open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,
I have a SIGSEGV error that comes up while starting a Telemac2D simulation which is a puzzle for me, and hope someone can help.

DOS Script from Listing Printout for the failed run is attached. If required, I can attach the model files, but my Bathymetry file over 370,000 nodes and is 14,500kB.


I admit, the mesh is large and not ideal (some nodes with >10 neighbours). But the model runs in certain situations with stable results. Also, the model used to work fine before I had to reinstall Telemac2D (V6P2).

Since reinstall, I cannot get the model to run with "COMPUTATION CONTINUED".

Is it possible that continuing a compuation with large SLF files for the Previous Computation File is the cause of this problem? But if so, why didn't I have this issue before my reinstall?

Also, I've noticed the line:
USING STREAMLINE VERSION 6.2 FOR CHARACTERISTICS in the Listing Printout which I do not remember seeing before. Is this a clue?

Regards,
Derek

Hello,

The maximum number of elements around a point is 15 in your mesh, and the maximum number of points around a point is 16. I am puzzled because I thought that we had a maximum of 10 for both. I'll check this tomorrow but it is probably a reason of the crash. However the program should issue a clearer message, so I am not sure. You could try "DEBUGGER = 1" in your steering file so that we see more precisely where it crashes.

With best regards,

Jean-Michel Hervouet

Hello,

Actually the limitation on neighbours is 11, and only for frontal matrix-vector product, which is seldom used. So this does not explain your crash.

JMH

Jean-Michel
Thank you for the response.

Do you know why this mesh works in some situations?

I can run a model with this mesh when initiated using INITIAL CONDITIONS; but when I try to run it with COMPUTATION CONTINUED I get the error provided above.

Hello,

No, it is the first time we have this problem, I would need to have the guilty case to investigate it, if you can provide it on a link where we can download it.

With best regards,

Jean-Michel Hervouet

Hello again.

Well this is puzzling:
When I run the model in its current directory, I get the error as listed above. But I was gathering the models into a single directory with simpler file names for sharing, and it runs OK. There is some user error I need to identify here!

But one question still: If I am running a model using a mesh with >10 neighbours, how will it affect my model results?

Hello,

Answer on the large number of neighbours : in this case you have very acute angles of yourtriangles, which means that they may have square or obtuse angles, which is not good (on triangles with obtuse angles, diffusion becomes anti-diffusion, we check this and cancel it but this is a hint that distorted triangles are not very good).

Regards,

JMH

OK... found the source of my error. User error in this case.

I did not provide the CULVERT (SIPHON) data correctly. I had listed 1 more culvert in the Steering File & Culvert Data File header than for which I had provided data.

But some interesting lessons learned here: The error message wasn't as I would have expected. It would be helpful if an error regarding # of Siphons was indicated.

I am still curious to know how my mesh with # of neighbours >10 may be impacting my model though.