Interactions of production strategies and a production system’s perfor-
mance
An analysis of the influence of production strategies on the final unit
costs of lithium-ion batteries
Tim Hettesheimer
Fraunhofer Institute for Systems and Innovation Research (ISI)
Tim.Hettesheimer@ isi.fraunhofer.de
Frank Schultmann
Institute for Industrial Production (IIP), Karlsruhe Institute of Technology (KIT)
Future-oriented industrial companies are reliant on the continuous expansion and ad-
justment of their business units to new markets. The profitability of such new business
units is therefore often linked with high investments and characterized by high uncer-
tainty regarding the competitiveness of the produced product and the company’s pro-
duction system. In this context, a company’s production strategy plays a crucial role
because it can be seen as a blueprint for the development of the production system. A
production strategy consists of four sub-strategies that address the location of a pro-
duction system, the way capacities are increased, the product portfolio manufactured
and the depth of vertical integration. Any modification or combination of these sub-
strategies affects the production system at multiple places and in different ways, mak-
ing it extremely difficult to predict the overall effect on the performance of the produc-
tion system. Similarly, the resulting unit costs of a product produced by this production
system are difficult to predict and it is hard to say whether the company would be
competitive in a new market.
This article analyzes the interactions between different production strategies and their
impacts on the final unit costs of lithium-ion batteries. A stock and flow model is con-
structed based on a qualitative system dynamics model that describes these interac-
tions. The model shows the impacts of the analyzed production strategies in different
scenarios. Finally, first insights into the benefits and disadvantages of specific sub-
strategies can be given that depend on the overarching market for electric vehicles.
System Dynamics, Manufacturing strategy, Production strategy, Lithium-ion battery,
Cost model, Technology strategy, Capacity strategy, Location strategy, Strategy of ver-
tical integration
1 Introduction
The world's increasing volume of road traffic is a catalyst for climate change and a
drain on already scarce resources. In order to curb these effects, electric mobility will
play an important role in the future. Successfully managing this transformation process
is a major challenge for the automotive industry. Internal combustion engines and re-
lated components will lose their relevance and be replaced by new components. In this
context, the lithium-ion battery (LiB) is particularly important for electric mobility be-
cause it determines the costs and range of electric vehicles (Schade et al. 2014). In
countries like Germany, the automotive industry plays a major role from an economic
perspective, accounting for about 21% of the total turnover of the manufacturing sec-
tor and directly employing about 700,000 people (German Federal Statistical Office
2014). It is therefore vital for Germany to navigate a successful transformation from
the conventional internal combustion engine to the electric motor. The LiB is a key
component here with very high added value. But compared to the USA, up to now, no
significant LiB cell manufacturing exists in Germany. Many players are reluctant to act
because of the high investments needed. A major reason for the lack of activity might
be that it is unclear which production strategy will achieve a competitive price level.
By using system dynamic analysis, the influence of different production strategies on
the resulting production costs of a company will be examined. A bottom-up approach
is applied to focus on the effects on the performance of the company’s production
system.
In the following, the sub-strategies making up a production strategy are described and
why it is difficult to foresee the effect of a production strategy on production system
performance. Then a short literature review will be given of existing similar approaches
in this field. Chapter 3 describes the main subsystems of the model, before two differ-
ent production strategies are analyzed under different scenarios in chapter 4. The pa-
per closes with conclusions and an outlook in chapter 5.
2 How production strategies influence the performance of a production
system
The performance of a production system as a central unit of a manufacturing company
is determined to a large extent by the production strategy used for its development
(Zaepfel 2000). How this performance is determined and how it is influenced by a pro-
duction strategy is described by the following illustration in the form of a causal dia-
gram of the dynamic problem behavior.
Technology strategy
Capacity strategy
Location strategy Production rate of LiB
+
a
: me
Production Materials and Employees
capacities components
+
Unit costs
\
Strategy of vertical
integration
Figure 2-1: Dynamic hypotheses for the model
The production of LiB can be seen as the driving force within the depicted qualitative
model and is triggered by the demand for LiB resulting from the sales market. On the
one hand, the planned production determines the need for employees, working mate-
rials and components as well as the necessary production capacities. On the other
hand, the feasible maximum production is determined by the availability of these key
elements. Furthermore, acquiring employees, materials and production systems re-
sults in costs which have a direct impact on the final unit costs and thus the competi-
tiveness (from an economic point of view) of the LiB product. Thereby, a comparative
increase in the price level leads to a reduction in the demand for LiB.
