Applied Energy 286 (2021) 116499

Contents lists available at ScienceDirect

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

Economies of scale in battery cell manufacturing: The impact of material

and process innovations
Lukas Mauler a, b, *, Fabian Duffner a, b, Jens Leker a, c
a
Institute of Business Administration at the Department of Chemistry and Pharmacy (IfbM), University of Münster, 48148 Münster, Germany
b
Porsche Consulting GmbH, 74321 Bietigheim-Bissingen, Germany
c
Helmholtz-Institute Münster (HIMS), 48149 Münster, Germany

H I G H L I G H T S

• Economies of scale are maximal when electrode roll-to-roll processes operate at full capacity.
• Balancing plant and electrode roll-to-roll capacities can reduce cost by >5 $ kWh− 1.
• Electrode roll-to-roll capacities depend on materials, cell design and production technology.
• Min. efficient plant sizes are currently below 2 and exceed 15 GWh year− 1 in future.
• Plant investments per GWh decrease, amounts for cost-efficient plant sizes increase.

A R T I C L E I N F O A B S T R A C T

Keywords: One key lever to reduce high battery cost, a main hurdle to comply with CO2 emission targets by overcoming
Battery cost generation variability from renewable energy sources and widespread electric vehicle adoption, is to exploit
Economies of scale economies of scale in battery production. In an industry growth currently supported by subsidies, cost-efficient
Minimum efficient scale
battery plant sizes are vital for the establishment of a self-sustaining industry and a transition into a long-term
Plant size
Battery manufacturing
climate-neutral society. For optimal plant sizing, no consensus has yet been achieved in the battery literature and
a detailed analysis of economies of scale is unavailable. To close this gap, a process-based cost modeling
approach is taken that reflects the determinants of economies of scale. In state-of-the-art, minimum viable plant
sizes are demonstrated to be below 2 GWh year− 1 but may exceed 15 GWh year− 1 in the future. This study finds
that economies of scale are related to the capacity of the roll-to-roll processes in electrode manufacturing and can
be maximized if the respective equipment operates at its capacity limit. This capacity depends on materials, cell
design and roll-to-roll process parameters. Since these parameters improve over time, increased plant sizes will
become necessary to achieve cost-efficient production levels. Required plant investments are found to decrease
on a per GWh basis, whereas significantly increased funds will become necessary to reach efficient plant sizes in
the future. Finally, implications are presented that support future battery cost reductions and a self-sustaining
market breakthrough of battery-powered products.

1. Introduction climate change either by paving the route to climate-neutral integrated

energy systems [1] or by supporting efficient storage of renewable en­
In recent years, battery technology has been identified as a key ergy [2] and replacing fossil fuels in vehicle traction [3] separately.
enabler for reducing CO2 emissions in the global endeavor to face However, under free market conditions, battery costs still constitute a

Abbreviations: CO2, Carbon dioxide; LFP, Lithium iron phosphate; LFP150, Lithium iron phosphate with practical discharge capacity of 150 mAh g− 1; LFP170,
Lithium iron phosphate with practical discharge capacity of 170 mAh g− 1; LMO, Lithium manganese oxide; LR-NMC, Lithium-rich lithium nickel manganese cobalt
oxide; NCA, Lithium nickel cobalt aluminum oxide; NMC, Lithium nickel manganese cobalt oxide; NMP, N-Methyl-2-pyrrolidone; PBCM, Process-based cost
modeling.
* Corresponding author at: Leonardo Campus 1, D-48149 Münster, Germany.
E-mail address: lmauler@uni-muenster.de (L. Mauler).

https://doi.org/10.1016/j.apenergy.2021.116499
Received 10 October 2020; Received in revised form 29 December 2020; Accepted 12 January 2021
Available online 25 January 2021
0306-2619/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
L. Mauler et al. Applied Energy 286 (2021) 116499

barrier to its widespread application and to its full support of the long- 2. Background
term goal to decarbonize energy supply [4,5] and transportation [3,6].
Nonetheless, demand forecasts for rechargeable batteries are surging 2.1. Economies of scale, minimum efficient scale and bottleneck processes
due to incentive schemes by policy makers [7,8] that support the market
penetration of battery-powered products [9]. For lithium-ion, the state- The sizing of production facilities is part of a company’s competitive
of-the-art technology for several years to come [10], annual global de­ strategy since economies of scale constitute a barrier for market entry
mand of 160 GWh in 2018 is expected to rise to more than 1000 GWh in [26,27]. Those economies of scale occur if the average cost curve de­
2030 [11]. In order to meet this demand, various large-scale plant creases with larger output volumes and arise from both, technical and
projects have recently been announced in the battery industry [12]. pecuniary advantages [28]. Technical economies of scale rely on gains
While these plants may operate profitably under current policy sce­ related to production activities and lead to a reduction of the ratio be­
narios, battery cost reductions are required to create sustainable busi­ tween manufacturing inputs and outputs. These advantages comprise a
ness models for industry players and minimize public financial support more efficient utilization of indivisible production elements, the use of
for climate change mitigation in the long-term. Apart from advances in advanced technology and more efficient deployment of labor [29].
material [10] and process [13] technology, one key lever to reduce Pecuniary economies of scale are related to administrative functions of a
battery cost is to exploit economies of scale by the optimal scaling of company and include higher bargaining power and favorable financing
manufacturing plants [14]. conditions [28]. Since this study intends to analyze the size of a single
Cost-optimal scaling of plants in the chemical and manufacturing manufacturing plant and pecuniary economies relate to the overall size
industry has been intensely discussed especially in the economic liter­ of a company’s activities, the following analysis focuses on technical
ature of the past century [15,16], revealing the importance of the pro­ economies of scale.
duction process for an accurate analysis [17,18]. In battery research, The minimum efficient scale is reached when the average unit cost
technical economies of scale have been mentioned in several publica­ curve becomes horizontal and substantial economies of scale are mate­
tions focusing on cost-efficient cell design [19], pack design [20], ma­ rialized [15]. To determine this volume threshold, several estimation
terial processing [21], production flexibility [22] and overall battery techniques have been developed such as engineering studies, accounting
cost estimation [23,24]. Thereof, Nelson et al., 2015; Sakti et al., 2015 studies and survivor tests [18,28]. Due to its focus on technical econo­
and Ciez and Whitacre, 2017 each mention cost-optimal volume mies of scale, an engineering study is used in this paper. In these studies,
thresholds, termed minimum efficient scales, ranging from 0.2 to 7.1 the production process is broken down into single steps and engineers
GWh of annual production, indicating a lack of consensus in battery estimate volume-dependent costs for individual process units. These
science. Furthermore, a detailed analysis of the impact of battery-related estimates are aggregated to an average cost curve of the entire plant
material, product and process parameters and respective advances on [17].
economies of scale has to date not been conducted, thus impeding a Economies of scale and minimum efficient scales are influenced by
future-oriented discussion of the effects of battery innovation from indivisibilities of production elements which create idle times repre­
research on industrial operations. senting the most relevant form of economic waste in manufacturing
The present paper represents the first study allowing for the discus­ [30]. To overcome these inefficiencies, the idle time of the equipment
sion of technical economies of scale by enabling an existing process- with the greatest capacity, referred to as the bottleneck, has to be mini­
based cost model [25] to analyze their underlying mechanics. Firstly, mized [31].
the model is adjusted to allow for variations in plant size and the inte­
gration of relevant parameters. Secondly, these parameters are investi­ 2.2. Economies of scale in energy technology literature and expected plant
gated in the academic literature regarding their state-of-the-art sizes in battery cell industry
characteristics and the potential of future innovations. Thirdly, in a
simulative approach, it is demonstrated that bottlenecks in the pro­ In line with the growing awareness of environmental and sustain­
duction process drive economies of scale in battery cell manufacturing ability issues throughout the society [32,33], the topic of economies of
and bottleneck processes are identified. Fourthly, the impact of in­ scale is gaining momentum within scientific literature about renewable
novations in processed materials, cell design and production processes energy conversion, energy distribution and energy storage. This can be
on the efficient production level is forecasted, ranges of minimum effi­ traced back to the reason that economies of scale are discussed as a main
cient scale are derived and compared between the current and potential lever for cost reduction strategies within these fields. When reviewing
future state of technology. Finally, the obtained results are set into the the available studies, they can be distinguished by their data foundation,
context of historical and expected battery plant sizes and implications the chosen background of the research question, and the objects studied.
for industry players are derived in order to support the exploitation of From a data foundation perspective, several studies analyze empirical
economies of scale. data such as accounting and plant performance data [34–36] and others
This study makes both, theoretical and practical contributions. With use engineering data [19,20,24]. With regard to the research question,
regard to scientific contributions, the study at hand provides a detailed some examine economies of scale as a component of an integrated cost
forecasting methodology for minimum efficient scales by combining reduction strategy [35,37] while others tackle the topic in a standalone
process-based cost modeling with plant-scaling literature. The field of approach [36,38,39]. In terms of the objects studied, the range reaches
battery cost research is enhanced by the introduction of state-of-the-art from biofuel production [37–39], to the production of renewable elec­
and future-oriented cost-optimal volume thresholds, the quantification tric power (e.g. wind [36] or solar photovoltaics [35]) and its storage
of material, product and process parameter effects and the impacts of [19,20,24] and distribution [34,40,41].
related innovations, all of which allowing for an informed discussion of The studies dealing with economies of scales within the field of en­
technological advances currently in research stage and their effect on ergy storage are mainly related to the topic of battery costs [42].
future plant projects. Furthermore, industry players are provided with Overarching, these studies can be categorized into those which focus on
measures aiming for the exploitation of economies of scale, thus the cost impact of state-of-the-art [43,44] and potential future battery
underlining the need to implement product and process innovations in materials [45,46], and those which take dedicated manufacturing cost
the industrial manufacturing process in order to complement the toolset into account. Studies of the latter category are most relevant for tech­
to further reduce battery costs in the future and support a self-sustaining nical economies of scale [28] and take two different approaches to
market breakthrough of battery-powered products. integrate economies of scale, namely regression-based and machine-
based methods.
The regression-based modeling of economies of scale has been

