Available online at www.sciencedirect.

com

ScienceDirect
Procedia CIRP 54 (2016) 215 – 220

6th CLF - 6th CIRP Conference on Learning Factories

Method for configuring product and order flexible assembly lines in the
automotive industry
Christian Kübera,*, Engelbert Westkämperb, Bernd Kellerc, Hans-Friedrich Jacobid *
a, b ,d
Graduate School of Excellence advanced Manufacturing Engineering, Nobelstr. 12, Stuttgart 70569, Germany
a ,c
DaimlerAG, Mercedesstr. 162, Stuttgart 70546, Germany

* Corresponding author. Tel.: +49 (0) 176/30922128. E-mail address: [email protected]; [email protected]

Abstract

In the last few years the number of offered vehicle derivatives in the multi variant serial production of the automotive industry increased. The
existing assembly lines have to manage many ramp ups. It is necessary to increase the product and order flexibility of existing assembly lines to
manage these challenges. This paper details the preconditions to learn, which assembly configurations fulfill the requirements of existing,
further and future products. Therefore the developed method uses degrees of freedom in the assembly order.
© 2016
© 2016The TheAuthors.
Authors. Published
Published by Elsevier
by Elsevier B.V.
B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the scientific committee of the 6th CIRP Conference on Learning Factories.
Peer-review under responsibility of the scientific committee of the 6th CIRP Conference on Learning Factories
Keywords: automotive industry; assembly; requirements; planning; flexibility; strategy

1. Introduction reconfigurations are necessary. Each reconfiguration causes

costs. These costs can be reduced, if a strategic configuration
The multi variant serial assembly of the automotive can be detected, which will allow to assemble different cars
industry is confronted with different challenges like a growing with different ramp-up dates.
number of car models and global demand changes. It can be This paper is based on the modularization of factories and
managed within the production network but only in a products, synchronizing of the assembly and product life
restricted corridor. Therefore, it is necessary to have product cycle and harmonizing of different assembly configurations
and order flexible assembly lines. [2, 3, 4]. The target is to learn, how existing assembly lines
Today normally one vehicle architecture can be realized on have to be reconfigured, that they can fulfill future product
an assembly line. Vehicle architectures are the platform for requirements. This is based on already existing learning
different cars, like a front or rear driven architecture. The factory approaches, which allow changing the order of
different car models, also called vehicle derivatives, are based different assembly elements [5, 6].
on these architectures. But even the number of the vehicle Before detailing the developed method to increase the
derivatives, which can be assembled on the same line, is flexibility of assembly lines, flexibility has to be defined. In
restricted. There are different reasons, like different assembly this paper the definition follows the view of WESTKÄMPER. A
times between the car models, so they cannot be assembled on system is called flexible, when it is reversible adaptable to
the same line economically [1]. The same assembly processes changing circumstances in the context of a principle
would be located at different positions in the line. preconceived scope of features [2].
An assembly line normally is planned and configured for In this context flexibility is the ability of an assembly line
one vehicle architecture and their derivatives. The product to be able to react to demand changes in quantity and
changes and the ramp-up of new products have an impact on derivatives in a specific scope.
the existing assembly configuration. It has to be analyzed, if An analysis in the automotive industry has shown that the
they can be assembled on the same line and which same vehicles can be assembled in a different order. These

2212-8271 © 2016 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the scientific committee of the 6th CIRP Conference on Learning Factories
doi:10.1016/j.procir.2016.03.051
216 Christian Küber et al. / Procedia CIRP 54 (2016) 215 – 220

degrees of freedom can be used to harmonize the assembly different vehicle derivatives are needed. Based on this
processes of vehicle architectures and an existing assembly information, decisions can be made, whether it has to be
configuration. A precondition to configure such product planned in a green- or in a brownfield.
flexible lines is the vehicle independent modularization of the A greenfield planning means to plan a new assembly line.
assembly processes. The results are assembly modules, which That is necessary, if there is not enough capacity in the
are identical for all the architectures and assembly lines production network or a new vehicle cannot be integrated in
(figure 1). an existing line. Otherwise it is a brownfield planning.
In a brownfield, the assembly time, the business operating
time and the area are already defined. Instead of adopting the
current conditions of the existing line, they can also be further
developed. In this paper, a brownfield planning is focused.
Before starting the planning of an assembly line, the basic
requirements need to be defined, which are shown in figure 2.
Those were considered in the past. But in the future, there is a
need to rethink them. The further development of today’s
preconditions will be amplified in the following chapters.

