Future Trends of the automotive Li-Ion Battery Supply Chain
in Germany
Dynamic effects on raw materials and employment
Tim Hettesheimer! & Christian M. Lerch!
1Fraunhofer Institute for Systems and Innovation Research ISI
Competence Center Industrial and Service Innovations
Breslauer Str. 48
76139 Karisruhe, Geamany
Phone: +49/721 6809-320
Email: Tim Hettesheimer@isi.fraunhofer.de
Researchers as well as car manufacturers expect a greater market penetration of elec-
tric vehicles (EV) in Germany and worldwide for the next years and decades. The Li-
thium-ion battery represents the core technology of electrically driven cars, because it
defines technical and economical limitations. Consequently, the knowledge about this
technology and the ability to produce it are the relevant drivers for a successful EV dif:
fusion. Due to the expected great market penetration, the battery supply chain has to
grow extrenely in future or otherwise the economic potential gets lost. But the prospec-
tive impacts of this fast growing demand on the supply chain and its environment can be
hardly estimated. Therefore, the aim of this paper is to build up a first simulation mod-
el, which demonstrates the development of the automotive Li-ion battery supply chain in
Germany over the next decades. By means of various EV diffusion scenarios we want to
show expected future trends and their impacts on the battery supply chain, on raw ma-
terials, production capacities and on employment effects.
System Dynamics, Electric Vehicles, EV, Closed Loop Supply Chain, Li-ion Batteries,
Diffusion
1 Introduction
Due to climate change and the global shortages of resources there has been a rethinking
of individual mobility basing on combustion engines. These extemal drivers on the one
hand and the strong technological improvements of batteries on the other hand led to the
introduction of first electric vehicles (EV) during the last years (see Kley et al. 2011).
But researchers as well as car manufacturers expect a greater market penetration for the
next years and decades.
The most promising energy storage technology at the moment is the Lithium-ion (Li-
ion) battery. This type of battery describes the core technology of electric vehicles, be-
cause it defines technical limitations, as power and range (e.g. Tate et al. 2008), and
economical limitations, as the purchase price (eg. Nemry et al. 2009; Brooker et al.
2010). Consequently, the knowledge about this technology and the ability to produce it
are relevant drivers for a successful EV diffusion. Beyond the technical know-how, this
technology holds high economic advantages as value and employment creation. Sum-
ming up, the battery supply chain will play a central role for the car manufacturing in
future.
The automotive industry is one of the most relevant sectors in Germany. The aim of the
German govemmoent is to establish Germany as lead market and provider for electric
mobility (see NPE 2011). If there will be such a great market penetration as expected,
this leads to a high demand of Li-ion batteries, too. Consequently, the battery supply
chain has to grow extremely or otherwise the economic potential gets lost.
The effects and impacts of this fast growing demand on the supply chain and their sub-
systems can be hardly estimated. Due to feedback loops, delays and the long time hori-
zon it seems impossible to give valid information about the existing economic potential,
the need of raw materials and recycling activities and the development of employment
effects. Moreover, there are uncertainties of the diffusion of electric cars and of a high
number of further influencing factors, such as labor productivity, recycling quotes and
production capacities. Finally, there exists a high variety of future trends for the battery
supply chain in Germany and its environment.
Therefore, the aim of this paper is to build up a first simulation model, which demon-
strates the automotive Li-ion battery supply chain in Germany over the next decades.
By means of various scenarios we want to show expected future trends and their im-
pacts on the supply chain and its environment, as
= the need of raw materials and the potential of recycling critical materials in fu-
ture and
= the development of employment in Germany and the involving need for quali-
fied staff.
To reach this aim, the paper is structured as follows: after the introduction, chapter 2
works out the dynamic hypotheses, which describes the linkage of the electric car diffu-
sion, the battery supply chain, raw materials and employment. Afterwards (chap. 3), the
causal loop is transferred into a first stock and flow model, which is filled with data
from actual studies afterwards. In chapter 4, we simulate various diffusion scenarios for
electric cars taken from actual research reports and analyze their influences on the bat-
tery supply chain and consequently, on raw materials and employment. Finally the pa-
per closes in chapter 5 with a conclusion and an outlook for further research.
2 Linking electric car diffusion and the Li-ion battery supply chain
As literature shows, System Dynamics is a practical tool for analyzing future trends of
closed loop supply chains or to demonstrate the market diffusion of innovative technol-
ogies. Moreover, system dynamics is already used for analyzing the dynamics of elec-
tric car diffusion. The basic ideas of those topics can be described by the following:
Closed loop supply chains are modeled and simulated in lots of system dynamics con-
tributions and publications. A high variety of topics is regarded under different points of
views, as e.g. replacement part management (e.g. Spengler/Schréter 2003; Lehr/Milling
2009), remanufacturing processes (e.g. Rodrigues et al. 2012) or stock management
(Mutallip/Y asarcan 2012). A basic structure for recycling processes, as eg. for re-
Placement parts, is delivered by Fleischmann 2001. The author describes the generic
idea of a dynamic recycling management: A new product generation is introduced to the
market and after a specific time horizon the first products of this generation become
obsolete. Consequently, these products may be recycled and delivered back to the mar
ket again. Those material feedbacks exist all over the whole life cycle of new product
generations and continue a long time after the last product was introduced to the market.
Accordingly, the recycling rates or rather the material feedbacks depend on the market
stock and may be used to decrease the need of raw materials. Following this basic idea,
the simulation of closed loop supply chains is an adequate instrument for analyzing the
material feedback during the battery production life cycle.
Also the diffusion of new technologies is a relevant topic in system dynamics research.
