World Academy of Science, Engineering and Technology 58 2009

Introducing Fast Robot Roller Hemming

Process in Automotive Industry
Babak Saboori, Behzad Saboori, Johan S. Carlson, Rikard Söderberg

enhance the capacities and capabilities of assembly processes

Abstract—As product life cycle becomes less and less every day, in automotive industry [2], [3].
having flexible manufacturing processes for any companies seems Press-and-die and table-top hemming processes are well
more demanding. In the assembling of closures, i.e. opening parts in suited in mass production due to their low cycle time. On the
car body, hemming process is the one which needs more attention.
other hand these two processes are product specific, i.e.
This paper focused on the robot roller hemming process and how to
reduce its cycle time by introducing a fast roller hemming process. A switching between one product to another is costly as well as
robot roller hemming process of a tailgate of Saab 93 SportCombi it increases production time. Hence the need of flexible
model is investigated as a case study in this paper. By applying task hemming processes is more demanding. In this respect roller
separation, robot coordination, and robot cell configuration hemming process could be a good alternative for assembly
principles in the roller hemming process, three alternatives are processes. Although beyond its high flexibility, it has higher
proposed, developed, and remarkable reduction in cycle times
cycle time compared to conventional methods; press-and-die
achieved [1].
and table-top. This paper tries to introduce a fast roller
Keywords—Cell configuration, cycle time, robot coordination, hemming process to tackle its high cycle time problem while
roller hemming. keeping its quality and flexibility. Before that it is good to
have an overview in the roller hemming process [1].
I. INTRODUCTION
II. ROLLER HEMMING PROCESS

N OWADAYS that customer demands fluctuate regularly,

companies need to change their manufacturing processes
in a way that they can stay alive and keep their market share.
Robot roller hemming process is a relatively new process
which was introduced at the market in the late 90s. This
It is also a case in the automotive industry that every now and process has great application areas in assembling closures, e.g.
then new processes are introduced. It is a must for company hoods, decks, lids, fenders, and tailgates. Closures are
owners to gradually improve their manufacturing processes. comprised of three main components including inner panel,
This improvement process should be in a way for companies reinforcement, and outer panel [4].
to be able to produce their new products in a more competitive The manufacturing of closure is done in three main steps. In
and efficient way. There are many ways to enhance the first step these three components are manufactured
production processes, not only in product quality aspect but separately by stamping, trimming, and flanging subsequently.
also in appearance, e.g. by substituting the welding process of In the flanging process of the outer panel a peripheral edge of
car body closures into assembling inner and outer panels [1]. an outer panel is formed to extend substantially
In this respect different hemming processes are introduced perpendicularly relative to the body of the outer panel. Then
such as press-and-die, table-top, and roller hemming to the closure’s components are transported to subassembly line
where the reinforcements are joined to the inner panels by
spot welding, riveting, or clinching. Finally the complete inner
This work is part of "Produktionsverifieringssystem för rullfalsning" panel is joined to the outer panel by adhesive bonding and
project, which is a collaboration project between Saab Automobile AB,
Svensk Verktygsteknik, and Fraunhofer-Chalmers research Centre. This work hemming processes [4]. In the roller hemming process a robot
was supported by VINNOVA (the Swedish Governmental Agency for guides the roller along the product to bends the outer panel
Innovation Systems) through the MERA program (Manufacturing Engineering over the inner panel, see Fig. 1 [2].
Research Area).
Babak Saboori is with the department of Product and production
development at Chalmers University of Technology (email:
saboori@student.chalmers.se).
Behzad Saboori is with the department of Product and production
development at Chalmers University of Technology (email:
behzads@student.chalmers.se).
Johan S. Carlson is with the department of Geometry and motion planning
at Fraunhofer Chalmers research centre (e-mail:
johan.carlson@fcc.chalmers.se).
Rikard Söderberg is with the department of Product and production
development at Chalmers University of Technology (email:
Fig. 1 Roller hemming process
rikard.soderberg@chalmers.se).

503
 World Academy of Science, Engineering and Technology 58 2009

The roller hemming process is usually accomplished in four different components; two FANUC robots, folding
three steps in which the roller orientations and TCPs, Tool fixture, loading/unloading fixture, and tool changers. The
Center Point, can vary [1]. The first two steps are called robot roller hemming cell is depicted in Fig. 3 [1], [3].
prehemming and the third one is called final hemming. In the
roller hemming process, three different rollers usually are
exploited, i.e. using three rollers in three different steps of
hemming process. In the prehemming steps the roller folds the
outer panel to get 60 degree and 30 degree respectively. But in
the final hemming step, roller would completely fold the outer
panel on the inner panel. This would happen by applying extra
forces, see Fig. 2 [2].