2
L. Mauler et al. Applied Energy 286 (2021) 116499

introduced to battery cost research by the publication of the battery seems to exist on optimal plant sizing in the industry. This can be
performance and cost model (BatPaC) [24] and has been applied in derived from Fig. 1 that provides an overview of selected projected
subsequent studies examining cost-efficient material processing [21], lithium-ion battery production capacities for the year 2025. Targeted
production flexibility [22] and overall cost estimation [23]. Based on a production volumes range from 7 to 76 GWh.
given pair of cost and output volume, this technique yields cost ap­
proximations resulting from variations in volume. In general terms, this 2.3. Cell manufacturing and roll-to-roll processes
cost can be expressed as a function C = C0 × (R/R0 )p where C is the unit
cost at a certain processing rate R, and C0 is the given unit cost at a given Battery cells consist of electrodes (cathodes, anodes) and separators
processing rateR0 . The power factor p defines the degree by which costs acting as a physical barrier to prevent electric shorting. Electrode sheets
decrease and economies of scale occur. Using this methodology, Nelson are connected by battery terminals, the cell is surrounded by a soft- or
et al., 2015 derive a minimum efficient scale at 202,000 battery packs of hard-case housing and soaked with liquid electrolyte to enable ionic
annual production, representing an energy output of 7.1 GWh year− 1. conduction. In the following, the state-of-the-art manufacturing process
Despite its ease of use, the regression-based method does not reflect is presented as described by Kwade et al., 2018 [13] to produce a
indivisibilities of production factors and does not provide accurate detail lithium-ion pouch cell and its individual steps are depicted in Fig. 2.
to measure technical economies of scale in a complex manufacturing In electrode production, the active electrode material slurry is mixed,
system with multiple interconnected discrete and continuous process coated on current collectors and dried. After a calendering step, elec­
steps. trodes are slit to the targeted width. In final drying, electrodes are dried
The machine-based method is applied by Sakti et al., 2015 and Ciez to limit residual moisture in the final cells. After cutting the electrodes to
and Whitacre, 2017, both using process-based cost-modeling (PBCM) to the desired length, discrete cathode and anode sheets are alternatingly
derive minimum efficient scales. PBCM is a widely accepted cost esti­ stacked on separator layers. The electrode-separator-assembly is inter­
mation method to translate technical parameters into manufacturing nally contacted and enclosed in a pouch cell housing. Subsequently, the
cost [47]. Sakti et al., 2015 derive minimum efficient scales in NMC cell is filled with liquid electrolyte, closed and stored. During cell con­
battery production of 0.2–0.3 GWh year− 1 [20]. Process-related tech­ ditioning, after final sealing of the cell, the formation procedure is
nical parameters and equipment cost are borrowed from BatPac [24] executed in order to build up the solid electrolyte interface protecting
and all types of manufacturing cost are driven by the number of required the anode from undesired side reactions. Finally, during aging, the cell is
equipment units. In reality, these numbers need to be integer values stored and monitored regarding unusual behavior to avoid quality
since further division results in the loss of purpose for most equipment issues.
types [29]. However, this criterion is relaxed for several process steps, In Section 4.1, it is demonstrated that the roll-to-roll processes
most notably for all roll-to-roll processes in electrode manufacturing represent the production bottleneck and that their cost is most sensitive
(Coating & Drying, Calendering, Slitting), leading to an underestimation to changes in production volume and hence, of highest relevance for the
of intermittent cost peaks and a lack of detail in the measurement of analysis of economies of scale. Therefore, these processes are described
economies of scale. Ciez and Whitacre, 2017 enhance the model of Sakti in more detail below and are graphically illustrated in Fig. 3.
et al., 2015 comparing LMO, NMC and NCA cells and derive a minimum The basis of the roll-to-roll process constitutes the active material
efficient scale for cylindric cell production of 1 GWh year− 1. An over­ slurry transferred from the mixing process. Its highest share represents
view of the described publication including an extract of the underlying the active material allowing for intercalation of lithium-ions during cell
parameter set is presented in Table 1. charging and discharging [48]. Commonly used active materials are
Within the available publications on the topic, three limitations have graphite for the anode and NMC, LFP and NCA for the cathode, ac­
been identified that impede an informed discussion and evaluation of counting for more than 90% respective market share in 2017 [3]. NMC
operating or future plant sizes. First, the range of minimum efficient and NCA cathodes are favorable for high cell energy density, whereas
scales from 0.2 to 7.1 GWh year− 1 indicates that no consensus has yet LFP has advantages regarding safety, cycle stability and state-of-charge
been achieved in the battery literature. Second, the quantification of the window [49]. Further, the slurry consists of polymer binders, con­
impact of processed materials, product and process parameters on ducting agents and solvents [50]. To allow for high processability, sta­
economies of scale remains unaddressed. Third, future advances in bility of the processed electrodes and performance of the final cell, the
technology have not yet been investigated regarding their effect on cost- slurry has to be balanced regarding viscosity, adhesion, corrosiveness,
optimal production levels in battery research. wetting behavior and homogeneity [51,52].
In addition to the lack of consensus in the literature, no agreement For Coating, the slurry is pre-metered by a pulsation free pump and,
most commonly, applied to the current collector using slot die coating
technology to allow for fast processing and high precision of coating
Table 1 properties [13,51]. This operation is usually conducted in two sequen­
Relevant publications, minimum efficient scales and extract of underlying tial steps as both sides of the collector require coating. In the coating
parameter sets. (and calendering) step, the desired electrode thickness is produced that
Unit Nelson et al. Sakti et al. Ciez and has a high impact on performance of the final cell [53]. In subsequent
Whitacre Drying, the solvent is evaporated using convective impingement dryers
Year of publication – 2015 2015 2017 divided into multiple zones with varying temperatures and drying rates
Integration of – Regression- Machine- Machine- to achieve porous electrode layers and reduce moisture [54,55]. During
economies of scale based based based Calendering, electrode porosity is reduced and adhesion between current
Minimum efficient GWh 7.1 0.2 – 0.3 1.0
scale obtained year− 1
collector and material film is strengthened by applying mechanical
Roll-to-roll speed m min− 1
10 10 10 pressure to the electrode. This is achieved by a roll press machine with
Electrode thickness µm 100 125 50–100 two counter-rotating rolls and leads to improved cell energy density and
(cathode) rate capability [50,56]. Since industrial electrode coils exceed common
Number of operating days 300 300 300
1 cell formats, foils are cut lengthwise in the Slitting process by roller
days year−
End-of-line scrap % 5 5 5 knives made of (non-conductive) ceramics [55,57].
rate
Cathode material – LMO, NMC NMC LMO, NMC, 3. Modeling approach
NCA
Cell format Prismatic Pouch Cylindrical
To quantify technical economies of scale, an existing model is
–

3
L. Mauler et al. Applied Energy 286 (2021) 116499

Fig. 1. Selected battery cell manufacturing plants announced for 2025 (see Appendix for related references).

