open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,I met a problem recently.
I use t3d to caculate a long river, with a fortran file in my case file to change some boundary condition on some specific mesh nodes.Firstly, I use telemac3d to caculate some time steps to gain a stable boundary conditions and want to use the result file as the initial conditions for the next caculation. But the result file cannot be the restart file. When I ues it as the restart file, it always shows 'exceeding maximum iterations'. By the way, this case can be run at previous telemac3d version with no mistake,but did not work on my current verison v8p2r1.Also, if I do not use the fortran file in the next caculation, there is no mistake.
I do not konw how solve this problem, need your help. Thanks very much.

Hello,

If there is no mistake without using your Fortran file, this Fortran file is suspicious. Can you upload it?

When you prepare your next computation, what do you exactly do?
It would be interesting to upload your initial steering file which generates your file to restart in addition to the 2nd steering file which continues your computation + Fortran files + every other input files to run your computation and reproduce your "mistakes".
Do you use the 3D RESULT FILE of the 1st computation for PREVIOUS COMPUTATION FILE (+ default value for PREVIOUS COMPUTATION FILE FORMAT) or do you use a RESTART FILE generated with default value for RESTART FILE FORMAT (i.e. SERAFIND) for PREVIOUS COMPUTATION FILE + PREVIOUS COMPUTATION FILE FORMAT = SERAFIND?

It would also be interesting to upload your listing file to read what bothers you exactly.

If you use a Fortran file for the 1st computation, which changes hydraulic conditions and then do not use it in the next computation continuing it, it is not so surprising that you do not have a perfect restart computation.

Hope this helps,

Chi-Tuan

HELLO,
Thanks!

The file I uploaded is not clear enough, now I will upload it again. The first calculation is to obtain stable boundary conditions, and the second calculation is to couple gaia

kind regards,
Shan

Hello Shan,

There is a problem with what you have implemented in your Fortran file source.f if running in parallel. You use global numbers of nodes before partitioning. After the partitioning step, the nodes are numbered differently depending on the subdomain. This kind of implementation is forbidden in parallel. You should rather implement things depending on the coordinates for example.

You can try and adjust the enclosed file which should better work.



Is it OK for you?

Hope this helps,

Chi-Tuan

Hello Chi-Tuan,
Thanks!!! I will adjust the file and try it again. Hope it works and I will tell you the results later!
Thanks again!!
Shan