Production strategies may directly influence one or more of the variables mentioned.
The capacity strategy offers the possibility to increase capacity in a rather defensive
(follow strategy) or aggressive way (lead strategy). While the first strategy is intended
to avoid overcapacities in production, the lead strategy aims at always having enough
capacity to satisfy demand and thus avoid any penalty for late or non-delivery.
The location strategy influences the production-relevant costs (e.g. labor costs), while
it has no effect on the overall demand for LiB, because the market for LiB is assumed
to be global.
The vertical integration strategy concerns the possibility to produce the whole LiB or to
buy LiB cells from a supplier. The first option is accompanied by the construction of an
entire cell production system, which is the most capital-intensive manufacturing step
in LiB production. This investment is not necessary if the cells are sourced. One disad-
vantage of purchasing the cells from outside means the margin of the supplier has to
be considered when calculating the unit costs.
Finally, the last production strategy - the technology strategy - addresses the product
portfolio to be produced. A company can concentrate on high energy batteries as only
used in BEV (battery electric vehicles) and PHEV (plug-in hybrid electric vehicles) or
additionally build high power batteries as used in HEV (hybrid electric vehicles). The
high demand for HEV and thus for high power batteries means there is a possibility for
a high production rate right from the outset (the demand for high energy LiB starts to
increase at a later time) and thus to gather experience with LIB production. High power
LiBs have a lower specific capacity (kWh) although the production steps are more or
less the same as for high energy LiB. This means that the same investments in produc-
tion systems have to be made as for high energy LiB (which have a higher specific ca-
pacity), which results in comparatively higher unit costs (€/kWh).
So far, there is no system dynamics approach to simulating the production costs of LiB
that considers the performance of a production system. This is even more the case if
performance is linked to the decision about production strategies. So while there are
hardly any approaches on the battery level, there are many more studies for the over-
arching market for electric vehicles (EV), which triggers the demand for LiB (e.g.
Struben & Sterman 2007, Keith et al. 2012 or Gomez et al. 2013).
3 System structure and basic elements
In the following, two subsystems will be described in detail. The “Sales market” and
the “Production system”. The sales market simulates the diffusion of three different
types of electric vehicles (EV) using a Bass diffusion model (adopted from Sterman
2000 and Bass 1969, based on Hettesheimer & Lerch 2013). The following Figure 3-1
shows the structure of the subsystem “Sales market”.
Change in marketshare _<Obsolence
Productionof hi ‘based on unit costs HEv>
Orders PHEV
BM ev \
Battery capacity
soancey
= HEV
Danae
per HEV Orders BEV
Innovation Innovation Imitation
coefficent HEV coefficient HEV coefficent PHEV coefficient PHEV
Imitation Battery capacity
coefficient BEV
needed BEV
Innovation
coefficent BEV
BEV
Battery capcacity
Demand Bk
<Obsolence
HEV> Timeto
obsolence BEV
‘Obsolence BEV
Figure 3-1: Structure of the sales market subsystem
The subsystem is further divided into three different parts, representing the different
types of EV: HEV, PHEV and BEV. In each part, EV diffusion is simulated with regard to a
specific customer potential and the specific coefficients of innovation and imitation.
The overall demand for LiB is determined considering the EV-specific battery capacity.
The three different parts simulate the global demand. The demand for a single compa-
ny results from this company’s market share in this global demand.
The “Orders HEV”, “Orders PHEV” and “Orders BEV” constitutes the central input to
the subsystem “Production system” and is the trigger for the simulated production of
LiB there. The model structure of the subsystem is depicted in Figure 3-2. The model
structure is inspired by the supply chain model depicted in Sterman (2000). For better
clarity and readability, the model is further separated into parts A and B and enlarged
in Figures 3-3 and 3-4.
