ben, N. Scho
F.T.C. Ro €ne, U. Bau et al. Journal of Cleaner Production 306 (2021) 127191

considered the most mature technology and has been available in for a warm-start and 1e2 h for a cold-start (Buttler and Spliethoff,
large system sizes for a long time. PEMEL is often considered for 2018). In contrast, load changes at nominal operating temperature
Power-to-Gas plants, as rapid load changes are possible, allowing are in the order of seconds (Buttler and Spliethoff, 2018). Since long
intermittent operation and provision of ancillary grid services shut-downs do not occur in our system due to the continuously
(Kopp et al., 2017). SOEL is currently the least mature technology operating copper production, we do not consider start-up times in
only available at high costs and small system sizes. However, SOEL the present study.
technology offers very high efficiencies with a high potential for
heat integration (Buttler et al., 2015). 2.3.2. Pressurized H2 storage
We base the parameters of our water electrolysis system on an To gain more flexibility in operating the water electrolyzer, a H2
AEL system because AEL is the most mature technology and storage tank is added to our model. We consider a steel storage tank
currently has the lowest investment cost (Buttler and Spliethoff, to store pressurized H2 at 200 bar, as such tanks are widely used in
2018; Parra et al., 2019; Proost, 2019). Current investment costs industry (Chardonnet et al., 2017). The investment costs cpxvar
H2st are
for AEL systems are stated in the range of 600-1500 EUR/kWel assumed to be 470 EUR/kgH2 with fixed operating costs of 2%, the
(Buttler and Spliethoff, 2018; Smolinka et al., 2018). We choose current parameters for stationary storage systems (Chardonnet
intermediate investment cost, also called capital expenditures et al., 2017).
(cpx), of the electrolyzer system (EL) cpxvarEL of 1000 EUR/kWel. It
must be noted that significant scaling effects are expected when 2.3.3. H2 compressor for pressurized storage
considering large water electrolysis systems (Bo € hm et al., 2020; To store H2 at 200 bar, a compressor is required to increase the
IRENA, 2020). Such investment cost reductions in line with ex- pressure from the 30 bar operating pressure of the electrolyzer. Due
pected future developments and scaling effects will be evaluated in to the wide range of H2 compressor configurations, investment
a detailed sensitivity analysis (cf. Section 4.3.2). costs vary widely in literature (van Leeuwen and Mulder, 2018). We
Equally important is the lifetime of the system. We differentiate use a variable cost function for H2 compressors which depends on
between the lifetime of the electrolyzer stack and the other com- the compressor capacity m_ H2, compression ratio, and output pres-
ponents. For the stack, we assume a lifetime nstk of 10 years which sure (Chardonnet et al., 2017). To keep the compressor cost function
is an average value for AEL stacks (Buttler and Spliethoff, 2018). For in our model linear, we fit a linear cost function to the model of
the other components of the Power-to-H2 system, we assume a Chardonnet et al. (2017) in the range from 0 to 40 MWH2
lifetime n of 20 years (Smolinka et al., 2018). Since the lifetime of throughput and a compression ratio of H2 from 30 to 200 bar. The
the electrolyzer stack is shorter than the lifetime of the system, we resulting investment cost for the compressor can be calculated as
include reinvestment cost for a new stack after 10 years. The cost-
share of the stack is assumed to be 45% of the electrolyzer system EUR,h
cpxtot
H2C ¼ 500; 000 EUR þ 2440 m_ H2 ; (9)
cost (Smolinka et al., 2018). The future reinvestment cost is dis- kgH2
counted by 10 years with an interest rate int of 5.2%, the weighted
average cost of capital (WACC) for energy projects in 2018/19 where cpxtot
H2C is the capital expenditure of the compressor system
(KPMG, 2019). The discounted reinvestment cost for the stack in EUR and m_ H2 is the maximum mass flow of H2 in kg/h. The
cpxvar is 271 EUR/kWel. corresponding fixed operating costs opxfix
stk H2C are estimated as 1.5% of
The operating costs include fixed and variable operating costs. the initial investment costs (Parra et al., 2019).
Fixed operating costs, including maintenance, are often defined as For the operation of the compressor, we assume an electricity
an annual share of the initial investment costs. We assume fixed demand of 1.7 kWhel/kgH2 based on a two-stage compression from
operational expenditures (opx) of 3%/a in our model (Buttler and 30 to 200 bar (Chardonnet et al., 2017), which is 0.051 kWhel/
Spliethoff, 2018). In contrast, the variable operating costs depend kWhH2 based on the LHV of H2 input (see hH2C;pH2;el in Table 4). We
on the energy consumption and thereby on electrolyzer system assume that there are no H2 losses during compression and
efficiency. We choose a typical value for AEL systems with a system decompression (see hH2C;pH2;H2 and hVL;H2;pH2 in Table 4).
efficiency hEL;H2;el of 60 %-LHV excluding the H2 compressor
(Buttler and Spliethoff, 2018; Smolinka et al., 2018). The production 2.4. Indirect GHG emissions
rate of the by-product O2 is calculated from the stoichiometric re-
action (R5) as Fossil fuels, electricity, and system components are associated
with indirect GHG emissions from upstream processes, e.g.,
   
kgO2 E_ H2;EL kgO2 extraction, transport, and production, typically summarized in CO2
m_ O2;EL ¼ 0:5 MO2 ¼ 0:238 E_ ; (8) equivalents.
h LHVH2 kWhH2 H2;EL
For the indirect GHG emissions, we consider only the upstream
where m_ O2;EL is the produced material flow of O2 by the electrolyzer emissions of commodities since the impact of the Power-to-H2
components is typically small compared to the impact of electricity
(EL), E_ H2;EL is the hydrogen output in terms of lower heating value,
supply (Koj et al., 2017; Zhang et al., 2017; Bareib et al., 2019). The
LHVH2 is the molar energy content of hydrogen, and MO2 is the
total GHG emissions then consist of the direct CO2 emissions of the
molar mass of O2. We assume that the products are supplied at an
copper production and the indirect GHG emissions. The assump-
operating pressure of 30 bar, a typical value for commercially
tions for the upstream GHG emissions of electricity supply and
available systems (Smolinka et al., 2018; Buttler and Spliethoff,
commodities are presented in Table 5.
2018).
Since the electricity supply is upstream of our copper produc-
A drawback discussed for AEL systems is the limited part-load
tion, we refer to them as indirect GHG emissions. However, the
operation. To account for that limitation in our model, we choose
GHG emissions of electricity supply itself are typically separated
a typical minimal part-load omin;EL of 20% (Buttler and Spliethoff,
into direct and indirect emissions. For the direct GHG emissions of
2018; Smolinka et al., 2018). It should be noted that for applica-
tions requiring fast response times and frequent system standby, electricity supply, we use the emission factor εind;dir
el
of 401 g CO2-
e.g., ancillary grid services, another limitation of AEL systems must eq/kWhel for grid electricity in Germany in 2019 (Icha and Kuhs,
ind;ind
be considered, namely the start-up times in the order of 1e5 min 2020). For the indirect GHG emissions, we add a factor εel of
7
€ben, N. Scho
F.T.C. Ro €ne, U. Bau et al. Journal of Cleaner Production 306 (2021) 127191

Table 5
Upstream GHG emission factors of the commodities.

Commodity Factor Unit Value Reference

Electricity mix directa εind;dir g CO2-eq/kWhel 401.0 Icha and Kuhs (2020)
el
Electricity mix indirect εind;ind g CO2-eq/kWhel 30.0 Wietschel et al. (2019)
el
Natural gas εind g CO2-eq/kWhCH4 46.1 Moro and Lonza (2018)
CH4
Metallurgical cokeb εind g CO2-eq/kWhc 98.9 own calculationb
C
Oxygen from ASUc εind g CO2-eq/kgO2 80.2 own calculationc
O2
a
Average direct emission factor of grid electricity in Germany in 2019.
b
Based on upstream coke plant emissions (Pardo et al., 2012) and LHV of coke (Suopaja €rvi et al., 2018).
c
ASU efficiency of 200 kWhel/tO2(Banaszkiewicz et al., 2014) and emission factor of electricity mix.

Table 6
Commodity prices.

Commodity Parameter Unit Value Reference

Electricitya pel;t EUR/MWh 34.2 Bundesnetzagentur SMARD.de (2020)


Natural gas pCH4;t EUR/MWh 28.6 Bundesnetzagentur and Bundeskartellamt (2019)
Coke pC;t EUR/MWh 35.2 €rvi et al., 2018)
(Steelonthenet, 2019; Suopaja
Oxygen pO2;t EUR/tO2 72 Pardo and Moya (2013)
a
The shown electricity price is the average price of the historic price time series used (year 2019, German day-ahead market).

