Professional Documents
Culture Documents
2005 - Near Net Shape Forging With FEM
2005 - Near Net Shape Forging With FEM
2005 - Near Net Shape Forging With FEM
Contents
1. INTRODUCTION .........................................................................................3
4. FE MODEL ................................................................................................12
8. CONCLUSION...........................................................................................21
1. Introduction
1.1 Background
The forge of Škoda in Mladá Boleslav produces every day 4000 pieces of
product 285Q. This product is a driving wheel of the third speed gear. In figure
1 the final product of 285Q is presented. The product is forged in 3 operation
steps. Each operation consists of a stroke between a male and female die.
The 3-operation method is based on the flashness die technology, developed
in the Škoda forge. Flashness forging technology means that the external
material during production is as less as possible.
Unfortunately, the result after the third and last operation is not perfect. The
die is not completely filled. That’s why a near-net shape study has been done
on this product to get a better result. The result could happen because the
numerical simulation program DEFORM 2D gave good results with the
conditions used in the forge. So this means that the material properties used
for the simulation are wrong. Before there can be made a good study, the
material properties have to be changed.
1.2 Goal
The goal of this project is to optimise the filling of product 285Q at the end of
operation two and three. With optimisation, the flashness forging technology
has to be preserved. The preference is to get easy geometry changes. This
means that an old die easily can be reproduced to a new one.
2. Production explanation
In this chapter, the complete production of product 285Q will be explained. All
the possible problems that appear in the production will be described. Each
operation is defined clearly so a good idea about the main problem of this
subject can be made.
2.1 Preparation
The begin product consist of a steel alloy cylinder with a diameter of 38 [mm]
and a height of 76.16 [mm]. The cylinder will call in the future piece. The piece
is preheated in an oven till it has the temperature of 1100°C. This means also
that the volume of the piece changes because of the big temperature
difference. With the formula the new volume can be calculated.
∆L = L ⋅ α ⋅ ∆T (1)
With expansion coefficient α= 0.000012 [1/K] and ∆T = 1100 [K], the diameter
increases to 38.5016 [mm] and the height increases to 77.165312 [mm]. So
these lengths have to be used in the FE model.
After heating, the cylinder cools for 18 seconds, because in that time the
cylinder is transported from the oven to the first operation place. After that, the
piece has to be placed in the centre of the lower die of the first operation. This
is very important for a good centring of the product. This takes at least 1.5
seconds. The 19.5 seconds of transportation and cooling has to be simulated
as well in the numerical simulation.
2.2 Operation 1
After positioning of the piece, the first stroke takes place. This operation is the
upsetting operation and has a couple of aims. First of all it removes the scale
and redistribute the material to the shape of the end product. Next to that, it
also rejects the piece if the weight tolerance is exceeded. Because if the
weight tolerance is exceeded, the piece will not fit in the die of the second
operation. Further it ensures the exact positioning in the die of the second
operation.
The final distance between the upper and lower die is 23.435 [mm]. In figure
2, the upper and lower dies of operation 1 are presented. The parameters,
which can be changed for the research, are given in figure 2. Only the height
can be changed, because by changing the height, also the angle shall
change. This will happens to keep the same volume between the two dies.
Figure 2: Upper and lower die operation 1 Figure 3: Filling of original geometry after operation 1
After the stroke, the upper die will go up and the piece will be transported to
the lower die of operation 2. This takes 1.5 seconds and in this time, the piece
will cool again.
2.3 Operation 2
After the transportation, the piece will be placed in the lower die of operation
2, which also takes 1.5 seconds. The second operation is the preforming
operation and has also a couple of aims. First of all it creates the central web,
so the principle of internal flash can be used. This deep punch in the middle
has to ensure that some place in the preform cavity in operation 3 remained
underfilled. The internal flash will finally happened in operation 3.
In figure 4, the upper and lower dies of operation 2 are presented. In this
operation a lot of parameters can be changed. This is also the reason that the
solution first has to be found in the second operation.
Now the second stroke takes place. The final distance between the upper and
lower die depends on the filling of the point where the upper and lower die
comes together. See the arrow at figure 4. The filling of the corner has to be
approximately the same as the filling in figure 5. A shape in the corner like
that, gives a good filling in the top of the die and also a good flashness filling
at the end of operation 3.
In figure 6 the filling of the original geometry with the original material
parameters is presented. As you can see, filling in the top is quite good and
so on not a representative simulation. This is because of using the wrong
material properties. In this case the properties of material AISI-5115 are used.
