Sunteți pe pagina 1din 14

Journal of Cleaner Production 90 (2015) 220e233

Contents lists available at ScienceDirect

Journal of Cleaner Production

journal homepage:

The role of electric vehicles for supply chain sustainability in the

automotive industry
H.-O. Günther a, *, M. Kannegiesser b, N. Autenrieb b
Dept. of Industrial Engineering, Seoul National University, 1 Gwanak-ro, Gwanak-gu, Seoul 151-744, Republic of Korea
Dept. of Production Management, Technical University of Berlin, Steinplatz 2, 10623 Berlin, Germany

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

Article history: Electric vehicles are perceived as key technology in the automotive industry to contribute to sustainable
Received 25 March 2014 development with lower greenhouse gas emissions, less air pollution for citizens and new job oppor-
Received in revised form tunities with positive social impact. So far the impact of electrified powertrains on the automotive in-
19 November 2014
dustry supply chain sustainability and the development of the passenger car fleet have not been
Accepted 19 November 2014
Available online 29 November 2014
investigated in an integrated way. In our study, we focus on the long-term industry supply chain design
end-to-end from the raw material exploration down to vehicle sales including the vehicle usage phase as
well as reverse supply chain activities. Specifically, we address the long-term dynamics in the industry
Sustainable supply chain
supply chain together with the fleet and powertrain development towards 2030 from a global
Electric vehicles perspective with a special focus on Germany and China. Based on industry data and using a sustainable
Automotive industry supply chain optimization framework covering economic, environmental and social objectives, the study
analyzes where jobs could be created or cut down, how total emissions in the supply chain including the
usage phase of passenger cars could be reduced and how supply chain wide costs would develop. In
addition, the impact of alternative market and technology scenarios as well as governmental regulations
on the mix of powertrains in the vehicle fleet is investigated.
© 2014 Elsevier Ltd. All rights reserved.

1. Introduction mobility. The main attempt to reach this goal in the next decades is
setting reduced CO2 emission limits and supporting the introduc-
As stated by Randers (2012) in his report to the Club of Rome tion of electrified vehicles such as pure electric vehicles or plug-in
greenhouse gas concentration in the atmosphere is expected to hybrids powered by energy generated from renewable resources.
continuously grow beyond 2030, which may well trigger self- Various governments introduced incentive programs for the pur-
reinforcing climate change exceeding the 2  C target by far. The chase and the use of electric vehicles (no taxes, free lanes and
automotive industry and its fossil fuel-based vehicles are seen as a parking in cities, etc.). Moreover, countries like Germany and China
major contributor to the emission of greenhouse gases. A major released plans to establish local electric vehicle prime markets. The
challenge is to replace fuel-based vehicles by alternative drive German chancellor announced plans to reach one million electri-
concepts like electric and hybrid vehicles. The automotive industry, fied vehicles on the road by 2020 (Federal Ministry of Economics
as one of the largest industries worldwide, thereby has a twofold and Technology of Germany, 2010), while China intends to obtain
scope of influence to the emission balance: reducing emissions this amount already in 2015 (KPMG, 2011) and to reach 5 million
when making vehicles including production, transportation and electrified vehicles in the fleet and 1 million annual production by
reverse supply chain activities, and reducing emissions when using 2020 (Reuters, 2010). Today, due to considerable production and
vehicles with cleaner powertrains. specifically high battery costs and minor driving ranges the electric
Recently, enforced by governmental regulations the European vehicle is still a niche product. However, further technological
automotive industry makes a fresh attempt towards zero emission progress, more advanced batteries, related battery control and
switching systems will enhance the range of these vehicles and
bring down the costs in the future.
* Corresponding author. Tel.: þ82 10 9195 7633. In this paper we address supply chains for conventional and
E-mail addresses: (H.-O. Günther), mkannegiesser@yahoo. electric vehicles from an industry perspective, i.e. we do not focus
com (M. Kannegiesser), (N. Autenrieb).
0959-6526/© 2014 Elsevier Ltd. All rights reserved.
H.-O. Günther et al. / Journal of Cleaner Production 90 (2015) 220e233 221

on individual enterprises or products, but integrate the various applications of the analytical hierarchy process. Several papers
company-specific supply chains into one composite supply chain reviewed are based on LCA type data providing either an assess-
model representing the entire automotive industry. We put ment of environmental impacts along the supply chain or deter-
particular emphasis on the long-term industry supply chain mining an optimum solution by balancing different objectives
development end-to-end from the processing of raw materials within a multi criteria decision making model. Another recent re-
down to the manufacture and sales of vehicles, as well as reverse view by Brandenburg et al. (2014) also focuses on quantitative
supply chain activities, e.g. scrapping and recycling of used parts. modeling for sustainable supply chain management. The authors
Contrary to classical supply chain models, we also include the evaluate the relevant literature along four dimensions, supply chain
vehicle usage phase allowing us to compare all impacts of making management (SCM), sustainability, modeling and research di-
vs. using vehicles. We investigate the long-term dynamics in the rections, and identify several gaps in the existing literature, e.g.
industry supply chain together with the fleet development towards insufficient consideration of the social sustainability dimension and
2030 from a global perspective with a specific focus on de- the integration of model-based methods with empirical research.
velopments in Germany and Europe vs. China. In their review Taticchi et al. (2014) screen 384 papers on decision
In our investigation we use mathematical optimization to model support tools and performance measurement in the domain of
the industry supply chain under environmental, social and eco- sustainable supply chain management. They conclude that these
nomic objectives. These three dimensions of the overall develop- areas evidently are isolated fields in the emerging literature and
ment of the industry are based on the so-called “triple bottom line” that there is a dearth of literature related to the implementation of
defined by Elkington (1998). This concept is specifically suited to integrated frameworks. A literature review to clarify the differences
derive sustainability performance indicators and to set related between green and sustainable SCM is provided by Ahi and Searcy
targets. The employed optimization model is fitted by empirical (2013). Their main conclusion is that green SCM is a subset of
data from automotive part and final manufacturers as well as public sustainable SCM focusing on narrow environmental aspects, while
sources. Numerical experiments are conducted for various sce- sustainable SCM aims to balance environmental, economic and
narios in order to predict possible developments of the industry social aspects.
supply chain for electric vehicles. In our investigation, the overall Also reverse supply chains, which close the loop after the end of
transformation processes of the automotive supply chain within product usage and include the reuse and recycling of materials,
the next 20 years are investigated, evaluating if electric vehicles gained increasing importance in the context of sustainable SCM.
could be the answer to the sustainable future mobility challenge. Ilgin and Gupta (2010) provide a literature overview of more than
Specifically, the following research questions are addressed. 500 papers focusing on environmentally conscious manufacturing
and product recovery. They classified the research into product
 How could capacities for the manufacture of electric vehicles design issues, reverse and closed-loop supply chains, remanu-
and related supply chain processes develop long-term and in facturing, and disassembly. Tang and Zhou (2012) emphasize the
which regions could these capacities be installed? importance to not only adopt the single-company perspective but
 What would be the effects of the required supply chain trans- apply a full industry supply chain view including upstream supply
formations on the labor market and social sustainability objec- stages and sustainability issues in emerging countries. Another
tives, e.g. minimizing job dismissals? comprehensive literature review on reverse logistics and closed-
 Would electric vehicles currently and prospectively be able to loop supply chain research is provided by Govindan et al. (2015).
compete against internal combustion engines from an economic They review 382 papers in that area and classify them into four
and an environmental perspective? classes: reverse logistics, closed loop supply chain, sustainability,
 Could total emissions in the make and in the usage phase be and green supply chain. Further factors such as problem type,
reduced towards sustainable levels and what could be the consideration of uncertainty, modeling approach, and solution
overall impact on costs? method are applied to characterize the reviewed articles. In addi-
 What would be the potential impact of governmental regula- tion, future research opportunities are identified. The paper by
tions like the implementation of tax measures or the introduc- Stindt and Sahamie (2014) contains a review of research on closed
tion of national grant programs on the fleet of electric vehicles? loop supply chain management in the process industry. This paper
aims to describe and analyze the main characteristics of closed loop
The remainder of this paper is organized as follows. In the next supply chain planning in the process industry in order to determine
section the relevant literature is briefly reviewed. This is followed in the evolution and gaps of this research area.
Section 3 by an introduction of the automotive industry supply Several authors propose specific frameworks and models in the
chain and the types of vehicles considered in our study. The in- context of sustainability in the supply chain. Sheu et al. (2005)
dustry model and the related data basis are outlined in Section 4. present an optimization framework which coordinates the logis-
Results of the numerical experiments are presented in Section 5. tic operations of the product flow and the respective reverse
Finally, conclusions are drawn and directions of further research are channel and also shows ways how to assess an equilibrium solu-
highlighted. tion. Quariguasi Frota Neto et al. (2008) developed a multi-
objective model for the design and evaluation of logistic net-
2. Relevant literature works that balanced economic and environmental impacts. Using
the European pulp and paper industry as an example they
In the last years a plethora of papers focusing on sustainable concluded that the adoption of environmentally preferred solu-
supply chain management has been published in the academic tions is generally bounded by an increase in costs. In their research
literature. We do not intend to comprehensively review these Hugo and Pistikopoulos (2005) address the environmentally
contributions but refer to Seuring (2013) who provides a review of conscious supply chain design problem based on LCA data. More
more than 300 papers related to sustainable supply chain research recently, Chaabane et al. (2012) presented a mixed integer linear
published over the last 15 years. Of the reviewed papers 36 focus on optimization model and, based on an example from the aluminum
quantitative modeling. Seuring classified the different mathemat- industry, demonstrated that for a long-term environmental sus-
ical models into four groups: life-cycle assessment (LCA) based tainability the numerous legislative carbon regulations have to be
models, equilibrium models, multi-criteria decision making and harmonized. Moreover, Chaabane et al. (2011) developed a multi-
222 H.-O. Günther et al. / Journal of Cleaner Production 90 (2015) 220e233