30 g CO2-eq/kWhel for electricity in Germany (Wietschel et al., This value fits into price ranges for O2 in Power-to-Gas studies
2019). ranging from 50 EUR/tO2 (Kuparinen and Vakkilainen, 2017;
Rosenfeld et al., 2020) to 150 EUR/tO2 (Guilera et al., 2018), with
2.5. Commodity prices and time of energy use several studies also lying in the middle of this range (Parra et al.,
2017; Breyer et al., 2015; Pardo and Moya, 2013; Graf et al., 2014).
The consumption of commodities generates operating costs. In For the electricity prices, we use the 2019 historic day-ahead
this section, we define the commodity prices and the demand spot market prices in Germany with hourly resolution
profile of our reference copper production. (Bundesnetzagentur d SMARD.de, 2020). The average electricity
price pel;t was 34.2 EUR/MWh. We use wholesale electricity prices
because Power-to-Gas systems in combination with energy-
2.5.1. Commodity prices
intensive industry are largely exempted from electricity tax and
The commodity prices used in our model are based on typical
levies in Germany (Chardonnet et al., 2017). Following Felgenhauer
prices for industrial consumers in Germany in 2019 (cf. Table 6).
and Hamacher (2015), we assume that the cost of water is
We assume a natural gas price of 28:6 EUR=MWh
negligible.
(Bundesnetzagentur and Bundeskartellamt, 2019) and a coke price
of 293 EUR=t, based on the average European coke price in 2019
(Steelonthenet, 2019). Assuming an average heating value of 30 MJ= 2.5.2. Energy and material demands
kg (Suopaja €rvi et al., 2018), the price for coke (C) amounts to 35.2 In this section, we present the assumed energy and material
EUR/MWh. demands of our reference copper production plant throughout the
To analyze the cost savings from using the electrolysis by- year (Table 7).
product O2, we define a price for substituted O2. We assume an The time to process one batch of 270 t of copper in an anode
oxygen price pO2;t of 72 EUR/t for 2020, based on a study of iron and furnace is 9 h (Schlesinger et al., 2011). The reduction with natural
steel production in Europe and an estimated annual growth rate of gas is one sub-task which lasts 3 h while consuming 120 m3 =h
1%/a compared to the reference year 2010 (Pardo and Moya, 2013). (Schlesinger et al., 2011). Since we consider an annual production of
473; 040 t copper (cf. Table 1) (Aurubis, 2019), the overall produc-
Table 7
tion consists of around 1; 752 batches. These batches are refined by
Parameters for the energy and material demands: process heat (phd), reducing two anode furnaces (Schlesinger et al., 2011). We assume that the
agent anode furnace (rad), reducing agent slag-cleaning furnace (rsd), process heat two anode furnaces operate in an alternating manner (cf. Fig. 2).
slag-cleaning furnace (phsd), process electricity demand (eld), and oxygen (O2d). The alternating operation causes an intermittent natural gas de-
The parameters are calculated based on the annual demands from Aurubis (2019)
and operating parameters from Schlesinger et al. (2011), as explained in Section 2.5. mand of D_ of 12.0 MW during the reduction.
rad;t
The slag-cleaning furnace continuously treats slag coming from
Demand Parameter Unit Value Profile
the smelter (Schlesinger et al., 2011). Thus, we assume that the slag-
Process heat D_ phd;t MW 47.7 Constant cleaning furnace has a steady-state demand for coke. Based on the
Reducing agent AF D_rad;t
MW 12.0 Intermittenta
annual coke demand (Table 1), the hourly coke demand D_ is rsd;t
Reducing agent SCF D_ rsd;t MW 5.5 Constant
5.5 MWh. Additionally, the slag-cleaning furnace has a power de-
Electricity for process heat SCF D_phsd;t
MW 3.0 Constant
mand of 3 MW to heat the slag.
Other electricity demand D_ eld;t MW 60.0 Constant
Due to a lack of public data, we assume a constant aggregated
Oxygen D_O2d;t
t/h 72.6 Constant
process heat demand, which is calculated from the annual natural
a
The intermittent demand profile is shown in Fig. 2. gas demand (cf. Table 1) minus the gas used as reducing agent. The
resulting annual process heat demand is 417.9 GWh, which
8
€ben, N. Scho
F.T.C. Ro €ne, U. Bau et al. Journal of Cleaner Production 306 (2021) 127191

Fig. 2. Intermittent natural gas demand profile as reducing agent in the anode furnace.

corresponds to a steady-state demand of process heat D_ phd;t of is written as


47.7 MW. X   X X
Oxygen is primarily used for the oxygen enrichment of the blast min CPXc , CRF þ OPXcfix þ var
OPXs;t ,Dt; (10)
in the flash furnace. Since the flash furnace runs continuously and c2C t2T s2S
due to the lack of detailed operating data, we assume a steady-state
oxygen demand. We calculate the O2 demand from the electricity where CPXc is the capital expenditure for the investment cost of
consumption of a modern air separation unit (Table 1) with an ef- component c with C as the set of all components, CRF is the capital
ficiency of 200 kWh/t of O2 (Banaszkiewicz et al., 2014). The cor- recovery factor which converts initial investment cost to annual
cost based on project lifetime n and discount rate int. Based on n
responding O2 demand D_ is 72.6 t/h.
O2d;t
and int, we calculate the capital recovery factor as
The annual electricity demand of our reference copper pro-
duction plant (Table 1) corresponds to an average steady-state
ð1 þ intÞn ,int
demand of 63 MW. Since we analyze the slag-cleaning furnace CRF ¼ : (11)
separately, we have to subtract the 3 MW electricity demand for ð1 þ intÞn  1
heating the slag. Thus, all other electricity consumers correspond to In the objective (10), OPXcfix are the fixed operating expendi-
an electricity demand D_ of 60 MW.
eld;t tures, i.e., maintenance cost. OPXs;tvar are the variable cost which

arise from consuming commodity s in time step t. Set S contains all


3. Method and problem description commodity sources and set T contains the 8760 hourly time steps
of the year. The operating costs are multiplied by Dt which is the
To evaluate the benefits of retrofitting a Power-to-H2 into cop- uniform duration of each time step and set to 1 h.
per production, we use mathematical optimization. The capital expenditure CPX or investment cost for a component
is calculated as
3.1. Optimization formulation
CPXc ¼ yc cpxfix N var
c þ Pc cpxc c c2C ; (12)
We use mixed-integer linear programming (MILP) to optimize
the design and operation of the Power-to-H2 system. MILP is an where the binary variable yc takes the value 1 if a component is
optimization method that minimizes a given linear objective installed, cpxfix N
c is the fixed investment cost, Pc is the component
function subject to a set of linear constraints (Nemhauser and size, and cpxvar is the size-specific investment cost.
c
Wolsey, 1988). The optimization variables include both discrete
The annual fixed operating expenditure OPXcfix of a component c
decisions, e.g., whether to install a certain component, and
is determined by
continuous decisions, e.g., the size of a component or the level of
operation at a certain point of time. Design problems where both
investment and the operating costs shall be optimized can often be OPXcfix ¼ CPXc ,opxfix
c c c2C ; (13)
stated as (piece-wise) linear models and solved efficiently to global
optimality by MILP (Grossmann, 1985; Voll et al., 2013). where CPXc is the initial investment cost of component c, and opxfix
c
is a factor for the annual operation and maintenance cost inde-
3.1.1. Objective: total annualized cost (TAC) pendent of operation.
var depends
In contrast, the variable operating expenditure OPXs;t
The objective of our optimization is to minimize the total
var arises when commodities are
on the actual operation. OPXs;t
annualized cost (TAC) of the retrofit. The TAC includes investment
and operating costs over a project lifetime of 20 years. The objective consumed and is calculated as
9
€ben, N. Scho
F.T.C. Ro €ne, U. Bau et al. Journal of Cleaner Production 306 (2021) 127191

var
OPXs;t ¼ ps;t S_s;t c s2S ; t2T ; (14) E_ Dphd;AF;t ¼ nAF;Dphd;H2 E_ H2;AF;t c t2T ; (20)

where ps;t is the price of commodity s in time step t, S_s;t is the E_ Dphsd;SCF;t ¼ nSCF;Dphsd;H2 E_ H2;SCF;t c t2T ; (21)
amount of commodity consumed.
where E_ Dphd;AF;t and E_ Dphsd;SCF;t are the amounts of process heat
reduction, and E_ H2;AF;t and E_ H2;SCF;t are the corresponding inputs of
3.1.2. Constraints
H2 as reducing agent based on LHV.
The copper production process has to be supplied by the
To enable a flexible operation of the water electrolyzer, a H2
necessary energy and materials of reducing agents, oxygen, process
storage tank (H2st) is part of the Power-to-H2 system. The state of
heat, and electricity (cf. Table 1). To this end, we formulate a
charge (SOC) of the H2 storage is obtained from
system-wide balance equation for each type of energy and material
in each time step as
SOCtþ1 ¼ SOCt þ E_ pH2;H2st;t ,Dt c t2T \f8760g; (22)
X
E_ e;c;t þ S_s;t ¼ D_ d;t c e2S ∪I ∪D ; s2S ; d2D ; t2T ;
c2C 1 ¼ xstoragein;t þ xstorageout;t c t2T ; (23)
(15)
SOC8760 ¼ SOC0 ; (24)
where E_ e;c;t describes the energy/material flow e to/from compo-
nent c in time step t, S_s;t are the flows from source s supplying the where SOCtþ1 is the state of charge based on the state of charge
system with a commodity, D_ d;t represents the energy or material SOCt from the previous time step and the in- or output E_ pH2;H2st;t of
demand d. pressurized H2 in time step t. The storage can only charge or
The demand set D contains demands that have to be fulfilled in discharge at a time (Eqn. (23)). The initial state of charge is a var-
every time step t to not hinder copper production. The six consid- iable and Equation (24) enforces that the final state of charge (end
ered demands are: process heat, oxygen, reducing agent for the of year) is equal to the initial state of charge (beginning of year).
anode furnace, reducing agent for the slag-cleaning furnace, elec-
tricity for process heat in the slag-cleaning furnace, and other
electricity demands. The set S contains four sources available in 3.1.3. GHG emissions
the model: electricity, natural gas, metallurgical coke, and oxygen. Direct CO2 emissions are emitted when fossil fuel is consumed
Components transform materials and energy from the sources by a component. The arising CO2 output contributes to the direct
to fulfill the demands of the system. The installation of components CO2 emissions of the overall system. The hourly system-wide direct
is described as CO2 emissions COdir
2t are calculated by summing over the different
components and sources, i.e.,
yc Pcmin  PcN  yc Pcmax c c2C ; (16)
X X
COdir εdir _
2t ¼ s;c Es;c;t c t2T ; (25)
where yc is the binary decision to install component c with Pcmin and s 2S c 2 C
Pcmax as the minimal and maximal installation size. Pcmin is set to
0 MW and Pcmax is chosen sufficiently high with 1000 MW to not and the annual direct CO2 emissions COdir
2 are calculated as
influence the optimal solution. X
O of a component in each time
To determine the operation rate Pc;t COdir
2 ¼ COdir
2t (26)
t 2T
step, we formulate operating constraints as

xc;t PcN omin O


 Pc;t  xc;t PcN c c2C ; t2T ; (17) COdir AL
2  CO2 ; (27)
c

where xc;t is the binary decision to operate the component in time where E_ s;c;t is the material consumption of component c in time
step t, and omin is the minimal part-load factor of component c. As step t, εc;s is the emission factor when component c consumes s, e.g.,
c
Eqn. (17) contains a bi-linear product of a binary variable and a natural gas. The parameter COAL
2 in Eqn. (27) limits the annual
continuous variable, i.e., xc;t and PcN , we reformulate the bi-linear direct CO2 emissions COdir AL
2 . We vary the limit CO2 to analyze
term based on Glover (1975). different degrees of decarbonization. The optimizer chooses how
The operation of a component is connected to the energy and and when to reduce CO2 to determine the most cost-efficient
material flows by design and operation for a given annual limit COAL
2 .
The indirect GHG emissions are calculated as
E_ eo ;c;t ¼ Pc;t
O
; c c2C ; t2T ; (18)
X X
GHGind ¼ εind
s S_s;t ; (28)
t 2T s 2S
E_ eo ;c;t ¼ hc;eo ;ei E_ ei ;c;t c c2C ; t2T ; (19)
where εind
s is the indirect emission factor corresponding to the
where E_ eo ;c;t is the main material/energy output and hc;eo ;ei denotes
commodity consumption S_s;t in time step t.
the efficiency of transforming ei into eo . Note that a component may
Finally, the total GHG emissions are computed by
have multiple in- and outputs, e.g., the water electrolyzer has one
input (electricity) and two outputs (H2 and O2).
GHGtot ¼ COdir
2 þ GHG
ind
: (29)
Based on the input of H2 as reducing agent, the correction fac-
tors nc;Deo ;H2 determine the reduction in process heat demand as
10
€ben, N. Scho
F.T.C. Ro €ne, U. Bau et al. Journal of Cleaner Production 306 (2021) 127191

