open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi,

I am having problems running a simulation of the British Isles. What is strange is that I am able to run a very simple domain with the same steering file fine. Both domains have 1 liquid boundary. The following error arises when trying to run the British Isles model...


CALLING LECLIM
BACK FROM LECLIM
CALLING INBIEF
_____________
runcode::main:
C:\opentelemac\Oceans\TEST:
|runCAS: fail to run
| C:\opentelemac\Oceans\TEST\Steering1.txt_2014-01-16-17h41min02s\out_dragfo2.exe


I am running Windows with 4 processors. I have attached the Steering file, the systel config file and the 2 geometries.

Any help on this would be greatly appreciated.

Cheers

Danny

Hello,

If it is not that the domain is too large for the computer, it could be an error in the mesh, that triggers a crash when preparing the data structure, or the boundary conditions file that does not correspond to the mesh. Can you see the large mesh with a post-processor ? If you cannot find what it is I can test it with all your files, with compiler options that give the line and reason of the crash.

Regards,

Jean-Michel Hervouet

Hi Jean-Michel

I am able to see the large mesh in Blue Kenue. I have attached the files you need to run job, if there is anything else you need to test it let me know.

Thanks for your help.

Danny

Danny,

Your mesh is very poor.
You must specify greater densities at the shore lines.

Hope this helps... Martin

Hi Martin,

I created a course mesh deliberately so that I could quickly check if teh model was working, do you think the low resolution mesh could be why teh model is not running?

Thanks for your help.

Danny

Hi Danny,

The answer is YES!
The mesh you posted will NOT work.

There are many unconnected elements.

Martin

Thanks Martin, I didn't realise, is there a way to check the mesh is ok before running the simulation?

Cheers

Danny

Try starting simply.

1. Start working with one of the Telemac examples to become familiar with Telemacs data requirements.

2. Look at the Blue Kenue mesher tutorial.

Martin

Hello Martin,

It is a question that was raised several times : can we be sure that if BK accepts to write a selafin file, the mesh is then topologically correct, even if not very nice. We have here the case of disconnected elements (in this latter case I'll try to see if we cannot have a clearer error message), but we have also the case of triangles in a corner ("over-constrained" triangles) which is not recommended. I plan also to swap the upside-down elements without stopping the program.

With best regards,

Jean-Michel

Hi Jean-Michel,

All the code behind the functions invoked by
"Edit->T3 Mesh->Analyse Mesh"
was intended to do just that.
It was my intention to have an interactive mesh analyser/editor that would allow the user to visualize and correct many of the various possible mesh errors, overconstrained elements, min angles, max number of segments per node, etc. etc.
Unfortunately that work remains on the back-burner.

Note: Blue Kenue already does guarantee that all elements are "right-side-up" (counter-clockwise) before writing the selafin file.

As for this particular mesh, it has many problems, including disconnects, over-constrained elements, no bathymetry,...

Cheers... Martin