Does MAZARS really workfor 3D with Non-local?

Sasan · 2023-09-07 08:38:33

Hi,

I have been struggling to get MAZARS to converge for a 3D concrete slab with rebar.

There are a lot of posts here that shows many people have also struggled with MAZARS model, and were not able to resolve the issue.

For example, it needs to be used along with a non-local model to eliminate its mesh dependency (like GRAD_EPSI). However, starting a few years ago, there is no non-local approach available that works with MAZARS.

I am just wondering if anyone here has had good success with this model?

Please share your experience.

Thanks,

Last edited by Sasan (2023-09-07 11:38:49)

MrSnail · 2023-09-07 22:07:09

I did but it was a long time ago.

First you should check what happens if you run it with elastic mat prop.
Does it run? Deformations, reaction forces as expected?
Second step I would change it to non linear, but use a deformation load similar to the original load you want to apply. If it runs again and all as expected than shift to Force load. Deformation loads usually converge better.

Non linear concrete materials are not very well converging as the growing cracks, hot-spots etc they can easily drop convergence.
(A sharp steel plate edge cutting into the concrete can be already enough for that as I remember)
Small time steps, first run with decreased tolerance for accuracy just to see what goes wrong.

Sasan · 2023-09-07 22:36:37

Thanks for the response.

For the linear material, the model works great with no problem at all.

For the nonlinear, I have also tried applying deformation (rather than force), but it did not converge either.

It looks like Code_Aster does not provide any NON_LOCAL approach for MAZARS model anymore, and that could be the problem. I find it hard to believe, because then it would be a backward step for the development of the program.

I have tried Secant matrix (rather than Tangent), changing RESI_GLOB_RELA, applying PETIT deformation, applying boundary conditions in different ways and locations, tried DEFI_LIST_INST with MODE_CALCUL_TPLUS, DECOUPE, etc., different LIST_INST parameters, different MAZARS material parameters. Nothing helps for the convergence of MAZARS model.
Can you think of any other reason?

Thanks,

AsterO'dactyle · 2023-09-16 07:51:36

Hello,

Mazars' model doesn't fit EDF's needs. It's not robust enough.
We prefer other models.
We haven't done away with it yet, but we probably will soon.

#1 2023-09-07 08:38:33

Does MAZARS really workfor 3D with Non-local?

#2 2023-09-07 22:07:09

Re: Does MAZARS really workfor 3D with Non-local?

#3 2023-09-07 22:36:37

Re: Does MAZARS really workfor 3D with Non-local?

#4 2023-09-16 07:51:36

Re: Does MAZARS really workfor 3D with Non-local?

Board footer