Fig. 2. Essential manufacturing process steps of a lithium-ion pouch cell in a state-of-the-art production plant.

expanded by the inclusion of plant production volume as a central 3.1. Applied cost model
variable as opposed to a fixed input parameter and by the introduction of
additional parameters, namely idle time and capacity of production For this study, an existing model [25] using a process-based cost
equipment. These two adaptations allow for the identification of modeling technique is applied. On the highest level of the model, total
bottleneck processes, a detailed analysis of cost impacts due to changes unit cost CTotal is composed of the cumulative cost incurred by each
in production volume and for transparency regarding their underlying process step of battery cell manufacturing (see Fig. 2). This specific cost
mechanics. To explain the functionality of the enhanced model, first, j j
of process step j, CProcess equals the sum of related variable CVariable and
relevant cost parameters of the original model are described and the j j
fixed cost CFixed . The variable cost CVariable being the sum of cost elements
newly introduced parameters as well as their interaction with existing j j j
parameters is presented. Second, the parameter set used in this study is for Material CMaterial , Labor CLabor and Energy CEnergy , the fixed cost
j j
outlined. CFixed being the sum of cost elements for Machine CMachine , Building
j j j
CBuilding , Maintenance CBuilding , and Overhead COverhead . The respective

4
L. Mauler et al. Applied Energy 286 (2021) 116499

Fig. 3. Sequence of the electrode roll-to-roll processes in cell manufacturing in state-of-the-art, based on [13] and [55].

relations in mathematical terms are formulated by Eqs. (1)–(4). j

As described in Section 2.2, the number of machines NMachine for each
∑n j process step j should take integer values to account for indivisibilities of
CTotal = C (1)
j=1 Process equipment and hence, technical economies of scale. It can be calculated
based on
j
CProcess j
= CVariable j
+ CFixed (2) ⌈ ⌉
PV Saleable xPCEnergy
j
NMachine = (12)
(3)
j j j j
CVariable = CMaterial + CLabor + CEnergy CapjEffective

j
CFixed j
= CMachine j
+ CBuilding j
+ CMaintenance j
+ COverhead (4) where PV Saleable denotes the number of annual saleable units (number of
cells) and is used as a variable to evaluate cost impacts due to variations
Since all variables CTotal , CjProcess , CjVariable , CjFixed and CjElement (in which in production volume. PCEnergy denotes the energy content per unit (kWh
Element denotes cost elements Material, Labor, Energy, Machine, cell− 1). The effective machine capacity for process step j, CapEffective ,
j
Building, Maintenance, Overhead and j the related process step,
transforms equipment and process properties into a common unit (kWh
respectively) are calculated on a per unit basis (in this study $ kWh− 1),
or GWh) which allows for a comparison between process steps and is
these cost are derived by dividing the annual cost for each cost element
necessary to identify bottleneck processes according to the definition in
ACjElement by the number of annual saleable units (number of cells)
Section 2.1. It can be calculated as shown in Eqs. (13) and (14).
PV Saleable and the energy content per unit (kWh cell− 1) PCEnergy as shown
in Eq. (5). availMT j ( )
CapjEffective = = Mixing)
xPCEnergy x 1 − xAging&Final Control (j ∕ (13)
CT j
ACjElement
j
CElement = (5)
PV Salable xPCEnergy availMT j ( ) ( )
CapjEffective = xPCEnergy x 1 − xMixing x 1 − xAging&Final Control (j
CT j
Variables ACjElement (Element ∕
= Material) for each process step j are
= Mixing) (14)
calculated by mathematical Eqs. (6)–(11):

ACjLabor = N jMachine xN jLaborers per Machine xU jLabor xAPOT j (6) where availMT j is the annual available operating time of the machine
that can be derived from Eq. (15), CTj is the respective cycle time,
ACjEnergy = (ACjLabor + ACjMaterial )xSREnergy (7) xAging&Final Control is the end-of-line cell scrap rate, xMixing is the machine-
specific scrap loss for mixing.
ACjMachine = NMachine
j
xRjMachine (8)
availMT j = DPYxNOSx(OHS − UB − PB − UDj ) (15)
ACjBuilding = RjBuilding xNMachine
j
xFPjMachine (9) where DPY denotes the operating days per year, NOS the number of
shifts per day, OHS the operating hours per shift, UB the unpaid breaks
ACjMaintenance = ACjMachine xSRMaintenance (10) hours per shift, PB the paid break hours per shift and UDj the process-
specific unplanned downtime hours per shift. The cycle time CTj de­
ACjOverhead = (ACjMachine + ACjBuilding + ACjMaintenance )xSROverhead (11)
scribes the amount of time necessary to produce the equivalent of one
j
unit of output and is specific for each machine type. For the roll-to-roll
where for each process step j, NMachine denotes the number of machines, processes, it can be calculated according to (16) for cathode processing
j j
NLaborers per Machine the number of laborers required per machine, ULabor the and can be equally applied to anode processing.
unit cost of labor, APOTj the annual paid operating time, SREnergy the cathode length x number of cathodes per cell
j CT j = ⌊ ⌋ (j
surcharge rate for energy, RMachine the annualized equivalent of machine
j
working speed x working width

investment, RBuilding ($ m− 2) the annualized equivalent of building in­

cathode width

vestment, FPMachine (m2 machine− 1) the footprint per machine,

j = Coating&Drying, Calendering, Slitting) (16)
SRMaintenance the surcharge rate for maintenance and SROverhead the sur­ To analyze the efficiency of equipment in use for different production
charge rate for overhead. scenarios, annual idle time idleMTj of process step j is defined as the
Hence, all types of manufacturing cost (excl. material cost) are difference between the annual available operating time of all machines
driven either directly in (6), Eqn. (8), (9) or indirectly Eqn. (7), (10),
j
availMT j of process step j and the annual time required to produce the
(11) by the number of machines NMachine required for each process step. target volume reqMT j and can be derived from

