Paper 80-2
Paper 80-2
Paper 80-2
Summary
Friction depends on numerous parameters such as temperature, pressure, sliding velocity,
and surface nature. The knowledge of intensity and distribution of the friction coefficient in
metal forming processes is of importance both from the theoretical point of view and the
practical applications. The first objective is to identify the friction coefficient, from a
simplified test of drawing such as the U-bending test. The comparison between
experimental die force evolution and numerical results form a numerical simulation gives
an approximate friction coefficient. The second objective is to rethink a friction model, with
a drawing test, and to connect local phenomenon with macroscopic physical parameters.
1 Introduction
The putting into service of deep drawing operations is very problematic. One reason is that
the parts have a complex geometry and the productivity is a crucial factor. The other
reason is that deep drawing is uneasy to modelize, due to the friction phenomenon.
The deep drawing conditions such as the blank holder force, the lubricant nature, the
sheet metal material, are very numerous and sometimes hard to control. They must be
chosen to make a part without defect in volume and in surface. The worst defects are the
necking, even cracks (figure 1), where the sheet metal is too stretched, wrinkling and the
shape defects due to spring back.
The most important physical parameter is friction which contains all the parameters of
deep drawing. Friction has effects on the roughness and the thickness of the final product.
An example of a study of a drawn part which was controlled in several points is given on
figure 2. The variation of the friction coefficient µ parameter, in the simulation, gives very
different results for the thinning process.
2,65
thickness (mm)
Thickness (mm)
2,4
Measured
measured thickness
thickness
2,15
µ=0.12
f=0.12
µ=0.05
f=0.05
µ constant per contact
f constant per zone
zone
1,9
0 20 40 60 80 100 120 140
abscisse (mm)
X-coordinate (mm)
Figure 2: Thickness of a drawn part in a section as a function of friction coefficient.
Up to a recent date, the putting into service of deep drawing processes was experimental,
with specialized technicians on industrial press. The objectives are to translate this know-
how into numerical simulation. As a consequence, the manufacturing range would be
optimized, the tools would be better defined, the mechanical properties of the final product
would be foreseeable.
3 Experiment
3.1 Material
The relevance of results simulation depends on the data precision. The material properties
are crucial in the process of drawing. In this case, we tested the HSLA steel sheet used in
car industry.
Behaviour law
We realized tensile tests in the three principle directions: 0°, 45° and 90°. Thanks to
extensometry gauges on test piece, the modulus of elasticity is 220 000 MPa whatever the
direction. The Ludwik law is chosen for the plastic law:
Plastic anisotropy
The results are: r0°=0.66, r45°=1.00 and r90°=1.01. The thickness reduction is specially
important in the rolling direction. The sheets will be cut in the transverse direction.
Force sensor
Die
Displacement
sensor
Sheet metal Blank holder
Punch
Press frame
Figure 3: Modular apparatus in the press for U-bending test.
The displacement of the mobile die and the die force are recorded with an acquisition
system. The difference between a test with and without a sheet, gives the deep drawing
force. We assure three parameters to be constant: the sheet material, the sliding distance
which is the U depth and the deep drawing velocity imposed by the press.
The sheets are not pre-oiled. So, for a lubricated test, we oil all the tools one time and the
sheets for each test.
Repeatability is verified with three tests and we obtained about 10% error at most, for the
same conditions.
Experimental observations
We study the measures from the beginning of drawing to the low break-even point.
140
Force (kN) Displacement (mm)
120
100
80 Die force
Displacement
60
Blank holder force
40 Deep drawing force
20
0
-20 0 0,1 0,2 0,3 0,4
Time (s)
Figure 4: Experimental results/ blank holder pressure=4 bars, sheet 180mm*100mm
lubricated, die radius=5mm.
The blank holder force graph is obtained during a no-load test. The die force graph is
obtained with a HSLA steel sheet. The difference between these two graphs is the deep
drawing force graph. Because of a blank holder force non constant, we had to collect its
evolution for each test. The die displacement is sinusoidal because of the press
mechanism.
We can see a static peak on the die force and the blank holder force graphs, at the
beginning of the test. It corresponds to the contact between the die and the blank or the
blank holder.
Influences on the deep drawing force.
120
120
100
Deep drawing force (kN)
Figure 5: Blank holder pressure influence Figure 6: Area sheet influence on deep
on deep drawing/ sheet 180*100 dry, drawing/ 5 bars, dry, die radius 5 mm.
die radius=5mm.
90
100
80
Deep drawing force (kN)
Deep drawing force (kN)
80 70
60
60
50
40 40
30
20 20
10
0
0
0 0,1 0,2 0,3 0,4 0 0,1 0,2 0,3 0,4
-20 -10
Each parameter has an influence more or less important on friction and so, on the deep
drawing force. Now, we want to reproduce these experimental tests on simulation.
4 Simulation
4.1 Data
We use the software ABAQUS v6 standard, based on the finite element method, which
can solve non-linear problems. The geometry is symmetric by the median plane of the
punch. The material is defined by the Ludwick law (3). The contact between parts is
separated in three cases: sheet/blank holder, sheet/die and sheet/punch. We choose the
model of tangential behaviour which corresponds to the Coulomb’s law (1). This law could
be refined with a dependence to pressure, velocity or temperature.
In this case, the second simulation, with the parameters µdie/sheet = µblank holder/sheet=0.1 and
µpunch/sheet=0.05, approaches the experimental graph, with 5% error. The friction coefficient
is smaller against the punch than these between the sheet and the die, or the blank holder.
This is due to the reduce sliding velocity on the punch. A constant friction coefficient
doesn’t often fit with the simulation. A constant coefficient per contact zone gives
approximate results.
160
140
120
Experiment
100
Die force (kN)
µfdie=0,1/
die=0.1 µfbh=0,1/
punch=0.05
80
µblank holder=0.1
fpunch=0,05
µdie= µpunchfbh=0,1/
fdie=0,05/ =0.05
60
fpunch=0,05
µblank holder=0.1
µ =0,03
f= 0.03
40
20
0
0
10,05 2
0,1 0,15 0,2
3
0,25 0,3
-20
Time (s)
Figure 10: Comparison between experiment and ABAQUS simulations/ 4 bars, sheet
without lubrication, 180*100, die radius=5mm.
4.3 Prospects
The results of friction coefficients will be gathered in tables for a kind of sheet metal. This
method remains quite long and can’t be applied generally to more complex shape of
drawing part. Moreover, several triplet of friction coefficient are acceptable for the same
experimental result.
To optimize the simulation, the friction mechanisms will be approached more precisely. A
friction coefficient more complex, which takes into account local phenomenon such as
pressure or sliding velocity, will be integrated. In that way, a new modular apparatus on the
press will be built. The plastic deformation will be separated from friction, in the drawing
process. This test consist in unrolling a sheet on a fix cylinder and then, on a mobile
cylinder. This test, used by H.Y. Kim [7] and by P. Terreaux [8], seems to be a
complementary test to plane friction.
5 Conclusion
In this study, U-bending tests were performed to analyse the forming and the friction
characteristics of HSLA steel sheets. A simulation of the drawing test, with a variable
parameter, the friction coefficient, is compared with experimental results to deduce the
friction coefficient. Different zones of contact associated to a friction coefficient with a
precision of 10-2 are taken into account. This method based on an inverse method gives
good results, with 10% error at most. Nevertheless, we must optimize our tests and
calculation strategy.
The deep drawing tribology remains delicate to control and to qualify, even for a simple
process like U-bending. But, it is necessary to better know and control it to modelize and
optimize deep drawing process.
References