Figure 1: Using degrees of freedom to enable the planning of product-flexible 2.1. Weight and dimensions of the vehicles
assembly lines
There are many restrictions, which are vehicle dependent:
In the following chapters, the preconditions for an x the weight
adaptable assembly line will be discussed. Further on the x the dimensions (length, width, height)
method to modularize and configure assembly lines, which x the pick-up points
are able to handle future products, will be detailed. During the planning process it has to be ensured that
present and future vehicles can be assembled on the lines. For
2. Preconditions of future assembly lines example, the conveyor system has to be compatible with the
vehicle specific pick-up points and it must be able to handle
An assembly line is comparable with the human immune the heaviest cars. Also the dimensions of the assembly
system. An immunization allows reacting to unknown threats. stations have to be big enough. Based on these restrictions,
Also an assembly line can be immunized. A company has to technical standards can be defined.
learn how an assembly line has to be configured to increase
the product flexibility [7]. Therefore it is necessary to 2.2. Technical standards
reconsider today´s preconditions (figure 2).
The degree of automatization is the most important part of
the technical standards for the assembly processes. Countries
with a high wage level are called high-cost-countries. In these
countries the degree of automatization is normally higher than
in low-cost-countries. That is why two different standards are
described, for the low- and the high-cost-countries.
Automatization is to ensure the product/process quality and/or
to improve the working conditions used in several locations.
The benefits of technical standards in automatization are:
x reduced construction costs and less planning time
(reason: one-time planning)
x higher transparency and better effects of improvements,
(reason: adoption of optimization results to each
identical process)
A premise therefore is the consideration of weight and
dimensions of the vehicles (chapter 2.1) and the same cycle
time of the different assembly lines.

2.3. Assembly cycle time

Today, the cycle time distinguishes between the assembly

lines, even if the same cars are assembled. Further on, the
Figure 2: Preconditions of future assembly lines
assembly cycle time is used to increase or reduce the output.
Different cycle times lead to a different assembly content of
At the beginning, the market demand has to be determined. the stations. It would be impossible to define technical
Afterwards it can be analyzed, which quantities of the standards like explained in chapter 2.2. Each assembly line
 Christian Küber et al. / Procedia CIRP 54 (2016) 215 – 220 217

has to be planned separately. That causes high costs, which output and the sequence of these different derivatives are
can be eliminated by harmonizing the cycle time. After this it restricted because the assembly times are not constant over all
will be possible to plan once and adopt the result to the other vehicles. The reason therefore is that the assembly tasks are
assembly lines. Changes in the output can be managed within configuration dependent. Different assembly times in the
the business operating time instead of changing the assembly stations, depending on the vehicles, can be compensated
cycle time. If the cycle time and the work content remain within the so-called “drifting”. When the vehicle specific
constant, higher learn curve effects will be realized [8]. assembly time is higher than the cycle time, the employee
floats to the next station. If the following car needs less time
2.4. Business operating time the employee can catch up the lost time. A flexible assembly
line would have the option to react to each model-mix
If the cycle time is constant, the increasing of the business scenario. If all vehicle architectures/derivatives have the same
operating time will lead to a higher output and the other way assembly time, there will not be any model-mix restrictions
around [1]. If all the assembly lines have the option to anymore. The cycle time spreading can be reduced by
increase the operating time, there will be a much higher modularizing the products and harmonizing the different
flexibility to increase the worldwide output or vice versa. The assembly times of these modules.
precondition is that all assembly lines have the same
flexibility. Otherwise the line with the lowest flexibility will 3. Cross-architectural modularization of assembly
be the restrictive factor. processes

