open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

I was trying to model Baltic Sea with the mesh I created. I'm not sure if it is the problem of my mesh or other reasons, the simulation didn't run due to an error "access violation".

The problem is found in the subroutine ELEBD (TELEMAC3D --> INBIEF --> ELEBD) during MESH ORGANISATION - 2D LEVEL.

To be more precise, the problem is caused by the following piece of code in ELEBD:
Further investigation shows that the error is caused by using the not-well-initialized matrix KP1BOR and NELBOR in the subroutine ELEBD (the maximum value of a single-precision variable 2147483647 is found in both KP1BOR and NELBOR).

For example, in the matrix KP1BOR, the following values are found:

It is similar in the matrix NELBOR.

Therefore, K2 and IEL may have the value of 2147483647, then N2 = NBOR(K2) will give the error "access violation".I tried to convert the mesh to double precision, but the problem remains.

Is this problem related to the mesh quality? Where are KP1BOR and NELBOR initialized in Telemac-3D?

Thanks in advance!

Best regards,
Qilong

Dear Qilong,

I suspect that there is a topology mistake in your mesh. NELEBD is so intensively used both in scalar and parallel modes that it is doubtful that we find uninitialised values all of a sudden. However it would be better if we can understand the mistake and issue a better message with some additional cross-check. Please send your case so that we can find out what the mistake is. I cannot check (here in Toamasina, Madagascar) where KP1BOR is built, it could be in subroutine voisin, or a similar subroutine not far from the call to voisin in inbief.f.

With best regards,

Jean-Michel Hervouet

Dear Jean-Michel,

Thanks for you suggestion. I have the same doubt. The subroutine ELEBD can handle other meshes without giving errors, so it's very likely to be related to the mesh I created.

Attached are the files for running the simulation. Meanwhile, I will try to make a better mesh and try again.

Have a nice stay in Madagascar!

Best regards,
Qilong

Hi Qilong,

There are definitely some mesh problems that you need
to sort out. I count at least 7 areas where you have sections
of the mesh joined by a single node.
See attached image

Hope this helps... Martin

Thanks a lot Martin!
I'm remaking the mesh to avoid this kind of sections.
I'll keep you updated when I have the new mesh.

Kind regards,
Qilong

Hi Qilong ,
Have you finally found what was the problem in your mesh ?

Was it only because of sections of the mesh joined by a single node?

I have the same problem with this message

"K2= 2147483647 K1= 1088"

Cordially,

Jean-Rémy

Hi Jean-Rémy,

I have solved this problem by remaking the mesh. Indeed like Martin said, there are some elements not formed correctly. Do you use BlueKenue for making a complex mesh?

Kind regards,
Qilong

Ok thank you for your response Qilong.
I use SMS for the mesh but i never had this problem before.
I think i will remake the mesh.

Kind regards
Jean-Rémy

I was using BlueKenue before. Bue due to my poor skills, I may accidentally messed up
during the local refining of the mesh. Then I shifted to Gmsh and then the mesh was fine.

Good luck!

Kind regards,
Qilong

Ok finally i did another mesh, not so far from the precedent, and now it's working.

Thank you
Kind regards,
Jean-Rémy