5
L. Mauler et al. Applied Energy 286 (2021) 116499

j
idleMT j = availMT j xNMachine − reqMT j (17) specific capacity, discharge potential vs. lithium and crystallographic
density affect the calculation of effective roll-to-roll machine capacities
The annual time required to produce the target volume reqMT j can be as shown in Section 3.1 in Eqs. (13) and (16). For a given cathode
∏
calculated from (18) where nj (1 − xi ) denotes the product of yield rates thickness and cell dimensions, the cathode material determines on the
of process steps between process step j and the finished cell. one hand the required anode thickness and hence the number of anodes
and cathodes per cell. On the other hand, the choice impacts the energy
PV Saleable
reqMT j = CT j x ∏n (18) content per unit. For given cell dimensions, the lower the number of
j (1 − xi )
electrodes per cell and the higher the energy content per cell, the more
annual output roll-to-roll machinery can produce in terms of energy.
3.2. Input and simulation parameters Therefore, cathode materials from the NMC, LFP and NCA category are
chosen since they are among the most common cathode materials and
Once a bottleneck process is running at full capacity, a further in­ are expected to further increase their importance, by 2030 accounting
crease in output, the number of annual saleable units PV Saleable , can be for more than 95% of the market [11]. For the Base Scenario, cathode
achieved by (i) an increase of the number of equipment employed or (ii) materials are assumed that are currently in use in the automotive in­
an increase of related process capacity by the use of advanced technol­ dustry (NMC-811, NCA, LFP150, see Table 2) and for the Optimized
ogy. While (i) is reflected in the model by the number of machines in the Scenario, potential future advances are taken into account (LR (lithium-
j
bottleneck process NMachine , (ii) is achieved by increasing the respective rich) NMC, NCA-955, LFP170) [10,45].
j An increased electrode thickness results, for a given amount of active
effective equipment capacity CapEffective . Since the roll-to-roll processes
material and a given capacity, in a lower number of electrode sheets
are identified as the bottleneck in cell manufacturing in Section 4.1, required in the final cell. Since electrode width and length remain
product, process and production strategy-related parameters affecting constant, cycle time to produce the equivalent of one cell, as shown in
their capacities are analyzed. In Eqs. (13), (15) and (16) five major Eq. (16), can be reduced and therefore, roll-to-roll machine capacities
determinants of the roll-to-roll capacity, and hence economies of scale, can be increased. The relationship between cathode and anode thick­
can be identified (details of this analysis are included in Appendix A). ness, number of respective electrodes, capacity and weight is derived
These determinants represent the product parameters cathode material from the model of Wentker et al., 2019. For state-of-the-art high energy
and electrode thickness, the process parameters roll-to-roll speed and end- cells, cathode thickness ranges from 65 to 80 µm [10]. Regarding energy
of-line scrap rate and the parameter number of operating days that is density and cost, it is desirable to further increase this parameter since
related to production strategy. These parameters and their impact on active material share can be increased and material cost for non-active
Eqs. (13), (15) and (16) are outlined in detail in this section and are used components (e.g. current collector, separator) can be reduced [20].
as simulation parameters in the model. In order to also identify impli­ However, due to a trade-off between energy density and rate capability,
cations of potential advances in technology on minimum efficient scale, there are limitations for electrode thickness. Since for cathode thick­
two scenarios are investigated. A Base Scenario, intended to reflect state- nesses exceeding 100 µm, discharge capacity rapidly decreases for C-
of-the-art cell and production technology and an Optimized Scenario rates greater than 1, this thickness is set as an upper bound in the
including improvement potentials expected to materialize in the future. Optimized Scenario [58,59]. Additionally, increasing electrode thickness
To derive product parameters, the CellEst battery cost model of has an effect on the Drying step. The required drying time depends on
Wentker et al., 2019 [45] is used. Based on cell format, active material multiple conditions including temperature, drying rate, airflow and
configuration, cell dimensions and electrode thickness, the model cal­ pressure [54,57]. Nevertheless, for electrode thickness, a linear rela­
culates properties such as energy content, weight and number of elec­ tionship can be assumed with regard to necessary drying time [60,61]
trodes. Product parameters underlying this study are listed in Table 2. and is reflected in a proportionally increased dryer length, energy con­
For the present study, an automotive cell format is used since most of sumption and respective equipment investments in the model.
recently announced battery cell plants are dedicated to traction batteries The relevant process parameters applied are either calculated in the
for electric vehicles. model or rely on values found in recent battery-related literature and are
To investigate the effect of processing different cathode materials on listed in Table 3.
the optimal scaling of plants, relevant cathode materials and their To account for advances in process and production strategy,
properties are taken into account. Their inherent characteristics such as

Table 2
Product parameters of battery cells in the Base and Optimized Scenario.
Base Optimized

Cell Design Specifications Unit NMC-811 NCA LFP150 LR-NMC NCA-955 LFP170

Practical discharge capacity of cathode active material mAh g− 1 200 200 150 225 205 170
Energy Wh cell− 1 198 199 122 220 215 143
Energy density Wh kg− 1 278 277 176 321 304 212
Format Pouch
Anode material Graphite
Number of two-sided electrodes
1
Cathode pieces cell− 29 29 35 19 20 24
1
Anode pieces cell− 29 29 35 19 20 24
Electrode Thickness
Cathode µm 65 100
Anode µm 78 78 47 134 124 82
Width of electrodes*
Cathode mm 93
Anode mm 95
Length of electrodes*
Cathode mm 294
Anode mm 296
*
based on ISO cell format Pouch VIFB-/99/300.

6
L. Mauler et al. Applied Energy 286 (2021) 116499

Table 3 proportionally increased available machine time, as shown in Eq. (15),

Relevant process parameters in battery cell manufacturing. and hence, in higher machine capacities. The value of this simulation
Parameter description Source parameter represents a part of a plant’s production strategy with various
impacts regarding organizational and operational processes. For
j
CProcess Cost for process stepj Calculated, see Eq. (2)
manufacturing processes involving high-investment machinery, com­
j
NMachine Number of machines required for Calculated, see Eq. (12)
panies are interested in maximizing the respective value to increase
process stepj
return on investment. However, there are limitations, since companies
j
CapEffective Effective machine capacity for process Calculated, see Eqs. (13),
stepj (14) need to take into account shut down days regarding overall maintenance
xj Machine-specific scrap losses for [24] periods and public holidays. For both scenarios, values from battery-
process stepj related literature have been applied in the present study. For the Base
NOS Number of shifts per day [19] and Optimized Scenario, values of 300 days year− 1 [24] and 360 days
OHS Operating hours per shift [19] year− 1 [62], respectively, have been used as parameter specifications.
UB Unpaid break hours per shift Derived from [68] A more extensive, categorized overview of input parameters
PB Paid break hours per shift Derived from [20,68] including those with no impact on technical economies of scale, is given
UD Unplanned downtime Derived from [20] alongside related sources in Appendix B.
CT Cycle time (excluding roll-to-roll Derived from
processes) [24,48,57,58,69–71] 4. Results and discussion
CTroll− to− roll Cycle time (Coating & Drying, Calculated, see Eq. (16)
Calendering, Slitting)
4.1. Identification of bottleneck process & minimum efficient scale in
reqMTj Annual required machine time for Calculated, see Eq. (18)
process stepj battery cell production
availMTj Annual available operating time per Calculated, see Eq. (15)
machine for process stepj To identify the bottleneck process in cell manufacturing, machine
idleMTj Annual idle time for process stepj Calculated, see Eq. (17) capacities and average idle times for each process step are analyzed.
Results for different annual production volumes are shown in Fig. 4 for
the production of NMC-811 cells in the Base Scenario. The effective
parameters affecting roll-to-roll equipment capacities are varied and machine capacities of the roll-to-roll processes for cathode and anode
their specifications in the Base and Optimized Scenario and respective production with more than 1.8 GWh year− 1 are highest compared to all
sources in literature are outlined in Table 4. other process steps and hence represent the bottleneck as defined in
Increasing the roll-to-roll speed decreases the cycle time of the roll-to- Section 2.1. The continuous nature of these process steps results in
roll process steps, as shown in Eq. (16), and hence increases related significantly lower cycle times (below 2 s) to produce one equivalent of
effective machine capacities. This is due to the fact that a larger amount cell output and therefore, the number of machines required per process
of electrode area can be produced per unit of time and hence, the time step is lowest. The difference between anode and cathode roll-to-roll
required to manufacture a given number of electrodes per cell is capacities is solely induced by cell design reasons since slightly over­
reduced. For state-of-the-art electrode processing, working speeds of sized anode areas are applied (see Table 2) to reduce lithium plating
more than 25 m min− 1 are reported in recent literature [13] and for during charging [23]. The higher equipment capacity in the anode
thick electrodes (approx. 100 µm) working speeds of up to 100 m min− 1 mixing step compared to cathode mixing results from a substantially
seem achievable without film break-ups during coating [55]. As shorter dilution phase for the water-based anode slurry in relation to
described for electrode thickness above, roll-to-roll speed is affecting the NMP-based cathode slurry [48]. Thus, cycle time for anode mixing is
Drying step. In the model, required drying time is assumed to remain significantly lower and is the only process step exhibiting likewise ma­
constant for variations in roll-to-roll speed and hence, dryer length and chine capacities above the 1-GWh-threshold. Machine capacities of Final
related investments are proportionally increased. Drying and those for process steps related to cell assembly and condi­
Decreasing the end-of-line scrap rate results in a lower electrode area tioning are substantially lower and do not represent bottlenecks. The
required for one saleable cell, and hence increases effective machine cost effect of the described machine capacities is revealed by variations
capacities of the roll-to-roll processes, as shown in Eq. (13). The rate is in plant production volume and their impact on average process step-
measured in the process step Aging & Final Control and represents the related idle times. In the case of a production volume of 0.5 GWh
amount of rejected finished cells. Cell cost are most sensitive to varia­ year− 1 (volume case A), which is significantly below the bottleneck
tions of this parameter since, compared to upstream process steps, more capacity, idle times of roll-to-roll equipment are considerable and
material cost and value add have been accumulated throughout the induce high cell cost of 113.2 $ kWh− 1. This changes substantially once
process. If a cell does not exhibit properties according to expected production volume is increased slightly below roll-to-roll capacity
product requirements, it is rejected and is not accounted for in annual (volume case B) when bottleneck idle times become negligible, overall
saleable output. This typically applies for 5% of final cells and is mean idle time is reduced, and cell cost of 98.1 $ kWh− 1 are calculated.
therefore set as parameter value in the Base Scenario [24]. Since a total However, marginal increases of production volume above roll-to-roll
elimination of scrap seems unlikely, a minimum value of 1% is assumed capacities to 1.9 GWh year− 1 (volume case C) require additional ma­
as future potential and applied in the Optimized Scenario [19]. chinery and additional resources (equipment and building investments,
An increased number of operating days per year results in a labor, energy, maintenance and overhead) and cell cost grow by 5.2 to a