objective model formulation for balancing the trade-off between triple bottom line concept are Lee et al. (2012), Kim et al. (2014) and
economic factors (logistics, procurement and production cost) and Zhang et al. (2014). A system dynamics model of a product-service
carbon emissions to determine different supply chain design op- system is developed in Lee et al. (2012). The authors put special
tions. Their research is applied to the steel industry in Canada emphasis on measuring the triple dimensions of sustainability and
facing new carbon cap regulations. Ferretti et al. (2007) demon- on modeling the relationships between the various system factors.
strate the potential of the introduction of environmental sustain- Kim et al. (2014) address supply profitability taking environmental
able practices into supply chain management, again using the and societal responsibilities into account. As a starting point to
aluminum industry as an example. analyze sustainable supply chains quantitatively the authors pre-
It can be concluded that meanwhile a number of studies have sent a theoretical model which provides insight into the possible
been published in the academic literature investigating sustain- surplus to companies resulting from the transition of conventional
ability issues of supply chain design with the help of mathematical to sustainable supply chains. In their paper Zhang et al. (2014)
models. However, apart from the economic dimension most of develop a framework for sustainable supply chain optimization
them only consider environmental aspects like carbon emissions considering the operations of a chemical company as an example of
and disregard the social impacts of sustainability. Many previous application. The proposed MILP model is based on LCA as a sus-
studies address one narrow aspect be it the trade-off of emissions tainability assessment method that considers the environmental
vs. costs or the reverse branch of the supply chain and closed-loop impact of a product during its entire lifecycle. For solving the multi-
structures, while for application in industry an integrated view is objective optimization problem an ε-constraint algorithm is
desired. Moreover, sustainable supply chain studies often focus just developed.
on the make but not on the usage phase of products and thus fall To conclude, an empirical research contribution is often missing
short to reflect the lifecycle perspective in the supply chain. In in sustainable supply chain research. In contrast, the specific
particular, in the automotive industry with its high environmental contribution of this paper is to provide an LCA-based empirical
footprints it is a crucial question to cover both the make and the study of the long-term development of the automotive industry
usage phase of the vehicles. There are additional studies in the supply chain putting special emphasis on the role of electric vehi-
academic literature which focus on specific aspects like emission cles and to include social sustainability measures into the numer-
reduction in freight transportation (e.g. Demir et al., 2014) or on ical investigation.
customer purchasing behavior for electric vehicles (e.g. Glerum
et al., 2014). However, these studies do not cover the entire in- 3. Automotive industry supply chain and vehicle
dustry supply chain with its carbon footprint and are not capable of characteristics
predicting the long-term sustainable development of the entire
automotive industry. Fig. 1 illustrates the basic structure of the automotive industry
There exist a number of model formulations that include supply chain considered in this study. Principally, supply chains can
corporate social responsibility (CSR) in the multi-criteria decision be separated into a forward and a reverse part. The actual vehicle
making process (see Cruz, 2008; Cruz and Matsypura, 2009). The manufacturing steps are performed within the forward supply
major objectives are the maximization of profit and minimization chain using either raw materials from different suppliers or sec-
of emissions and risk factors. However, as Seuring (2013) states, ondary raw materials, which are produced in the recycling facilities
contrary to the title of these studies, they are not modeling social of the reverse channel. It is supposed that both the primary and re-
sustainability issues, but rather examine environmental measures cycled raw materials meet quality standards and therefore are
considering only the risk associated with not having enough CSR considered equivalent. The parts and components manufactured by
(see Cruz and Matsypura, 2009). Examples of publications which the first tier suppliers are assembled to final vehicles at the OEM
cover the different dimensions of sustainability according to the production sites. Thereafter, the finished products are distributed

forward material flow Reverse supply chain Waste

reverse material flow

Recycler Scrap Yard

Forward supply chain Use

Parts Export
suppliers markets

Raw material Component

OEMs Distributors Customers
suppliers suppliers

Electricity supply chain according to local energy mix

Fuel supply chain (well-to-tank and tank-to-wheel)

Energy supply chain

Fig. 1. Structure of the automotive industry supply chain.

H.-O. Günther et al. / Journal of Cleaner Production 90 (2015) 220e233 223

by the dealers on the primary market to the customers and to Table 1

export markets, respectively. In the lower part of the figure the Reference vehicles of different powertrain technologies.

energy supply chain is depicted. Electricity for making and using Category Small Compact Premium
electrified vehicles is supplied according to the local energy mix. Old Smart fortwo (C 450) Volkswagen Golf IV 1.6 Mercedes Benz E200
The fuel supply chain consists of the well-to-tank stage which (W210)
covers the supply from crude oil exploration to the distribution of IC Smart fortwo (C 451) Volkswagen Golf VI 1.6 Mercedes Benz E200
fuel, and the tank-to-wheel stage which refers to the internal use of CDI Blue Efficiency
EV Smart fortwo ED Volkswagen Golf blue Tesla Model S
fuel by the powertrain. The fleet of vehicles consists of new as well
as used vehicles. During their life time used vehicles can be sold HV Toyota Yaris Hybrid Toyota Prius Hybrid Lexus GS 450h
back to distributors and retailed on the secondary market. Finally, PHV Audi A1 e-tron Toyota Prius Plug-in Volvo V60 Plug-in
end-of-life vehicles are delivered to the scrap yards where they are Hybrid Hybrid
dismantled. Lastly, the dismantled parts are either recycled or Old: old internal combustion engine vehicle; IC: new internal combustion engine
wasted. Out of scope is the supply chain for single spare parts and vehicle; EV: Electric vehicle; HV: Full hybrid vehicle; PHV: Plug-in hybrid vehicle.
repair materials that are needed during the vehicle's life time. Not
shown in this figure are the transportation lanes which connect the
various geographical diverse locations by means of different According to Meyr (2004) a vehicle consists of around 10,000
transportation modes. parts. Given the huge variety of different car models and the
Generally, this paper investigates the long-term development of tremendous number of optional parts, considerable aggregation of
automotive supply chains of Europe and China where Europe with its parts is inevitable. Therefore, in our study a total of 12 different
very different economies is sub-divided into four locations, namely components are modeled. Thereby we put special emphasis on the
Germany, North-West Europe, South-West Europe and East Europe. structural differences of the considered vehicle types as illustrated
All manufacturing processes are assigned to one of these locations. in Fig. 2. The various vehicle types differ especially with regard to
Next to manufacturing two other basic processes are included in our the transmission systems that include the gearbox, the differential
model: raw material extraction and product usage. These processes and the drive shaft. Moreover, vehicles differ in the amount of
not necessarily have to be located in one of the manufacturing re- necessary electronic components, the assembled power engine and
gions. For instance, the extraction of specific raw materials can also the implemented energy storage system. However, the principal
be located in Australia or Chile. Moreover, on the demand side there body and chassis-structures are assumed to be the same for all
is also an export market located in North America (USA). Hence, powertrain technologies in the same vehicle segment, cf. Helms
European as well as Chinese OEMs are able to export their assembled et al. (2011). With the chosen component aggregation, we are
vehicles to this region. Other global automotive markets such as Asia able to model the differences in the key technologies that are
(except China) or Latin America are out of scope. investigated in our study.
On the manufacturing side the focus is on Europe and China
while Japan and Korea as well as the USA are left out. Reason for this 4. Industry model and supply chain data basis
limitation is the enormous data complexity which is extremely
challenging for establishing a comprehensive, close-to-reality in- 4.1. Sustainable supply chain optimization framework
dustry data basis. Therefore, we had to cut the scope of the study.
Europe and especially Germany were considered as the technology In our investigation we focus on the long-term sustainable
leaders in automotive production. China as a key country in the development of an industry supply chain. To reflect the relation-
investigation was chosen because of several reasons. First, China ships in the considered industry we apply linear optimization
nowadays is the biggest automotive manufacturer by volume modeling which has proven to be a powerful tool to optimize the
worldwide. Second, it will represent the largest automotive market structural design of company supply chains. The optimization
in the future. Third, China is the world largest contributor of green framework applied in our study has been developed in
house gases. Hence, the global warming and climate change chal- Kannegiesser and Günther (2014) and tested in Kannegiesser et al.
lenge will be decided in China. (2014). For the purpose of this study, the optimization model was
The vehicle fleet considered in this study consists of five pow- extended in order to include specifics related to electric vehicle
ertrain technologies: an old and less fuel-efficient internal com- supply chains.
bustion engine and a new more fuel-efficient internal combustion
engine (“Old” and IC), hybrid vehicles (HV) using fuel, plug-in 4.1.1. Overview of the sustainable supply chain optimization model
hybrid vehicles (PHV) using fuel but can be charged with elec- Basically, the model consists of three sub-models as illustrated
tricity from the grid, and pure electric vehicles (EV) only charged in Fig. 3. The process model is used to model the production and
with electricity from the grid. Each of them is subdivided into three other transformation processes within a defined location with the
different product segments: small, compact and premium sized relevant relationships, such as production capacity constraints,
vehicles. These segments match customer demand whereby cus- availability of input factors and the change and relocation of pro-
tomers first look for a vehicle in one of the segments independent duction capacity over time. The major decision variable used in the
of the powertrain option. Thus, next to the three existing old ve- process model is xslt defining the quantity of units processed in
hicles types, 12 different types of cars are modeled as shown in process s at location l in period t. Note that processes are used to
Table 1. Each vehicle type refers to a specific model chosen as the model different types of transformations that can occur in a loca-
reference vehicle for the respective segment, e.g. Tesla Model S as tion, e.g. production processes, raw material exploration, recycling
the model representing a premium EV. Reference models were or disposal. In addition to pure processing activities further deci-
chosen such that they fit well with the specific powertrain tech- sion variables are used to express the increase or decrease of ca-
nology and market segment. The reference models provide char- pacities at locations. Transport relations between locations in the
acteristic parameters, such as energy consumption, to be used in network are modeled in the transportation model. The major deci-
our investigation. In addition to the different powertrain technol- sion variable used in the process model is ypllmt defining the
ogies, the vehicles differ with respect to emissions, energy con- quantity of product p shipped from location l to location l using
sumption and costs for service and infrastructure. transportation mode m in period t. Constraints address the balance
224 H.-O. Günther et al. / Journal of Cleaner Production 90 (2015) 220e233

