Code_Aster forums

Hello,

I would like to resume the problem reported here http://www.code-aster.org/forum2/viewtopic.php?id=16566
Now I am using the code-aster version 11.1.20.

Again, I attached the described example. Some hints for the files: the file disc-spring.py generates the mesh when executed in SALOME (you'll have to modify the script, line 31 to define your working directory). It will produce two meshes: a discspring.med and a support.med; The discspring_contact.comm expects the support as unit 21 and the discspring as unit 20. The creation of the *.export file is then straight forward.

As apparent from the *.comm file, there are two models: with and w/o contact. One can select the desired model with the boolean in line 170: sans_contact=False or  sans_contact=True

Some explanations to the limit conditions of the model (axis-symmetric):

inner top circumference            |      ____
inner bottom circumference      .     |___   -------  _____   outer top circumference
                                                  |              -------- _____|   outer bottom circumference
                                                  .
                                                  | (support not shown)

For the case w/o contact, there are the following limit conditions (model A):
support:
  bottom line fixed (DX=DY=0; line 121)
disc-spring:
  inner top circumference: imposed deflection DY=smax (to be set in lines 120 and 122)
  outer top circumference: DX=0 (line 125)
  outer bottom circumference: DY=0 (line 124)

These limit conditions correspond very much to the effect of the support when calculating with contact (model B):
support:
  bottom line fixed (DX=DY=0; line 121)
disc-spring:
  inner top circumference: imposed deflection DY=smax (to be set in lines 120 and 122)
The outer top circumference and the outer bottom circumference are subject to the contact and do thus not require a limit condition.

The simulation increases in a first step the imposed deflection at the inner top circumference from 0 to smax. The number of steps can be adjusted in line 132. The pressure at the top face of the disc-spring is 0 in that first step.
In a second step, I applied a follower pressure from 0 to the value set in line 126. The imposed deflection at the inner top circumference is smax.

Solving the system does unexpectedly not work for large deflections and/or pressures. It performs even worse for model A, hence, the case w/o contact (i.e. Model B, the contact-model works better sic.) If smax=-0.8 both, model A and model B work for the first step (increasing imposed deflection). When smax=-0.9, the model A (i.e. w/o contact) fails. It enters a flip-flop at node 269 (which is a node close to the inner bottom circumference; no limit condition there) thus diverges. It is surprising to encounter troubles already at such small deflections since the radius of the disc-spring is 125.
The contact version (model B) does not have any problems with really large imposed deflections (until smax=-15, which is quite well beyond the height of the disc-spring).

Both models fail, however, when adding a follower pressure (to compare the two models, only an imposed deflection smax=-0.8 is used since model A does not work beyond that). The contact model (model B) works until a pressure of 0.9. It fails at a pressure of 1.0 (line 126). There again, it enters a flip-flop at node 2 (which is the outer top circumference). Model A fails already at a pressure of 0.7 at node 22 which is close to the outer top circumference. A pressure of 0.7 and 0.9 is still quite low considering the geometry of the disc-spring.

Does anyone have an idea why this happens? I'd be grateful for any hints.

al

Last edited by alberich (2012-05-22 21:58:16)