Generic structures are shown for example in Sterman 2000, Morecroft 2007 or Warren
2008. Other authors transferred the famous diffusion model of Bass (1969) into a sys-
tem dynamics structure and are then able to analyze the influences of single factors on
the diffusion progress (see Maier 1998; Milling, Maier 1996). Moreover, those models
give recommendations for the innovation management of companies aiming at a faster
or greater diffusion of new products under competition (e.g. Milling 2002). As a conse-
quence of this system dynamics diffusion research, there exist some simulation models,
which deal explicitly with the diffusion of electric vehicles, but under various points of
views (eg. Keith et al. 2012; Wachter/Spengler 2012).
But up to now, there seems to be a gap considering the diffusion of electric vehicles and
their impacts on the supply chain of batteries and its environment. Because of the high
relevance of this topic for policy makers and manufacturing companies in Germany, we
will combine existing approaches and present a first automotive Li-ion battery supply
chain model for Germany.
As a first step, this paper works out the linkage between electric car diffusion, the de-
mand and production of batteries and the involving dynamic effects on raw materials
and employment. Therefore, Figure 2-1 shows a causal loop diagram describing the
dynamic hypotheses of the battery supply chain model with sub-sectors.
At first it gets obvious, that the model uses four sub-systems: the sales market, for de-
fining the diffusion of electric cars, the production and supply chain and the supply
markets, for analyzing the dynamic effects on raw and recycled materials as well as on
employment.
Production &
z 1
!
Recycling 1
adi 1
Supply Chaii
+ 1) ; \\ supply Chain i
Batteries. i
( =
N
- A
¢ + 4
I Market I
' diffusion I
I
1
i
+ Asi) - Sales market :
i
7 1
Figure 2-1: Dynamic hypotheses for the battery supply chain model
The sales market describes the diffusion of electric cars. Therefore, this sub-system
holds the market potential, defining the saturation limit of the new technology. The
market diffusion represents the number of sold EV. The higher the market potential or
rather the saturation limit, the higher the number of sold electric cars over time and the
smaller the remaining potential of customers. Consequently, this balancing loop (B1)
describes the market diffusion by means of an S-shaped growth.
Depending on the progress of the market diffusion, the order rate as well as the produc-
tion rate increase over time (Sector “Production & Supply Chain"). The manufacturing
of batteries is a time-consuming process, but a few weeks or months later they ran
through the whole supply chain and are delivered to the market. This reinforcing loop
(R1) demonstrates the interconnection between the demand of batteries and the produc-
tion rate. Afterwards the battery is in usage and after a longer time of maybe a few years
the obsolete battery gets recycled. Consequently, the growth of the market diffusion
leads to a few year time-delayed increase of the recycling rate.
The supply market of materials is separated into two sectors: the raw materials and the
recycled materials. This leads to the opposing trend that a higher recycling rate leads on
the one hand to a higher availability of recycled materials and on the other hand to a
decreasing demand of raw materials. Consequently, these connections close two loops
(R2 & B2). While recycled materials are pushed by an increasing production rate (R2),
the need for raw materials is slowed down by a higher recycling rate.
Finally, there exists a linkage to the employee market. This supply market is directly
linked with the production rate. A higher production rate leads to a higher demand of
employees. This demand gets slowed down by the employee potential, which describes
the number of available persons with an adequate qualification. The growth of the pro-
duction rate leads to an increase of the demand of employees and hence, to an exhaus-
tion of the employee potential. The only possibility to raise the pool of qualified persons
is to modify the extemal driven qualification rate. Accordingly, the pool gets exhausted
by a greater production rate and gets raised by a higher qualification rate. The linkage
between battery supply chain, employees and qualification is shown by the loops R3
and B3.
3 Stock and flow structure and interaction of the sub-systems
The simulation model has a simple system structure and consists, as already mentioned
above, of four sub-systems. The sub-systems and their interaction are illustrated in Fig-
ure 3-1.
Raw material
Sales market —) market
Sa 1
Employee > Production &
market Supply chain
Figure 3-1: Overview of the sub-systems and their interaction
The “Sales market” represents the superior sub-system, which triggers the development
in the other three sub-systems by the customers demand for electric vehicles (see Figure
3-2). The model distinguishes between three types of electrically driven cars: The hy-
brid electric vehicle (HEV), the plug-in hybrid electric vehicles (PHEV) and the battery
electric vehicles (BEV). All types of cars influence the Li-ion battery supply chain. For
this purpose, Bass's diffusion model for demand development was used as initial point
for model and modified to take repeat purchases into consideration (adopted from Ster
man 2000; Bass 1969). A certain customer potential exists at product introduction,
which is exhausted over time by the demand. "Customers" describes the client stock.
“Obsolescence”, the period in which an electric vehicle ages and must be
when the client becomes a potential customer again (see Sterman 2000; Stummpfe 2003).
Alpha and beta are factors for describing the diffusion process (see Bass 1969) for ini-
tial purchases. Furthermore, the model shows two specifics compared to the origin
models of Sterman 2000 and Stumpfe 2003. The “Obsolence” does not only lead to the
repurchase of an electric vehicle’s battery, it also represents the input for the battery
recycling of aged batteries. Another special character of the sub-system is the decline of
the HEV life-cycle (“Potential customers”). This is based on the assumption, that HEV
are only seen as an interim solution.