Fig. 2. Vehicle components depending on the powertrain.

of transshipments in the network and the capacities for different factors like social and environmental responsibility. Giving
transportation modes. These two models are similar to the ones emphasis to sustainable development and in expectation of
developed in Kannegiesser and Günther (2014) and thus not pre- stronger environmental and social regulations a composite sus-
sented here because of space limitations. tainability objective function seems to be adequate in our long-
Finally, sustainability issues during the lifecycle of technical term industry-related study.
products are reflected in the product-in-use model (Section 4.1.1).
This model extends the basic model formulation of Kannegiesser
and Günther (2014) for the purpose of the study at hand, e.g. by 4.1.2. Product-in-use model
introducing the vehicle powertrain options defined in the previous Technical products like passenger cars have a lifecycle of several
section and by developing equations which model the replacement years. Hence, in order to cover the related effects on costs, emis-
of batteries during the vehicle's lifecycle. In particular, the devel- sions and labor, the development of the vehicle fleet at the various
opment of the fleet of individual vehicle types at the various lo- locations is measured through a set of equations. First, the growth
cations is modeled through a set of equations. This sub-model is of the vehicle fleet is modeled in equation (1), where wtot
linked to the processes which introduce new vehicles into the fleet the total fleet size at demand location l2Ldem and period t2T with
and replace old ones for dismantling. To assess the multiple di- mfleet
expressing the annual growth of the fleet at location l.
mensions of sustainability optimization for the case of alternative  
powertrains sustainability key performance indicators, such as costs, wtot tot
lt ¼ wl;t1 $ 1 þ ml cl2Ldem ; t2T : t > 0 (1)
jobs created or CO2 emissions are defined in Section 4.1.2 sub-
stantiating the basic performance indicators of Kannegiesser et al. Second, equation (2) represents the allocation of the total fleet
(2014). size wtot
between different vehicle types f2Fg used by customer
Furthermore, to reflect the impact of selected external factors group g. The respective fleet size of vehicle type f at demand
such as governmental measures or different technologies of electric location l in period t is modeled through decision variables wflt, and
vehicle powertrains corresponding scenarios are defined. The sce- ugl is the given share of customer group g in demand location l.
narios also address the future uncertainty and allow testing the X
sensitivity of the system. wflt ¼ wtot
lt $ugl cl2Ldem ; g2G; t2T (2)
f 2Fg
It should be noted that the application of linear optimization
models in our study is based on the implicit assumption that the Finally, equation (3) is introduced to balance the development of
entire industry is driven by a single fictitious decision maker who the specific fleet size wflt from period to period. In this equation
executes normative decisions with respect to a multi-dimensional decision variables ypl1 lmt represent the quantity of products p2Pfþ
objective function, cf. Kannegiesser and Günther (2014). This gen- which increase the fleet size of vehicle type f and are shipped from
eral assumption is rectified by the long-term development of in- location l12Ll to demand location l2Ldem by transportation mode
dustries which is driven by economic objectives and influencing m2Ml1 l on lane (l1,l) in period t2T. Similarly, decision variables

Fig. 3. Sustainable supply chain optimization framework (adopted from Kannegiesser and Günther, 2014).
H.-O. Günther et al. / Journal of Cleaner Production 90 (2015) 220e233 225

proc fleet
ypll2 mt indicate the quantity of dismantled products p2Pf which Elt , transportation activities Elttr and product usage Elt . In this
decrease the fleet size of vehicle type f and thus are shipped from study we use carbon dioxide equivalent (CO2e) as common basis for
demand location l2Ldem to the linked scrapping destination l22Ll greenhouse gas emissions and as measure for greenhouse warming
by transportation mode m2Mll2 on lane (l,l2) in period t2T. In our potential (GWP) translating various greenhouse gases on a com-
model formulation we do not use explicit scrapping rates but derive mon CO2 factor basis, e.g. GWP of 1 t of methane emission is
scrapping quantities indirectly from the change of the specific translated to the GWP equivalent of 25 t of CO2. Depending on the
vehicle fleet size. use of fuel, coal or natural gas in the electricity generation process,
X X X X X X location-specific energy factors are determined and introduced in
wf lt ¼ wf l;t1 þ ypl1 lmt  ypll2 mt the calculation of total emissions.
p2Pfþ l1 2Ll m2Ml1 l p2Pf l2 2Ldem