Fig. 3 Robot cell layout at Saab Automobile AB

The roller hemming process of a tailgate is divided into two

different parts, i.e. roller hemming of the upper and lower part
of a tailgate. One robot is responsible to load and unload a
tailgate as well as hem the lower part of a tailgate while
Fig. 2 Three steps in the roller hemming process another robot is only responsible for hemming the upper part
of it. One robot should wait until the other one completely
As roller hemming process is accomplished in three steps, hem the tailgate and leave the hemming surface. It is
the hemming process cycle time is higher than conventional important to mention that the robots are not allowed to stop in
hemming processes. Hence it is crucial to reduce the roller the middle of hemming process. Any stops on the hemming
hemming process cycle time. surfaces would cause quality problems.
The roller hemming process was simulated in IPS software,
III. CYCLE TIME REDUCTION
i.e. Industrial Path Solution software is simulation software
To be able to shorten the high cycle time of roller hemming developed by Fraunhofer-Chalmers research Centre. IPS
process, three principles are introduced which can be applied software mainly entangles with robot path planning and cable
in any roller hemming processes. These principles are simulation [7], [9], [10]. After simulating the roller hemming
described in the following. process in IPS software, the initial cycle time of 90 s, is
acquired [1].
A. Task separation
Task separation and robot coordination principles are
The whole roller hemming process should completely be applied in the roller hemming process and simulated in the
investigated, i.e. necessary tasks should be defined and IPS software. By applying task separation principle, three
allocated to different resources [1]. different tasks are identified for the hemming process of a
B. Robot coordination tailgate; loading, unloading, and hemming process. Then
Coordinate robots to work simultaneously. This would loading and unloading tasks are allocated to a conveyor
minimize the robots’ idle time during the whole roller system while both of the robots would only be responsible to
hemming process [6], [7]. do the hemming process, see Fig. 4 [1].

C. Cell configuration
Configure robot cell layout, i.e. change the orientation of
the robot cell components [8].

In this respect task separation, robot coordination, and cell

configuration principles are applied in a case study of
automotive industry.

IV. CASE STUDY

This case study is a robot roller hemming process at Saab
Automobile AB, Sweden. Particularly robot roller hemming
process of a tailgate of Saab 93 SportCombi model is Fig. 4 Robot cell layout after applying task separation and robot
investigated in this paper [1]. This robot cell is comprised of coordination principles

504
 World Academy of Science, Engineering and Technology 58 2009

Furthermore robot coordination principle is applied to make Task separation impact: 34 – 17 = 17 %

robots work simultaneously. By applying this principle, some
collisions between the robots were figured out. The problem Cell configuration is another principle which can reduce
was solved by exploiting a collision free path planner in IPS cycle time of roller hemming process [8]. In this respect it is
software as well as defining specific time lags for each robot important to first identify whether the robot cell components
during the hemming process [7], [11], [12]. are in the robot operability region or not. Robot operability
By applying task separation and robot coordination region is an area that each robot with respect to its physical
principles three different alternatives are proposed for the and operational constraints has the possibility to reach and do
roller hemming process of tailgate. These alternatives are the task [13]. In order to specify the robot operability region,
common in robot tasks, see Fig. 4, and the only difference is robot specifications and cell layout dimensions need to be
in the way of hemming the tailgate, i.e. both robots start to investigated. By using robot specifications, number of robot
hem the tailgate from the same or different point. These three links, their lengths, and robot operability region are identified.
alternatives are described in the following [1]. As in Fig. 5 a circle with radius of 2405 mm, can resemble
robot operability region [1], [3].
A. Alternative A
One robot hems the tailgate right after the other one finishes
with its task.
B. Alternative B
Two robots start from the same point on the tailgate.
C. Alternative C
Each robot starts from different side of the tailgate.

Alternative A has no robot coordination consideration

whereas it is applied in alternative B and C. These alternatives
are simulated in IPS software and different cycle times are
acquired based on robot coordination and path planning Fig. 5 Robot specifications and operability region
solutions. Furthermore production rate for each alternative is
calculated in (1). The whole cell layout including robots, folding fixture, tool
changers, loading/unloading fixture, and robot operability
Production rate (1/hr) = 3600 / Cycle time (1) region are shown in Fig. 6 [1].

Cycle time and production rate of roller hemming process

for each alternative are given in Table I [1].

TABLE I
CYCLE TIME AND PRODUCTION RATE FOR DIFFERENT ALTERNATIVES

Alternative Cycle time (s) Production rate (1/hr)

A 71 51
B 61 59
C 59 61

As it is shown in Table I, cycle time for alternative A, B,

and C are 71, 61, and 59 s, respectively. Moreover, by
comparing the initial cycle time of roller hemming process
and the one achieved from alternative C, the overall cycle time
reduction of 34 percent is acquired. Among 34 percent
reduction in cycle time, 17 percent is due to applying the task Fig. 6 Robot cell layout and operability region
separation and 17 percent is because of the implementation of
robot coordination principle. The corresponding calculations As in Fig. 6, the folding fixture is in the operability region
on cycle time reduction, robot coordination, and task of robots. It means that there is a possibility for robots to
separation impacts are as following [1]. reach different points of the tailgate. But as it was recognized
from the simulation of roller hemming process in IPS
Cycle time reduction: ( 90 – 59 ) / 90 = 34 % software, one robot spent a lot of time in the corners of lower
part of the tailgate. Hence it is tried to change the folding
Robot coordination impact: ( 71 – 59 ) / 71 = 17 % fixture orientation in terms of rotation and displacement [1].