3.2. Successive decarbonization build on each other.


As a key figure of each scenario, we define the direct CO2
Our reference scenario is defined as conventional production abatement cost (AC) as
and energy supply based on fossil fuels as well as electricity from
the grid. We then investigate different degrees of decarbonization DTAC
AC ¼ ; (31)
by imposing limits on the allowed direct CO2 emissions (COAL 2 )
DCOdir
2
corresponding to emission reductions from 5% to 100% in in-
crements of 5% (see Fig. 3). The CO2 limit is defined as where DTAC is the increase of total annualized cost (TAC) of the
retrofitted production in comparison to the reference case, and
DCOdir
2 is the annual reduction of direct CO2 emissions compared to
2 ¼ a,CO2
COAL conv
; (30) the reference case. Analogously, we calculate the total GHG
abatement cost based on the total GHG emissions.
where a is the share of CO2 emissions allowed in relation to the
emissions from conventional operation COconv 2 which correspond to
an a of 1 (100%). The CO2 limitation makes the retrofit with the 4. Results and discussion
Power-to-H2 system necessary as it allows to replace coke and/or
natural gas by H2 (Section 2.2). To chart a successive design path for In this section, we evaluate the potential of Power-to-H2 to
decarbonization, the results of the previous optimization are used decarbonize copper production for various degrees of decarbon-
ization. First, we analyze the economic potential and system design
as lower bound of the design decisions yc and Pcmin (Eqn. (16)) in the
in Section 4.1. Second, we assess the environmental impact with
successive optimization with tighter CO2 limit. Thereby, the designs
regard to GHG emissions in Section 4.2. Last, in Section 4.3, the
sensitivity of the results towards several techno-economic pa-
rameters is evaluated and set into relation to expected future
technological advancements.

4.1. Decarbonizing copper production

We analyze the optimal results for decarbonizing a copper


production with a Power-to-H2 system under today's technical
parameters and prices. To rank the economic efficiency of decar-
bonizing the sub-tasks of our copper production, we analyze the
results of optimized systems under increasingly strict CO2 limits.
With increasing CO2 emission reduction from 5% to 100%, the
resulting direct CO2 abatement cost for the retrofit increases from
112 to 201 EUR/t CO2, see Fig. 4(a).
These abatement costs vastly exceed current European Union
Allowances (EUA) that had an average price of 24.6 EUR/t CO2 in
2019 on the European Energy Exchange (EEX, 2019). However, CO2
prices are expected to increase in the future as governments take
more action on climate change. While the expected future CO2
prices vary strongly between different scenarios and models
(Rogelj et al., 2018), a European scenario projects up to 310 EUR/t
CO2 in 2050 (EU Commission, 2011). Such a price is consistent with
studies considering the 1.5  C climate change mitigation goals
(Rogelj et al., 2018).
Meanwhile, costs for CO2 higher than the current EUA prices are
already in place in some countries today, e.g., in Sweden with 110
EUR/t CO2 (Government Offices of Sweden, 2020) or in Switzerland
with 90 EUR/t CO2 in 2018 (Bundesamt für Umwelt, 2019).
Breakdowns of the direct CO2 abatement costs are shown in
Fig. 4(a). As expected, the electricity cost for the Power-to-H2 sys-
tem is the largest cost factor of the retrofit with 70% for the com-
plete elimination of direct CO2 emissions. Therefore, we analyze the
impact of different electricity prices in a sensitivity analysis.
Further, we find that the utilization of the O2 by-product has a
substantial positive economic impact on the direct CO2 abatement
cost, reducing costs by 81 EUR/t CO2 in the 100% reduction case.
Thus, the utilization of the water electrolysis by-product O2 reduces
the direct CO2 abatement cost by 29% compared to a system that
emits the O2. The results make a strong case for retrofitting Power-
Fig. 3. Solution procedure of the successive decarbonization of copper production. The to-H2 on-site at a plant with large O2 demand. Furthermore, it can
rhomboid represents input data. Rectangles without a heading describe actions. The be seen that the investment cost of the H2 storage system is
rectangle with heading represents the design optimization. The rhombi mark branches
in the procedure with the corresponding condition written at the outgoing arrow. The
negligible. The electricity cost of the compressor is small as well,
rectangles with rounded corners represent the collected results: the optimal system compared to the water electrolyzer with only 0.15% of the elec-
designs. tricity cost of the supplied H2.
11
€ben, N. Scho
F.T.C. Ro €ne, U. Bau et al. Journal of Cleaner Production 306 (2021) 127191

Fig. 4. Results of the decarbonization of copper production by retrofitting a Power-to-H2 system. (a) The resulting direct CO2 abatement cost for increasingly tighter direct CO2
emission limits ranging from the conventional process (0% reduction) to the fully-decarbonized process (100% reduction). (b) The degree of substitution with H2. The substitution is
defined as the share of H2 in the total energy demand.

Fig. 5. Electrolyzer peak capacity in terms of hydrogen output (LHV). The left axis
shows the total peak capacity and the right axis shows the peak capacity in relation to
the annual copper production volume.

Table 8
Power-to-H2 system design in the fully-decarbonized case.

Component Unit Value

Water electrolyzer MWH2 63.1


H2 compressor MWH2 8.7
H2 storage capacity MWhH2 77.8
H2 initial filling level (t ¼ 0) MWhH2 24.5
Fig. 6. Influence of the electricity prices (a) on the electrolyzer operation (b) and state
of charge of the H2 storage over the first week of the year (c). The solid ellipse (c) marks
the effect of the intermittent demand of the anode furnace reducing agent. The dashed
Fig. 4(b) shows the degree of H2 usage which is defined as the ellipses mark an example of discharging the storage in hours of high electricity prices
share of H2 in the annual energy demand of the respective process. followed by slow recharging of the storage. The dashed line in the electrolyzer oper-
When imposing a CO2 limit, the first process to switch to H2 is the ation (b) marks the minimal active part-load.

slag-reduction task because of the large direct CO2 emissions from


the reduction with coke. The use of H2 in the slag-reduction quickly
process heat (Fig. 4(b)). The stagnation of the H2 use for process
reaches 99% and only in the fully-decarbonized case reaches 100%.
heat at around 85e90% reduction is caused by a change in system
The 99% threshold occurs because the water electrolyzer is shut
design: When the copper refining in the anode furnace starts to use
down during the highest electricity price peaks in the year and coke
H2, a larger H2 storage is introduced to balance the uneven demand
is used instead.
profile of the copper refining (Fig. 2). Therefore, it becomes cheaper
When almost all coke is substituted, H2 is also used to supply
12
€ben, N. Scho
F.T.C. Ro €ne, U. Bau et al. Journal of Cleaner Production 306 (2021) 127191

to use H2 for copper refining than for increasing H2 usage in process


heat supply during times of high electricity prices.
Fig. 5 shows the optimal electrolyzer capacity for varying degree
of decarbonization. To enable comparisons with other electrolysis
applications, we set the electrolyzer capacity in relation to the
annual copper production volume (Fig. 5, right axis).
The resulting design for the Power-to-H2 system in the fully-
decarbonized case shows that the optimal storage is small
compared to the electrolyzer output with a capacity corresponding
to 1.23 h of operation, see Table 8. This design decision is attributed
to the relatively constant energy demand of the process, the addi-
tional costs for the storage system, and most importantly, the
necessary over-sizing of the electrolyzer for intermittent produc-
tion of stock H2 for later use.
Fig. 6 shows the state of charge (SOC) for the H2 storage during
the first week in the year. The storage is slowly charged during Fig. 8. The additional total annualized cost per ton of copper.
times of low electricity prices and discharged within few hours of
high electricity prices (e.g. hours 64 to 68). The pattern of rapid
discharging followed by slow charging repeats itself throughout the
year. The electrolyzer operation (Fig. 6(b)) shows the economic
advantage of adding H2 storage: the system can react to variable
electricity prices. Beyond this pattern, the state of charge also re-
veals a shorter charging and discharging pattern which is caused by
the intermittent demand for reducing agent in the copper refining
(Fig. 6(c)). In our model, a potential energy demand of the elec-
trolyzer during system standby is neglected, as we observe that
standby occurs only in 69 h of the year and never lasts longer than
1 h. The additional annual electricity demand for hot-standby mode
would amount to just 0.008e0.04% of the annual electrolyzer
electricity demand, assuming typical values for standby energy
demand of 1e5% of nominal load (Smolinka et al., 2018). We want
to note, however, that in systems with more frequent or extensive
standby operation, electrolyzer technology allowing for fast startup
times and thus complete shutdowns avoiding standby losses might Fig. 9. Direct CO2 plus indirect GHG emissions of copper production with Power-to-H2
retrofit. The GHG emissions of electricity include only the direct emissions of grid
be advantageous.
electricity in Germany in 2019 (Icha and Kuhs, 2020). The other GHG emissions further
An analysis of the H2 cost structure (Fig. 7) shows that initial include upstream emissions from electricity supply and other commodities, i.e., natural
investment and electricity costs dominate. The resulting cost for gas and coke. The left axis shows the annual GHG emissions and the right axis shows
the H2 is around 2.65 EUR/kgH2 which is similar to the study by the annual GHG emissions in relation to the annual copper production volume.
Chardonnet et al. (2017), which finds H2 costs of 2.4 EUR/kg for a
large-scale water electrolysis system supplying an oil refinery. Note
to decarbonization by Power-to-H2. While the cost per ton of
that the system in Chardonnet et al. (2017) considers an oversized
copper may appear small compared to typical copper market pri-
water electrolyzer which enables even lower electricity prices.
ces, the actual revenue realized for smelting and refining (around
However, the main reason for the relatively low H2 price are the low
435 EUR/t Cu in February 2020 (Aurubis, 2020)) is significantly
electricity prices both in our investigation and their study. The
smaller than the copper market price. This revenue is calculated
electricity prices are based on direct procurement on the wholesale
from treatment and refining charges as well as premiums for the
day-ahead market and tax exemption (Section 2.5.1).
sale of copper cathodes (Aurubis, 2020), as these make up the main
Besides the cost of hydrogen, the additional cost related to the
revenue of a smelter (Díaz-Borrego et al., 2019). Therefore, the
final product copper are most relevant for the economic operation
shown additional cost corresponds to up to 11% of the main reve-
of a plant. Fig. 8 shows the increase in copper production cost due
nue when the production is fully decarbonized, thus constituting a
relevant cost factor.