cf 2F;l2Ldem ;t2T
(3) X X proc fleet

Etot ¼ Elt þ Elttr þ Elt (6)
Currently, the lifecycle of lithium-ion batteries does not achieve t2T l2L
the life time of other basic car components. Thus it is necessary to
replace battery modules within the life cycle of a vehicle. Conse- Total process emissions Elt are mainly related to the con-
quently constraint (4) is introduced to approximate the number of sumption of energy which is generated by electric power plants.
battery renewals rfbt of battery type b2B used for electric vehicles These emissions either emerge directly from the processes, e.g. in
f2Fb in period t2T. In (4) the coefficient afb indicates the number of case of raw material exploitation or metal production, or they can
additionally needed battery modules within the vehicle's life time. be caused indirectly through the use of electricity as an energy
X X X X X source for the production equipment. In our investigation, we as-
rfbt ¼ ypllmt $afb cb2B; f 2Fb ; t2T sume a slight relative decrease in process-related greenhouse gas
g2Gf p2Pfþ l2Lout
p l2Ll m2Mll emissions over time due to technological advancements. Total
transport related emissions Elttr are calculated in the transportation
model based on the transportation quantities, the mode of trans-
Depending on the specific fleet size of vehicles wflt total fleet size portation and the respective emissions per weight and distance
dependent costs Ct
can be calculated considering service costs unit. Vehicle fleet related emissions Eltfleet are obtained from the
for vehicles, costs for the establishment of the necessary infra- product-in-use model.
structure, e.g. loading stations for electric vehicles, location-specific The final key performance indicator determines supply chain
fuel costs for either gasoline or electricity consumption, and costs wide employee dismissals Ntot defined in (7) as sum of dismissals
for the renewal of battery modules. Likewise total emissions Elt
Nltproc caused by production processes over all periods t2T and lo-
for the fleet of vehicles used at location l2L in period t2T can be cations l2L. In our study we use the number of dismissals as an
calculated depending, in the case of electric vehicles, on the indicator to measure the social dimension of sustainable develop-
location-specific energy mix. Detailed calculations for both fleet ment of global supply chains. Avoiding dismissals and creating new
size dependent costs and emissions are omitted here because of jobs is perceived to have a positive social impact. While causing
space limitations. high dismissals and confronting employees with instabilities for
their personal life and families with job losses is perceived as
negative for social sustainability in societies. The number of dis-
4.1.3. Sustainability key performance indicators and objective
missals per period and location is determined in the process model
based on the process quantities and the respective labor re-
To measure the sustainability of the industry supply chain
quirements. Dismissals result from the transfer of process capacity
different performance indicators are defined and, for the applica-
to another location.
tion of a linear optimization model, integrated into one composite
objective function. The first key performance indicator used in our
study consists of supply chain wide costs Ctot defined in (5) as sum
of costs over all periods t2T, caused by production processes Ctproc , XX
transportation activities Cttr and product usage Ctfleet . Ntot ¼ Nlt
X proc fleet
 t2T l2L
C tot ¼ Ct þ Cttr þ Ct (5) The integrated objective function (8) is obtained by summing up
the scaled performance measures using weights ucost, uem and ulab
Total process related costs Ctproc are calculated in the process for costs, emissions and labor, respectively. Weights are used to
model for each location l2L and period t2T based on the amount of express the various measures, namely supply chain wide costs,
process activities of a specific type (e.g. production or scrapping) emissions and the number of dismissals. In decision theory,
using the installed capacity, the change of the capacity level, the weights are used to express the decision maker's preferences.
workforce employed, the amount of input material processed and Further, weights serve as parameters which are systematically
the electricity consumption. Total transport related costs Cttr are varied in a sensitivity analysis. All performance measures are scaled
calculated in the transport model based on the transport quantities in a range from zero to one using a base value, which is determined
and the realized changes of the transportation capacity. The by exclusive optimization of one single measure, a span value,
calculation of total vehicle fleet related costs Ct , e.g. service and which is the difference between the maximum (worst) value ob-
fuel costs and costs for establishing the infrastructure like charging tained in all the single-measure optimization runs and the base
stations for electric vehicles, is accomplished in the product-in-use (best) value for that measure, and the actual value obtained
model. resulting from an optimization run using the composite objective
The second key performance indicator expresses supply chain function. In (8) the superscripts tot, base and span indicate the
wide emissions Etot defined in (6) as sum of emissions over all reference values used in the scaling of the various performance
periods t2T and locations l2L, caused by production processes measures.
226 H.-O. Günther et al. / Journal of Cleaner Production 90 (2015) 220e233

! ! Since the major focus in our study is on electric vehicles specific

C tot  C base
cost em Etot  Ebase data reflecting the costs of batteries are essential. Particularly,
Min u $ þu $
C span Espan Lithium-ion batteries are seen as a major cost driver of electric and
! plug-in hybrid vehicles. Today the production of the positive elec-
lab N tot  N base trode (cathode) and the current underutilization of battery
þu $ (8)
N span manufacturing plants are the main reasons for their high produc-
tion costs. In order to examine current battery technologies, at first
out of the various kinds of cathode materials a specific type has to
be selected. According to Dinger et al. (2010) and Saft (2012) the
lithiumenickelecobaltealuminum (NCA) technology offers suit-
4.2. Supply chain data basis able features regarding energy density, specific power density,
cycling capability and service life and is market applicable today.
This section gives a brief overlook of the most essential data Future innovations such as lithium-air batteries are not available
used in our study. Considering the intention of the study to evaluate before 2020 and final parameterization is unclear. Therefore, they
the long-term development of an entire industry supply chain, the are not considered in the study. Data assumptions for batteries used
investigation has to be based on an accordingly high degree of in our study are summarized in Table 3.
aggregation. In our study, products are related to specific processes As for the external vehicle demand at the various locations our
which are characterized by a set of parameters, e.g. electricity calculations are derived from base values and assumed demand
consumption, labor input, capacity costs for the established pro- growth rates. In addition, each demand location consists of three
duction facilities, and process lead times. As for capacity utilization different customer groups, using small, compact and premium cars.
the minimal profitable utilization rate is assumed to be 50%, cf. Moreover, these segments differ with respect to the assumed
Reed (2012), meaning that otherwise capacity consolidation takes annual driving range: 12,000, 17,000, and 25,000 km per year for
place. Furthermore, based on a report by Grimes et al. (2008) direct small, compact and premium cars, respectively (cf. Klink et al.,
emission factors for the production of raw materials are introduced. 2012). The location-specific demand parameters for the various
Processes are assigned to specific locations, which vary widely with locations are summarized in Table 4.
respect to specific parameters such as labor costs (data from Customer shares are based on data from 2011 and are assumed
Schro€der, 2010), labor productivity (data from Yueh, 2009 and the to be constant within a region over all periods. However, the total
Bureau of Labor Statistics, 2012), CO2e emissions due to electricity shares can change due to region-specific growth rates. Respective
production (data from European Environment Agency, 2009) and location-specific data can be found in Follmer et al. (2008), Huynh
electricity costs (data from European Commission, 2012). Specific (2012) and Toh (2011). The conservative estimates of the demand
input data used in the transportation model, e.g. related to costs growth rates are based on historical values from OICA (2012).
and emission for different transportation modes, were obtained Finally, an overview of new vehicle registrations in Europe is
from Helms et al. (2011) and UMBReLA (2014). published annually by ACEA (2011), while data for the Chinese and
An overview of respective data is given in Table 2. It is important American market are adapted from Volkswagen Group (2012).
to note that locations represent regions which contain several Furthermore, due to the European financial crisis especially con-
countries. Hence, for example, data for South-West Europe repre- cerning Southern Europe, the demand growth rates for SWE are
sent an average of the specific values from Spain, France, Portugal, reduced further below the conservative estimate.
Italy and Austria weighted by the respective production volumes.
The necessary labor input per product unit is supposed to be equal
for all production locations, but labor productivity is assumed to 5. Numerical experiments
differ by location. Reference values are taken from Daimler (2011)
and Damme (2011). Furthermore, these parameters are matched A large number of numerical experiments have been conducted
with respective information published in the Harbour Report (see to investigate the dynamics of the industry supply chain and to
Oliver Wyman, 2008). Finally, it should be noted that for 2011 assess the impact of changing powertrain conditions in the vehicle
Chinese production sites are assumed to operate at a productivity fleet in the various locations. It should be noted that all experi-
level of about 30 percent compared to respective European plants ments have been completed by use of the optimization model
(see Yueh, 2009; Schro €der, 2010). However, this productivity gap is outlined in Section 4.1. In total, the optimization model comprises
expected to shrink until 2030 to six percent. 199,688 decision variables, 101,606 constraints, and 1,387,960 non-
zero coefficients. Solution times varied between 650 and 1100 s. As
optimization software CPLEX Optimization Studio 12.2 was used.
Table 2
Location-specific parameter values.

2011 Chinab EEc Germany NWEd SWEe Table 3

Battery data assumptions.
Production capacity 14.49 3.88 5.87 2.14 4.51
[Mio Cars/year] Battery Lithium ion battery Nickel-metal hybrid battery
Unit labor costs [V/h] 2.69 7.45 34.47 28.79 27.52
Sa Cb Pc S C P
Labor productivity [%]a 25 67 90 87 85
Industry electricity costs [V/kWh] 0.07 0.0901 0.1 0.09 0.0835 Life cycle [cycles] 750 750 750 30,000 30,000 30,000
Customer electricity costs [V/kWh] 0.14 0.18 0.21 0.18 0.15 Weight [kg] 175 350 525 22.5 45.0 67.5
Gasoline price [V/l] 1.0 1.38 1.56 1.6 1.48 Capacity [kWh] 18 26 40 0.94 1.30 1.87
Electricity CO2e [gCO2e/kWh] 845 546 503 391 277 Cost [V] 5000 10,000 15,000 700 1300 200
Electricity NOx [gNOx/kWh] 1.86 1.35 0.49 0.63 0.30 a
Battery for small sized vehicles.
a b
Measured in GDP per worked hour (USA 100%). Battery for compact vehicles.
b c
China excl. Hong Kong. Battery for premium vehicles; S: Small car; C: Compact car; P: Premium car.
c d
EE: East Europe incl. CZ, HU, PL, RO, SI, SK, TR. Number of charging/discharging cycles before the battery capacity falls to 80
NWE: North-West Europe incl. BE, DK, FI, NL, NO, SE, UK. percent, cf. Vezzini (2009).
SWE: South-West Europe incl. AT, ES, FR, IT, PT.
H.-O. Günther et al. / Journal of Cleaner Production 90 (2015) 220e233 227

Table 4 5.2. Baseline experiment

Demand assumptions.

