Optimisation of a deep drawing process with experimental validation: Applied to an automotive deep drawing process of a B-pillar
Wijlen, E.V. ter (2007)
Each car contains between the 200 and 300 sheet metal formed parts. They can be found
amongst others in the body and the chassis of vehicles. Sheet metal parts are generally
manufactured by a deep drawing process. In the past designing such a deep drawing process
was done by experimental trial-and-error in the factory. Since this trial-and-error process is
very time-consuming and costly, Finite Element simulations have been developing to move
the trial-and-error procedure from the factory to the computer which makes the process
design much faster and cheaper. The next step is to optimise the manufacturing process in
the automotive industry to obtain a robust process with no scrap and low costs.
The assignment is to apply optimisation techniques to a real automotive deep drawing
process of a B-pillar. This will be done by executing four tasks:
• Applying the optimization strategies to determine the variables with the most influence
on the B-pillar.
• To evaluate the material models which are included in AutoForm with the ones provided
by Corus.
• To determine the numerical trends with the design variables.
• To verify the numerical trends with experimental tests.
The optimisation strategy is applied to the manufacturing process of the B-pillar. First
the 7 step methodology is applied to model the manufacturing process of the B-pillar.
The outcome is a mathematical optimisation problem with one implicit constraint and ten
design variables. The objective is to maximize the distance of the strains to the forming
limit curve but stay above the wrinkling line, constraint. The most influential variables,
for this problem, are determined with OptForm and AutoForm Sigma which yielded to one
process variable, the blank holder force and three geometrical variables.
Several different materials can be used in the deep drawing process. In cooperation
with Corus, three materials were chosen out of their range, namely TRIP700, H340LAD
and DP600. The difference between the Corus material models and the ones which are
implemented in AutoForm are specified. The difference can be found in the FLC, yield
surfaces and hardening curves. This is the reason why the material models in the database
of AutoForm are, “stronger” then the ones provided by Corus.
The four design variables are varied one by one while the others are set to the reference
setting, while recording their effect on the responses, necking and wrinkling. The outcome
is the effect of the variable on the response in a graphical display, the so called scatter
plots. Also two variables are varied while the other two are set to there reference setting,
the outcome is recorded in a surface plot. A remarkable conclusion was drawn during this
investigation, a non continuous trend was discovered in a certain region in the B-pillar with
the process variable, blank holder force.
To validate the trends, which are found with the FEM simulations, experimental tests
are conducted at the University of Dortmund, with H340LAD and DP600. The first objective
was to determine the process window of the different materials, the upper (necking)
and lower (wrinkling) limit of the blank holder force. The other tests that are conducted
were to determine the effect of the geometrical variables on the deep drawing of the B-pillar.
The tests were conducted during two days. With the material DP600 a large difference,
between the two days, was found. With the same blank holder force, the B-pillar showed a
crack at day 2. Possible explanations for this difference are: the position of the blank and
friction. To determine if these possibilities are responsible for the difference, in blank holder
force, the press data was evaluated. To determine the effect of the different possibilities the
effective punch force is evaluated. The outcome is that the position had some influence on
effective punch force. If one compares the experiments, with the same blank holder force
for DP600 at day 1 and day 2, the effective punch force increased. This is the reason that
the B-pillar started to neck at a lower blank holder force at day 2 in comparison with day 1.
The explanations for the increase in effective punch force is the increase of friction. What
did cause this friction was unknown at that moment. To evaluate the effect of position
and friction additional FEM simulations are preformed. This endorses the statement of the
large influence of friction.
The conclusion that can be drawn is that the trends of the FEM simulations are similar
with the trends found with the experiments. But the exact values of the trends are different.
This is caused by the variation in experiments and the limitation of FEM simulations.
msc2007_terwijlen.pdf