Welcome to the forums. Please post in English or French.

You are not logged in.

#1 2019-08-05 17:23:23

stephanscheidl
Member
From: Austria
Registered: 2017-10-24
Posts: 11

[SOLVED] Contact converges based on load multiplier definition

UPDATE:
======

I tried with Code Aster for Windows Version 13.4.0 and the the simulation is running and converging very fast without problems. So it seems to be a version problem.

Could somebody reproduce the convergence problems? Do they also exist on Linux?

ORIGINAL QUESTION:
===============

Hello,

I have a weird problem with a nonlinear contact. I simulate a 2D rotor cross section. I have one non-linear contact with initial shrink fit of 3mm.

Offset0=DEFI_CONSTANTE(VALE=3,);
Frics=DEFI_CONTACT(MODELE=model, FORMULATION='CONTINUE', INFO=2, FROTTEMENT='COULOMB', ALGO_RESO_GEOM='NEWTON', ZONE=(
    _F(GROUP_MA_ESCL='Tar_Spider_Rim', GROUP_MA_MAIT='Con_Spider_Rim', COULOMB=0.2, COEF_FROT=10, CONTACT_INIT='OUI', SEUIL_INIT=1, DIST_MAIT=Offset0)) );

The first time step is calculated at 0 angular velocity. At the next time steps I am increasing the angular velocity. At some point in time the shrink fit contact will open and get inactive. The simulation is running perfectly if I omit one of the time steps, if I want to take all, it fails to converge at time step 3 even though the load is the same as in the previous case.

WORKING:
mult0=DEFI_FONCTION(NOM_PARA='INST', VALE=(0, 0, 1.0, 0.0000, 2.0, 0.8000, 3.0, 1.0000));
rotation=AFFE_CHAR_MECA(MODELE=model, ROTATION=_F(TOUT='OUI', VITESSE=39.2699081698724, AXE=(0,0,1)) );

NOT WORKING:
mult0=DEFI_FONCTION(NOM_PARA='INST', VALE=(0, 0, 1.0, 0.0000, 2.0, 0.5000, 3.0, 0.8000, 4.0, 1.0000));
rotation=AFFE_CHAR_MECA(MODELE=model, ROTATION=_F(TOUT='OUI', VITESSE=39.2699081698724, AXE=(0,0,1)) );

Can sombody tell me why it fails if I use all load steps?

Comm and mesh files are attached.

Code Aster Version 14.02.00

Best regards,
Stephan

Last edited by stephanscheidl (2019-08-06 08:37:24)


Attachments:
fem.zip, Size: 300.83 KiB, Downloads: 18

Offline

#2 2019-08-06 08:40:35

jeanpierreaubry
Guru
From: nantes (france)
Registered: 2009-03-12
Posts: 3,606

Re: [SOLVED] Contact converges based on load multiplier definition

hello

without looking any further at your files
i would say that is every day life with non linear analysis

on has to look at the convergence table in the .mess file and adjust the steps according to what is printed

and quite often i happens that one has to make larger steps instead of short ones

there is an example of such a situation in my book, chapter 12.3.8

jean pierre aubry

Offline

#3 2019-08-06 08:55:08

stephanscheidl
Member
From: Austria
Registered: 2017-10-24
Posts: 11

Re: [SOLVED] Contact converges based on load multiplier definition

Hello Jean Pierre,

thank you for your reply and even more for your wonderful book that helped me a lot in mastering Code Aster. Your example in chapter 12.3.8 was a template for setting up the present simulation.

I am just confused why the simulation is running VERY fast (only few Newton iterations necessary) without problems with version 13.4 and does not converge under certain circumstances with version 14.2. I have to admit that I am using the Windows version of Code Aster. I will also give the Linux version a try.

By the way, do you have any experience in setting up a ParaView server and javascipt client or can you recommend somebody I can contact? I am also willing to pay for any help on this topic.

Best regards,
Stephan

Offline

#4 2019-08-06 09:14:42

jeanpierreaubry
Guru
From: nantes (france)
Registered: 2009-03-12
Posts: 3,606

Re: [SOLVED] Contact converges based on load multiplier definition

By the way, do you have any experience in setting up a ParaView server and javascipt client or can you recommend somebody I can contact? I am also willing to pay for any help on this topic.

no i do not have
as i am using Gmsh and not at all Paraview

Offline

#5 2019-08-06 09:58:20

stephanscheidl
Member
From: Austria
Registered: 2017-10-24
Posts: 11

Re: [SOLVED] Contact converges based on load multiplier definition

Is it possible to use offscreen rendering in Gmsh also on Windows or only on Linux? I need offscreen rendering because I want to create the images in a web service. As far as I have seen I need to compile a special version of Gmsh in order to use offscreen rendering. Do you have any experience with that? Does it work as expected?

I am also looking for a way to add probe values to a print in Gmsh. I attached an image that shows what I am looking for in ANSYS.

Sorry for asking so many questions. I promise this will be the last one ;-)


Attachments:
ANSYS_probe.png, Size: 33 KiB, Downloads: 21

Offline

#6 2019-08-06 13:00:28

jeanpierreaubry
Guru
From: nantes (france)
Registered: 2009-03-12
Posts: 3,606

Re: [SOLVED] Contact converges based on load multiplier definition

i do not understand what is off screen rendering
for the probe value maybe annotation in Gmsh does what you want

Offline

#7 2019-08-06 13:15:31

Anirudh
Member
Registered: 2017-09-11
Posts: 245

Re: [SOLVED] Contact converges based on load multiplier definition

Hi,
You could use paraview to do that. Paraview packs a lot of functionality. You may want to check this tutorial here https: / / \youtu.be/c0w3U5nzo0s
What you want to do is covered somewhere around 19 minutes. You could select any result instead of ID to display it graphically on the geometry.

Regards
Anirudh

Last edited by Anirudh (2019-08-06 13:38:32)

Offline