Segments Customer locations In a first baseline experiment the moderate trend scenario is
Chinaa EEb Exportc Germany NWEd SWEe
assumed and all terms in the objective function, namely network-
wide costs, emissions and job dismissals receive equal weight.
Small [%] 25.53 47.19 13.21 21.00 42.31 42.31
The results of the experiment were evaluated with respect to the
Compact [%] 58.82 42.70 56.60 60.00 44.87 44.87
Premium [%] 17.65 10.11 30.19 19.00 12.82 12.82 following effects:
2011 [Mio. units] 12.3 1.34 3.00 3.17 3.50 5.27  development of supply chain wide costs and emissions (Section
Annual growth [%] 10.0 2.0 4.0 1.5 1.5 0.8
Fleet 2010 [Mio. units]f 43.6 40.6 40.0 44.25 51.75 104.5
 change in the powertrain mix of vehicle fleets in Europe and
China excl. Hong Kong. China (Section 5.2.2),
EE ¼ East Europe incl. CZ, HU, PL, RO, SI, SK, TR.
Export focused on NAFTA only with CA, MX, US.
 restructuring of supply chains in the automotive industry
NWE ¼ North-West Europe incl. BE, DK, FI, NL, NO, SE, UK. (Section 5.2.3).
SWE ¼ South-West Europe incl. AT, ES, FR, IT, PT.
Starting fleet 2011 assumptions based on ACEA (2011).
5.2.1. Costs and emissions
The left part of Fig. 4 shows the development of the total fleet
5.1. Scenario definition
size as well as costs and emissions over the next 20 years assuming
the moderate trend scenario. Costs are annual costs for production
Future market and technology developments are subject to
processes, transportation activities in the supply chain and for
uncertainty. In the applied sustainability optimization framework,
product usage as defined in Section 4.1.3. The fleet size in scope
external scenario drivers are an important element to reflect these
grows towards 700 million vehicles which is primarily caused by
uncertainties. Hence, three future scenarios are defined to reflect
the development in China. Note that already today the global
best, worst and average case assumptions: the moderate trend
vehicle fleet size is estimated to be above 1 billion vehicles
(“average case”), the economic growth (“best case”) and the stag-
including other continents such as the Americas and other Asian
nation scenario (“worst case”). The scenario-dependent trends are
countries (Sousanis, 2011), so that 700 million vehicles at the lo-
assumed to be global trends relevant to all locations given the fact
cations in scope appear to be a realistic figure. At the same time
that technology and respective progress is available to all regions.
total costs in the considered industry network are also increasing,
In the definition of the scenarios eight different factors are used
which indicates economic growth for the locations participating in
which are considered important for the future development of the
the automotive value chain.
automotive industry supply chain and specifically the adoption of
As can be seen from the right part of Fig. 4 total emissions and
electrified powertrains.
costs in relation to vehicles in the fleet drop continuously from
Table 5 summarizes the scenario definitions. For each factor
4789 kg CO2e and 3313 EUR per fleet vehicle per year in 2012 to-
such as fuel cost growth or battery cost reduction an annual per-
wards 3512 kg CO2e and 2907 EUR per fleet vehicle in 2030,
centage change rate is defined reflecting different economic growth
respectively. Note that costs per vehicle include the costs for
conditions. It is assumed that higher economic growth leads to
making new vehicles plus costs for operating the fleet with existing
higher investments and respective growth rates in green technol-
and new vehicles. As a result, the specific costs and emissions in
ogies. Moreover, energy demand is correlated with economic
relation to the total vehicle fleet can be reduced. From the view-
growth, i.e. in an economically more optimistic scenario energy
point of global climate however, prospects are less favorable since
demand is expected to be higher than in scenarios with lower
the reduced emissions in relation to the vehicle fleet are over-
economic growth rates. Consequently, the prices for oil and energy
compensated by the significant increase of the fleet. In this regard
are expected to be correlated with the economic growth rates.
the obtained experimental results do not support the attainment
In our experiments, we use the moderate trend as standard
towards the internationally agreed 2  C climate protection target
scenario. Results obtained from the other scenarios support the
foreseeing a peak in absolute emissions before 2020 and then a
principal results for the moderate trend. It was observed that these
continuous reduction.
effects are merely slightly shifted to earlier or later periods. For this
Fig. 5 shows experimental results for costs and emissions as well
reason and because of space limitations we do include the full set of
as respective details by sub-categories. Overall results for “making”
scenario results except for the electric vehicle scenarios addressed
vs. “using” vehicles indicate that ca. 53% of the estimated 26 trillion
in Section 5.3.
EUR costs and 55% of the 34 billion tons of CO2e emissions are
linked to the product-usage phase, while the remaining shares are
Table 5 linked to production processes for making the whole car and
Scenario overvieweannual parameter change rates. transport. Further finding is that transport costs and related
Drivers S1a S2b S3c S4d S5e S6f S7g S8h emissions play a minor role. In contrast, production processes in the
make phase of the supply chain determine costs and emissions
Zero 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%
Stagnation 0.5% 0.0% 2.0% 1.0% 0.5% 1.0% 1.0% 1.5%
depending on the specific locations at which the production pro-
Moderate trend 1.5% 0.5% 3.0% 3.0% 1.0% 2.0% 2.0% 3.0% cesses are carried out. As can be seen from the left part of the figure,
Growth 4.0% 1.0% 4.0% 5.0% 3.0% 4.0% 4.0% 5.0% labor costs, process capacity costs and raw material costs are the
Fuel cost growth. main contributors to total process costs in manufacturing. This ef-
Electricity cost growth. fect corresponds with industry practices, where in spite of an
Internal combustion engine cost growth. increasing share of automation still many jobs involving manual
Battery cost reduction. work can be created specifically in emerging countries. Remarkably,
Reduction in indirect emissions due to electricity generation.
Reduction in direct vehicle in use emissions.
electricity and transportation costs have only minor influence on
Reduction in vehicle fuel consumption. total costs. However, detailed optimization results reveal that some
Increase in battery life cycles. highly energy-intensive processes are moved to locations such as
228 H.-O. Günther et al. / Journal of Cleaner Production 90 (2015) 220e233

Baseline results Baseline results

Emissions absolute vehicles, costs, emissions Emissions, specific costs and emissions
Costs Fleet size Costs
Emissios per Fleet Vehicle [kg CO2e/vehicle]
5.000 800 6.000
Emissions [mt CO2e] Costs per Fleet Vehicle [EUR/veh.]
4.500 Costs [bn EUR] 700 5.000 4.789
4.000 Fleet Size [mio.]
600 3.959
3.500 4.000
500 3.512
3.000 3.313
3.057 2.907
2.500 400 3.000
2.000 300
500 100

0 0 0
12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 2012 2020 2030
Periods 20xx Periods

Fig. 4. Total baseline results e costs, emissions and fleet size developmente moderate trend.

China or SWE with lower energy costs. Regarding the vehicle usage experimental results are based on the moderate trend scenario
phase, fuel and service costs show the highest contribution. with equal weights for costs, emissions and job dismissals. First
The right part of Fig. 5 visualizes the emission details occurring insight is that the number of vehicles in China catches up to Eu-
in the closed loop supply chain. Again the product-in-use stage ropean levels at the end of the 20-year horizon. This development
accounts for the majority of the total supply chain emissions. As is due to the 10% growth rate in Chinese demand, while European
expected, tank-to-wheel emissions, which are caused when driving locations have rather low growth rates. Taking a closer look at the
the vehicle, represent the main emission factor with more than 40% powertrain mix (see the reference models introduced in Table 1)
of total emissions. However, the well-to-tank emissions from oil the first striking result is that pure electric vehicles (EV) play almost
exploration to the gas station pump as well as the emissions related no role in China and only a minor role in Europe. In contrast, plug-in
to the generation of electricity for electrified vehicles also show a hybrid (PHV) and hybrid vehicles (HV) gain major shares in Europe,
high degree. All-in-all, well-to-tank plus emissions for generating while in China they remain at a low level. As for combustion en-
electricity have similar relevance compared to fuel-related mis- gines, old models (Old) are continuously replaced by more efficient
sions. In this regard, fuel efficiency of combustion engine vehicles vehicles with internal combustion engines (IC) in all regions. Sur-
can help only in part to reduce total emissions, as least as long as prisingly, ICs are still the growing and dominant powertrain in
crude oil exploration and fuel processing remain very emission- China towards 2030 assuming that no governmental regulatory
intensive. On the manufacturing side, emissions linked to elec- actions are taken. Even hybrid powertrains will not replace ICs.
tricity play a larger role. These emissions are driven by the location- At first glance, the above results appear counterintuitive given
specific energy mix. For instance, a large amount of energy in China the fact that costs, emissions and social impact are equally
is still generated by coal-fired power plants leading to higher weighted in the objective function of the optimization model.
emissions per kWh. This effect is reflected in the experimental Hence, it could have been expected that EVs as the “cleanest”
results. Direct process emissions are mainly caused by raw material vehicle would replace other powertrains. Explanations on the mix
processing steps using specific technologies for steel, aluminum or of powertrains can be derived from Fig. 7. Taking into account the
plastics production. energy mix and emission intensity in a country, pure EV emissions
per km exceed those of plug-in hybrids specifically in China due to
5.2.2. Powertrain mix development of vehicle fleets in Europe and the coal-dominated energy mix. Since it is assumed in the under-
China lying moderate trend scenario that emissions related to energy
Next we analyze the resulting powertrain mix for the vehicle generation are reduced by 1% p.a. and fuel is still very cheap in
fleets in Europe and China towards 2030 as shown in Fig. 6. Again China, EVs do not outperform fuel-based vehicles in terms of the