Table 4
Simulation parameters to vary effective machine capacity of roll-to-roll processes.
Category Simulation parameter Unit Base Scenario Optimized Scenario Reflected in model parameter Source
1
Process Roll-to-roll speed m min− 25 100 working speed Base: [13]
Optimized: [55]
Process End-of-line scrap rate % 5 1 xAging&Final Control Base: [24]
Optimized: [19]
1
Production strategy Number of operating days days year− 300 360 DPY Base: [24]
Optimized: [62]

7
L. Mauler et al. Applied Energy 286 (2021) 116499

Fig. 4. Machine capacities, percentage of idle time and cell cost for different plant production volumes for NMC-811 cells in the Base Scenario.

total of 103.3 $ kWh− 1. This is mainly due to considerable idle times in material cost. In the present study, due to its focus on a single plant and
the roll-to-roll processes where one of two equipment units produces technical economies of scale, material cost (excluding scrap cost which
only marginal volume and costly resources are not efficiently utilized. is allocated to respective process cost) are assumed linear in relation to
The fact that cell cost sensitivity to variations in production volume production volumes, since economies of scale in this category arise from
mainly is caused by roll-to-roll processes, can be observed in Fig. 5 bargaining power instead of higher efficiency [63]. In contrast, the costs
where, for cells using NMC-811 chemistry, cell cost is split into unit cost for value-adding process steps are driven by the number of respective
dedicated to value-adding process steps, supporting processes and machines and follow a saw-toothed trend. Distinct intermittent peaks in

Fig. 5. Cell cost development as a function of production volume of NMC-type cells.

8
L. Mauler et al. Applied Energy 286 (2021) 116499

recurring steps are mainly caused by roll-to-roll processes as indicated effect of simulation parameters on effective machine capacities of the
by the solid black box on the bottom representing the calculated roll-to- roll-to-roll-processes as depicted in Fig. 6, in which the difference be­
roll machine capacities for NMC-811 chemistry. For these cells, of the tween Base (NMC-811 cells) and Optimized Scenario (LR-NMC cells) is
above-described cost delta between volume case B and C, 4.6 $ kWh− 1 explained. The process simulation parameter increased roll-to-roll speed
alone are dedicated to their process cost increase. However, moderate results in a fourfold increase in effective roll-to-roll machine capacities
peaks at lower production volumes between 0.6 and 1.3 GWh year− 1 can and therefore implies the highest shift in minimum efficient scale. The
be observed in cell assembly where contacting, enclosing and filling second process simulation parameter decreased end-of-line scrap rate only
reach the limits of their first and second equipment unit, respectively. has a marginal impact on machine capacities. The effects of the product
The described peaks flatten with increased production and the curve parameters increased electrode thickness and material change from NMC-
tends to take a constant shape since average idle times decrease and 811 to LR-NMC and enhance roll-to-roll capacities by the factor of
associated cost are distributed among a larger amount of output units. 1.54 and 1.10, respectively. The simulation parameter increased number
This constant value represents an ideal cost level and can be calculated of operating days as part of a company’s production strategy is enhancing
by eliminating the ceiling (round up) function in the number of ma­ minimum efficient scale by a multiple of 1.2. Although the individual
chines formula (see Eq. (12) in Section 3.1), thereby relaxing the impacts of simulation parameters on minimum efficient scale are
requirement for integer numbers of equipment machinery. For NMC-811 considerable, their aggregate effect in the Optimized Scenario is even
in the Base Scenario, this results in a value of 96.8 $ kWh− 1. Since de­ more significant and results in an 8.5-fold increase of minimum efficient
viations from these cost levels are reduced to 1% already at plant pro­ scale between both scenarios. This is due to the fact that variations in
duction levels equal to the effective roll-to-roll unit capacity (volume single input parameters of the underlying effective capacity formula
case B, 1.8 GWh year− 1), this first local cost minimum is defined as the amplify in a multiplicative manner (see Eq. (13) in Section 3.1). For
minimum efficient scale in the present study. Meaning on the one hand, additional information, the material change from LFP150 to NMC-811 with
that below this volume threshold substantial economies of scale are an almost two-fold increase in machine capacity for the Base Scenario, is
forgone and cell production therefore is inefficient and uncompetitive also included in the figure.
and, on the other hand, further upscaling (preferably in integer multi­ The resulting cost curves for all investigated cells and their minimum
ples of the roll-to-roll capacity) generates only moderate process cost efficient production scale in the Base and Optimized Scenario are pre­
improvements and is connected to significantly higher investments and sented in Fig. 7. The cost curves for cells in the Base Scenario are plotted
associated risks. Following this approach for the NMC-811 cell in the as solid lines with distinct peaks in the graph. Their respective minimum
Base Case, a five-fold increase in plant size to 9 GWh year− 1 yields a efficient scales span the range in which cell manufacturing becomes cost
process cost improvement of <1 $ kWh− 1 enabled by additional in­ efficient in the state-of-the-art oriented Base Scenario (red shaded area
vestments in the order of 500 $ million. As can be seen from the dotted on the left). Likewise to NMC-811 cells, a local cell cost minimum for
cost curve in Fig. 5, representing LR-NMC cells in the Optimized Sce­ NCA cells of 100.9 $ kWh− 1 can be identified at production volumes of
nario, again, the respective local cost minimum can be identified slightly 1.8 GWh year− 1, which can be attributed to their similarity regarding
below the roll-to-roll capacity represented by the white box at the bot­ energy content and number of electrodes per cell (see Table 2). In
tom. The improvement of all parameters described in Section 3.2 leads contrast, the significantly lower energy content of LFP150 cells and their
to a significant shift of the minimum efficient scale to 15.3 GWh year− 1, higher number of electrodes lead to reduced roll-to-roll capacities and
where cell cost of 75.7 $ kWh− 1 can be reached. hence, a local cost minimum of 144.9 $ kWh− 1 occurs already at lower
This difference in minimum efficient scales can be attributed to the production volumes of 0.9 GWh year− 1. Again, the correlation between

Fig. 6. Minimum efficient scales for different simulation parameter configurations.

9
L. Mauler et al. Applied Energy 286 (2021) 116499

Fig. 7. Cell cost development as a function of production volume in Base and Optimized Scenario.