2.5. Shop-independent area The explained premises are the frame conditions of
adaptable assembly lines. Before using the degrees of freedom
Today the area of a plant is separated in different shops. If to configure cross-architectural assembly lines, the assembly
the area can be used for each shop, there will be the option to processes have to be modularized.
adjust the area distribution retroactively. Therefore, fixed Products and processes can be modularized. The product
installed assembly stations need to be mobile to avoid structure represents the parts of the whole end-product. It
relocation activities, which cause high costs and too much begins with the final product and ends on the component
time. A flexible area distribution enables an expansion of level. The components can be listed on more than one level,
single shops. If vehicles with a longer assembly time are depending on the product structure. To reach a higher level,
launched, the assembly line has to be extended. A constant the parts before have to be assembled. On the upper levels
cycle time makes an extension of the assembly line necessary. there are product modules, which are the result of assembling
The whole assembly time divided by cycle time is the amount submodules [3].
of needed stations. The needed number of stations multiplied
by the length of the stations equals the length of the whole
assembly line. The needed area of the assembly line is the
multiplication of the length and the width. Further areas have
to be added: pre-assemblies, logistics, rework, quality checks,
buffers, inverters, driveways, strategic meaningful open area
and so on. It will be optimal, if all vehicles have the same
assembly time.

2.6. Vehicle specific assembly time

The time needed to assemble a car on the final line

distinguishes between the vehicle derivatives and their
configurations. Within line balancing, time differences can be
harmonized. But that is not possible, if the deviation between
the assembly times is too high. That is the reason why today
only a specific number of vehicle derivatives (based on the
same architecture) are assembled on one line. With a constant
cycle time, a higher assembly time leads to a longer assembly Figure 3: Harmonizing of the product and assembly structure
line (chapter 2.5). The target is to realize comparable
assembly times over all vehicle architectures, derivatives and There will be some benefits, if the assembly lines have the
configurations. Today differences in the assembly time are same structure as the product. In this case, the final line is on
managed within a restrictive model-mix. the highest level, where the final product is assembled and the
submodules are assembled in the pre-assembly lines (figure
2.7. Model-mix 3). The benefits are:
x changes in a product module only affects the linked
Today an assembly line is planned for a defined number of assembly module
vehicle derivatives of the same architecture. The maximum
218 Christian Küber et al. / Procedia CIRP 54 (2016) 215 – 220

x only the pre-assembled parts are assembled in the main 4. Configuration of a product and order flexible assembly
line, so the length and complexity of the final line is line by using degrees of freedom
minimized
x there is less cycle time spreading in the final line, On this consideration level, the pre-defined and
because only the main parts with comparable assembly standardized assembly modules apply for each vehicle. This is
times are mounted, which are independent of the a precondition to analyze the ability to integrate new products
customer individual configuration on existing assembly lines. The degrees of freedom can be
Compared to a product module, an assembly module is considered within an adjacency matrix, a possibility to save
characterized by an assembly time and technical operating precedence graphs [4].
resources.
In the multi variant serial assembly the pre-assembled 4.1. Determining the degrees of freedom within adjacency
product modules are delivered within the logistics to the matrices
assembly modules in the final line. The product modules are
directly linked to the assembly modules [7]. The modules are listed in the column and the row of a
After harmonizing the product and the assembly structure, matrix. In the cells of the matrix, the following question has to
the assembly modules can be defined. be answered: whether there is a directional dependency
between the different modules or not [4]. Such a matrix
represents all degrees of freedom. The matrices have to be
filled for all architectures and the existing assembly line, on
which they have to be assembled. Afterwards, the matrices
have to be harmonized. The result is a matrix with the lowest
degrees of freedom, which fulfills the minimum of identical
degrees of freedom of all the architectures and the assembly
line. The path dependency between the modules makes an
assessment of possible configurations impossible. The degrees
of freedom are only considered between two modules. That
ensures an objectively filling of the matrix by the experts. The
premise is that all possible configurations can be generated,
even if they are completely different from already existing
assembly orders. It will not be manageable anymore, if all the
assembly parts are modularized. There will be too many
modules for which the degrees of freedom have to be filled in.
The numbers of cells, which have to be filled in, are
represented in the following formula:

Table 1: Formula to calculate the number of degrees of freedom to fill in

y amount of cells, for which the degrees of freedom

Figure 4: Modularizing and configure the assembly line has to be determined
x amount of modules
To minimize the complexity in the final assembly line,
only the pre-assembled product modules on the highest level For example: If there are 30 modules, the experts have to
(level 1 in figure 3) in the product structure are assembled. fill in 420 cells. In the classic adjacency matrix, there would
Examples are the flaps (doors, hood, trunk lid), which are be 870 (30²-30) cells to be filled in. This can be halved, within
linked to the chassis. These parts have to be mounted in the a higher information concentration per cell. Instead of giving
main line, even if they cause cycle time spreading. The pre- a binary answer the experts have to choose one of the
assembly modules (PAM) can be linked to the final assembly following three possibilities:
modules (FAM). The modularization can be started by x A: to go from x to y,
beginning at the product structure or at the assembly structure. x B: to go from module y to x,
It is independent if they are identical like shown in figure 3. x C: to go from modules x to y and from y to x.
The results are the different assembly modules. The degrees Afterwards it can be transformed into the classic adjacency
of freedom allow different possible assembly configurations. matrix, but the process to fill in is much leaner and needs less
They have to be detected and evaluated. This will be detailed time (figure 5).
in the following chapter. Instead of evaluating the degrees of freedom between
different architectures also the degrees of freedom of an
already existing assembly line can be detected, because the
assembly modules are the same. It is possible to match them.
This allows answering the question, whether a new or further
 Christian Küber et al. / Procedia CIRP 54 (2016) 215 – 220 219

architecture can be assembled on an existing assembly line in enumeration). Then the process to generate the alternatives
a brownfield planning and which reconfigurations are stops (step 1 in figure 7). The already generated
necessary. That is the transformation from a static assembly configurations are evaluated. Therefore, criteria like the
line to a flexible, learning line. conveyor position, have to be defined before. A change in the
height causes costs. That is the reason why the configuration
with less as possible changes in height should be chosen.
There are lots of further possible criteria. The best rated ones
are chosen (step 2 in figure 7). Based on these ones the
configuration process continues (step 3 in figure 7).

Figure 5: Table to fill in the degrees of freedom

The cross-architectural degrees of freedom allow

generating all theoretical possible configurations. If there are
too many modules and/or degrees of freedom, there are some
problems, which are explained in the following chapter.

4.2. Challenges during the configuration of possible

alternative assembly orders

Today´s computing power is a restriction, if there are too

many configuration alternatives to generate, especially for
exact solving methods like the full enumeration. Further
Figure 7: An objective pre-modularization of assembly modules
alternatives are heuristic solving algorithms, but they cannot
ensure, that the optimal configuration is found [4].
All these possible configurations are also rated based on
the criteria. The result is an approximate optimal assembly
configuration. The benefit is that there is no subjectively pre-
modularization of the assembly modules necessary. These
ones are generated fact-based. The negative aspect is that the
result can be the non-optimal configuration. The optimal
configuration has to take all possible configurations in
account. Otherwise an alternative can be preferred, which is
not the best solution; it is good in the first assembly steps, but
not over all existing ones. That is also the reason, why
approaches to search the shortest way cannot be used. If there
are fewer degrees of freedom, it is possible to skip step 2. The
Figure 6: The interdependence between the amount of modules and degrees
of freedom result would be the optimal configuration.
This method allows answering the question, if a product
A solution might be the summarization of the modules, but can be integrated on an existing assembly line and which
the disadvantage is, that the process is influenced by personal reconfigurations will be necessary. The company learns, to
preferences of the experts. The configuration would be manage the assembly configurations.
comparable with already existing ones. Also the degrees of After the choice of the assembly configuration, the
freedom have to be rated objectively. Thereby the experts assembly circumstances, which are not included in the
have to answer the question, if there is theoretical order modules, can be located in the assembly line to get a more
flexibility between two modules. This is done for each detailed assembly configuration.
module pair. The benefit is that the experts have not to think
about the path dependency. Path dependency can be explained 5. Locating the flexible time blocks
within the following example: When module 1 has to be
before module 2 and module 3 has to be before module 1 then There are two different types of modules; the defined and
module 3 automatically has to be before module 2. the flexible ones. The defined modules were explained before
(see chapter 4). Those are the modules, which are assembled
4.3. The solution to configure an approximate optimal to the basic product directly. But there are also other parts,
assembly order which have to be assembled in the final line. Examples
therefore are doormats, operating instructions, grommets and
The solution is to build all possible assembly temping. It does not make sense to modularize the whole
configurations until a critical level is reached (full assembly line, especially not for the flexible parts, which can
be realized on different positions in the line. The defined
220 Christian Küber et al. / Procedia CIRP 54 (2016) 215 – 220