Recyclable
Tame to Loman sli g
absolence ——| v a “Savona Tire to
cS cine | ballets batieries | PHIEV batt Pieites
recyelmg rate
ng rate
[ee
[ce] ‘Obsolence FEV
Customers} Demand for
EV batteries im kWh j .
oy HEV 7K aN
cs: a =) capt
Alpha < ‘Production ‘AiphspHEy Beta PHEY
volume PHIEV
Beta Z
Production
volure BEV Capacity BICV
Alpha BEY Beta BEV
Potential
cuestarners
Demand BEV gas BEV
Customers
BEV
—
[Reqyenpie
BEV batteries Datteries
recycling rate
Figure 3-2: Development of the demand for electric vehicles and the required battery
Capacity
The demand for each kind of electric vehicle, together with their specific battery capac-
ity (in kWh) results in the total demand for produced battery capacity. This value also
determines the required working force in the “Employee market” sub-system as well as
the required production capacity in the “Production & supply chain’. Currently there are
different approaches to construct an electric vehicle battery. Besides the design, they
differ in the materials used for the cathode, which constitutes the core element of all
cells. In the model it is assumed that so called NMC-cathodes are used for the batteries.
They contain the materials: Nickel, manganese and cobalt and are seen as one of the
most promising cathodes for e-mohility applications.
The sub-system “Production & supply chain’ has a common structure (e.g. see Stemman.
2000) and is kept quite simple due to the fact that the main emphasis of the model lies
on the sub-systems “Employee market” and “Raw material market’. Nevertheless it is
to transform the production orders into the final battery. Therefore the sub-
system consists of three parts which are depicted in Figure 3-3.
The actual transformation process takes place in the “Production & supply chain”
whereat the production rate is dominated by three possible constraints: The existing
material capacity, the production capacity of the staff and the installed production ca-
pacity. Last named is determined in the part “Capacity adjustment” and follows the de-
mand for batteries in kWh. Because of the novelty of the production processes for EV
batteries, production rejects are playing a prominent role and cannot be ignored. This
fact is taken into account by the “Production yield” which improves over the course of
time from 70 % up to 95%. The rejects are transferred into the sub-system “Raw mate-
rial market” and are there available for recycling. Finally the third part, “Order rate”
offers information about the gap between the current production and the incoming or-
Missing orders
| Orders Ff a
Shipment rate
oF
Ordering rate
Production potential
direct battery staff
Cell ; Battery
Production rate Inventory Productionrate inventory
cells batteries
Production yield_ 7
Installed production Installed production gq se
capacity cells OO
capacity battery ate
Neue capacity Production capacity ff
scaling cell aling battery
Figure 3-3: Parts of the sub-system "Production and supply chain"
The third sub-system “Employee market” represents one of the major parts of the
model. As seen above, the production potential of the staff is one of the limiting factors
for the production rate. So it is evident, that for a successful production of electric vehi-
cle batteries, a sufficient availability and productivity of employees is essential. To
achieve the necessary reliability for the employment rates, current data from the Ger
man “Federal Institute for Employment” were evaluated for the professional groups (see
Destatis 2012). It was taken into account, that cell production and battery production are
different knowledge areas and hence, different qualifications are needed. While the em-
phasis in the cell production lies on chemistry close-by qualifications, the demand for
the battery production is by nature from the field of electronics. To consider these dif-
ferent demands for qualification the sub-system includes for each group one adjustment
part. The part for the cell staff is illustrated representatively in Figure 3-4.
Figure 3-4: The cell staff employment as part of the sub-system "Employee market"
The cell staff adjustment is regulated by the demand for batteries, the available qualified
staff and by the “Scale factor productivity’. Each employee is able to process a certain
amount of kWh. This could lead to the assumption, that with a rising amount of kWh
the number of employees also rises in linear manner. But by scale effects and the ongo-
ing advances in processing technology, each employee can handle a growing amount of
kWh over the time. The initial values for the scale factor and the required employees are
based on information from the Argonne Laboratory (see Nelson et al. 2011).
The last sub-system “Raw material market’ represents the process of material procure-
ment and the recycling of batteries, from obsolete cars as well as from the production
rejects. Also some approaches for modelling the effects of recycling in system dynamics
exist (see Vlachos et al. 2007 or Schréter 2006). But the authors are not aware of any
approach conceming the recycling of batteries from EV. The sub-system attends to the
cathode materials already mentioned above. The different amount of these materials in a
cell as well as the possible yield from recycling is based on the “LithoRec’”-process
(Buchert et al. 2011). Currently there exists, with the exception of some pilot lines, no
facility that is able to recycle matenials in a way to use them as primary materials again.
Nevertheless with a rising amount of electric vehicles and recyclable batteries this is
only a matter of time. But due to the lack of required information conceming the reuse
as primary materials, there is, although this option is in the model integrated (indicated
by the dotted arrow in Figure 3-1), no feedback from recycled material back into pro-
duction. However, the influence of the recycled material on the future material pro-
curement can be depicted and the recycled material is available to be sold on the stock
market, which enables an analysis of the net present value. For this analysis, informa-
tion from an existing recycling centre by the Belgian enterprise “Umicore” are used (see
EUWID 2011). Furthermore it is possible to examine the share of the required material
in its annual global production. The according data is taken from U.S. Geological Sur
vey (see USGS 2013). Figure 3-5 depicts representatively the recycling of lithium.
Figure 3-5: Lithium recycling as part of the sub-system "Raw material market"
Based on the “Recyclable kWh’, the “yield” - and “collecting rate’, the recyclable ma-
terial can be determined. The “material rate lithium” is designed in a way to use all of
the recycled material before ordering additional material.
The four sub-systems are already sufficient for a first model to depict the dynamic hy-
pothesis, carry out simulation runs and gain first insights in the dynamic effects on raw
materials and employment.
4 Dynamic effects on raw materials and employment
In order to test the dynamic hypothesis and to generate a deeper understanding of the
system's behaviour, various simulation runs and tests were conducted. Some exemplary
Tuns are shown in the following and interpreted to build first hypotheses. The relevant
variables changed for the tests are listed in Table 1.
The information about the “Potential customers” is derived from a model developed for
the “Office of Technology Assessment at the Geman Bundestag” which is based on the
European transport model “ASTRA” (see Fermi et al. 2012 and Schade et al. 2012).