Figure 5: Example of filling after operation 2 Figure 6: Filling of original geometry after operation 2
After the stroke, the upper die will go up and the piece will be transported to
the lower die of operation 3. This takes again 1.5 seconds and in this time, the
piece will cool again.
2.4 Operation 3
After the transportation, the piece will be placed in the lower die of operation
3, which also takes 1.5 seconds. In figure 7, the upper and lower dies of
operation 3 are presented. In this operation only one parameter can be
changed. This because of the fact that the shape of the end product is
prescribed. D can change because after operation 3 a hole will be punched in
the piece.
Now the third stroke takes place. The final distance between the upper and
lower die depends on the filling of the top of the die. If the whole die is filled, it
isn't necessary to press further. In DEFORM 2D this is easily to see with the
graphical function of the normal pressure. If there consist a normal pressure
between the piece and the die, the piece hit the die. In figure 8 the difference
between an unfilled and filled die is presented.
In figure 9 the filling of the original geometry with the original material
parameters is presented. As presented, also here the filling in the top is quite
good and so on not a representative simulation.
After this stroke the simulation is ended and the final result is there. Of course
in reality, the last operation is the punch of the hole in the middle, but no
problem will occur here. The punch of the hole is also the reason why it
doesn’t matter what the shape of the piece is around the area of the internal
space. All the material here will be removed.
First thing what has to do is to get new material properties, instead of the
properties of AISI-5115 because otherwise it is impossible to make a
geometry research on this product. This will be explained in chapter 3.
3. Material research
In this chapter, the material research will be explained and finally the new
material properties will be given. There have to be said that the most work is
done by ing. D. Kesner, because the knowledge was not enough to get the
good results.
3.1 AISI-5115
First the material properties of AISI-5115 will be given. AISI-5115 has no
elastic material properties and has so on only plastic and thermal properties.
The plastic properties are given in figure 10 and the thermal properties are
given in table 1. This data is based on the information DEFORM 2D has in his
program. The plastic properties consist of a flow stress-strain curve with to
strain velocities variated at different temperatures between 400°C and
1100°C. Only the graph at T=400°C is given.
strain velocity = 1.6 [1/s] strain velocity = 40 [1/s]
570
560
550
flow stress [MPa]
540
530
property value
520
thermal 32 [W/mK]
510
conductivity
500
heat 3.9 [J/kgK]
490
0 0,2 0,4 0,6 0,8 1
capacity
strain [-]
emissivity 0.7 [-]
Figure 10: Plastic properties AISI-5115 at T=400°C Table 1: Thermal properties AISI-5115
For each material in the material map, the composition is given. By comparing
the composition of CSN 14220 with the other materials it was easy to see that
there where no better options than AISI-5115.
3.3 Extrapolation
After a look in the data of AISI-5115 in the DEFORM 2D program, the
conclusion was that the extrapolation of the plastic data was wrong. For two
strain velocities, έ=0.1 and έ=40 the flow stress-strain curve is given. But
looking at the post processing data, it was clearly to see that the maximal
strain velocity was around έ=250. DEFORM 2D has to extrapolate this data by
his own and maybe that gives the bad results.
strain velocity = 1.6 [1/s] strain velocity = 40 [1/s] strain velocity = 250 [1/s]
1000
900
800
flow stress [MPa]
700
600
500
400
300
200
100
0
0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9
strain [-]
Figure 11: Plastic material properties of AISI-5115 at T=400°C
So after this conclusion, there is drawn a new curve for έ =250 which is
presented in figure 11. Unfortunately, the results were worse than the original
results. The material was getting weaker instead of stronger. Reason of this is
of course that the new data is not right, especially for the data around ε=0.1 -
0.3.
250 1,40E-05
200 1,35E-05
150 1,30E-05
100 1,25E-05
50 1,20E-05
0 1,15E-05
0 500 1000 1500 0 200 400 600 800 1.000
strain rate = 1.6 [1/s] strain rate = 8 [1/s] strain rate = 40 [1/s]
900
800
700
flow stress [MPa]
600
500
400
300
200
100
0
0 0,2 0,4 0,6 0,8 1 1,2
strain [-]
Figure 13: Plastic material properties of new material at T=20°C
8 24,8
heat capacity [J/kgK]
The most remarkable about the new material data is the inconstant elastic
and thermal properties. In table 3 the only constant material data is presented
and it is obvious that this is quite less if you compare it with the properties of
AISI-5115. An explanation for the new data cannot be given because it’s
complete in the hand of ing. Kesner.
property value
poisson’s ratio 0.3 [-]
emissivity 0.7 [-]
Table 3: Constant material data new material
3.4 Comparison
It is important to find a good way to compare the two materials with each
other. These comparisons also will be used for the geometry study.