4.2. Total GHG emissions

Fig. 9 shows the impact of direct CO2 emission reductions on the


GHG emissions of our copper production retrofitted by Power-to-
H2.
A major part of the total GHG emissions originates from up-
stream processes even in the conventional system (74%), most
notably from electricity generation. For grid electricity, we differ-
entiate between direct emissions from electricity supply with 401 g
CO2-eq/kWhel (Icha and Kuhs, 2020) and indirect emissions from
processes upstream of electricity supply with 30 g CO2-eq/kWhel
(Wietschel et al., 2019). The remaining indirect GHG emissions in
Fig. 7. Cost structure of the produced H2 in the fully-decarbonized scenario. our analysis are from the upstream processes for natural gas and
13
€ben, N. Scho
F.T.C. Ro €ne, U. Bau et al. Journal of Cleaner Production 306 (2021) 127191

Fig. 10. Impact of the emission factor of electricity on the indirect (a) and total GHG emissions (b). The dashed line shows the point of intersection between the conventional
production and the fully-decarbonized production. The left axes show the annual GHG emissions and the right axes show the annual GHG emissions in relation to the annual copper
production volume.

coke supply. The specific emissions of the conventional operation of 1% and 0.7% Cu, to account for 2.6 and 2.7 t CO2-eq/t Cu,
(0.92 t CO2-eq/t Cu) in our model agree well with the reported respectively. These climate impacts are in the lower range of values
direct and indirect CO2 emissions (0.96 t CO2-eq/t Cu) of the of copper production reported by Dong et al. (2020). While the
reference plant (Aurubis, 2019). These climate impacts are small analyzed plant thus already represents a modern and very efficient
compared to the average climate change impacts reported for plant in Europe, it remains important to further reduce emissions in
copper production (Dong et al., 2020), since only the production view of the growing global copper demand.
plant is considered and not the mining operations for copper ore For the Power-to-H2 system, GHG emissions from electricity
concentrate which make up a large part of the overall climate supply exceed the avoided emissions from natural gas and coke of a
change impact, especially at low ore grades (Norgate and Haque, decarbonized system (Fig. 9). Thus, running the Power-to-H2 sys-
2010). For a production process using the same technologies as tem using today's German electricity mix would increase total GHG
our model, Reuter et al. (2015) and Abadías Llamas et al. (2019) emissions.
determined the climate change impact of the complete produc- In view of the evolving decarbonization of the electricity supply,
tion chain, including the mining operations with copper ore grades we analyze the impact of lower electricity emission factors in in-
crements of 50 g CO2-eq/kWhel (Fig. 10).
A decreasing total emission factor for grid electricity also re-
duces the total GHG emissions of the conventional copper pro-
duction due to the high electricity demand of the conventional
process. The scenario with 150 g CO2-eq/kWhel is the first instance
where total GHG emissions of the fully-decarbonized copper pro-
duction are lower than for conventional copper production. A linear
interpolation shows that the crossover point between the con-
ventional and the fully-decarbonized copper production lies at
160 g CO2-eq/kWhel. Thus, only if the GHG emissions from the
electricity supply are below 160 g CO2-eq/kWhel, decarbonization
of copper production with Power-to-H2 has not only a local but also
a positive global impact on GHG emissions.
Since the total GHG emissions strongly depend on the upstream
emissions of electricity supply, we further evaluate the abatement
cost in relation to avoided total GHG emissions (Fig. 11). The copper
plant is assumed to have reduced all direct emissions. Because total
GHG emissions are reduced only if the emission factor of electricity
is below 160 g CO2-eq/kWhel (Fig. 10), we focus our total GHG
abatement cost analysis on values below this threshold. As ex-
Fig. 11. Total GHG abatement cost for complete decarbonization of copper production
pected, the total GHG abatement cost strongly depends on the total
by Power-to-H2, i.e., 100% reduction of direct emissions. The remaining indirect GHG emission factor of electricity. Further, we see that the total GHG
emissions of electricity are based on the total emission factor of electricity and the abatement cost is much higher than the direct CO2 abatement cost
fossil fuels consider the avoided upstream GHG emissions from natural gas and coke and only reaches similar cost at low-emission electricity with
supply.

14
€ben, N. Scho
F.T.C. Ro €ne, U. Bau et al. Journal of Cleaner Production 306 (2021) 127191

Fig. 13. Total direct CO2 abatement cost under varied parameters: (a) electrolyzer
investment cost and (b) electrolyzer efficiency.

EUR/t CO2 (6.1%).


Increasing the O2 price by 10% decreases the abatement cost by 8
EUR/t CO2 (4.1%) due to the higher economic benefit from the
Fig. 12. Sensitivity analysis for the fully-decarbonized system. The values are set in avoided O2 procurement cost.
relation to the reference case with direct CO2 abatement cost of 201 EUR/t CO2. The shown investment cost of the hydrogen storage system
includes both costs for H2 storage and compressor. However, the
influence of the H2 storage on the direct CO2 abatement cost is
around 30 g CO2-eq/kWhel. Beyond that value, the total GHG negligible. An increase of the storage system cost of 10% increases
abatement cost can drop below the direct CO2 emission abatement the direct CO2 abatement cost by only 0.2 EUR/t CO2 (0.1%). Inter-
cost due to the avoided emissions from upstream natural gas and estingly, the H2 storage is designed larger in scenarios with low
coke supply, which are not considered in the direct CO2 emissions. electrolyzer investment cost as over-sizing becomes more attrac-
tive. When the electrolyzer investment cost is reduced by 10%, the
4.3. Future technology developments optimal storage size increases by about 53%.

Some parameters in our model are likely to change in the future 4.3.2. Impact of water electrolysis development
due to technological progress. Most importantly, water electrolysis Since technical and economic performance of water electrolysis
systems are expected to be further improved leading to lower in- systems are expected to improve in the future, we analyze the
vestment cost (Thema et al., 2019; Buttler and Spliethoff, 2018; impact of such developments on the direct CO2 abatement cost by
Smolinka et al., 2018) and higher efficiency, e.g., by commerciali- Power-to-H2.
zation of large SOEL systems (Buttler and Spliethoff, 2018). Com- To analyze the impact of decreasing water electrolyzer invest-
modity prices, most notably for grid electricity and O2, are also ment cost, we analyze the direct CO2 abatement cost over a typical
subject to change and influence the economic feasibility of the range of investment costs (Fig. 13(a)). The investment costs range
Power-to-H2 retrofit. We, therefore, conduct a sensitivity analysis from 1500 to 500 EUR/kWel, which represents the range from an
on the influence of prices for O2 and electricity, water electrolyzer expensive alkaline-electrolyzer system today down to future AEL
efficiency and investment cost, as well as H2 storage investment and PEMEL systems (Smolinka et al., 2018). Assuming investment
cost on the direct CO2 abatement cost. cost of 500 EUR/kWel for the electrolyzer system, the direct CO2
abatement cost reaches 136 EUR/t CO2 to avoid all direct emissions.
4.3.1. Sensitivity analysis The slight non-linearity in the direct CO2 abatement cost curve
Fig. 12 shows that the direct CO2 abatement cost is very sensitive is a result of the electrolyzer becoming larger in scenarios with
to changes in the electricity cost. This result is not surprising as lower investment cost. The larger size allows for more operational
electricity costs dominate the cost structure (Fig. 7). Any change in freedom to exploit fluctuating electricity prices and thus can save
electricity prices through power generation, procurement strate- operating cost.
gies, taxes or tax exemptions can have a huge influence on the The efficiency of electrolyzer systems is also expected to
profitability of decarbonization by Power-to-H2. Thus, the eco- improve in the future and has a large impact on the direct CO2
nomics of decarbonization depend strongly on the local electricity abatement cost (cf. Fig. 13(b)). The costs decrease to 127 EUR/t CO2
market conditions. An electricity price increase of 10%, for instance, for an electrolyzer system efficiency of 80%, a value corresponding
results in additional abatement cost of 29 EUR/t CO2 (14.5%). to existing SOEL systems with heat integration (Buttler and
The second most influential parameter is the efficiency of the Spliethoff, 2018; Sunfire GmbH, 2020), e.g., a system with satu-
electrolyzer system, which influences both investment cost and rated steam input at 150 C and 3 bar achieves an electric efficiency
operating cost by reducing the necessary electrolyzer size as well as of 82% (LHV) (Sunfire GmbH, 2020). However, the higher invest-
the electricity demand. Thus, any potential improvement in the ment cost of SOEL systems counteracts cost reductions in electricity
efficiency has a large influence on the direct CO2 abatement cost. costs achieved by higher efficiencies.
Assuming an electrolyzer system can improve its efficiency by 5% in The non-linearity in the direct CO2 abatement cost curve as a
absolute terms, e.g., from 60 to 65% system efficiency (corre- function of electrolyzer efficiency is caused by the efficiency
sponding to a relative increase of 8%), the abatement cost decrease influencing both investment and operating costs: a higher effi-
by 23 EUR/t CO2 (11.3%). ciency allows for a smaller electrolyzer and less electricity con-
The investment cost of the electrolyzer system has the third sumption during operation.
strongest influence on the direct CO2 abatement cost. A decrease of As both investment cost and efficiency of water electrolysis
100 EUR/kWel (10%) reduces the direct CO2 abatement cost by 12 systems are expected to improve, we analyze a combined
15
€ben, N. Scho
F.T.C. Ro €ne, U. Bau et al. Journal of Cleaner Production 306 (2021) 127191

Fig. 14. Direct CO2 abatement cost under decreasing electrolyzer investment cost and increasing efficiency. Highlighted estimates for the electrolyzer system parameters are taken
from Smolinka et al. (2018). To display the estimations on our parameter grid, we rounded each value to the nearest increment in our sensitivity analysis. The stated efficiency of the
solid oxide electrolysis (SOEL) does not consider the energy for steam generation (Smolinka et al., 2018).