Numbers for the initial employees as well as the productivity factor are calculated from
information given in Nelson et al. 2011. The rates for the collection of batteries until the
year 2016 are published in the EU’s “battery directive 2006/66/ec” (see European Union
2006). For the model it is assumed, that after 2016 the development of the rates contin-
ues in the same manner as before. Finally, the information about the qualification rates
is published in the German statistical yearbook (see Destatis 2012). It is assumed that
for the cell staff, apprentices and graduates from chemical professions match the re-
quirement profile, while for the battery production electrical professions are more
suited. Furthermore machine and systems operators are able to work in both areas with
an equal share.
Potential | Potential | Potential | Initial fuarncauon| | rmscueaany
customers | customers | customers | employees sess poe
HEV PHEV BEV cell ele leet s
0.25 (2012)/ 0.45 (2016)/
Baserun 12.055.516 4.987.941 3.605.359 (0.65 (2020)/ 0.85 (2024)/
0.95 (2028)
0.25 (2012)/ 0.45 (2016)/
Oil age 14,459.089 6.189.727 98.786 198 6 158 509 0.1 0.65 (2020)/ 0.85 (2024)/
0.95 (2028)
0.25 (2012)/ 0.45 (2016)/
Electric age 14.293.346 6.652.062 4.870.639 198 683 158 509 1 (0.65 (2020)/ 0.85 (2024)/
(0.95 (2028)
‘Adjustments
ee 287 - as 0.95
in simulation
Table 1: Overview of the variables and constants used for the simulation runs
Due to the fact that not all graduates are going to work in the field of battery production,
it is expected, that every fifth graduate is willing to work in the battery production.
However, it is considered that from these potential employees only further 20 % possess
the required qualification level. In the following the model and its behaviour will be
shown by means of the “Base run” scenario.
4.1 Results of the , Base run“ scenario
The “Base rum” scenario starts, as shown in Table 1, with a lower number of potential
customers for HEV and PHEV compared to the other two scenarios. Nevertheless it is
estimated that BEV will also play a dominant role in the observed period. Figure 4-1
depicts the diffusion of the different kinds of electric vehicles as well as the resulting
demand for produced kWh battery capacity, which is needed to equip the vehicles with
an according battery. It can be observed, that the diffusion of the HEV shows the fastest
growing curve. This is reasonable because of the lower costs for HEV batteries. The
HEV reach their peak between 2019 and 2020 before they steadily deplete. The PHEV
are growing more slowly and peak in 2022. Their number will finally level off at a
value around 450.000 vehicles. The demand for BEV is beginning to rise at the same
time when the HEV growth curve begins to tum. They develop slowly but constantly,
until they reach the same level as the PHEV in 2030.
Figure 4-1 shows the impact of the diffusion of all types of electric vehicles on the total
demand of batteries and hence, on the battery capacity measured in kWh. The demand
follows the rise of the Hybrid electric vehicles but while their number declines, the de-
mand for batteries stays on comparable level. The reason for that lies in the higher
amount of kWh needed for PHEV or BEV. While HEV only contain a battery with 3
kWh, the PHEV are equipped with a 15 kWh hattery and BEV even with 30 kWh. By
means of that, the lower number of those is able to compensate the effect of falling de-
mand for HEV.
To serve the market with the desired amount of batteries an equivalent number of en+
Ployees is needed to produce them. The potential for a high degree of automation is
10
higher in the battery production than in the cell production. Due to that fact we assume,
that an employee in the battery production is able to produce a higher amount of kWh
than an employee in the cell production. This effect results in different numbers of em-
Ployees for each stock. The employees required to produce the desired kWh shown in
Figure 4-1 are depicted in Figure 4-2.
Diffusion of EV in scenario "Base run!
3M
AR
§ A
f *
‘ A \
z 15M y \
Pam
Aloe aS +233
0 {aot pt 3 3 aaah F
2012 "2015" 2018 2021 2024 2027 2030
Production volume HEV : Baserun -}-——-}-—_4-—
Production volume PHEV : Basenm —2——2———2-
Production volume BEV : Baserun. ——3——-3——
Demand for battery capacity in kWh
25M
0 vl 44
2012-2015 «2018 »=—«-2021:»=«2024-=«-2027'-—«- 2030
Demand battery capacity kWh: Basernm —i——-1—
Figure 4-1: Diffusion of electric vehicles and the involving demand for batteries meas-
ured in kWh, forthe “Base run”.
The effect of different productivity levels gets clear by analyzing at the relation between
the cell staff and the battery staff. The required number of employees in cell production
is with a value of 770 more than three times higher. The required indirect staff levels at
around 80 employees. While the graph for the battery and indirect staff flattens and fol-
lows the curve of demand, the curve for the cell employees is still constantly rising. The
reason for that anomaly can be detected on the lower diagram of Figure 4-2. While there
11
are no recruitment deficits for the battery and indirect staff, the cell production suffers
under an insufficient recruitment rate beginning with the year 2016.
Development of the employee stock in scenario "Base run!
2—2— 22
3-3-3
2024 2027 2030
2344537453 23 33153 53 9
2nd ~ “2615 ~ “2018 ~~2021 ~ 2024“ 2027 ~ 2030
Gap cell staff : Base run 4
Gap battery staff : Base run
Gap indirect staff : Base rn.
nn
a
ny
nly
Figure 4-2: Development of the employee stock and the deficits in recruitment
Besides the dynamic effects on employment, the recycling constitutes the second em-
phasis of the model. To generate a better understanding, the required material for the
battery production is compared to the annual global production in the year 2012. Three
materials which are all contained in the cathode of the observed cell type were closer
examined: Lithium, cobalt and nickel. Their share in production according to Figure 4-1
is depicted in Figure 4-3.
