Advice needed on <EXCEPTION> <DVP_1> (DYNA_NON_LINE)

hkboondoggle

Hello Sameer, thanks for your reply.

However, your suggestion is exactly what I did as in my previous description:

4.) ... and disable "Merge coincident nodes and elements".

So right now, I'm unsure if the latest convergence issue is in fact mesh related or not.

Did you also used compound mesh as well in the reference case you shared on 08/25?

I notice that the mesh size of the substrate is larger than that of the ball in the reference case. Does this factor plays a role in the convergence of the calculation?

sameer21101970

if i am not wrong, the slave surface has to be fine mesh then master.

i will go through your zip file and reply later.

give some time.

sameer21101970

It is converging, do below.

Make Sphere (ball)Mesh Fine.

Reduce Time Step to 100

hkboondoggle

Sorry Sameer, but I have to try out your latest suggestion at a later time.

My office desktop, where Code_Aster is installed, is now show BSOD issues. (Thanks Microsoft)

Could you please explain the reason why slave surface mesh has to be finer then that of master?

Also, I find your suggestion of reducing time steps surprising. Because I thought have a smaller time increment could aid the calculation in achieving convergence. Am I mistaken, or just this concept is not applicable in this calculation?

sameer21101970

i had done with your 10000 time step....it shown material law failure.

so with 100...it was smooth...

regarding convergence,,,,,define as finer the mesh more accurate the results....there is a limit to fineness.

mib

hkboondoggle wrote:

Could you please explain the reason why slave surface mesh has to be finer then that of master?

maybe this

code-aster.org/doc/default/en/man_u/u2/u2.04.04.pdf (copy and paste in your browser)

from 2.2.3 Case general

to 2.2.7 Quality of the grid

hkboondoggle

sameer21101970 wrote:
It is converging, do below.

Make Sphere (ball)Mesh Fine.

Reduce Time Step to 100

Sammer, how small was the ball mesh size in your setting?

I ended up using this setting:

[ball = 0.1; substrate = 2.5] & number of time steps = 100

But even with this master-to-slave mesh size ratio in 1.5, it still gave out convergence issue.

I have attached the log of this latest trial FYI.

In which there was a multiple occasion of occurred:

!-----------------------------------------------------------------!

! !

! !

! Les dÃ©formations deviennent trop grandes : !

! => GROT_GDEP sous COMPORTEMENT n'est plus valide. !

! !

! Pour un calcul en !

! grandes dÃ©formations !

! il faut utiliser GDEF_LOG ou SIMO_MIEHE. !

! !

! Pour un calcul hyper-Ã©lastique, utiliser ELAS_HYPER. !

! !

! !

! This is a warning. If you do not understand the meaning of this !

! warning, you can obtain unexpected results! !

!-----------------------------------------------------------------!

Is it something related to the convergence issue?

I am unable to test further shrinking the ball mesh size for the time being, as somehow Salome-Meca is giving out fatal error whenever loading up the mesh of all previous case files:

Any suggestions other than re-installing Salome-Meca?

mib wrote:
maybe this

code-aster.org/doc/default/en/man_u/u2/u2.04.04.pdf (copy and paste in your browser)

from 2.2.3 Case general

to 2.2.7 Quality of the grid

Thanks for the info Mib.

So I suppose this is a hard-coded limitation of CA?

Then this is not a good news for me, because the physics in my case study includes deformation mechanism & local strain hardening effect at the ball / substrate under the same length order.

Are you suggesting that I should look for other simulation tools instead?

Attached file:
stderr_command_salome.log, 157921kb

sameer21101970

Does your system have Graphic Card.

You need to download sm2019 or use CA 14.4 or now 14.6

You are using Windows version, i am using Linux version.

It is better to do Partition of HardDisk, Download Lubuntu 18.04, Sm2019 Linux Version.

Windows, i am not using because of unexpected errors, as you are getting.

refer salome meca installation forum, installation procedure given.

konyaro

Hello,

it is possible to simulate the beginning of the crash of your ball on the plate with Code_Aster but once the distortion of elements gets too high the simulation will stop. You should investigate other ways:

1) Use of linear tetrahedrons can help but it won't be enough anyway.

2) Adaptive meshing would be an option but doesn't exist in Code_Aster. Homard can refine a mesh but not remesh a deformed mesh. Different commercial software handle it.

3) An explicit solver is a good option but I don't know any good opensource ones.

4) SPH could be an option with good opensource codes.

Konyaro

sameer21101970

spheric-sph.org/sph-projects-and-codes

above link for SPH will be helpful.

« Previous Page