First of all there is chosen for the filling of the top after operation 2. This filling
is the main problem and has to be compared with each other. A good filling of
the top in operation 2 means also a good filling in the top after operation 3. In
figure 15 this result for the old and new material is presented.
Furthermore the height between the two dies after operation 2 is given (h2).
This height is presented in figure 16. Operation 2 ended at the moment of the
filling presented in figure 5, so this height gives a good idea when the corner
is filled well.
material h2 H3 h3 ∆h3
old 5.096 13.116 12.676 0.44
new 5.144 12.746 12.786 -0.04
Table 4: Comparison old with new material. Everything is in [mm]
As you already could see in table 4 and figure 15, the new material is a better
and representative one. So from now on, all the research and simulations will
be done with the new material.
4. FE model
The FE model consists of 3 bodies: the upper die, lower die and the piece.
The two dies are rigid bodies and the piece is a plastic body. This is quite
obvious because the dies have to keep their shape and the piece not. In
figure 17 the model is presented at the time of forging of operation 2. As
already mentioned only the half of an axi-symmetric through cut is given.
In table 5 and 6 the friction parameters are presented. There has to be said
that the friction parameters used in the first simulation are other parameters
then with the new material.
The upper die is the only die with a movement in the form of a mechanical
pressure. The total stroke distance of the pressure machine is 280 [mm]. This
means that the movement is formulated with the following formula:
h − 280 = H (2)
h [mm] is the height that the upper die has to go down and H [mm] is the
height that you have to fill in the DEFORM 2D program. The angle of the
movement has to be -90°. This to be sure that the upper die goes vertical
downwards.
5.1 Preparation
As already mentioned in chapter 3, the comparison of all the different
geometries, are on the same way as the comparison of the old and new
material. With these data, there could be made a good choice between all the
available geometries.
The last condition needs some explanation. Easy way of production means
that the new geometry could be made very easily from the old geometry. This
lead to the fact that only material of the die can be taken away. This is
necessary to produce easily a prototype.
In table 8 the information about the height between the die is presented and in
figure 18, the top filling is presented. Looking at these results, there can be
seen that geometry 5 has the best filling and geometry 1 has the highest ∆h3.
Geometry 2 gives no flash, but this is nonsense and the simulation is most
probably not representative. Looking at the geometry changes, it is quite
obvious that geometry 4 has the easiest shape to produce.
geometry h2 H3 h3 ∆h3
original 5.144 12.746 12.786 -0.04
1 5.235 12.906 12.787 0.119
2 5.215 12.685 no flash -
3 4.938 12.851 12.841 0.01
4 4.767 12.805 12.725 0.08
5 4.507 12.738 12.728 0.01
6 4.812 12.817 12.867 -0.05
Table 8: Results global geometry research
The conclusion is that geometry 4 is the best result and already a good result.
This because geometry 4 has the best results compare with the three
conditions. The reason why geometry 4 gives a good result is difficult to say.
The following explanation is a possible one.
The corridor, which connects the top of the piece with the main part, is bigger
now because of increasing the two angles a1 and a4. A bigger corridor leads to
a better flow of material, which leads to a better filling of the top.
The next research is based on geometry 4. This because is done because
easily better results can be made and maybe to get an answer of the reason
why geometry 4 is a good result.
geometry a1 a4
original 6 10
4 10 20
7 15 20
8 10 25
9 15 25
10 10 30
11 12 22
12 8 18
13 14 24
Table 9: Geometry changes accuracy research
In table 10 the information about the height between the die is presented and
in figure 19, the top filling is presented. As presented in table 10, there are
some geometries in a gray field. This means that the complete filling now
happens at least in the other corner of the top after the third operation.
Looking at these results, there can be seen that geometry 11 has the highest
∆h3. It’s maybe better not to choose for a gray result, because it is better to
find the result as close as possible to the original state. Figure 18 shows that
the difference of the filling at the end of operation 2 is small.
geometry h2 H3 h3 ∆h3
4 4.767 12.805 12.725 0.08
7 4.462 12.905 12.815 0.09
8 4.705 12.869 12.771 0.098
9 4.441 12.803 12.744 0.059
10 4.647 12.717 12.796 -0.079
11 4.624 12.955 12.797 0.158
12 4.888 12.889 12.79 0.099
13 4.494 12.883 12.783 0.1
Table 10: Results accuracy geometry research
The maximal stress, force and flow velocity of both operation 2 and 3 are
given. As presented table 11 the maximal stress and force in operation 2 is in
both new geometries higher than the original one, but these values aren’t
worrisome. The maximal stress of geometry 11 is only 16% higher than the
original one. Also the maximal force of geometry 11 is only 14% higher than
the original one.