The development for required material is similar to the demand curve. But while the
share of nickel constantly lies under 1%, the share of lithium and cobalt constitute
nearly 10% of the annual production, only for Germany. The lower diagram of Figure
12
4-3 explores how this big share can be in influenced by establishing recycling activities
for used and rejected batteries.
Although collecting rates were introduced before, values between 25% and 45% show
no effects on the share of the annual production. The development of both curves pro-
ceeds similar until 2020 when the collecting rate rises to 65%. The steadily rise is cut
down and peaks in 8% share of the annual production. Afterwards, further two changes
in the collecting rate take place and lead to an instantly drop in share. Finally, in the
year 2030, the share e.g. for cobalt dropped down on 2 %, compared to 8.2 % without
recycling measurements before.
Share of required material in its annual production
10
Es
0 al [3l fo +3 3 3
2012 2014 2016 2018 2020 2022 2024 2026 2028 2030
Share of required lithium: Base run. ——t——1—_4-—_
Share of required cobalt : Baserun -2——2——2—2—-
Share of required nickel : Baserun. —3——3——_3——3
Share of required material in its annual production with recycling
8
Percent
>
0) 23. 3 3 3 3 3.3]
2012 2014 2016 2018 2020 2022 2024 2026 2028 2030
Share of required lithium with recycling: Basen —t—
Share of required cobalt with recycling : Baserun ——-2—
Share of required nickel with recycling: Baserunn —-3——3
Figure 4-3; Share of required material in annual production with and with recycling
Consequently, we can determine that recycling slows down the demand for raw materi-
als for the battery production and may help saving resources. But this is just one side of
the medal. Recycling not only has to fulfil the claim to be ecological reasonable but
13
economical, too. To analyse recycling from an economical point of view, it is necessary
to take most battery materials or those with the biggest share in value into account.
Therefore the observed materials were complemented by aluminium, copper and
stainless steel (type 1.4301). Their share in the battery (see Buchert et al. 2011) was
converted into the specific share per kWh. Together with current data from the stock
market the value of the recycled material can be calculated. To be able to recycle these
materials, recycling centres and -capacity is required. The required invest and the in-
stalled recycling capacity is derived from a recycling centre from “Umicore” in Hobo-
ken. The investment is calculated with 25M € and the capacity with an equivalent of
150.000 car batteries (see EUWID 2011). Based on the described data and assumption,
the net present value for recycling is determined in Figure 4-4.
The net present value keeps falling until the year 2024 due to high number of HEV bat-
teries, which only contain a small number of materials but require additional invest for
recycling capacities. After 2024 the effect of batteries with a higher capacity becomes
visible. The curve of the net present value tums and reaches a positive value in the year
2028.
Net present value for recycling
300 M
Fa
160M Pi
/
20M
€ E art. Al
-120M Peat He
-260M
-400 M
2012-2015 2018 = 2021 = 2024S 2027'—Ss_ 2030
NPV: Basenm —t + it 5 a £
Figure 4-4: Net present value for the recycling of EV batteries in scenario "Base run"
However the general development of the electromobility is hardly to predict. To exam-
ine the effects of altemative futures on raw materials and employment we will take fur
ther two scenarios into account: “Oil age” and “Electrical age”. The scenarios will be
shortcut in following with: OA (Oil age), EA (Electrical age) and BR (Base run).
14
4.2 Effects on employment
The input values for the following scenarios are listed in Table 1. The difference in po-
tential customers for EV results in a different development of the overall demand for
battery capacity. As shown in Figure 4-5 the OA scenario shows a volatile course of the
graph. While the FA scenario rises in a constant manner. Settled between both scena-
tios, conceming amount and development, lies the BR scenario.
Demand for battery capacity in kWh in different scenarios
30M Iaeto_
24M
i ee :
4y
g 18M 7
12M oe ae
vd f ey pal
6M Pl
2) or”
ty) ae al
2012 2014 2016 2018 2020 2022 2024 2026 2028 2030
Demand battery capacity kWh : Baserun. —1-———~1—_1—_}.-—_
Demand battery capacity kWh : Oil age. ——2——2_-—_
Demand battery capacity kWh : Electrical age —3——3—_3
Figure 4-5: Overall demand for kWh under different scenarios
The following Figure 4-6 shows the evolution of the cell staff. While the battery- and
indirect staff showed no deficits in recruitment the cell employees need to be closer ex-
amined.
Figure 4-6 shows that the development of employees in the OA scenario follows the
graph for battery demand. In contrast, the graphs of the scenarios BR and EA converge
to a specific amount. The lower part of Figure 4-6 gives the reason for that behaviour.
While there is only a low number of missing employees for the OA scenario, a big gap
exists between desired and recruited employees for the other two scenarios.
15
Direct cell staff in different scenarios
2012 2014 2016 2018 2020 2022 2024 2026 2028 2030
Direct staff cell : Base run —14 4 t t 4
Direct staff cell : Oil age 2? 3 3 3
Direct staff cell : Electrical age
Gap in cell staff employment
350 = ail
=e
280 waa “Dt aN
SPR.
210 wil
7 AN
aah i aiAty | NIT Tol Tol TG
2015 Bota™ 2016~ 2018 2020 “2022 3024 4026 2028 2030
Gap cell staff : Base run. + + a . t
Gap cell staff : Oil age —2 2 2 3 2 2
Gap cell staff : Electrical age
Figure 4-6: Direct cell staff and recruitment deficits under scenarios
To reduce or even negate the gap in recruitment the influence of two possible options
are tested in the following. First in Figure 4-7 the effect of an increased productivity of
the cell staff is analysed. This could come from a higher level of automatisation or scale
effects and is expressed by the “Scale factor productivity”. This factor will be raised
from 0.1 to 0.5.