cell cost and roll-to-roll unit capacities is graphically revealed and local product technology and the willingness to undertake high investments
cost minima can be identified slightly below their full capacity. For the in progressive equipment technology.
Optimized Scenario, that includes potential advances in materials and
production technology described in Section 3.2, three phenomena can
be observed in the cost curves represented by dotted lines. First, cost 4.2. Comparison with the literature & industry
minima can be observed at an increased production scale (e.g. for NCA-
955 cells at a plant production volume of 14.4 GWh year− 1). This effect To put obtained results into context with the literature and industry,
is directly connected to the impact of parameter improvements that the derived minimum efficient scales are compared to values from
increase roll-to-roll unit capacities and has been explained above. Sec­ previous scientific publications on the topic and to historical and fore­
ond, intermittent cost peaks are less distinct compared to the Base casted plant-specific production volumes of cell manufacturing
Scenario, originating from the fact that, due to increased roll-to-roll companies.
capacities, process cost for idle times can be distributed among a In Fig. 8, values from previous battery-related literature are marked
larger amount of output units. Third, lower cell costs can be identified as grey solid polygons in their year of publication and compared to the
for each material category between Base and Optimized Scenario. This derived minimum efficient scale obtained in the state-of-the-art oriented
can be attributed to the effect of the described simulation parameters on Base Scenario, indicated by the lower red horizontal bar. The mentioned
the ideal cost level such as material and process cost improvements but values for minimum efficient scale in publications of Sakti et al. in 2015
are not the focus of the present study. The range of minimum efficient (0.2 to 0.3 GWh year− 1) and Ciez and Whitacre (1 GWh year− 1) in 2017,
scale in the Optimized Scenario is defined by the respective volume both investigating NMC cells, are significantly lower than the obtained
thresholds for LFP170 and NCA-955 at annual plant production volumes value of for NMC-811 cells in our Base Scenario (1.8 GWh year− 1).
of 8.0 and 15.3 GWh year− 1, respectively and is represented by the red Considering these values in our model yields significantly higher cell
shaded area on the right side of the graph. Since the achievement of cost of 140.9 and 102.8 $ kWh− 1, respectively, resulting in substantial
future advances in the defined simulation parameters involves uncer­ cost deviations of + 44% and + 5% compared to our cell cost at mini­
tainty, a sensitivity analysis is conducted showing that variations of mum efficient scale (98.1 $ kWh− 1). While these volume thresholds
parameter improvements of 10% imply variations of less than 8% for the certainly have been valid for factories at the time of publication, newly
derived minimum efficient scale (see Appendix C). established plants using updated equipment technology and processing
Consequently, advances in product technology, production processes more advanced materials cannot produce at sufficiently low cost on
and production strategy strongly effect roll-to-roll unit capacities and these production levels and are therefore not competitive. In contrast,
hence imply the necessity of increased plant sizes to fully exploit Nelson et al., 2015 derive a higher minimum efficient scale of 7.1 GWh
economies of scale in the future. Therefore, in order to outperform their year− 1 that yields slightly lower cell cost of 97.4 $ kWh− 1 coinciding
competitors, companies need both, the ability to industrialize innovative with a local cost minimum at four instead of one roll-to-roll units in
electrode processing. This fact confirms our finding that further

10
L. Mauler et al. Applied Energy 286 (2021) 116499

Fig. 8. Development of plant production volumes and comparison with derived minimum efficient scales.

economies of scale can be materialized when increasing plant produc­ Since pecuniary economies of scale, most notably the bargaining power
tion volume by multiples of the roll-to-roll capacity. However, since for the costly battery raw materials, are related to a company’s overall
further cost improvement by this volume quadrupling remains below size of operations as opposed to the size of a single plant, we deem the
1%, we consider the minimum threshold reached earlier. following reasons more important for higher plant-specific scaling de­
In order to validate the derived results with regard to industry re­ cisions. First, in the ongoing global expansion of the industry, a com­
ality, the development of the average plant size in lithium-ion industry is pany’s number of key experts required for a successful plant set-up and
plotted as a dashed line in Fig. 8. Average plant sizes have significantly ramp-up is limited and can be more productive if appointed to one single
increased from 1.3 GWh in 2015 to 5.6 GWh in 2020 and are expected to site. Second, operating multiple roll-to-roll equipment lines at one site
further increase to a value of 12.1 GWh in 2025 [64]. This trend towards decreases the risk of full production outages and allows for local risk
higher manufacturing capacity per plant reflects the desire to exploit pooling regarding spare parts and maintenance activities. Third, in­
technical economies of scale due to advances in product and equipment vestments in higher automation in cell assembly and cell conditioning
technology and confirms the shift of minimum efficient scale over time become more attractive at higher production volumes. However,
reflected in our findings for Base and Optimized Scenario (the latter increasing plant sizes can in turn lead to higher transportation cost, since
indicated by the upper red horizontal bar). Further, the average battery the average distance to customers becomes longer [63]. Regarding
plant is currently operating above minimum efficient scale whereas, additional automotive-related plant projects in 2025, most of the plan­
depending on the timing of technological improvements, future plants in ned production volumes are located above the minimum efficient scale
their entirety might not be able to fully exploit economies of scale in the of the Optimized Scenario with a stronger concentration at the top end
manufacturing process. of the red marked range. Even though detailed plant planning assump­
In addition to the average plant size, several plants mentioned in the tions and further expansion plans are not completely available, ac­
context of the manufacture of automotive traction batteries have been cording to our results, most of these plants will have the ability for cost-
considered in the analysis depicted in Fig. 8. On the one hand, historical efficient production.
and expected developments of plant production volume of the largest
battery plant per region already in operation (in terms of output volume
2019) are plotted as solid black lines. On the other hand, selected 4.3. Implications for industry players
automotive-related battery plant projects and their expected production
volume in 2025 are indicated by stars. Further details regarding meth­ In order to identify implications for players in the battery industry,
odology and related references can be found in Appendix D. Regarding the derived results of the Base and Optimized Scenario are compared.
the plants already in operation, a clear trend towards increasing plant Additionally, a third scenario based on already outdated product and
production volumes can be observed that is expected to continue in the process parameters is set up and calculated to illustrate the chronolog­
future. The four regionally largest plants in operation are already ical development of minimum efficient scales in the industry. To
exceeding the derived minimum efficient scale of the Base Scenario in examine financial and competitive consequences of this development for
2019 and are all expected to surpass the threshold of the Optimized companies, the plant investments are discussed for both, a normalized
Scenario until 2023. The currently largest three plants of CATL, Tesla/ per GWh level and the required amount to reach minimum efficient
Panasonic and LG Chem with forecasted production volumes of 76, 76 scale. The derived minimum efficient scales, required investment
and 62 GWh in 2025 may in fact produce several times the volume amounts and an extract of the underlying parameter set are shown in
necessary to exploit the largest share of technical economies of scale. Fig. 9.
The plant sizes necessary to exploit economies of scale have been

11
L. Mauler et al. Applied Energy 286 (2021) 116499

Fig. 9. Development of minimum efficient scales and required plant investments.

steadily increasing until today and this trend is expected to continue as reduction in cell costs, promote a market breakthrough of battery-
discussed in Section 4.1. This development is induced by advances in powered products less reliant on incentive schemes by policy makers
product and process technology that are increasing bottleneck capacities and hence support an economically viable transformation into a climate-
of the roll-to-roll processes in electrode manufacturing. These techno­ neutral society.
logical improvements have two opposed effects on required investments
for future plant projects. 5. Conclusion
On the one hand, on a per GWh basis, the required investments are
decreasing by 36% from 85 to 54 million $ GWh− 1 between the state-of- The present study applies a process-based cost modeling technique to
the-art-oriented Base and the future-oriented Optimized Scenario. This identify cost-efficient plant sizes in battery cell manufacturing. By taking
cost advantage mainly results from the increase in electrode thickness, into account advances in battery materials, equipment technology and
the processing of advanced materials, both resulting in higher energy production strategy, it is shown that from a technical perspective, cost-
output for all process steps, and the increase of roll-to-roll speed efficient volume thresholds are below 2 GWh year− 1 in state-of-the-art
resulting in higher capacities in investment-intensive electrode plants but may exceed 15 GWh year− 1 in the future. Thereby, the
manufacturing equipment. This more favorable utilization of equipment study contributes to the linkage between theoretical research in energy
outweighs the investment increase induced by longer dryers. Therefore, technologies with its practical application in the industry.
cell manufacturers should closely monitor technological advances Regarding theoretical contributions, the study at hand combines
currently in research stage in order to early implement innovative process-based cost modeling with the field of minimum efficient plant
product technology and progressive equipment technology and hence scales, thereby complementing existing techniques of engineering
create competitive cost advantages by an increased exploitation of studies by a methodology enabling first, a detailed analysis of the rela­
economies of scale. tionship between scale and cost in complex production processes and
On the other hand, the plant investment required to reach the min­ second, an analytical translation of technological advances into shifts in
imum efficient scale and to allow for the exploitation of economies of cost-efficient production levels. By this means, it allows for an ex-ante
scale is significantly increasing by 440% from 152 to 820 million $ in the impact analysis of innovation on the industrial manufacturing process
future. In an already concentrated industry dominated by a few com­ and the forecasting of optimal plant sizes that is less subjective to expert
panies, this trend raises barriers for market entry with regard to capital opinions, can be conducted without prior knowledge of detailed ac­
requirements and economies of scale [26,49]. Especially new entrants counting data and historical plant performance, and can be replicated
and smaller players, with lower potential for internal financing, should for the manufacturing of various products.
therefore seek long-term investors allowing for plant investments of Regarding practical contributions, the present study applies the
sufficient scale or co-invest in respective plant sizes with original developed methodology to battery cell manufacturing and transforms
equipment manufacturers in order to share financial risks. This tendency knowledge of material, cell design and process innovations gained in
is already observable in the industry where partnerships have recently academia into implications for cost-optimized plant scaling decisions in
been announced between Northvolt AB and Volkswagen AG [65], industry. This is achieved by first, the quantification of minimum effi­
Farasis Energy and Daimler AG [66], and SAFT and PSA [67], cient scales in cell production based on processed materials, product and
respectively. process parameters, second, the identification of electrode roll-to-roll
The described implications can help market leaders and smaller in­ processes as major drivers of technical economies of scale in the
dustry players in fully materializing economies of scale by the imple­ manufacturing process, third, the determination of state-of-the-art and
mentation of innovative technology, thereby creating more profitable the forecasting of future minimum efficient scales based on respective
business models and safeguarding their economic survival. Further, the parameter configurations and fourth, the derivation of proposed mea­
establishment of mutually beneficial strategic partnerships, allowing for sures for industry players to allow for the implementation of cost-
long-term capacity utilization and secured supply, creates a direction efficient production levels.
towards less concentrated market structures. Both, economies of scale In our quest for measures to decarbonize society in the long-term,
and increased competition in the battery market will contribute to a reduced battery cost can play a pivotal role in supporting the efficient