The flow velocity of geometry 4 and 11 in operation 2 is much higher than the
original one, but this is only good and gives a reason of the good filling of the
top after operation 2.
Looking at the results, operation 3 gives no problem at all. The maximal stress
and force are even lower than the original one. Conclusion is that both new
geometries can be used for the geometry of the new dies. Taken into account
that the results of DEFORM 2D are representative.
In table 13 and figure 20 the results are presented. To compare the results, all
the information of geometry 11 is presented. As you can see, the effect of
changing the height is quite huge. Especially the effect on the filling of the top
after operation two. Looking at geometry 15, there can be concluded that
reducing the height gives a bad result.
geometry h2 H3 h3 ∆h3
11 4.624 12.955 12.797 0.158
14 4.873 12.568 12.62 -0.052
15 4.742 12.988 12.849 0.139
16 4.734 12.829 12.405 0.444
Table 13: Results geometry research operation 1
Raising the height gives a good result. There will be a good filling of the top
and ∆h3 will increase too. This finally results in complete filling in the other
corner, as you can see at geometry 14. The disadvantage of a high hill is the
chance that the piece after the stroke will stick in the upper die. This is
possible because of absent of an ejector at operation 1.
the production of this kind of die is also not easy so it’s certainly not
recommended.
D also defines the space of internal flash. But this will give no problem at all,
because the internal flash space is huge compare to the small differences of
volume of the begin product.
In table 15 the results are presented. To compare the results, all the
information of geometry 11 is again presented. The value h2 is always the
same. This is because till operation 3, all the geometries have the same
simulation. That’s also why there are no figures presented.
geometry h2 H3 h3 ∆h3
11 4.624 12.955 12.797 0.158
17 4.624 14.637 14.587 0.05
18 4.624 11.875 11.795 0.08
19 4.624 10.955 10.866 0.089
20 4.624 9.103 9.025 0.078
Table 15: Results geometry research operation 3
Don’t get confused of the big difference between the values of H3 and h3. This
is only because these two values are measured at D. Only ∆h3 is the useful
parameter in this research. And the highest value of ∆h3 has geometry 11.
So the recommendation now is quite easy: don’t change the shape of the dies
of operation 3. The height of D is now optimal.
7. Elastic research
After the complete geometry research, now a small elastic research follows.
The upper and lower dies were described as rigid bodies in all the previous
simulations. Of course a rigid body is a good approach of the real world. But a
rigid body doesn’t exist. The dies are a little bit elastic in the real world so
that’s why there are made some simulations with elastic dies.
This is done, only to compare the elastic simulation with the normal simulation
and to find out if the rigid die simulation is a good approach. Only the original
geometry and geometry 4 are simulated with elastic dies.
The difference between both kinds of simulations shall not be big, because
the shape of the whole production has the shape of a square, so the bending
moment will be small.
7.1 Results
In table 16 and figure 21 the results are presented. As presented in table 15,
no special results are described. It only says that with the elastic simulation,
you get better results. H2 is smaller and so on the stroke is bigger, because
the dies also will be impressed.
geometry h2 H3 h3 ∆h3
original 5.144 12.746 12.786 -0.04
original elastic 5.07 12.782 12.762 0.02
4 4.767 12.805 12.725 0.08
4 elastic 4.672 12.874 12.765 0.109
Table 16: Results elastic research
Looking at the results of figure 21, it is clearly that the filling of the top in the
elastic simulation isn’t as good as the normal simulation. This will have
something to do with the fact that elastic dies aren’t as stiff as the rigid dies
and can’t fill the piece as good as the rigid dies. But the difference here is also
quite small.
this gives no problem at all, there can be concluded that geometry 4 also don’t
give any problem.
The conclusion is that it is not necessary with this product to create an elastic
simulation. The bending moments are too small in this product and the
calculation time of an elastic simulation is maybe as twice as long as the
normal simulation.
8. Conclusion
The aim of this project is reached. At the Škoda factory the shape of geometry
11 will be used and a prototype will be built. If the geometry has the same
results as in the simulation, this geometry will be used in production.
The only changes of geometry 11 compare to the original geometry are the
two angles a1 and a4 from the die in operation 2. Both angles got higher. A1
changed from 6° to 12° and a4 changed from 10° to 22°. This change is also
easy to produce because only little material has to take away.