modules include the needed assembling time. But the sum of Finally, the architecture specific adjacency matrices can be
the assembly times of the defined modules is less than the used to reduce the throughput time within blocking,
whole vehicle needs. Therefore, also the other parts, the interchanging and parallelizing the modules.
flexible time blocks, have to be positioned on the line. That
allows the exact positioning of the defined modules. 6. Conclusion and Outlook

The developed method is based on the modularization of

final assembly processes. These assembly modules are valid
for existing assembly lines and products. This is the
precondition to research, which reconfigurations are necessary
Figure 8: The sum of the assembly time of all defined modules and flexible
to integrate new products on an existing assembly line.
time blocks represents the assembly time for the whole vehicle
Modifications on product and process level increase the
degrees of freedom and offer a wider solution area.
There are four possibilities to locate the flexible time
The rating of the possible configurations is based on
blocks shown in table 2.
strategic relevant criteria. This method supports the learning
Table 2: Locating of the flexible time blocks factory by focusing on strategic dimensions during the
assembly planning process. The benefit is that the planning
process is with less human bias and allows checking possible
assembly line configurations in a short period of time. There
is no standardized precedency graph anymore. The impact of
changes in the degrees of freedom and the strategic criteria
are transparent and influence the configurations. That allows a
permanent learning and immunization of existing final
assembly lines. The result is an assembly line, which is
product and order flexible and fulfills strategic goals.

7. References

[1] Askar G, Sillekens T, Suhl L, Zimmerman, J, Flexibility

Planning in Automotive Plants, Physica Verlag 2007,
235–255.
It is not possible to modularize the whole assembly line, [2] Westkämper E, Factory Transformability: Adapting the
because the flexible modules would have too many degrees of Structure of Manufacturing, in Reconfigurable
freedom, there would be too many possibilities to configure Manufacturing Systems and Transformable Factories,
the lines (chapter 4.1). 2006, 371-381.
A constant allocation of the flexible time blocks also is not [3] Wiendahl, HP, Reichardt J, Nyhuis P, Handbook factory
possible. The reason therefore is the time of the flexible time planning and design. Heidelberg: Springer, 2015.
blocks is dependent of the order of the defined assembly [4] Scholl A., Balancing and sequencing of assembly lines.
2nd ed. Heidelberg, New York: Physica-Verlag, 1999.
modules. The result is not representing the reality. The
[5] Riffelmacher P, Kluge S, Kreuzhage R, Hummel V,
locations of the modules in the line are wrong.
Westkämper E, Learning factory for the manufacturing
The definition of the starting point for each module is a
industry. Digital learning shell and a physical model
complex task, because the starting points are dependent on the factory. CAPE Proceedings 2007,120-131.
prefixed modules. [6] Westkämper E, Digital - Real Learning Factory for
The only usable solution is to allocate the flexible time Manufacturing Engineering. 3rd Conference on
blocks after pre-configuring the assembly line. That allows Learning Factories : Increasing resource efficiency
using the flexible time blocks to harmonize the time through education and training, 2013, 170-181
differences between the vehicle architectures. [7] Küber C, Westkämper E, Keller B, Jacobi H. Planning
method for the design of flexible as well as economic
assembly and logistics processes in the automotive
industry. In: CIRP Conference of Manufacturing
Systems 2016, 556-561.
[8] Löffler, C.; Westkämper, E.; Unger, K. Changeability in
Structure Planning of Automotive Manufacturing.
Procedia CIRP 3, 2012,167–172.

Figure 9: Using flexible time blocks to harmonize assembly times