16
Direct staff cell with higher productivity
La
=a ay =
°
2012 2014 2016 2018 2020 2022 2024 2026 2028 2030
Direct staff cell : Base run with higher productivity ———1—
Direct staff cell : Oil age with higher productivity —-2——2
Direct staff cell : Electrical age with higher productivity —— 3
Gap in cell staff employment with higher productivity
ah
70 A Sa 3
0 334 9 3195/5 ie } é | 2 ee Y a 4
2012 2014" 2016 2018 “2020 “2022 2024” 3026 “2628 “2030
Gap cell staff : Base rn with higher productivity —-——1—
Gap cell staff : Oil age with higher productivity ——2——2
Gap cell staff : Electrical age with higher productivity —3——
Figure 4-7: Effect of an increased productivity on the recruitment deficit
By an increased productivity the gap in recruitment can be cleared in the OA scenario.
Furthermore, the deficit in the scenarios BR and EA can be reduced be a third. But nev-
ertheless there still exists a deficit in those scenarios.
The second option constitutes an increased qualification rate for cell employees. There-
fore the rate is adjusted according toTable 1. It is now assumed, that every fourth gradu-
ate is willing and qualified to work in the cell production. The effect of this measure is
depicted in Figure 4-8. The productivity is reduced back to the initial value.
Figure 4-8 shows the major effect of this measure. All graphs are following now the
curve of demand. This already indicates, what the lower part of Figure 4-8 confinns.
There no longer exists a deficit in recruitment.
7
Direct staff cell with higher qualification rate
1,200
0
2012 2015 2018 = 2021 2024 2027 2030
Direct staff cell : Base run with higher qualification rate ———}——
Direct staff cell : Oil age with higher qualification rate —-2——2—
Direct staff cell : Electrical age with higher qualification rate ———3
Gap in cell staff employment with higher qualification rate
100
20 4 2 2 2 8
Gap cell staff : Base run with higher qualification rate —_-——-}—
Gap cell staff : Oil age with higher qualification rate ——2——2—
Gap cell staff : Electrical age with higher qualification rate —3——3
Figure 4-8: Effect of an increased qualification rate on the employment deficit
4.3 Effects on raw materials
Similar to the extended analyses done in chapter 4.2, this chapter focuses on the effects
on raw materials under different scenarios. Therefore Figure 4-9 depicts the share of
required cobalt in its annual production, with and without recycling. The reason why the
emphasis lies on cobalt is on the one hand, that its share in production is very high (see
Figure 4-3) and on the other hand, cobalt not only posses an economical value, it is ad-
ditionally characterized as a “critical raw material” by the European Union.
As already examined in the BR scenario, the recycling has a major effect on the share of
required cobalt on its annual production. The impact of this effect varies according to
the scenarios. While in the OA scenario the share is reduced by nearly 3,5 %, the impact
in the EA scenario is with a reduction of 7,8 % down to a value of 3,65 % unmistakable.
18
By introducing recycling measurements the peak share in the EA scenario (2027) can be
reduced from 12,2% to 6 %.
Share of required cobalt in annual production.
15
12
3 9 78%
Z 6
615%
3 3,43 %
6 Zo
2012, «2014S. 2016) 2018) 2020. 2022.) 2024S 2026)» 2028 =. 2030
Share of cobalt with recycling : 1 4 3
Shareof cobalt with recycling : Oil agesteps. — — - — ~) PessSS Qo aa
Share of cobalt with recycling : Electrical age steps 3 3 3
Share of required cobalt : 4 4 4 4
Shareof required cobalt: Oil agesteps —-- - 5------- §------- 5 ------
Share of requited cobalt: Electrical age steps
Figure 4-9: Share of cobalt in the annual production under different scenarios
In order to achieve insights about the economic appropriateness of the recycling, the net
present value for the different scenarios is depicted in
Figure 4-10. It can be observed, that the NPV of the EA scenario declines faster than in
the BR scenario. The reason for that can be found in the high initial number of HEV.
However, after the curve tums, the NPV reaches the same value in 2030 like in the BR
scenario. In contrast, the NPV of the OA scenario stays constantly negative and ends up
in the year 2030 with a value of -430 million €.
‘Net present value of recycling with collecting steps
NPV:Oillage —2 2 2 2 2 2 2
NPV: Electrical age 3 3 3
Figure 4-10: Net present value of recycling with collecting steps
19
As shown in the two figures above, recycling offers a great ecological potential, while
the economical is in the medium term not given. For that reason the following Figure
4-11 compares the default collecting rate with a rate starting in the year 2012 witha
value of 95%. The fundamental idea behind that assumption is that the customers lease
the batteries. After the leasing period, the battery goes back to the manufacturer. By this
means the battery stays in the property of the enterprise, by what such a high collecting
rate can be achieved.
Share of required cobalt with early recycling
2012 2014 2016 2018 2020 2022 2024 2026 2028 2030
Shareof cobalt with recycling poe aap withealy ————
Figure 4-11: Effect of an early high collecting rate
On the first glance, it becomes obvious, that the early recycling concept leads to a lower
share in the middle of the observed period. Both concepts end up with the same value.
This is reasonable because both have a final value of 95%. A specific occurs in the year
2020. While the share drops in the “collecting step” concept, it raises in the “early
recycling” concept.
Tt was mentioned above, that rejected batteries are available for recycling. With the
progress in production technology, the yield increases. This happens in the years 2016
and 2020. However, in the year 2020 the yield is only increased by 10 % to its final
value of 95%. But in contrast to the year 2016 where no effect is visible, this increase
coupled with the higher production rate leads for the “early recycling” concept to a step
in the share. This effect also takes place in the “collecting step” concept, but there it is
negatetd by the increased collecting rate which is introduced at the same time.