Cost details Emission Details

Cost share by sub-categories in % Emission share by sub-categories in %

Raw 26.488
Opening & 18.968
2% Process
Fleet Direct
Process 14.052 Product
Service 15%
Capacity In-Use
11% Tank-to-
Transport 584
892 42%
Process Electricity
Labor 28%
Fleet Fuel 17%
35% 14.893
11.544 Well-to-
Electricity 13%
4% Transport
Transport 2%
Costs Emissions
[bn. EUR] [mio t. CO2e]

Fig. 5. Total baseline costs and emissions result e moderate trend.

H.-O. Günther et al. / Journal of Cleaner Production 90 (2015) 220e233 229

Fig. 6. Fleet powertrain mix development in Europe and China.

weighted objective function. Since the fuel efficiency of ICs and HVs 5.2.3. Supply chain implications
is the same in all countries, the reason for the extensive use of HVs In total, the optimization model covers 135 processes in all steps
in Europe compared to China is due to the price for gasoline which of the supply chain from the exploration of natural resources down
is about 56 percent lower in China compared to Europe. In Europe, to reverse steps. Fig. 8 indicates the development of aggregate
however, the additional costs and emissions arising from the pro- process volumes in the supply chain steps for the moderate trend
duction of the electric components are overcompensated by the scenario. To better highlight the development trend the various
reduced usage costs and emissions of HVs. Particularly, in some process volumes are indexed at 100 despite their difference in
European regions, EVs appear most favorable due to a cleaner en- absolute measures. The first finding is that, driven by the total
ergy mix, e.g. in South West Europe where France generates most of vehicle growth, process volumes in nearly all supply chain steps
its electric energy by use of nuclear power plants which is by far grow continuously following similar trends. One exception is the
less CO2e-emission intensive compared to fossil energy-powered growth of recycling volumes, which grow at higher rates than the
electricity. other supply chain steps. This exceptional trend can be explained

Fig. 7. Comparison of vehicle in-use emissions by powertrain Germany and China.

Fig. 8. Process volume development by supply chain step (index: 2012 ¼ 100).
230 H.-O. Günther et al. / Journal of Cleaner Production 90 (2015) 220e233

of jobs would be created in Eastern Europe (þ3.9 million) followed

by NWE and SWE (each þ0.3 million) and China (þ0.1 million). In
Germany, the existing 0.8 million jobs are maintained. The minor
increase of labor units compared to the significant growth in
external demand is due to the supposed considerable productivity
increases in China and EE during the next twenty years. Specifically
for China these numbers would be not satisfactory considering the
huge population and the large number of people looking for jobs.
From the German perspective, given the aging population and the
existing lack of qualified workers, stagnating job numbers would be

5.3. Electric vehicle scenarios influenced by market drivers

Fig. 9. Distribution of process volume by supply chain steps and regions.
In the following, we analyze specific drivers that may push the
electrification of vehicles. In the related experiments we focus on
by the fact that recycling as an innovative technology starts from a specific external factors, but do not consider governmental inter-
considerably lower base level compared to other processes. Thanks vention. First, in the left part of Fig. 10 we demonstrate the size of
to recycling, reverse disposal volumes drop continuously and can the total fleet of electrified vehicles in all considered regions for the
be nearly avoided by 2030. Still exploration of natural resources final period of 2030 under different market scenarios as introduced
continues to grow with the vehicle fleet and OEM components and in Table 5. All scenario results range between 200 and 250 million
assembly follow very much the same trend. electrified vehicles. This represents around 30%e35% of the total
Next the regional implications are investigated. Fig. 9 shows the projected fleet size of 700 million vehicles in 2030 as shown in
share of process units in the various supply chain steps by major Fig. 4. In the scenario-based results it gets obvious that the no
regions cumulated until 2030. The experimental results reveal that change (“zero”) and the stagnation scenario will not lead to a sig-
the different supply chain steps are unequally distributed between nificant fleet of pure electric vehicles (EVs). Only in the moderate
the considered regions. For instance, China shows particularly high trend and in the growth scenario the share of EVs increases, since it
shares in labor intensive tier 1 component manufacturing and in is assumed that the economic growth will drive innovation and
OEM assembly processes as well as in dismantling processes. Under investment in better batteries and in renewable energy. At the
the balanced weighting of the different sustainability objectives of same time prices for fossil energy will increase. As a result, EVs
costs, environmental and social aspects, however, other European outperform fuel-based powertrains in terms of costs and emissions.
regions and specifically Eastern Europe gain considerable process Given the high cost importance of the battery for electric vehi-
unit shares as costs are relatively low and productivity rates cles, we investigate in a second experiment annual Li-ion battery
improving. In addition, the energy mix in Eastern Europe is similar cost reduction rates and their impact on the EV fleet size consid-
to Germany, but better than in China in terms of CO2e-emissions. ering the moderate trend scenario (see the right part of Fig. 10). The
Associated with the growth of process figures is the impact on assumed maximum 5% annual cost reduction is equivalent to a total
jobs and dismissals. Growing process units do not automatically cost decline from about 400 EUR to approximately 240 EUR per
mean more jobs, since productivity gains and automation could kWh in 2020. Stepping from 1% to 5% annual battery cost reduction
compensate the use of labor. The job results related to process unit results in additional ca. 50 million electrified vehicles, most of
numbers, however, are positive on a total location perspective since which are PHVs. Only in the 5% annual battery cost reduction sce-
dismissals and job transfers to other locations are completely nario, the share of pure EVs is increasing slightly, as they get
avoided until 2030 when applying equal weighting of economic, economically more competitive compared to partially fuel-based
social and environmental objectives. The assumed overall demand PHVs. Nevertheless, the experiment reveals that reducing battery
growth even creates between 4 and 5 million new jobs until 2030 in costs should be the main goal for the industry to support electric
Europe and China. However, in the assumed scenario the majority mobility in the future.

Fig. 10. Impact of scenario parameters on the electric vehicle fleet size.
H.-O. Günther et al. / Journal of Cleaner Production 90 (2015) 220e233 231