AEL and PEMEL systems are also expected to improve in the


short-term (2030) and long-term (2050) (Smolinka et al., 2018).
The estimates for 2030 already imply strong reductions in the
direct CO2 abatement cost. The AEL reaches direct CO2 abatement
cost of around 98 EUR/t CO2 in 2030. While the abatement cost for
the PEMEL is noticeably higher in 2030, it reaches a value of 105
EUR/t CO2 in 2050, similar to the 89 EUR/t CO2 for the AEL in 2050.
The expected future improvement of the electrolyzer efficiency
also has an impact on the total GHG emissions (Fig. 15). With
increasing efficiency, the system consumes less electricity which
reduces the indirect GHG emissions from the electricity supply.
Since the emission factor of the electricity supply is expected to
improve in the future, we also analyze its impact.
While an increase in electrolyzer efficiency has a positive impact
on the total GHG emissions, it becomes clear that the emission
factor of the supplied electricity is absolutely crucial. The dashed
line in Fig. 15 marks the threshold at which the retrofit becomes
overall positive in comparison to the conventional copper pro-
duction in terms of total GHG emissions. Thus, a drastic reduction of
GHG emissions in the electricity supply is of utmost importance to
Fig. 15. Total GHG emissions under decreasing emission factor of the electricity supply
and increasing electrolyzer efficiency. The dashed line marks the threshold where the support the decarbonization of copper production by Power-to-H2.
total GHG emissions of the fully-decarbonized process decrease in comparison to the In summary, the results emphasize that the supply with renewable
conventional production process. electricity is by far the most important factor to reduce total GHG
emissions even if considering improvements in electrolyzer
efficiency.
development of both parameters. The parameter pairs and the
resulting direct CO2 abatement cost are presented in Fig. 14. We
highlight typical estimates for future electrolyzer systems based on 4.4. Discussion and implications
the parameterized data set given in Smolinka et al. (2018). The
highlighted results show three electrolyzer technologies: alkaline We now discuss the significance and implications of our work
(AEL), polymer electrolyte membrane (PEMEL), and solid oxide from an economic and environmental perspective, as well as the
electrolysis (SOEL) in 2030 and 2050. importance for different stakeholders.
The potential commercialization of SOEL systems is often While our model is based on a specific process design (Section
considered to result in high efficiencies and large cost reductions, 2.1), the process setup consists of the most commonly applied
with some estimates even below 500 EUR/kWel in the long-term technologies in copper production. Thus, we expect our main
(Bertuccioli et al., 2014; Smolinka et al., 2018). We find that the findings in combination with the given sensitivity analysis to apply
combination of both parameters can drastically reduce the direct to a broad range of copper plants. In addition, the disclosure of our
CO2 abatement cost to a value of 54 EUR/t CO2 for a system expected model equations, parameters, and assumptions enables readers
in 2050. with access to plant-specific data to adapt our analysis to their
16
€ben, N. Scho
F.T.C. Ro €ne, U. Bau et al. Journal of Cleaner Production 306 (2021) 127191

particular plant and site. With access to additional data not publicly system to eliminate direct CO2 emissions. The total greenhouse gas
available, the model could also be updated to account for potential (GHG) emissions are determined to evaluate the climate change
impacts beyond those we could assess, e.g., non-ideal fuel utiliza- impact of the retrofit.
tion rates. The results show direct CO2 abatement cost of around 201 EUR/t
The determined potential provides industry stakeholders with a CO2 assuming current Power-to-H2 technologies and commodity
first estimate of the cost of decarbonizing copper production by costs in Germany (2019). This direct CO2 abatement cost vastly
hydrogen and shows the beneficial co-utilization of the electrolysis exceeds the average price of European Union Allowances (EUA) of
by-product O2. The direct CO2 abatement cost of 201 EUR/t CO2 24.6 EUR/t CO2 in 2019. However, future prices of up to 310 EUR/t
gives an indication under which cost of CO2 the retrofit becomes CO2 in 2050 are estimated in a scenario by the European Union to
profitable. Vice versa, the findings show policy makers the in- mitigate climate change. Other studies find that similar prices for
centives or carbon prices that need to be put in place to encourage CO2 emissions are necessary for climate change mitigation in the
industry to decarbonize. future, indicating that the Power-to-H2 route might become viable
Our results show that the cost of decarbonization constitutes a in the future.
relevant cost factor in copper production. To prevent carbon While the profitability depends strongly on the price of elec-
leakage by making production regionally unprofitable, border car- tricity, the savings from utilizing the electrolysis by-product O2 in
bon adjustment might be necessary. Preserving regional copper the copper plant have a substantial positive impact on the direct
production capacities with low GHG footprints can be especially CO2 abatement cost. This result makes on-site integration of Power-
important for the goal of a circular economy since copper produc- to-H2 and copper production more attractive than separate oper-
tion is of great importance for the recycling of electronic waste and ation sites. Co-utilization of H2 and O2 also reduces GHG emissions
recovery of valuable minor elements (Reuter et al., 2019). as less O2 has to be supplied by cryogenic air separation. While the
Future improvements in water electrolysis technology would hydrogen storage system has little impact on the abatement cost,
greatly reduce the economic cost of decarbonization. High- storage allows the water electrolyzer to react to the electricity
temperature electrolysis with heat integration offers particular prices to some extent. While the storage capacity is sized rather
high efficiencies. Cooperation between potential customers of small compared to the peak power of the electrolyzer in our study,
large-scale electrolyzers such as copper producers and manufac- we expect the storage to have a larger economic impact in case of
turers of such systems is beneficial as the estimated cost reductions more severe electricity price fluctuations.
are in parts based on learning effects from installed systems (Bo€ hm While direct CO2 emissions can be avoided by retrofitting a
et al., 2020). The expected cost reduction together with higher CO2 Power-to-H2 system, our results show that the retrofit reduces total
certificate prices might lead to a reinforcing effect by making Po- GHG emissions only if the emission factor of electricity is below
wer-to-H2 system for decarbonization of industries economically 160 g CO2-eq/kWh.
feasible. The sensitivity analysis shows that expected future de-
As our analysis has once again confirmed, the electricity price is velopments of water electrolysis systems in terms of technical and
a critical factor for cost-efficient decarbonization of energy- economic performance can drastically reduce the direct CO2
intensive industries by Power-to-H2. Moreover, the required elec- abatement cost. Reduced investment cost and efficiency gains of
tricity must come from low-carbon sources, as the overall emis- alkaline electrolyzers expected already for 2030 halve the direct
sions are highly dependent on the emission factors of electricity CO2 abatement cost to around 98 EUR/t CO2. Most promising is the
supply and in case of copper production only decrease with elec- commercialization of large solid oxide electrolyzer systems (SOEL)
tricity emission factors below 160 g CO2-eq/kWh. Therefore, with heat integration, enabling direct CO2 abatement cost of around
assessment of the local electricity supply conditions is always 90 EUR/t CO2 expected in 2030 and 54 EUR/t CO2 expected in 2050.
required to determine the overall GHG emissions reduction po- Based on the progress of water electrolysis technology and ris-
tential for such a retrofit. Future works should therefore include ing future CO2 certificate prices, it seems likely that the decar-
specific local emission factors as the local situation can vary bonization of copper production by a Power-to-H2 system retrofit
strongly from the case for Germany laid out here. Many smelters will become profitable in the next decades. However, decarbon-
are located near the sea to receive ore shipments (Schlesinger et al., ization of the electricity supply will be crucial to lower the total
2011) and proximity to off-shore wind parks could potentially offer GHG emissions of copper production.
opportunities to access low-carbon power. Further research should focus on the local conditions of the
While the on-site integration of a Power-to-H2 system shows a energy supply as we observe strong impacts of electricity prices
great economic benefit by reducing the direct CO2 abatement cost and emission factors. Our model and analysis can easily be adapted
by 81 EUR/t CO2, imported H2 that is produced in optimal locations to a specific plant or site by updating the relevant model parame-
with high full-load hours for wind and solar power generation ters. The model should also be refined by learnings from experi-
represents an interesting alternative that should be investigated in mental implementations of hydrogen in copper production. Since
a future study. Considering H2 production facilities in remote lo- renewable energy is key for a successful decarbonization but
cations may also become necessary due to the high demand for intermittent in nature, copper production equipped with a Power-
renewable electricity when decarbonizing industry, as shown for to-H2 system including H2 storage may also represent a potentially
the chemical industry (Ka €telho
€n et al., 2019), which could readily valuable source of demand-side flexibility.
exceed regional renewable energy potentials.
CRediT authorship contribution statement
5. Conclusions
Fritz T.C. Ro€ ben: Conceptualization, Methodology, Software,
This work analyzes the techno-economic potential of decar- Formal analysis, Validation, Visualization, Writing e original draft.
bonizing copper production with H2 by retrofitting a Power-to-H2 Nikolas Scho € ne: Methodology, Software, Formal analysis,

17
€ben, N. Scho
F.T.C. Ro €ne, U. Bau et al. Journal of Cleaner Production 306 (2021) 127191