In part A, it can be seen that the orders simulated in the “sales market” are fed into
the subsystem in step 1. In step 2, the corresponding number of battery systems is
determined by the orders for LiB. Then, based on the orders for battery systems, the
number of battery modules, battery cells and finally battery electrodes is submitted.
After that, in step 3 (part B), a comparison is made between the numbers of orders for
electrodes and the resulting desired production rate for electrodes. The maximum
possible production rate is determined restricted by the installed production capacity.
Penalty total
Penalty delivery
Orders total
delay
ete ata
Orders (cee
system Ordering rate alias = Ordering rate
battery module battery cell
fags
WP battery J stock battery | wip (ee
production rate pa Delivery rate ‘module | production rate Delivery rate
battery system battery module battery module
ae, battery call
*
-<Maxlmum production rate
module production
“Maximum production rate capadty>
‘employee module>
“Nordering rate L_system
battery system
‘Stock battery
Delivery rate
battery system
Maximum production
<<Maximum production rate
system producti
-<Maximum production rate peeaty>
employee systems>
Show restriction
Show restriction
system eae
Figure 3-3: Subsystem production system - Part A
While a penalty has to be paid for orders which cannot be met due to limited capacity,
those orders which can be theoretically produced form the inputs to step 4. This step
represents the stages of production from the battery electrode to the final battery
system as an aging chain. The maximum production rate of each stage is determined
by a potential lack of production capacities, materials or employees. This comparison
happens in step 5. If buying the cells from a supplier is chosen as the production strat-
egy, then the comparison starts at the stage of battery modules and the previous steps
are not taken into account when determining any restrictions. The final battery system
is assembled at the end of the aging chain. Any delivery delays between the incoming
order and the delivery of the final product are calculated. If the delivery time is longer
than the targeted delivery time of 1 month, an additional penalty payment is included
depending on the length of the delay.
> penalty tora
Delay battery
cll
‘Orders battery
call Deliveries ‘Ordering rate
battery cell battery electrode
Penalty lack of
capacty
Orders battery
electrode
‘Ordering rate Deliveries battery
battery call
i
Stock
Delivery rate production rate peliveryrate electraden| production ate
battery cell battery cell electrode battery electrode rae
oe chin
Production rat -
besiming QA
Maximum pr
‘Maximum produétion
a electrode
-<Maximum production rate
bgttery material>
Max. production rate
< -<Maximum production rate
‘eal production capacity
electrode production
capadty>
«<Max. production rate,
employee cel -<Maximum production‘rate
employee electrode>
Show restriction
Show restrictor
te electrode
Figure 3-4: Subsystem production system - Part B
Thus the final unit costs are calculated by taking into account the costs for the depreci-
ations for the production system, for materials and additional components, the labor
costs, other costs (such as for warranty, Selling, general and administrative expenses,..)
and the penalty payments for late or non-delivery.
4 Simulation runs and tests
Various simulation runs and tests are conducted in order to test the dynamic hypothe-
sis and the effects of different production strategies as well as to obtain a deeper un-
derstanding of the system’s behavior.
The underlying data about the potential customers for each type of EV (HEV, PHEV and
BEV) are derived from a model developed for the “Office of Technology Assessment at
the German Bundestag”, which is based on the European transport model “ASTRA”
(see Fermi et al. 2012 and Schade et al. 2014). The specific amount of production ca-
pacities, employees and materials needed to produce a specific number of LiBs are
based on the BatPaC model (ANL 2012). The materials, production systems and em-
ployees are calculated bottom-up from 18 materials or components and 23 production
steps for the two different kinds of LiB (high power or high energy batteries). The simu-
lation period covers 15 years (180 months) until the year 2030. Post 2030, other bat-
tery technologies might be more favorable (Thielmann 2012).