20
As a result from Figure 4-11 in the “early recycling” concept the share of required
cobalt is lower. In Figure 4-12 the economical effect of this higher recycling rate in the
midium tem is depicted.
‘Net present value for early recyding
800M
580M
360M
e
140M
80M BPP Eee nae Coo
-300M
2012 2014 2016 2018 2020 2022 2024 2026 2028 2030
NPV: Base nm with eally recyding ——t——_1——_-—__-—
NPV: Oil age with early recyding —2-—2—_2—__2-—__2_
NPV: Electrical age with ealy recyding ——3——_3—_3—_3-
‘Net present value for "oil age" sceanrio with an increase in the cost of raw materials
20M
-39M
-98M
ico)
-157M
-216M
-275M.
2012 2014 2016 2018 2020 2022 2024 2026 2028 2030
NPV: Oil age with early recyding ——t+>—__—__—__4
NPV: Oil age costs increase 5% =2———2———_2——_ 2
NPV: Oil age costs increase 10% —3———3——__3—__3—_
NPV: Oil age costs increase 15% ——4—_4—4—4—
Figure 4-12: Net present value for an early recycling concept and the influence of rising
costs for raw materials
The higher and earlier recycling rate inflicted by the “early recycling” concept has a
strong impact on the net present value of the scenarios. Both, the BR and the EA scena-
tio reach an economical level in the medium term of the observed period. While the BR
scenario reaches that level some years earlier because of the constant level of demand,
the EA scenario reaches a higher final net present value due to the still steadily rising
demand. Only the OE scenario stays negative. In a further step the effect of rising costs
for raw materials and their influence on the NPV of the OA scenario are examined.
21
While in the previous simulations was assumed that the costs stay on today’s level, they
are now considered to rise with a rate of 5 %, 10 % or 15 % per year.
The net present value of the OA scenario as shown in the lower part of Figure 4-12
stays negative. Only if the costs for raw materials would increase with a rate of 15 %
per year, the net present value would reach the positive range just before the period
ends.
5 Conclusions and Outlook
This paper presents a simulation model which demonstrates the automotive Li-ion bat-
tery supply chain in Germany for the next decades. The model is able to show the dy-
namics and long-ranging impacts of different EV diffusion scenarios on the battery sup-
ply chain and its environment. Consequently, the model gives first insights into future
trends and their effects on employment and on raw materials.
First findings show, that there may be a shortage of chemists for the production of lith-
ium-ion cells in future. The expected growth of labour productivity and improvements
of the economies of scale may countervail an employee shortage for small or mediun+
sized sales volumes. Rapid or great market diffusions however will probably lead to
lack of qualified staff in Germany.
The demand for lithium and cobalt will be on a very high level in future. This could be
especially a problem for cobalt, which is classified as a critical raw material from the
EU (European Union). A possibility for slowing down this effect is a faster and earlier
increase of the defined recycling quotes. If those quotes are too low the demand of raw
material increases inexorable, which may lead to a shortage and an increase of prices of
some relevant resources. Higher recycling quotes ensure solid production rates of batter-
ies in the next decades.
Summing up, we can formulate the following findings as hypotheses:
> A fast and great diffusion leads to a gap of available employees in cell produc-
tion, whereas employees in battery production seem to be no limiting factor
> The employment gap in cell production may be slowed down by increasing the
productivity level until the year 2020
> The remaining gap may be closed by increasing the qualification rate up to 150
percent compared to the value of 2012
> Depending on the scenario, the demand for raw materials may be reduced up to
50 percent in the year 2025 and 75 percent until the year of 2030 by means of
recycling activities
> An early introduction of high recycling quotes is essential to avoid the “recy-
cling gap” during the stages of great diffusion between the years of 2020 and
2028
22
> An early introduction of high recycling quotes rises the economic potential of
battery production extremely
However, this study gives just some insights, basing on a first system dynamic simula-
tion model. The system structure of the supply chain in reality is much more complex
than the structure of our model. Model boundaries may be changed and their impacts on
results should be checked. Purchase prices for batteries were not considered, nor were
the strategic decisions of companies taken into account. This model should be regarded
as a first step and as a base for improved and enhanced battery supply chain models.
Acknowledgment
The authors would like to thank:
“» the German Federal Ministry of Education and Research (BMBF) for funding
this study as part of the project ,,Energiespeicher- Monitoring fiir die Elektromo-
bilitét (EMOTOR)’’
«» the Fraunhofer Gesellschaft and state ministries of Baden-Wirttemberg for
funding this study as part of the project ,,regional eco mobility 2030” (REM
2030)
23
6 References
Bass, F. M. (1962): A New Product Growth Model for Consumer Durables, in: Man-
agement Science, Vol. 15, No. 5, pp. 215-227.
Buchert, M., Jenseit, W. Merz, C, Schiller, D. (2011): Okobilanz zum_,,Recycling von.
Lithium Ionen-Batterien” (LithoRec), Darmstadt.
Brooker, A., Thomton, M., RughJ. (2010): Technology improvement pathways to cost
effective vehicle electrification. Technical Report 2010-01-0824, National
Renewable Energy Laboratory.
Destatis (2012): Statistisches Jahrbuch 2012 - Deutschland und Intemationales, Wies-
baden.
European Union (2006): Directive 2006/66/EC of the European parliament and of the
council.
EUWID (2011): Umicore erdffnet neue Anlage fiir Batterierecycling. Online:
http://www.euwid-
recycling.de/news/intemational/einzelansicht/Artikel/umicore-eroeffnet-neue-
anlage-fuer- batterierecycling. html (12.3.2013).
Femi 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 in+
pacts of past and future sustainable transport policy in Europe). Project co-
funded by European Commission 7th RTD Programme. Fraunhofer-ISI,
Karlsruhe, Germany.
