open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi all,

I have encounter a problem with culvert in parallel with V6P2 version.
I have updated my version from V6P0 to V6P2 and I have tried a simulation with a model with culverts (after replacing the numbers of sources by the global numbers of nodes in the culvert data file). The simulation ran well in parallel with V6P0 but, with V6P2 I have a flow of -1.10^10 m3/s for all the culverts. I have tried with another model, same problem. (No problem in scalar mode).
Then I've tried with the test case 033_siphon, no problem in scalar, but in parallel, telemac computes a flow through the culvert but the flow stays in the left reservoir (you can see the results on attached files).
I've concluded than the problem comes from the nodes numbering.
I've tested with tubes (039_bridge test case))both in scalar and parallel and the same problem doesn't appear.
Well, finally, it seems that the pb comes from lecsip.f subroutine.
I've attached the modified lecsip.f subroutine which runs well now on my case in parallel with V6P2. Can you please check it (and validate it if it's good)??
Thank you

Olivier

Hello,

You are perfectly right, the part you have commented out is certainly wrong, and what you did will work in scalar mode.

Now we have to correct the guilty part for parallelism...

With best regards,

Jean-Michel Hervouet