Two different types of production strategies will be examined within the scope of this
paper. On the one hand, strategy 1 produces high energy LiB as well as high power LiB,
production capacities are constructed according to a lead strategy, the site is situated
in the USA and the whole battery is produced (including the battery cell). On the other
hand, strategy 2 focuses more on avoiding investments that would be necessary for
the production of two types of batteries or a whole battery system. Capacities are in-
creased carefully and, in line with this cost-oriented thinking, the plant is situated in
China. According to the composition of the production strategy, the two strategies will
be defined as follows: Strategy Scenario_location_Capacity strategy Technology
strategy Vertical integration. Thus strategy 1 is shortened as:
S1_X_USA_Lead_HEV_All and strategy 2: S2_X_CN_Follow_NoHEV_NoCell.
The production strategies are tested under two alternative scenarios for the diffusion
of EV and thus for the market demand for LiB: “Oil age”, a rather pessimistic scenario
(from the point of view of electric mobility) and “Electrical age”, a quite optimistic one.
The two scenarios differ in the number of potential customers for EV and the degree of
diffusion (represented in the Bass diffusion model by the coefficient of innovation and
imitation).
41 Results of the oil age scenario
Figure 4-1 shows the diffusion results for the different EV types on the left-hand side
and the resulting orders on the right-hand side. It is apparent that the diffusion of the
different types of EV takes place at different times and with different intensities.
Figure 4-1: Diffusion of EV (left) and orders for LiB (right) in the oil age scenario
In the oil age scenario, the overall demand and thus the orders for LiB are rather small
compared to the electrical age scenario depicted in Figure 4-3. In strategy 1, the effect
of the additional production of HEV batteries is visible in the early increase in orders.
But this effect is soon outpaced by the orders of strategy 2. The comparatively low unit
costs of strategy 2 depicted in Figure 4-2 are the reason for this strong increase. The
costs decrease rapidly so the produced LiB soon reach a competitive cost level. This
makes it possible to increase the company’s market share and thus the orders.
€/kWh
2,000
90 180
Time (Month)
S1_OA_USA_LEAD_HEV_All —————— $2_OA_CN_Follow_NoHEV_NoCell
Figure 4-2: Unit costs per kWh in the oil age scenario
In strategy 1, the unit costs do not decrease as rapidly. This is because production ca-
pacities have to be established to meet the increasing demand for high power batter-
ies (see Figure 4-1). Therefore investments in the production system are necessary so
depreciation increases as do the unit costs. The decrease in unit costs in strategy 2
continues until around month sixty, when the demand for PHEV and BEV begins to
take off so that new production systems have to be implemented. This leads to the
peak in the unit costs. This effect is enhanced by the defensive follow-capacity strate-
gy. Because of the lack of sufficient production capacity, penalty payments have to be
made which raise unit costs even more. In the long term, however, both strategies
show a comparable cost level.
4.2 Results for the electrical age scenario
In the electrical age scenario, the demand for LiB and therefore the orders are approx-
imately ten times higher than in the oil age scenario. The development of orders for
strategy 1 and strategy 2 follow the same pattern as in the oil age scenario. Strategy 1
has a higher number of orders at the beginning due to the HEV-batteries and is soon
outpaced by strategy 2.
Figure 4-3: Diffusion of EV (left) and orders for LiB (right) in the electrical age scenario
With regard to the unit costs depicted in Figure 4-4, both strategies show the same
pattern as in the oil age scenario at the beginning. But in contrast to the oil age scenar-
io, the effect of missing production capacities in strategy 2 around month sixty is much
greater due to the stronger rise in market demand.
€/kWh
2,000
1,500
1,000
500
90 180
Time (Month)
S1_EA_USA_LEAD_HEV_All —————— S2_EA_CN_Follow_NoHEV_NoCell
Figure 4-4: Unit costs per kWh in the electrical age scenario
Additionally in this scenario, which is characterized by strong and rapid diffusion of EV,
the effect of the vertical integration strategy becomes obvious: While the unit costs
under strategy 1 decline continuously, the unit costs under strategy 2 stay more or less
on the same level (in the long term). A major reason for this (besides the penalty pay-
10
ments) is that, for each battery system sold, the supplier margin for cells has to be
considered. Thus strategy 1 becomes more and more favorable with an increasing
number of orders.
5 Conclusions and outlook
This paper presents a system dynamic model to simulate and analyze the impact of
different production strategies on the unit costs of a lithium-ion battery system. Two
different strategies were examined under a rather pessimistic and an optimistic sce-
nario. The model can capture the interactions between market development, the in-
fluence of the different strategies and the performance of the production system in
the form of the resulting unit costs.