Still the obtained experimental results do not indicate a break- electric vehicle industry. For China similar to the German market,
through for the electrification of vehicles and pure EVs, in partic- grant incentives help to accelerate the establishment of capacities
ular. One should remember that the baseline experiment foresees a but, according to the detailed model results, most of the produced
fleet of 700 million vehicles in 2030 and even in the most optimistic vehicles during that time are delivered to other regions in Europe
scenarios, the fleet of electrified vehicles never exceeds 250 million. e.g. SWE and NWE.
Hence still 550 million vehicles will be fossil fuel based.
5.4.2. Governmental energy measures
5.4. Electric vehicle scenarios influenced by governmental measures In a second experiment the influence of energy policies and
costs on fleet outcomes is investigated. Fig. 12 illustrates the
In the following, we investigate the impact of governmental number of electrified vehicles added to the fleet due to energy
measures, specifically: measures in Germany and China compared to the baseline. For
Germany a decrease in electricity costs from 0.21 EUR/kWh to 0.15
 Paying direct grants to customers for buying an electric vehicle EUR/kWh and 0.10 EUR EUR/kWh as well as the scenario of electric
to lower the hurdle of still high battery costs and make buying vehicles powered only with green electricity were considered. It
the car more affordable (Section 5.4.1). Countries like Norway or turns out that with little less electricity costs of 0.15 EUR/kWh
Japan already offer this incentive to their population. electrified vehicles are not yet much more cost competitive and
 Regulate the change towards renewable energy, e.g. Germany's there would be little add-on to the baseline. Interesting to note is
“Energiewende”, by making fossil energy more expensive and that the fleet impact of 0.15 EUR/kWh is similar to paying the
pushing the development of renewable energy in the energy governmental grant of 25% for 5 years. However, the governmental
mix at decreasing costs (Section 5.4.2). grant would be much more expensive from a tax impact perspec-
tive, as the 5000 EUR direct grant payment is equivalent to 1360
EUR reduced taxes on electricity. Hence, governments should
5.4.1. Governmental incentives closely validate, if it is more efficient to reduce taxes on electricity
First, the impact of different grant model options in Germany vs. for e-mobility than paying direct grants which may have the same
China on the electrified vehicle fleet is investigated. In the mod- fleet effect but at much less taxes. At 0.10 EUR/kWh the picture
erate trend scenario as basis the number of electrified vehicles in would change and 10 million more electrified vehicles would be
2030 is around 25 million both in Germany and in China summing added to the fleet. Finally, pushing green energy usage for electric
up to 50 million electrified vehicles out of the 200 million total vehicles would generate the highest fleet results of up to 50 million
electrified vehicles shown in Fig. 10. This means that in a moderate vehicles.
scenario, the majority of electrified vehicles would not be used in Due to its high demand growth rates the Chinese market basi-
Germany and China, but in other locations such as South West cally has substantial potentials in founding an electric vehicle
Europe and North West Europe. The impact of different grant prime market. However, since gasoline is comparatively inexpen-
models on the size of the fleet of electrified vehicles is compared in sive and electricity generation comes with significant CO2e emis-
Fig. 11. The models differ by the percentage paid as grant on the sions, from an economic and environmental perspective China
product value ranging from 15% vs. 25% as well as the period for would be not prepared for a sustainable electric vehicle fleet.
which the grant is offered to vehicle buyers (5 years vs. 10 years). In Nevertheless, we simulated options of higher fuel taxes and lower
addition, different market scenarios (stagnation, moderate trend, electricity costs of 0.07 EUR/kWh as well as powering e-mobility
and growth) are considered. It is obvious that the highest fleet just with green energy. Experimental results show that lower en-
growth can be achieved with a grant of 25% paid over 10 years, ergy costs would add some 5 million vehicles and higher fuel costs
leading to some 5 and 10 million additional electrified vehicles in another 5 million vehicles by 2030. Green electricity would be a
Germany and China, respectively. All other grant models are less breakthrough stepping up the number of electrified vehicles from
effective. All-in-all, direct grant payments do achieve the desired 35 to 85 million in 2030, since electrified vehicles no longer have
effect despite their considerable costs. Moreover, the various stra- the emission burden of the coal-heavy energy mix. However, as the
tegies start to make an impact in Germany after 2019 even if they Chinese fleet in 2030 reaches more than 300 million vehicles in
are introduced in 2011. This is due the fact, that during the start-up total, even 85 million electrified vehicles would still represent only
periods only minor production capacities are available in the 25% of the fleet by 2030. A detailed analysis of the experimental

Fig. 11. Impact of governmental incentives on the electrified vehicle fleet in Germany and China.
232 H.-O. Günther et al. / Journal of Cleaner Production 90 (2015) 220e233

Fig. 12. Impact of governmental energy measures on the fleet of electrified vehicles in Germany and China.

results reveals that battery production capacities become a Governmental regulations turned out to have potential larger
bottleneck. Allowing more drastic capacity expansions of battery impact on the fleet, specifically, making sure that electricity costs
production would certainly result in larger shares of electrified are significantly lower than fuel costs. These governmental levers
vehicles in the fleet. appeared to be more sustainable than direct grant payments for
Hence, to found a relevant market for electric mobility, gov- electric vehicles. Ensuring that electric vehicles are powered only
ernment in both countries should focus on transforming the by renewable energy gives also a push and turns around the fleet in
mobility-related energy mix towards renewables and create a Germany with a majority of electrified vehicles by 2030. In China,
stronger cost spread between electricity vs. fuel used to push however, the majority could be still fuel-based vehicles in 2030,
electrified vehicles as more cost-efficient technology. These mea- even in a best-case renewable energy scenario for e-mobility.
sures are economically and environmentally more sustainable than According to our numerical investigation, jobs could be created
paying direct grants. If grants are provided then it is more efficient in all regions driven by market growth. The establishment of an
to give smaller incentives for a longer time than vice versa. electric vehicle industry sector shows no correlation with possible
job cuts. Rather, due to the manufacture of additionally required
6. Summary and outlook electric components and especially due to the production mix of
different powertrain technologies, a further job creation is more
This study investigates future development scenarios for the likely. Germany would secure its current large work base in the
automotive industry supply chain with a focus on electric vehicles industry, but without further growth provided that social objec-
in Europe and China within the next 20 years. A model-based tives are considered by decision makers. China could grow its
quantitative analysis using a comprehensive linear optimization current workforce slightly, but this might not be sufficient for the
framework was conducted based on real world data and specific large population. Generally, all supply chain steps experience vol-
scenarios. The model investigates the possible sustainability per- ume growth with an increasing share of recycling.
formance of the industry by evaluating making vs. using the vehicles In conclusion, the investigation delivered insights and raised
based on an end-to-end supply chain including reverse steps questions to be addressed by policy makers and the industry to be
considering all three sustainability dimensions, namely economic, investigated in more detail when aiming for a sustainable auto-
social and environmental. Key findings are the following. motive industry in line with climate protection targets. As
Over the considered 20-year horizon total greenhouse gas mentioned in the literature reviews referenced in Section 2,
emissions would steadily grow as efficient fuel-based cars are empirical supply chain research considering all three dimensions of
dominating in the fleet especially in China, where electrified ve- the triple bottom line concept is scarce. While a few papers pre-
hicles are still powered by a large share of energy generated in sented integrated quantitative modeling approaches for company-
emission-intensive power plants. Even in the best-case scenario, specific supply chains, a practical framework for analyzing the
emissions do not get on the 2  C climate protection trajectory, but long-term sustainable development of an entire industry supply
still grow again after 2025 due to the large fuel-based fleet. In this chain is still missing in the current literature. The aim of this paper
respect specifically Chinese but also European governments may was to push sustainability supply chain research in this direction
need to put further measures in place such as regulating vehicle and to provide insight into the opportunities for sustainable
usage in cities (zero-emissions only), greening the energy mix even development of the automotive industry considering innovative
harder to also decarbonize the make part of the supply chain as well powertrain technologies.
as bringing more mobility into public transport in order to limit Inevitably, a model-based analysis builds upon specific as-
demand growth for vehicles overall. sumptions and data input. For instance, further technologies such
Electrification of vehicles can be improved with lower battery as fuel cells, sustainable biofuels or natural gas-powered vehicles
costs and, in an even more favorable economic growth scenario, could be included in the investigation. Another extension could be
allowing more investments in clean technologies. However the to explicitly reflect regions like the Americas, other Asian Pacific
total fleet size of electrified vehicles does not grow substantially. countries and Africa for a complete picture. In addition, specific
The main reason is that battery production capacity could be a key scenarios on consumer preference changes in cities, e.g. requesting
bottleneck which may be resolved by drastic expansions towards green vehicles or increasingly using car sharing and public trans-
2020 and 2030. port, could be investigated. Finally, the impact of future battery
H.-O. Günther et al. / Journal of Cleaner Production 90 (2015) 220e233 233