Visualization, Writing e review & editing. Uwe Bau: Conceptuali- P2H_Full_Study_FCHJU.pdf. Accessed: 2020-04-23.
de Coninck, H., Revi, A., Babiker, M., Bertoldi, P., Buckeridge, M., Cartwright, A.,
zation, Supervision, Writing e review & editing. Markus A. Reuter:
Dong, W., Ford, J., Fuss, S., Hourcade, J.C., Ley, D., Mechler, R., Newman, P.,
Funding acquisition, Writing e review & editing. Manuel Dahmen: Revokatova, A., Schultz, S., Steg, L., Sugiyama, T., 2018. Implementing and
Conceptualization, Supervision, Writing e review & editing. Andre strengthening the global response. In: Masson-Delmotte, V., Zhai, P.,
Po€ rtner, H.O., Roberts, D., Skea, J., Shukla, P.R., Pirani, A., Moufouma-Okia, W.,
Bardow: Funding acquisition, Conceptualization, Supervision,
Pean, C., Pidcock, R., Connors, S., Matthews, J.B.R., Chen, Y., Zhou, X., Gomis, M.I.,
Writing e review & editing. Lonnoy, E., Maycock, T., Tignor, M., Waterfield, T. (Eds.), Global Warming of
1.5 C. An IPCC Special Report on the Impacts of Global Warming of 1.5 C above
Declaration of competing interest Pre-industrial Levels and Related Global Greenhouse Gas Emission Pathways, in
the Context of Strengthening the Global Response to the Threat of Climate
Change, Sustainable Development, and Efforts to Eradicate Poverty. Intergov-
The authors declare that they have no known competing ernmental Panel on Climate Change, pp. 313e443.
financial interests or personal relationships that could have Cusano, G., Gonzalo, M.R., Farrell, F., Remus, R., Roudier, S., Sancho, L.D., 2017. Best
available techniques (BAT) reference document for the non-ferrous metals in-
appeared to influence the work reported in this paper. dustries. Report. European Comission, Industrial Emissions Directive. https://
doi.org/10.2760/8224.
Acknowledgements Deetman, S., Pauliuk, S., van Vuuren, D.P., van der Voet, E., Tukker, A., 2018. Sce-
narios for demand growth of metals in electricity generation technologies, cars,
and electronic appliances. Environ. Sci. Technol. 52, 4950e4959. https://
The present contribution is supported by the Helmholtz Asso- doi.org/10.1021/acs.est.7b05549.
ciation under the Joint Initiative ‘Energy Systems Integration’. Degel, R., Lux, T., Hecker, E., Filzwieser, A., Krassnig, H.J., Filzwieser, I., 2019. Inno-
vative solutions in non-ferrous metals production. In: Proceedings of the 58th
Annual Conference of Metallurgists (COM) Hosting the 10th International
References Copper Conference 2019, Vancouver, Canada, 18-21 August 2019. The Canadian
Institute of Mining, Metallurgy and Petroleum.
Abadías Llamas, A., Valero Delgado, A., Valero Capilla, A., Torres Cuadra, C., Dong, D., van Oers, L., Tukker, A., van der Voet, E., 2020. Assessing the future
Hultgren, M., Peltomaeki, M., Roine, A., Stelter, M., Reuter, M.A., 2019. Simula- environmental impacts of copper production in China: implications of the
tion-based exergy, thermo-economic and environmental footprint analysis of energy transition. J. Clean. Prod. 274, 122825 https://doi.org/10.1016/
primary copper production. Miner. Eng. 131, 15. https://doi.org/10.1016/ j.jclepro.2020.122825.
j.mineng.2018.11.007. Díaz-Borrego, F.J., Miras-Rodríguez, M.d.M., Escobar-Pe rez, B., 2019. Looking for
Aurubis, 2019. Environmental protection in the Aurubis group. Report. Aurubis AG. accurate forecasting of copper TC/RC benchmark levels. Complexity 1e16.
https://www.aurubis.com/binaries/content/assets/aurubis-en/dateien/responsi https://doi.org/10.1155/2019/8523748, 2019.
bility/environmental-statement/2019_aurubis_ag_environmental_statement_e EEX, 2019. Emission spot primary market auction report 2019. Report. European
n.pdf. Accessed: 2019-12-18. Energy Exchange. https://www.eex.com/en/market-data/environmental-marke
Aurubis, 2020. Copper mail no. 181. Report. Aurubis AG. https://www.aurubis.com/e ts/eua-primary-auction-spot-download. Accessed: 2020-02-19.
n/investor-relations/news-and-reports/copper-mail/copper-mail-update. Elshkaki, A., Graedel, T.E., Ciacci, L., Reck, B.K., 2016. Copper demand, supply, and
Accessed: 2021-02-16. associated energy use to 2050. Global Environ. Change 39, 305e315. https://
Bailera, M., Lisbona, P., Romeo, L.M., 2015. Power to gas-oxyfuel boiler hybrid sys- doi.org/10.1016/j.gloenvcha.2016.06.006.
tems. Int. J. Hydrogen Energy 40, 10168e10175. https://doi.org/10.1016/ EU Commission, 2011. Energy Roadmap 2050. Impact Assessment and Scenario
j.ijhydene.2015.06.074. Analysis. Report SEC, 1565. EU.
Banaszkiewicz, T., Chorowski, M., Gizicki, W., 2014. Comparative analysis of oxygen Felgenhauer, M., Hamacher, T., 2015. State-of-the-art of commercial electrolyzers
production for oxy-combustion application. Energy Procedia 51, 127e134. and on-site hydrogen generation for logistic vehicles in South Carolina. Int. J.
https://doi.org/10.1016/j.egypro.2014.07.014. Hydrogen Energy 40, 2084e2090. https://doi.org/10.1016/
Bareiß, K., de la Rua, C., Mo €ckl, M., Hamacher, T., 2019. Life cycle assessment of j.ijhydene.2014.12.043.
hydrogen from proton exchange membrane water electrolysis in future energy Fischedick, M., Marzinkowski, J., Winzer, P., Weigel, M., 2014a. Techno-economic
systems. Appl. Energy 237, 862e872. https://doi.org/10.1016/ evaluation of innovative steel production technologies. J. Clean. Prod. 84,
j.apenergy.2019.01.001. 563e580. https://doi.org/10.1016/j.jclepro.2014.05.063.
Bertuccioli, L., Chan, A., Hart, D., Lehner, F., Madden, B., Standen, E., 2014. Study on Fischedick, M., Roy, J., Acquaye, A., Allwood, J., Ceron, J.P., Geng, Y., Kheshgi, H.,
development of water electrolysis in the EU. Report. E4tech Sa rl with Element Lanza, A., Perczyk, D., Price, L., Santalla, E., Sheinbaum, C., Tanaka, K., 2014b.
Energy Ltd for the Fuel Cells and Hydrogen Joint Undertaking. https://www. Industry. In: Edenhofer, O., Pichs-Madruga, R., Sokona, Y., Farahani, E.,
fch.europa.eu/node/783. Accessed: 2020-02-26. Kadner, S., Seyboth, K., Adler, A., Baum, I., Brunner, S., Eickemeier, P., nad
€hm, H., Zauner, A., Rosenfeld, D.C., Tichler, R., 2020. Projecting cost development
Bo Kriemann, B., Savolainen, J., Schlo €mer, S., von Stechow, C., Zwickel, T., Minx, J.C.
for future large-scale power-to-gas implementations by scaling effects. Appl. (Eds.), Climate Change 2014: Mitigation of Cimate Change. Contribution of
Energy 264, 114780. https://doi.org/10.1016/j.apenergy.2020.114780. Working Group III to the Fifth Assessment Report of the Intergovernmental
Bonnin, M., Azzaro-Pantel, C., Domenech, S., 2019. Optimization of natural resource Panel on Climate Change. Cambridge University Press, Cambridge, United
management: application to French copper cycle. J. Clean. Prod. 223, 252e269. Kingdom and New York, NY, USA, pp. 739e810.
https://doi.org/10.1016/j.jclepro.2019.03.081. Friedmann, S.J., Fan, Z., Tang, K., 2019. Low-carbon Heat Solutions for Heavy In-
Breyer, C., Tsupari, E., Tikka, V., Vainikka, P., 2015. Power-to-gas as an emerging dustry: Sources, Options, and Costs Today. Report. Columbia SIPA Center on
profitable business through creating an integrated value chain. Energy Procedia Global Energy Policy. URL: https://energypolicy.columbia.edu/sites/default/file
73, 182e189. https://doi.org/10.1016/j.egypro.2015.07.668. s/file-uploads/LowCarbonHeat-CGEP_Report_100219-2_0.pdf. (Accessed 30
Brunner, C., Michaelis, J., Mo €st, D., 2015. Competitiveness of different operational October 2019).
concepts for power-to-gas in future energy systems. Z. Energiewirtschaft 39, Glover, F., 1975. Improved linear integer programming formulations of nonlinear
275e293. https://doi.org/10.1007/s12398-015-0165-0. integer problems. Manag. Sci. 22, 455e460. https://doi.org/10.1287/
Bundesamt für Umwelt, 2019. CO2 Abgabe. Web page. https://www.bafu.admin.ch/ mnsc.22.4.455.
bafu/de/home/themen/klima/fachinformationen/klimapolitik/co2-abgabe.html. Government Offices of Sweden, 2020. Sweden's carbon tax. https://www.governme
Accessed: 2020-03-19. nt.se/government-policy/taxes-and-tariffs/swedens-carbon-tax. Accessed:
Bundesnetzagentur, Bundeskartellamt, 2019. Monitoringbericht 2019. Report. 2020-03-19.
Bundesnetzagentur und Bundeskartellamt. https://www.bundesnetzagentur. Goyal, P., Themelis, N.J., Zanchuk, W.A., 1982. Gaseous refining of anode copper. JOM
de/SharedDocs/Mediathek/Berichte/2019/Monitoringbericht_Energie2019.pdf. The Journal of The Minerals, Metals & Materials Society (TMS) 34, 22e28.
Accessed: 2019-12-16. https://doi.org/10.1007/BF03338157.
Bundesnetzagentur | SMARD.de, 2020. SMARD Strommarktdaten. Web page. Bun- Graf, F., Krajete, A., Schmack, U., 2014. Abschlussbericht - Techno-o €konomische
desnetzagentur. https://www.smard.de/. Accessed: 2020-02-13. Studie zur biologischen Methanisierung bei Power-to-Gas-Konzepten. Report.
Buttler, A., Koltun, R., Wolf, R., Spliethoff, H., 2015. A detailed techno-economic Deutscher Verein des Gas- und Wasserfaches e.V. https://www.dvgw.de/theme
analysis of heat integration in high temperature electrolysis for efficient n/forschung-und-innovation/forschungsprojekte/dvgw-forschungsbericht-g-
hydrogen production. Int. J. Hydrogen Energy 40, 38e50. https://doi.org/ 30113/. Accessed: 2019-04-30.
10.1016/j.ijhydene.2014.10.048. Grossmann, I.E., 1985. Mixed-integer programming approach for the synthesis of
Buttler, A., Spliethoff, H., 2018. Current status of water electrolysis for energy integrated process flowsheets. Comput. Chem. Eng. 9, 463e482. https://doi.org/
storage, grid balancing and sector coupling via power-to-gas and power-to- 10.1016/0098-1354(85)80023-5.
liquids: a review. Renew. Sustain. Energy Rev. 82, 2440e2454. https://doi.org/ Guilera, J., Ramon Morante, J., Andreu, T., 2018. Economic viability of SNG pro-
10.1016/j.rser.2017.09.003. duction from power and CO2. Energy Convers. Manag. 162, 218e224. https://
Chardonnet, C., Giordano, V., Rapoport, S., De Vos, L., Genoese, F., Roig, G., Bart, F., doi.org/10.1016/j.enconman.2018.02.037.
Lanoix, J.C., Vanhoudt, W., De Lacroix, T., Ha, T., Van Genabet, B., 2017. Study on Henckens, M.L.C.M., Worrell, E., 2020. Reviewing the availability of copper and
early business cases for H2 in energy storage and more broadly power to H2 nickel for future generations. The balance between production growth, sus-
applications. Report. FCH 2 JU. https://www.fch.europa.eu/sites/default/files/ tainability and recycling rates. J. Clean. Prod. 264, 121460. https://doi.org/

18
€ben, N. Scho
F.T.C. Ro €ne, U. Bau et al. Journal of Cleaner Production 306 (2021) 127191