It becomes obvious in both scenarios that a technology strategy including the produc-
tion of high power batteries results in comparatively higher unit costs to start with.
Furthermore, it can be stated that a defensive capacity strategy is not recommendable,
because the savings due to avoided overcapacities are more than compensated by
penalty payments for late or non-delivery. With regard to the strategy of vertical inte-
gration, buying the cells from a supplier offers an advantage when the market is still
young, because no investments have to be made. But in the long term, especially un-
der strong market growth, this strategy does not permit cost reductions below a cer-
tain level, because the supplier’s margin has to be factored in.
This study provides a first insight into the complex interactions between the decision
to pursue a particular production strategy and the resulting performance of a compa-
ny’s production system. Obviously, under real life conditions, the interactions are
much more complex and the decision for or against a specific production strategy is
not purely an economic one but also takes soft criteria into account (e.g. protection of
intellectual property,..). Furthermore, the impact of a production strategy on the other
two strategic production targets of time and quality has to be considered. These as-
pects should be the subject of further investigation when trying to improve the meth-
odology.
6 References
ANL (Hg.) (2012): Modeling the Cost and Performance of Lithium-lon Batteries for Elec-
tric-Drive Vehicles. Final Report. Argonne National Laboratory (ANL). Chicago,
USA.
Bass, F. M. (1962): A New Product Growth Model for Consumer Durables, in: Manage-
ment Science, Vol. 15, No. 5, pp. 215-227.
Fermi F., Fiorello D., Krail M., Schade W. (2012): The design of the ASTRA-EC model.
Deliverable D4.1 of ASSIST (Assessing the social and economic im-pacts of past
and future sustainable transport policy in Europe). Project co-funded by Euro-
pean Commission 7th RTD Programme. Fraunhofer-ISI, Karlsruhe, Germany.
11
German Federal Statistical Office (2014): Produzierendes Gewerbe. Beschaftigte, Um-
satz und Investitionen der Unternehmen und Betriebe des Verarbeitenden Ge-
werbes sowie des Bergbaus und der Gewinnung von Steinen und Erden.
Gomez, J.; Jochem, P.; Fichtner, W. (2013): EV market development pathways—An
application of System Dynamics for policy simulation. In: Electric Vehicle Sym-
posium and Exhibition (EVS27), 2013.
Hettesheimer, T.; Lerch, C. (2014): Future Trends of the automotive Li-lon Battery Sup-
ply Chain in Germany — Dynamic effects on raw materials and employment. In:
31st international conference of the System Dynamics Society 2013. Cambridge,
Massachusetts, USA, 21 - 25 July 2013. Red Hook, NY: Curran, S. 1392-1420.
Keith, D.; Sterman, J.; Struben, J. (2012): Understanding Spatiotemporal Patterns of
Hybrid-Electric Vehicle Adoption in the United States, Proceedings of the 30th
International System Dynamics Conference, St. Gallen, Switzerland.
Schade, W.; Zanker, C.; Kihn, A.; Hettesheimer, T. (2014): Sieben Herausforderungen
fiir die deutsche Automobilindustrie. Strategische Antworten im Spannungsfeld
von Globalisierung, Produkt- und Dienstleistungsinnovationen bis 2030.
Sterman, J. D. (2000): Buisness dynamics. Systems thinking and modeling for a complex
world. Boston: Irwin/McGraw-Hill.
Struben, J.; Sterman, J. (2007): Transition Challenges for Alternative Fuel Vehicle and
Transportation Systems. In: SSRN Journal. DOI: 10.2139/ssrn.881800
Thielmann, A.; Sauer, A.; lsenmann, R.; Wietschel, M. (2012): Technologie-Roadmap
Energiespeicher fur die Elektromobilitat 2030. Hg. v. Fraunhofer-Institut fiir Sys-
tem- und Innovationsforschung ISI. Karlsruhe.
Zaepfel, G. (2000): Strategisches Produktions-Management: Oldenbourg Verlag.