12
L. Mauler et al. Applied Energy 286 (2021) 116499

storage of renewable energy and replacing fossil fuels in vehicle traction. decisions. Appl Energy 2020:116269. https://doi.org/10.1016/j.
apenergy.2020.116269.
The study at hand provides transparency on and guidance to the
[7] Zhang J, Cho H, Luck R, Mago PJ. Integrated photovoltaic and battery energy
exploitation of economies of scale in battery manufacturing, thereby storage (PV-BES) systems: An analysis of existing financial incentive policies in the
supporting a key lever for the battery cost reductions that are required US. Appl Energy 2018;212:895–908.
for a self-sustaining market breakthrough of battery-powered products. [8] Feng S, Magee CL. Technological development of key domains in electric vehicles:
Improvement rates, technology trajectories and key assignees. Appl Energy 2020;
The presented approach has several limitations that should be 260:114264.
addressed in future research. First, only technical economies of scale [9] Hache E, Seck GS, Simoen M, Bonnet C, Carcanague S. Critical raw materials and
have been investigated since they are deemed decisive for the scaling transportation sector electrification: A detailed bottom-up analysis in world
transport. Appl Energy 2019;240:6–25.
decision of a single plant. However, on the level of overall company size, [10] Schmuch R, Wagner R, Hörpel G, Placke T, Winter M. Performance and cost of
pecuniary economies of scale such as higher bargaining power for costly materials for lithium-based rechargeable automotive batteries. Nat Energy 2018;3:
raw materials and process equipment are expected to play a substantial 267–78.
[11] Pillot C. Impact of the EV Market Growth on Lithium-ion Batteries and Raw
role and have not been considered in the underlying study. Second, this Materials Supply 2019-2030; 2020.
study has focused on technological advances affecting roll-to-roll pro­ [12] Duffner F, Krätzig O, Leker J. Battery plant location considering the balance
cesses in electrode manufacturing since they have been identified as, between knowledge and cost: A comparative study of the EU-28 countries. J Clean
Prod 2020;264:121428.
and in the opinion of the authors will remain, bottlenecks in the fore­ [13] Kwade A, et al. Current status and challenges for automotive battery production
seeable future. Yet, since technology development is certainly not technologies. Nat Energy 2018;3:290–300.
limited to these process steps, changes in bottlenecks are possible and [14] Nykvist B, Nilsson M. Rapidly falling costs of battery packs for electric vehicles. Nat
Clim Chang 2015;5:329–32.
are not reflected in the present study. Third, for processed materials,
[15] Silberston A. Economies of Scale in Theory and Practice. Econ J 1972;82:369–91.
product and process parameters, only evolutionary advances have been [16] Moore FT. Economies of Scale: Some Statistical Evidence. Q J Econ 1959;73:232.
considered. However, potential leaps to more revolutionary technology [17] Pratten CF. A Survey of the Economies of Scale. in Research on the costs of. ‘Non-
such as dry electrode processing and the manufacture of all-solid-state or Europe’ 1988;2:11–166.
[18] Haldi J, Whitcomb D. Economies of Scale in Industrial Plants. J Polit Econ 1967;75:
post-lithium-ion batteries have not been taken into account. Fourth, this 373–85.
study has focused on battery cell dimensions for automotive applica­ [19] Ciez RE, Whitacre JF. Comparison between cylindrical and prismatic lithium-ion
tions. While the general findings of the study can be transferred to other cell costs using a process based cost model. J Power Sources 2017;340:273–81.
[20] Sakti A, Michalek JJ, Fuchs ERH, Whitacre JF. A techno-economic analysis and
cell formats and applications, a detailed analysis of battery cells for optimization of Li-ion batteries for light-duty passenger vehicle electrification.
other battery-powered products such as stationary energy storage or J Power Sources 2015;273:966–80.
consumer electronics has not been conducted. In summary, these gaps [21] Ahmed S, Nelson PA, Gallagher KG, Susarla N, Dees DW. Cost and energy demand
of producing nickel manganese cobalt cathode material for lithium ion batteries.
will also affect the competitiveness of plants and companies and should J Power Sources 2017;342:733–40.
therefore be further investigated in future studies. [22] Nelson PA, Ahmed S, Gallagher KG, Dees DW. Cost savings for manufacturing
lithium batteries in a flexible plant. J Power Sources 2015;283:506–16.
[23] Gallagher KG, Nelson P. A. Manufacturing Costs of Batteries for Electric Vehicles.
Funding In Lithium-Ion Batteries. Elsevier; 2014, 97–126. https://doi.org/10.1016/B978-0-
444-59513-3.00006-6.
This research did not receive any specific grant from funding [24] Nelson P, Gallagher K, Bloom I, Dees DW. Modeling the Performance and Cost of
Lithium-Ion Batteries for Electric-Drive Vehicles, 2nd ed.; 2012, 116. http://www.
agencies in the public, commercial, or not-for-profit sectors. cse.anl.gov/batpac/files/BatPaC ANL-12_55.pdf.
[25] Duffner F, Mauler L, Wentker M, Leker J, Winter M. Large-scale automotive battery
CRediT authorship contribution statement cell manufacturing: Analyzing strategic and operational effects on manufacturing
costs. Int J Prod Econ 2021;232:107982.
[26] Porter ME. Competitive Strategy: Techniques for Analyzing Industries and
Lukas Mauler: Writing - original draft, Writing - review & editing, Competitors. Simon and Schuster; 2008. p. 3–33.
Methodology, Investigation, Visualization, Data curation, Software, [27] Powell DM, et al. The capital intensity of photovoltaics manufacturing: barrier to
scale and opportunity for innovation. Energy Environ Sci 2015;8:3395–408.
Formal analysis. Fabian Duffner: Conceptualization, Investigation.
[28] Shepherd WG, Shepherd JM. The Economics of Industrial Organization. Waveland
Jens Leker: Supervision, Project administration. Press; 2003.
[29] Morroni M. Knowledge, Scale and Transactions in the Theory of the Firm.
Cambridge University Press; 2006, p. 163–176. https://doi.org/10.1017/CBO9780
Declaration of Competing Interest 511617232.
[30] Georgescu-Roegen N. Process in farming versus process in manufacturing: a
The authors declare that they have no known competing financial problem of balanced development. In: Economic problems of agriculture in
Industrial Societies. Palgrave Macmillan; 1969. p. 497–533.
interests or personal relationships that could have appeared to influence [31] Mir-Artigues P, González-Calvet J. Funds, flows and time: An alternative approach
the work reported in this paper. to the microeconomic analysis of productive activities. Springer Science & Business
Media; 2007. p. 87–144.
[32] Marcelino-Sádaba S, González-Jaen LF, Pérez-Ezcurdia A. Using project
Appendix A. Supplementary material management as a way to sustainability. From a comprehensive review to a
framework definition. J Clean Prod 2015;99:1–16.
Supplementary data to this article can be found online at https://doi. [33] Olawumi TO, Chan DWM. A scientometric review of global research on
sustainability and sustainable development. J Clean Prod 2018;183:231–50.
org/10.1016/j.apenergy.2021.116499. [34] Mydland Ø, Haugom E, Lien G. Economies of scale in Norwegian electricity
distribution: a quantile regression approach. Appl Econ 2018;50:4360–72.
References [35] Pillai U. Drivers of cost reduction in solar photovoltaics. Energy Econ 2015;50:
286–93.
[36] Dismukes DE, Upton GB. Economies of scale, learning effects and offshore wind
[1] Mathiesen BV, et al. Smart Energy Systems for coherent 100% renewable energy
development costs. Renew Energy 2015;83:61–6.
and transport solutions. Appl Energy 2015;145:139–54.
[37] de Jong S, et al. Cost optimization of biofuel production – The impact of scale,
[2] Luo X, Wang J, Dooner M, Clarke J. Overview of current development in electrical
integration, transport and supply chain configurations. Appl Energy 2017;195:
energy storage technologies and the application potential in power system
1055–70.
operation. Appl Energy 2015;137:511–36.
[38] Sanchez DL, Callaway DS. Optimal scale of carbon-negative energy facilities. Appl
[3] Ballinger B, et al. The vulnerability of electric vehicle deployment to critical
Energy 2016;170:437–44.
mineral supply. Appl Energy 2019;255:113844.
[39] Skovsgaard L, Jacobsen HK. Economies of scale in biogas production and the
[4] Coester A, Hofkes MW, Papyrakis E. Economic analysis of batteries: Impact on
significance of flexible regulation. Energy Policy 2017;101:77–89.
security of electricity supply and renewable energy expansion in Germany. Appl
[40] Mydland O, Kumbhakar SC, Lien G, Amundsveen R, Kvile HM. Economies of scope
Energy 2020;275:115364.
and scale in the Norwegian electricity industry. Econ Model 2020;88:39–46.
[5] Parra D, Patel MK. The nature of combining energy storage applications for
[41] Tovar B, Javier Ramos-Real F, de Almeida EF. Firm size and productivity. Evidence
residential battery technology. Appl Energy 2019;239:1343–55.
from the electricity distribution industry in Brazil. Energy Policy 2011;39:826–33.
[6] Burd JTJ, Moore EA, Ezzat H, Kirchain R, Roth R. Improvements in electric vehicle
battery technology influence vehicle lightweighting and material substitution