10.1016/j.jclepro.2020.121460. Parra De Lazzari, C., Capocchi, J.D.T., 1997. Copper deoxidation by bubbling
Icha, P., Kuhs, G., 2020. Entwicklung der spezifischen Kohlendioxid-Emissionen des hydrogen/nitrogen mixtures through the melt. In: Mishra, B. (Ed.), EPD
deutschen Strommix in den Jahren 1990 - 2019. Report Climate Change 13/ Congress 1997: Proceedings of Sessions and Symposia Sponsored by the
2020. Umweltbundesamt. https://www.umweltbundesamt.de/en/publikatione Extraction and Processing Division, Orlando, Florida, USA, 9-13 February 1997,
n/entwicklung-der-spezifischen-kohlendioxid-6. Accessed: 2020-07-15. the Minerals. Metals & Materials Society.
IRENA, 2020. Green Hydrogen Cost Reduction: Scaling up Electrolysers to Meet the Proost, J., 2019. State-of-the art CAPEX data for water electrolysers, and their impact
1.5 C Goal. Report. International Renewable Energy Agency, Abu Dhabi. on renewable hydrogen price settings. Int. J. Hydrogen Energy 44, 4406e4413.
Ka€telho€ n, A., Meys, R., Deutz, S., Suh, S., Bardow, A., 2019. Climate change mitigation https://doi.org/10.1016/j.ijhydene.2018.07.164.
potential of carbon capture and utilization in the chemical industry. Proc. Natl. Qu, G., Wei, Y., Li, B., Wang, H., Yang, Y., McLean, A., 2020. Distribution of copper and
Acad. Sci. Unit. States Am. 116, 11187. https://doi.org/10.1073/pnas.1821029116. iron components with hydrogen reduction of copper slag. J. Alloys Compd. 824,
Koj, J.C., Wulf, C., Schreiber, A., Zapp, P., 2017. Site-dependent environmental im- 153910. https://doi.org/10.1016/j.jallcom.2020.153910.
pacts of industrial hydrogen production by alkaline water electrolysis. Energies Ramachandran, V., Díaz, C., Eltringham, T., Lehner, T., Mackey, P.J., Newman, C.J.,
10, 860. https://doi.org/10.3390/en10070860. Tarasov, A., 2003. Primary copper production - a survey of operating world
Kopp, M., Coleman, D., Stiller, C., Scheffer, K., Aichinger, J., Scheppat, B., 2017. copper smelters. In: Díaz, C., Kapusta, J., Newman, C. (Eds.), Proceedings of the
Energiepark Mainz: technical and economic analysis of the worldwide largest Copper 2003-Cobre 2003, the 5th International Conference, Volume IV e Py-
Power-to-Gas plant with PEM electrolysis. Int. J. Hydrogen Energy 42, rometallurgy of Copper (Book 1), the Hermann Schwarze Symposium on Py-
13311e13320. https://doi.org/10.1016/j.ijhydene.2016.12.145. rometallurgy, Santiago, Chile, 30 November - 3 December 2003. Canadian
KPMG, 2019. Cost of capital study 2019. Report. KPMG AG. https://home.kpmg/de/e Institute of Mining, Metallurgy and Petroleum, pp. 3e106.
n/home/insights/2019/10/cost-of-capital-study-2019.html. Accessed: 2020-03- Reuter, M.A., van Schaik, A., Gediga, J., 2015. Simulation-based design for resource
02. efficiency of metal production and recycling systems: cases - copper production
Kuipers, K.J.J., van Oers, L.F.C.M., Verboon, M., van der Voet, E., 2018. Assessing and recycling, e-waste (LED lamps) and nickel pig iron. Int. J. Life Cycle Assess.
environmental implications associated with global copper demand and supply 20, 671e693. https://doi.org/10.1007/s11367-015-0860-4.
scenarios from 2010 to 2050. Global Environ. Change 49, 106e115. https:// Reuter, M.A., van Schaik, A., Gutzmer, J., Bartie, N., Abadías-Llamas, A., 2019. Chal-
doi.org/10.1016/j.gloenvcha.2018.02.008. lenges of the circular economy: a material, metallurgical, and product design
Kulczycka, J., Lelek, Ł., Lewandowska, A., Wirth, H., Bergesen, J.D., 2016. Environ- perspective. Annu. Rev. Mater. Res. 49, 253e274. https://doi.org/10.1146/
mental impacts of energy-efficient pyrometallurgical copper smelting tech- annurev-matsci-070218-010057.
nologies: the consequences of technological changes from 2010 to 2050. J. Ind. Rogelj, J., Shindell, D., Jiang, K., Fifita, S., Forster, P., Ginzburg, V., Handa, C.,
Ecol. 20, 304e316. https://doi.org/10.1111/jiec.12369. Kheshgi, H., Kobayashi, S., Kriegler, E., Mundaca, L., Se  fe
rian, R., Vilarin
~ o, M.V.,
Kuparinen, K., Vakkilainen, E., 2017. Green pulp mill: renewable alternatives to fossil 2018. Mitigation pathways compatible with 1.5 C in the context of sustainable
fuels in lime kiln operations. BioResources 12, 4031e4048. https://doi.org/ development. In: Masson-Delmotte, V., Zhai, P., Po € rtner, H.O., Roberts, D.,
10.15376/biores.12.2.4031-4048. Skea, J., Shukla, P.R., Pirani, A., Moufouma-Okia, W., Pe an, C., Pidcock, R.,
van Leeuwen, C., Mulder, M., 2018. Power-to-gas in electricity markets dominated Connors, S., Matthews, J.B.R., Chen, Y., Zhou, X., Gormis, M.I., Lonnoy, E.,
by renewables. Appl. Energy 232, 258e272. https://doi.org/10.1016/ Maycock, T., Tignor, M., Waterfield, T. (Eds.), Global Warming of 1.5 C an IPCC
j.apenergy.2018.09.217. Special Report on the Impacts of Global Warming of 1.5 C above Pre-industrial
Li, B., Wang, X., Wang, H., Wei, Y., Hu, J., 2017. Smelting reduction and kinetics Levels and Related Global Greenhouse Gas Emission Pathways, in the Context of
analysis of magnetic iron in copper slag using waste cooking oil. Sci. Rep. 7, Strengthening the Global Response to the Threat of Climate Change, Sustainable
2406. https://doi.org/10.1038/s41598-017-02696-y. Development, and Efforts to Eradicate Poverty. Intergovernmental Panel on
Li, B., Wei, Y., Wang, H., Yang, Y., 2018. Reduction of magnetite from copper smelting Climate Change.
slag using petro-diesel and biodiesel. ISIJ Int. 58, 1168e1174. https://doi.org/ Rosenfeld, D.C., Bo €hm, H., Lindorfer, J., Lehner, M., 2020. Scenario analysis of
10.2355/isijinternational.ISIJINT-2017-723. implementing a power-to-gas and biomass gasification system in an integrated
Luderer, G., Vrontisi, Z., Bertram, C., Edelenbosch, O.Y., Pietzcker, R.C., Rogelj, J., De steel plant: a techno-economic and environmental study. Renew. Energy 147,
Boer, H.S., Drouet, L., Emmerling, J., Fricko, O., Fujimori, S., Havlík, P., Iyer, G., 1511e1524. https://doi.org/10.1016/j.renene.2019.09.053.
Keramidas, K., Kitous, A., Pehl, M., Krey, V., Riahi, K., Saveyn, B., Tavoni, M., Van Sallee, J.E., Ushakov, V., 1999. Electric settling furnace operations at the Cyprus
Vuuren, D.P., Kriegler, E., 2018. Residual fossil CO2 emissions in 1.5e2 C path- Miami Mining Corporation copper smelter. In: George, D.B., Chen, W.J.,
ways. Nat. Clim. Change 8, 626e633. https://doi.org/10.1038/s41558-018-0198- Mackey, P.J., Weddick, A.J. (Eds.), Copper 99-Cobre 99, pp. 629e644.
6. Schlesinger, M.E., King, M.J., Sole, K.C., Davenport, W.G., 2011. In: Extractive Met-
Marin, T., Utigard, T., 2010. Deoxidation of liquid copper with reducing O2/CH4 allurgy of Copper, fifth ed. Elsevier. https://doi.org/10.1016/C2010-0-64841-3.
flames. Metall. Mater. Trans. B 41, 535e542. https://doi.org/10.1007/s11663- Segura-Salazar, J., Lima, F.M., Tavares, L.M., 2019. Life cycle assessment in the
010-9346-8. minerals industry: current practice, harmonization efforts, and potential
Moreno-Leiva, S., Haas, J., Junne, T., Valencia, F., Godin, H., Kracht, W., Nowak, W., improvement through the integration with process simulation. J. Clean. Prod.
Eltrop, L., 2019. Renewable energy in copper production: a review on systems 232, 174e192. https://doi.org/10.1016/j.jclepro.2019.05.318.
design and methodological approaches. J. Clean. Prod. 246, 118978. https:// Siemens, AG, 2018. SILYZER 300. Die na €chste Dimension der PEM Elektrolyse.
doi.org/10.1016/j.jclepro.2019.118978. Report. Siemens AG. https://new.siemens.com/global/de/produkte/energie/e
Morgenthaler, S., Ball, C., Koj, J.C., Kuckshinrichs, W., Witthaut, D., 2020. Site- rneuerbare-energien/hydrogen-solutions.html#Portfolio. Accessed: 2019-04-
dependent levelized cost assessment for fully renewable power-to-methane 30.
systems. Energy Convers. Manag. 223, 113150. https://doi.org/10.1016/ Smolinka, T., Wiebe, N., Sterchele, P., Palzer, A., Lehner, F., Jansen, M., Kiemel, S.,
j.enconman.2020.113150. Miehe, R., Wahren, S., Zimmermann, F., 2018. Studie IndWEDe - Industri-
Moro, A., Lonza, L., 2018. Electricity carbon intensity in European member states: alisierung der Wasserelektrolyse in Deutschland: Chancen und Herausforder-
impacts on GHG emissions of electric vehicles. Transport. Res. Part D 64, 5e14. ungen für nachhaltigen Wasserstoff für Verkehr, Strom und Wa €rme. Report.
https://doi.org/10.1016/j.trd.2017.07.012. NOW-GmbH. Berlin. http://publica.fraunhofer.de/documents/N-519494.html.
Nemhauser, G.L., Wolsey, L.A., 1988. Integer and Combinatorial Optimization. John Accessed: 2019-12-16.
Wiley & Sons, United States of America. Song, X., Pettersen, J.B., Pedersen, K.B., Røberg, S., 2017. Comparative life cycle
Nguyen, T., Abdin, Z., Holm, T., Me rida, W., 2019. Grid-connected hydrogen pro- assessment of tailings management and energy scenarios for a copper ore
duction via large-scale water electrolysis. Energy Convers. Manag. 200, 112108. mine: a case study in northern Norway. J. Clean. Prod. 164, 892e904. https://
https://doi.org/10.1016/j.enconman.2019.112108. doi.org/10.1016/j.jclepro.2017.07.021.
Norgate, T., Haque, N., 2010. Energy and greenhouse gas impacts of mining and Soulier, M., Pfaff, M., Goldmann, D., Walz, R., Geng, Y., Zhang, L., Tercero
mineral processing operations. J. Clean. Prod. 18, 266e274. https://doi.org/ Espinoza, L.A., 2018. The Chinese copper cycle: tracing copper through the
10.1016/j.jclepro.2009.09.020. economy with dynamic substance flow and input-output analysis. J. Clean.
Otto, A., Robinius, M., Grube, T., Schiebahn, S., Praktiknjo, A., Stolten, D., 2017. Po- Prod. 195, 435e447. https://doi.org/10.1016/j.jclepro.2018.04.243.
wer-to-steel: reducing CO2 through the integration of renewable energy and Steelonthenet, 2019. European met coke prices. https://www.steelonthenet.com/fi
hydrogen into the German steel industry. Energies 10, 451e471. https://doi.org/ les/blast-furnace-coke.html/. Accessed: 2020-04-14.
10.3390/en10040451. Sunfire GmbH, 2020. Sunfire-hylink factsheet. https://www.sunfire.de/files/sunfi
Pardo, N., Moya, J.A., 2013. Prospective scenarios on energy efficiency and CO2 re/images/content/Produkte_Technologie/factsheets/Sunfire-HyLink_FactSheet.
emissions in the European iron & steel industry. Energy 54, 113e128. https:// pdf. Accessed: 2020-03-18.
doi.org/10.1016/j.energy.2013.03.015. Suopaja €rvi, H., Umeki, K., Mousa, E., Hedayati, A., Romar, H., Kemppainen, A.,
Pardo, N., Moya, J.A., Vatopoulos, K., 2012. Prospective Scenarios on Energy Effi- Wang, C., Phounglamcheik, A., Tuomikoski, S., Norberg, N., Andefors, A.,
ciency and CO2 Emissions in the EU Iron & Steel Industry. European Commis- €
Ohman, M., Lassi, U., Fabritius, T., 2018. Use of biomass in integrated steel-
sion - Joint Research Centre - Institute for Energy and Transport. https://doi.org/ making e status quo, future needs and comparison to other low-CO2 steel
10.2790/64264. Report. production technologies. Appl. Energy 213, 384e407. https://doi.org/10.1016/
Parra, D., Valverde, L., Pino, F.J., Patel, M.K., 2019. A review on the role, cost and j.apenergy.2018.01.060.
value of hydrogen energy systems for deep decarbonisation. Renew. Sustain. Thema, M., Bauer, F., Sterner, M., 2019. Power-to-gas: electrolysis and methanation
Energy Rev. 101, 279e294. https://doi.org/10.1016/j.rser.2018.11.010. status review. Renew. Sustain. Energy Rev. 112, 775e787. https://doi.org/
Parra, D., Zhang, X., Bauer, C., Patel, M.K., 2017. An integrated techno-economic and 10.1016/j.rser.2019.06.030.
life cycle environmental assessment of power-to-gas systems. Appl. Energy 193, Thyssenkrupp, 2019. World first in Duisburg as NRW economics minister Pinkwart
440e454. https://doi.org/10.1016/j.apenergy.2017.02.063. launches tests at thyssenkrupp into blast furnace use of hydrogen. https://