505
 World Academy of Science, Engineering and Technology 58 2009

Different values for displacement, rotation, as well as for these three alternatives to further improve the cycle time
combination of these two are investigated. Thereafter different achieved by robot coordination and task separation principles.
robot cell configuration alternatives achieved. Finally the best In this respect folding fixture is configured and new cycle
configuration with the optimum orientation for the folding time is gained. It is demonstrated the cycle time reduction of
fixture is chosen and applied in the robot cell simulation, see 26 percent after applying the cell configuration improvement
Fig. 7 [1]. for different alternatives. Consequently alternative C is
recommended for hemming the tailgate of Saab 93
SportCombi model. This alternative has the least cycle time of
19.5 s, i.e. excluding loading and unloading process. As a
conclusion by applying task separation, robot coordination,
and cell configuration principles the overall cycle time
reduction of 42 percent for the roller hemming process is
achieved.

REFERENCES
[1] Babak Saboori, Behzad Saboori, “Development of roller hemming
process in automotive industry – Saab Automobile AB,” M.S. thesis,
Dept. Product and Production Development, Chalmers University of
Technology, Gothenburg, Sweden, 2009.
[2] B. Jonkers, “Voestalpine Polynorm Automotive,” M.S. thesis,
Universiteit Twente, Bunschoten, Netherlands, 2006.
Fig. 7 Robot cell layout after applying cell configuration principle [3] Andreas Engdahl, “En jämförande studie av rullfalsning kontra
konventionell falsning,” B.S. thesis, Dept. Mechanical Eng., Chalmers
By applying cell configuration principle, better cycle time University of Technology, Gothenburg, Sweden, 2006.
[4] Mats Sigvant, “The hemming process, a numerical and experimental
for each three alternatives mentioned earlier in this paper was study,” Dept. Mechanical Eng., Chalmers University of Technology,
achieved. This makes robots to hem the tailgate in an easier Gothenburg, Sweden, 2003.
and quicker way. Overall cycle time reduction is then [5] Allan Tubaileh, Ibrahim Hammad, and Loay Al Kafafi, “Robot Cell
calculated for alternative A, B, and C, see Table II. All the Planning,” In Proc. the 2007 WASET, vol. 20 pp. 250-255, 2007.
[6] R. Bohlin, L. E. Kavraki, “Path Planning Using Lazy PRM,” In Proc. the
cycle time values in this table are excluding loading and 2000 IEEE International Conference on Robotics & Automation San
unloading time which tailgate spend in the conveyor system. Francisco, 2000.
[7] “Automatic Path Planning crash course,” Fraunhofer-Chalmers research
TABLE II Centre, Gothenburg, Sweden, 2009.
CYCLE TIME REDUCTION FOR DIFFERENT ALTERNATIVES [8] V. MATA and A. TUBAILEH, “The machine layout problem in robot
cells,” INT. J. PROD. RES., vol. 36, no. 5, pp. 1273-1292, 1998.
Cycle time (s) [9] Marcus Andersson, “Effektiv simulering och analys av flexibla
cell configuration Cycle time Production rate slangkomponenter,” M.S. thesis, Dept. Product and Production
Alternative
reduction (s) (1/hr) Development, Chalmers University of Technology, Gothenburg,
Before After Sweden, 2008.
[10] VINNOVA, the Swedish Govermental Agency for Innovation Systems
A 54.7 39.5 15.2 27.8 2007, MORE Program Projects Catalog, “Simulation and path planning
B 27.8 21.0 6.8 24.5 for Flexible Materials,” ISSN: 1650-3120, Sweden, 2007.
C 27.2 19.5 7.7 28.3 [11] Johan S. Carlson, Rikard Söderberg, Robert Bohlin, Lars Lindkvist, and
Tomas Hermansson, “Non-nominal path planning of assembly
processes,” In Proc. the 2005 ASME International Mechanical
As in Table II, the cycle time reduction for different Engineering Congress and exposition November 5-11, Orlando, FL,
alternatives is approximately 26 percent. It means that folding 2005.
fixture configuration was quite effective. Furthermore among [12] Johan Segeborn, Johan S. Carlson, Robert Bohlin, and Rikard
Söderberg, “Geometry Design Supported by Minimizing and Visualizing
different alternatives proposed, alternative C has the least
Collision in Dynamic Packing,” In Proc. the 2007 WASET, vol. 25 pp.
cycle time, 19.5 s, hence this alternative is recommended to be 164-171, 2007.
used for the roller hemming process. [13] Dandan Zhang, Liwei Qi, “Virtual engineering II, Optimal cell layout for
industrial robots improves manufacturing productivity,” Special report
dancing with the dragon, ABB review, pp. 90-92, China, 2008.
V. CONCLUSION
In this paper cycle time reduction for the roller hemming
process is being considered to introduce a fast roller hemming
process. By investigating robot coordination and task
separation principles, three different alternatives are suggested
for the roller hemming process. In general, by comparing the
current situation with alternative C, the overall cycle time
reduction of 34 percent is achieved, from which 17 percent is
due to robot coordination and 17 percent by applying task
separation. A robot cell configuration principle is also applied

506