technologies for electric vehicles could be another topic of research Hugo, A., Pistikopoulos, E.N., 2005. Environmentally conscious long-range planning
and design of supply chain networks. J. Clean. Prod. 13 (15), 1471e1491.
as well as examining the possible weight reduction and resulting
Huynh, Q.-N., 2012. The Automobile Industry e Pocket Guide. European Automobile
decrease in fuel consumption through the use of light-weight Manufacturers’ Association (ACEA), Brussels.
chassis exclusively made from materials such as carbon fiber. The uploads/files/ACEA_POCKET_GUIDE_2012_UPDATED.pdf. checked on
mathematical modeling framework used in our study is very flex- 10.02.2013.
Ilgin, M., Gupta, S., 2010. Environmentally conscious manufacturing and product
ible and can thus be applied to other industries. It further enables recovery (ECMPRO): a review of state of the art. J. Environ. Manag. 91, 563e591.
the investigation of other transport technologies such as rail, ship, Kannegiesser, M., Günther, H.-O., 2014. Sustainable development of global supply
aircraft, bus or truck, which all show costs and emissions for make chains e Part 1: sustainability optimization framework. Flexible Serv. Manuf.
26, 24e47.
vs. use and offer alternative fuel and powertrain options. Kannegiesser, M., Günther, H.-O., Gylfason, O., 2014. Sustainable development of
global supply chains e Part II: Investigation of the European automotive in-
References dustry. Flexible Serv. Manuf. 26, 48e68.
Kim, K., Jeong, B., Jung, H., 2014. Supply chain surplus: comparing conventional and
sustainable supply chains. Flexible Serv. Manuf. 26, 5e23.
ACEA, 2011. New Vehicle Registrations e by Country. Technical report. European
Klink, G., Krubasik, S., Weber, T., Mickelson, S., Dyer, S., Mathur, M., 2012. Powertrain
Automobile Manufacturers’ Association (ACEA).
2025 e a Global Study on the Passenger Car Powertrain Market towards 2025.
news_detail/new_vehicle_registrations_by_country/. checked on 10.10.2012.
Technical report. A.T. Kearney.
Ahi, P., Searcy, C., 2013. A comparative literature analysis of definitions for green and
sustainable supply chain management. J. Clean. Prod. 52, 329e341.
checked on 21.11 2012.
Brandenburg, M., Govindan, K., Sarkis, J., Seuring, S., 2014. Quantitative models for
KPMG, 2011. China's 12th Five-year Plan: Overview. URL.
sustainable supply chain management: developments and directions. Eur. J.
Operat. Res. 233 (2), 299e312.
Documents/China-12th-Five-Year-Plan-Overview-201104.pdf. checked on
Bureau of Labor Statistics, 2012. International comparisons of GDP per capita and
per hour, 1960e2011. Technical report. United States Department of Labor,
Lee, S., Geum, Y., Lee, h., Park, Y., 2012. Dynamic and multidimensional measure-
Washington. checked on
ment of product-service system (PSS) sustainability: a triple bottom line (TBL)-
based system dynamics approach. J. Clean. Prod. 32, 173e182.
Chaabane, A., Ramudhin, A., Paquet, M., 2011. Designing supply chains with sus-
Meyr, H., 2004. Supply chain planning in the German automotive industry. OR
tainability considerations. Prod. Plan. Control 22 (8), 727e741.
Spectr. 26 (4), 447e470.
Chaabane, A., Ramudhin, A., Paquet, M., 2012. Design of sustainable supply chains
OICA, 2012. Sales Statistics 2012. checked
under the emission trading scheme. Int. J. Prod. Econ. 135 (1), 37e49.
on 17.03.2013.
Cruz, J.M., 2008. Dynamics of supply chain networks with corporate social re-
Quariguasi Frota Neto, J., Bloemhof-Ruwaard, J.M., van Nunen, J.A.E.E., van Heck, E.,
sponsibility through integrated environmental decision-making. Eur. J. Operat.
2008. Designing and evaluating sustainable logistics networks. Int. J. Prod. Econ.
Res. 184 (3), 1005e1031.
111 (2), 195e208.
Cruz, J.M., Matsypura, D., 2009. Supply chain networks with corporate social re-
Randers, J., 2012. 2052: a Global Forecast for the Next Forty Years. Chelsea Green
sponsibility through integrated environmental decision making. Int. J. Prod. Res.
Publishing, Vermont.
47 (3), 621e648.
Reed, J., 2012. Car Industry's Resistance to Change Comes Tome to Roost. Financial
Daimler, 2011. Daimler Weltweit e Standorte in Europa.
Times. June 7th 2012, p. 15.
unternehmen/daimler-weltweit/europa. checked on 02.02.2013.
Reuters, 2010. China electric Vehicles to Hit 1 Million by 2020. report. http://www.
Damme, G., 2011. Volkswagen Umwelterkla €rung 2011. Technical report. Volkswagen
AG, Braunschweig.
checked on 23.07.2014.
Saft, 2012. Lithium-ion Technology.
Lithium_Liion_301/Language/en-US/Default.aspx. checked on 03.02.2013.
pdf. checked on 02.02.2013. €der, C., 2010. Industrielle Arbeitskosten im internationalen Vergleich. IW
Demir, E., Bektaş, T., Laporte, G., 2014. A review of recent research on green road
Trends Anal. Dok. Progn. 37 (3), 19.
freight transportation. Eur. J. Operat. Res. 237, 755e783.
Seuring, S., 2013. A review of modeling approaches for sustainable supply chain
Dinger, A., Martin, R., Mosquet, X., Rabl, M., Rizoulis, D., Russo, M., Sticher, G., 2010.
management. Decis. Support Syst. 54 (4), 1513e1520.
Batteries for Electric Cars e Challenges, Opportunities, and the Outlook to 2020.
Sheu, J.-B., Chou, Y.-H., Hu, C.-C., 2005. An integrated logistics operational model for
Technical report. Boston Consulting Group.
green-supply chain management. Transp. Res. Part E: Logist. Transp. Rev. 41 (4),
php?ht¼a/GetDocumentAction/id/27906. checked on 12.09.2012.
Elkington, J., 1998. Cannibals with Forks: the Triple Bottom Line of 21st Century
Sousanis, J., 2011. World Vehicle Population Tops 1 Billion Units. WardsAuto. URL:
Business. New Society Publishers, Stoney Creak. checked on
European Commission, 2012. Electricity Household and Industry. Europe's Energy
Portal, November 2012. checked on
Stindt, D., Sahamie, R., 2014. Review of research on closed loop supply chain
management in the process industry. Flexible Serv. Manuf. 26, 268e293.
European Environment Agency (EEA), 2009. CO2 electricity per kWh. Technical
Tang, C., Zhou, S., 2012. Research advances in environmentally and socially sus-
report. European Union.
tainable operations. Eur. J. Operat. Res. 223, 585e594.
co2-electricity-g-per-kwh/co2-per-electricity-kwh-fig-1_2010_qa.xls. checked
Taticchi, P., Garengo, P., Nudurupati, S.S., Tonelli, F., Pasqualino, R., 2014. A review of
on 12.12.2012.
decision support tools and performance measurement and sustainable supply
Federal Ministry of Economics and Technology of Germany, 2010. Establishment of
chain management. Int. J. Prod. Res.
the National Platform for Electric Mobility on 3 May 2010. BMWI press release.
00207543.2014.939239.,did¼346562.html. checked on
Toh, H., 2011. China auto Sector. Technical report. Credit Suisse. https://doc.
Ferretti, I., Zanoni, S., Zavanella, L., Diana, A., 2007. Greening the aluminum supply
chain. Int. J. Prod. Econ. 108 (1), 236e245.
Follmer, R., Gruschwitz, D., Jesske, B., Quandt, S., 2008. Mobilit€ at in Deutschland.
2BJjwGbGzUzhwjw%3D. checked on 12.12.2012.
Ergebnisbericht, infas Institut für angewandte Sozialwissenschaft GmbH and
UMBReLA, 2014. Umweltrechner. last checked on
Deutsches Zentrum für Luft- und Raumfahrt e.V., Institut für Verkehrsfor-
Vezzini, A., 2009. Lithiumionen-Batterien als Speicher für Elektrofahrzeuge. Bulletin
Abschlussbericht_I.pdf. checked on 18.03.2013.
SEV/AES, 06/2009, p. 1.
Glerum, A., Stankovikj, L., The mans, M., Bierlaine, M., 2014. Forecastig the demand
for electric vehicles: accounting for attitudes and perceptions. Transp. Sci. 48
pdf. checked on 23.07.2014.
(4), 483e499.
Volkswagen Group, 2012. Global Automotive Industry Performs Better than Ex-
Govindan, K., Soleimani, H., Kannan, D., 2015. Reverse logistics and closed loop
pected. Technical report.
supply chain: a comprehensive review to explore the future. Eur. J. Operat. Res.
managementreport/businessdevelopment/globaldemand.html. checked on
240 (3), 603e626.
Grimes, S., Donaldson, J., Cebrian Gomez, G., 2008. Report on the Environmental
Oliver Wyman, 2008. The Harbour Report. Technical report. http://www.
Benefits of Recycling. Technical report. Bureau of International Recycling, (BIR).
HarbourMedia08.pdf. checked on 12.11.2012.
pdf. checked on 10.01.2013.
Yueh, L., 2009. China's productivity challenge. Bus. Strategy Rev. 20 (1), 68e71.
Helms, J., Jo € hrens, J., Hanusch, J., Ho € pfner, U., Lambrecht, U., Pehnt, M., 2011.
Zhang, Q., Shah, N., Wassick, J., Helling, R., van Egerschot, P., 2014. Sustainable
UMBReLA Umweltbilanzen Elektromobilita €t Wissenschaftlicher Grund-
supply chain optimization: an industrial case study. Comput. Ind. Eng. 74,
lagenbericht. Technical Report. ifeu - Institut für Energie- und Umweltfor-
schung, Heidelberg.