Fleischmann, M. (2001): Quantitative models for reverse logistics, Springer, Bedin.
Keith, D.; Sterman, J.; Struben, J. (2012): Understanding Spatiotemporal Pattems of
Hybrid-Electric Vehicle Adoption in the United States, Proceedings of the 30th
Intemational System Dynamics Conference, St. Gallen, Switzedand.
Kley, F.; Lerch, C.; Dallinger, D. (2011): New business models for electric cars - A
holistic approach, in: Energy Policy, Vol. 39, No. 6, pp. 3392-3403.
Lehr, C.; Milling, P. (2009): From Waste to Value - A System Dynamics Model for
Strategic Decision Making in Closed-Loop Supply Chains, Proceedings of the
27th Intemational System Dynamics Conference, Albuquerque, NM, USA.
Maier, F.H. (1998): New product diffusion models in innovation management - A sys-
tem dynamics perspective. In: System Dynamics Review, 14 (4), S. 285-308.
Milling, P.M.; Maier, F. (1996): Invention, Innovation und Diffusion - Eine Simulation-
sanalyse des Managements neuer Produkte, 48, Bedin: Duncker & Humboldt.
24
Milling, P.M. (2002): Understanding and managing innovation processes. In: System
Dynamics Review, 18 (1), S. 73-86.
Morecroft, J. (2007): Strategic modelling and business dynamics - A feedback systems
approach, Chichester: John Wiley & Sons.
Mutallip, A.; Yasarcan, H. (2012): A Parametric Analysis of the Effect of a Material
Supply Line Delay in Stock Management; Proceedings of the 30th Intema-
tional System Dynamics Conference, St. Gallen, Switzedand.
Nelson, P., Bloom, K., & Dees, D. (2011). Modeling the performance and cost of
lithium-ion batteries for electric-drive vehicles: Argonne National Laboratory
(ANL), Argonne, IL (United States).
Nenny, F., Leduc, G., Mun’oz, A. (2009): Plug-in hybrid and battery-electric vehicles:
State of the research and development and comparative analysis of energy and
cost efficiency.
NPE - Nationale Plattform Elektromobilitat (2011): Zweiter Bericht der Nationalen
Plattform Elektromobilitat, Berlin.
Rodrigues, L. L. R.; Motlagh, F. G.; Ramesh, D.; Kamath, V. (2012): System Dynamics
Model for Remanufacturing in Closed-Loop Supply Chains, Proceedings of the
30th Intemational System Dynamics Conference, St. Gallen, Switzerland.
Schade, W., Hettesheimer, T., Kinkel, S., Kihn, A. Zanker, C. (2012): Zukunft der Au-
tomobilindustrie. Vorlaufige Fassung, Berlin.
Schréter, M.: Strategisches Ersatzteilmanagement in Closed-Loop Supply Chains - Ein
systemdynamischer Ansatz. Dissertation, Universitat Braunschweig. Wiesba-
den: Deutscher Universitats-V edlag 2006
Spengler, T.; Schréter, M. (2003): Strategic management of spare parts in closed-loop
supply chains - a system dynamics approach. Interfaces, Vol. 33, No. 6, pp. 7-17
Sterman, J. D. (2000): Business Dynamics, Modeling and Simulation for a complex
World, New Y ork.
Stumpfe, J. (2003): Interdependenzen von Produkt- und Prozessinnovationen in indust-
riellen Untemehmen - Eine System-Dynamics-basierte Analyse, Peter Lang,
Frankfurt a M. etal.
Tate, E. D., Harpster, M. O., Savagian, P. J. (2008): The electrification of the automo-
bile: From conventional hybrid, to plug-in hybrids, to extended-range electric
vehicles. In: 2008 SAE World Congress, Detroit, Michigan, April 14-17, 2008.
General Motors Corporation, SAE 2008-01-0458
25
USGS (2013): Commodity Statistics and Information. Online:
hittp://minerals.usgs.gov/minerals/pubs/commodity/ (13.03.2013)
Vlachos, Dimitrios, Patroklos Georgiadis, and Eleftherios Iakovou: A system dynamics
model for dynamic capacity planning of remanufacturing in closed-loop supply
chains. Computers & Operations Research 34.2 (2007): 367-394.
Wachter, K; Spengler, T. S. (2012): Innovation diffusion under competition - simula-
tive strategy development for the introduction of altematively powered vehi-
cles, Proceedings of the 30th Intemational System Dynamics Conference, St.
Gallen, Switzedand.
Wann, K. (2008): Strategic management dynamics, West Sussex: John Wiley & Sons,
Ltd.
26
ul Appendix
Cases per ton
sine
olen sls ne Melati ese
1 Waderocel
sane mad
neeelg center Recyelale
a aluniniuce
Casts per ra : aie LY
recycling center Thvest recycling Value mickel ysfue Heh
ecg ke be. =
Fast Teey cling meter te
ee ovement
©:
Investment ail
a
Figure A-1: Net present value
Diffusion of EV in the scenario "Oil age"
if
2012 2015 2018 2021 2024 2027 3030
Production volume BEV : Diffusion Oil age re cE = t Cars/year
Production volune HEV : Diffusion Oil age =
Cars/year
Production volune PHEV : Diffusion Oil age 4545 45 4S Cars/year
Figure A-2: Diffusion in the "Oil age" scenario
27
Diffusion of EV in the scenario "Electrical age"
ol w |
2012 2015 2018 2021 2024 2027 2030
Production volune BEV : Diffusion Hectrical age —_—}_}#_}+_ Cars/year
Production volume HEV : Diffusion Hectrical age SS Cars/year
Production volune PHEV : Diffusion Hectrical age 33 Cars/year
Figure A-3: Diffusion in the "Electrical age" scenario