19
€ben, N. Scho
F.T.C. Ro €ne, U. Bau et al. Journal of Cleaner Production 306 (2021) 127191

www.thyssenkrupp-steel.com/en/newsroom/press-releases/world-first-in-dui Weigel, M., Fischedick, M., Marzinkowski, J., Winzer, P., 2016. Multicriteria analysis
sburg.html. Accessed: 2019-12-18. of primary steelmaking technologies. J. Clean. Prod. 112, 1064e1076. https://
Van der Voet, E., Van Oers, L., Verboon, M., Kuipers, K., 2018. Environmental im- doi.org/10.1016/j.jclepro.2015.07.132.
plications of future demand scenarios for metals: Methodology and application Wietschel, M., Kühnbach, M., Rüdiger, D., 2019. Die aktuelle Treibhausgase-
to the case of seven major metals. J. Ind. Ecol. 23, 141e155. https://doi.org/ missionsbilanz von Elektrofahrzeugen in Deutschland. Report working paper
10.1111/jiec.12722. on sustainability and innovation No. S 02/2019. Fraunhofer ISI. http://publica.
Voll, P., Klaffke, C., Hennen, M., Bardow, A., 2013. Automated superstructure-based fraunhofer.de/documents/N-537432.html. Accessed: 2020-11-03.
synthesis and optimization of distributed energy supply systems. Energy 50, Zhang, W., Li, Z., Dong, S., Qian, P., Ye, S., Hu, S., Xia, B., Wang, C., 2021. Analyzing the
374e388. https://doi.org/10.1016/j.energy.2012.10.045. environmental impact of copper-based mixed waste recycling-a LCA case study
Wang, S., Davenport, W., Yao, S., Walters, G., Gonzales, T., Siegmund, A., George, D.B., in China. J. Clean. Prod. 284, 125256. https://doi.org/10.1016/
2019. Copper smelting: 2019 world copper smelter data. In: Proceedings of the j.jclepro.2020.125256.
58th Annual Conference of Metallurgists (COM) Hosting the 10th International Zhang, X., Bauer, C., Mutel, C.L., Volkart, K., 2017. Life cycle assessment of power-to-
Copper Conference 2019. Canada, Vancouver, pp. 18e21. August 2019. gas: approaches, system variations and their environmental implications. Appl.
Warczok, A., Riveros, G., 2007. Slag cleaning in crossed electric and magnetic fields. Energy 190, 326e338. https://doi.org/10.1016/j.apenergy.2016.12.098.
Miner. Eng. 20, 34e43. https://doi.org/10.1016/j.mineng.2006.04.007.

20
Update
Journal of Cleaner Production
Volume 322, Issue , 1 November 2021, Page

DOI: https://doi.org/10.1016/j.jclepro.2021.129144
Journal of Cleaner Production 322 (2021) 129144

Contents lists available at ScienceDirect

Journal of Cleaner Production


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

Corrigendum to “Decarbonizing copper production by power-to-hydrogen:


A techno-economic analysis” [J. Clean. Prod. 306 (2021) 127191]
Fritz T.C. Röben a, b, Nikolas Schöne a, Uwe Bau a, Markus A. Reuter c, d, Manuel Dahmen a,
André Bardow e, a, f, g, *
a
Institute of Energy and Climate Research - Energy Systems Engineering (IEK-10), Forschungszentrum Jülich GmbH, 52425, Jülich, Germany
b
RWTH Aachen University, 52062, Aachen, Germany
c
SMS Group GmbH, 40237, Düsseldorf, Germany
d
Helmholtz Institute Freiberg for Resource Technology (HIF), 09599, Freiberg, Germany
e
JARA-ENERGY, 52425, Jülich, Germany
f
Institute of Technical Thermodynamics, RWTH Aachen University, 52062, Aachen, Germany
g
Energy & Process Systems Engineering, ETH Zürich, 8092, Zürich, Switzerland

The authors regret that our recent paper (Röben et al., 2021) con­ investment costs are always converted based on the reference efficiency
tains an error in our calculations presented in section 4.3, where we (60%). The error affected the results shown in Figs. 12–14 of the original
analyze varying water electrolyzer efficiencies. The water electrolyzer manuscript but does not concern model equations and parameters stated
efficiency was not correctly updated in our code when converting the but was limited to the implementation. The updated results are shown in
investment costs from EUR/kWel to EUR/kWH2. In the results, the the updated Figs. 12–14 below.

DOI of original article: https://doi.org/10.1016/j.jclepro.2021.127191.


* Corresponding author. Energy & Process Systems Engineering, ETH Zürich, CLA F19.1, Tannenstrasse 3, 8092, Zürich, Switzerland.
E-mail address: [email protected] (A. Bardow).

https://doi.org/10.1016/j.jclepro.2021.129144

Available online 24 September 2021


0959-6526/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
F.T.C. Röben et al. Journal of Cleaner Production 322 (2021) 129144

Fig. 12. Sensitivity analysis for the fully-decarbonized system. The values are set in relation to the reference case with direct CO2 abatement cost of 201 EUR/t CO2.

In Fig. 14, the impact of the combined development of water elec­


trolysis efficiency and investment costs on the CO2 abatement cost is
Fig. 12 shows the impact of parameter changes on the direct CO2 shown, highlighting the projected future development (Smolinka et al.,
abatement cost for the decarbonized copper production. Comparing the 2018). The erroneous calculations overestimated the CO2 abatement
new Fig. 12 to the erroneous Fig. 12 from the original manuscript, it can cost for water electrolyzer efficiencies that exceeded the reference effi­
be noted that the impact of the water electrolyzer efficiency is larger and ciency of 60%. Particularly for high efficiencies, significantly lower CO2
now affects the CO2 abatement costs as strong as the electricity price. abatement costs are found, compared to the erroneous Fig. 14. CO2
The abatement cost decrease by 32 EUR/t CO2 (16.0%) when improving abatement cost are now expected to lie at 31 EUR/t CO2 for system
the electrolyzer efficiency by 5% in absolute terms, e.g., from 60 to 65% parameters expected for SOEL systems in 2050. The expected values for
system efficiency (corresponding to a relative increase of 8%). AEL systems achieve direct CO2 abatement cost of around 91 EUR/t CO2
The corrected Fig. 13(b) shows the impact of the water electrolyzer in 2030 and 80 EUR/t CO2 in 2050. The PEMEL system achieves direct
efficiency on the total direct CO2 abatement costs, which now decrease CO2 abatement cost of around 176 EUR/t CO2 in 2030 and
to 97 EUR/t CO2 for an electrolyzer efficiency of 80%. 98 EUR/t CO2 in 2050.

Fig. 13. Total direct CO2 abatement cost under varied parameters: (a) electrolyzer investment cost and (b) electrolyzer efficiency.

2
F.T.C. Röben et al. Journal of Cleaner Production 322 (2021) 129144

Fig. 14. Direct CO2 abatement cost under decreasing electrolyzer investment cost and increasing efficiency. Highlighted estimates for the electrolyzer system pa­
rameters are taken from Smolinka et al. (2018). To display the estimations on our parameter grid, we rounded each value to the nearest increment in our sensitivity
analysis. The stated efficiency of the solid oxide electrolysis (SOEL) does not consider the energy for steam generation (Smolinka et al., 2018).

Instead of the erroneous value of 54 EUR/t CO2 given in the abstract, References
highlights, and conclusions of the original manuscript, the abatement
cost reduce to 31 EUR/t CO2 for the projected future water electrolysis Röben, F.T.C., Schöne, N., Bau, U., Reuter, M.A., Dahmen, M., Bardow, A., 2021.
Decarbonizing copper production by power-to-hydrogen: a techno-economic
development. Further, in the conclusions, the results for the SOEL and analysis. J. Clean. Prod. 306, 127191. https://doi.org/10.1016/j.
AEL in 2030 need to be updated to 61 EUR/t CO2 and 91 EUR/t CO2, jclepro.2021.127191.
respectively. Thus, the corrected calculations show the potential for Smolinka, T., Wiebe, N., Sterchele, P., Palzer, A., Lehner, F., Jansen, M., Kiemel, S.,
Miehe, R., Wahren, S., Zimmermann, F., 2018. Studie IndWEDe - Industrialisierung
lower CO2 abatement costs for Power-to-H2 in copper production. der Wasserelektrolyse in Deutschland: Chancen und Herausforderungen für
The authors would like to apologise for any inconvenience caused. nachhaltigen Wasserstoff für Verkehr, Strom und Wärme. Report. NOW-GmbH,
Berlin. http://publica.fraunhofer.de/documents/N-519494.html. (Accessed 16
December 2019).

You might also like