13
L. Mauler et al. Applied Energy 286 (2021) 116499

[42] Duffner F, Wentker M, Greenwood M, Leker J. Battery cost modelling: A review [57] Kaiser J, et al. Prozess- und Produktentwicklung von Elektroden für Li-Ionen-
and directions for future research. Renew Sustain Energy Rev 2020;127. Zellen. Chemie Ing Tech 2014;86:695–706.
[43] Kalhammer FR, Kopf BM, Swan DH, Roan VP, Walsh MP. Status and Prospects for [58] Wood DL, Li J, Daniel C. Prospects for reducing the processing cost of lithium ion
Zero Emissions Vehicle Technology. Report of the ARB Independent Panel 2007; batteries. J Power Sources 2015;275:234–42.
2007. prepared for the California Air Resources Board, April 13. http://www. [59] Zheng H, Li J, Song X, Liu G, Battaglia VS. A comprehensive understanding of
ehcar.net/library/rapport/rapport045.pdf. electrode thickness effects on the electrochemical performances of Li-ion battery
[44] Amirault J, et al. The Electric Vehicle Battery Landscape : Opportunities and cathodes. Electrochim Acta 2012;71:258–65.
Challenges; 2009. http://large.stanford.edu/courses/2011/ph240/zhong [60] Wood DL, et al. Technical and economic analysis of solvent-based lithium-ion
1/docs/BatteryBrief_final.pdf http://www.funginstitute.berkeley.edu/sites/defaul electrode drying with water and NMP. Dry Technol 2018;36:234–44.
t/files/The Electric Vehicle Battery Landscape - Opportunities and Challenges.pdf. [61] Westphal BG, Kwade A. Critical electrode properties and drying conditions causing
[45] Wentker M, Greenwood M, Leker J. A Bottom-Up Approach to Lithium-Ion Battery component segregation in graphitic anodes for lithium-ion batteries. J Energy
Cost Modeling with a Focus on Cathode Active Materials. Energies 2019;12:504. Storage 2018;18:509–17.
[46] Berckmans G, et al. Cost Projection of State of the Art Lithium-Ion Batteries for [62] Schnell J, Knörzer H, Imbsweiler AJ, Reinhart G. Solid versus Liquid—A Bottom-
Electric Vehicles Up to 2030. Energies 2017;10:1314. Up Calculation Model to Analyze the Manufacturing Cost of Future High-Energy
[47] Field F, Kirchain R, Roth R. Process cost modeling: Strategic engineering and Batteries. Energy Technol 2020;8:1901237.
economic evaluation of materials technologies. JOM 2007;59:21–32. [63] Morroni M. Production Process and Technical Change. Cambridge University Press;
[48] Yoshio M, Brodd RJ, Kozawa A. In Yoshio M, Brodd RJ, Kozawa A, editors, 1992, p. 141–64. https://doi.org/10.1017/CBO9780511599019.
Lithium-Ion Batteries. New York: Springer; 2009, p. 163–79. https://doi.org/10. [64] Roskill. Personal Communication. Roskill Inf. Serv. Ltd.; 2020.
1007/978-0-387-34445-4. [65] Financial Times. VW and Goldman lead $1bn investment in Swedish battery
[49] Zubi G, Dufo-López R, Carvalho M, Pasaoglu G. The lithium-ion battery: State of project; 2019. https://www.ft.com/content/f94de008-8d0c-11e9-a1c1-51b
the art and future perspectives. Renew Sustain Energy Rev 2018. https://doi.org/ f8f989972.
10.1016/j.rser.2018.03.002. [66] Financial Times. Daimler to take stake in Chinese battery cell maker Farasis; 2020.
[50] Bockholt H, Indrikova M, Netz A, Golks F, Kwade A. The interaction of consecutive https://www.ft.com/content/baf2f5d9-a485-4e3a-9412-14d7c89c5ed9.
process steps in the manufacturing of lithium-ion battery electrodes with regard to [67] SAFT. SAFT Company Website; 2020. https://www.saftbatteries.com/media-resou
structural and electrochemical properties. J Power Sources 2016;325:140–51. rces/press-releases/launch-pilot-plant-manufacture-european-batteries-electric
[51] Hawley WB, Li J. Electrode manufacturing for lithium-ion batteries—Analysis of -vehicles.
current and next generation processing. J Energy Storage 2019;25:100862. [68] Fuchs ERH, Bruce EJ, Ram RJ, Kirchain RE. Process-based cost modeling of
[52] Dreger H, Bockholt H, Haselrieder W, Kwade A. Discontinuous and Continuous photonics manufacture: the cost competitiveness of monolithic integration of a
Processing of Low-Solvent Battery Slurries for Lithium Nickel Cobalt Manganese 1550-nm DFB laser and an electroabsorptive modulator on an InP platform. J Light
Oxide Electrodes. J Electron Mater 2015;44:4434–43. Technol 2006;24:3175–86.
[53] Zhao R, Liu J, Gu J. The effects of electrode thickness on the electrochemical and [69] Petri R, Giebel T, Zhang B, Schünemann J-H, Herrmann C. Material cost model for
thermal characteristics of lithium ion battery. Appl Energy 2015;139:220–9. innovative li-ion battery cells in electric vehicle applications. Int J Precis Eng
[54] Font F, Protas B, Richardson G, Foster JM. Binder migration during drying of Manuf Technol 2015;2:263–8.
lithium-ion battery electrodes: Modelling and comparison to experiment. J Power [70] Tagawa K, Brodd RJ. In Yoshio M, Brodd RJ, Kozawa A, editors, Lithium-Ion
Sources 2018;393:177–85. Batteries. New York: Springer; 2009, p. 181–94. https://doi.org/10.1007/978-0-38
[55] Schmitt M. Slot die coating of lithium-ion battery electrodes. Karlsruhe Institute of 7-34445-4.
Technology; 2015, p. 140–54. https://doi.org/10.5445/KSP/000051733. [71] Knoche T, et al. In situ visualization of the electrolyte solvent filling process by
[56] Meyer C, Bockholt H, Haselrieder W, Kwade A. Characterization of the calendering neutron radiography. J Power Sources 2016;331:267–76.
process for compaction of electrodes for lithium-ion batteries. J Mater Process
Technol 2017;249:172–8.

14