open TELEMAC-MASCARET The mathematically superior suite of solvers

Dear Users,

Could anyone please advise me on my problem. Searching through the Forum Posts, I have seen the similar error messages. I have tried to applly the some given suggestions (decrease time step, use constant viscosity as turbulence model etc) but it did fix the problem unfortunately. The case, which contain multiple layers stratification, tides, very steep bathymetry is working fine, but when I add Coriolis Force and its coefficient it gives the error message as below, right from the first step:


================================================================================
ITERATION 0 TIME 0 D 0 H 0 MN 0.0000 S ( 0.0000 S)
================================================================================

---

MASS BALANCE
INITIAL MASS OF WATER IN THE DOMAIN : 954613276780.86072
USING STREAMLINE VERSION 6.3 FOR CHARACTERISTICS
INITIALISATION BASED ON TPXO:
- ACQUIRING LEVELS
- INTERPOLATING LEVELS
- ACQUIRING VELOCITIES
- INTERPOLATING VELOCITIES
END OF TPXO INITIALISATION
GRACJG (BIEF) : EXCEEDING MAXIMUM ITERATIONS: 500 RELATIVE PRECISION: NaN
GRACJG (BIEF) : EXCEEDING MAXIMUM ITERATIONS: 500 RELATIVE PRECISION: NaN
GRACJG (BIEF) : EXCEEDING MAXIMUM ITERATIONS: 500 RELATIVE PRECISION: NaN
GRACJG (BIEF) : EXCEEDING MAXIMUM ITERATIONS: 500 RELATIVE PRECISION: NaN
GRACJG (BIEF) : EXCEEDING MAXIMUM ITERATIONS: 500 RELATIVE PRECISION: NaN
GRACJG (BIEF) : EXCEEDING MAXIMUM ITERATIONS: 500 RELATIVE PRECISION: NaN
GRACJG (BIEF) : EXCEEDING MAXIMUM ITERATIONS: 500 RELATIVE PRECISION: NaN
GRACJG (BIEF) : EXCEEDING MAXIMUM ITERATIONS: 500 RELATIVE PRECISION: NaN
GRACJG (BIEF) : EXCEEDING MAXIMUM ITERATIONS: 500 RELATIVE PRECISION: NaN
GRACJG (BIEF) : EXCEEDING MAXIMUM ITERATIONS: 500 RELATIVE PRECISION: NaN


I have attached the steering, result and fortran files, just in case. Thanks in advance.

Kind Regards!

Violeta

Hello Violeta,

I did not look at the Coriolis force yet but for the characteristics in parallel I corrected a bug. In the rare case where a trajectory touches a boundary just before going to another subdomain, the coordinates of the current point on the trajectory were not updated before being sent to the other subdomain, and an infinite loop resulted. I enclose a corrected module streamline.f for version 6.3 and for your case (other subroutines in the modules will be corrected in the next version). You can add it to your Fortran file first and if OK you can even put it in the library BIEF and recompile it.

With best regards,

Jean-Michel Hervouet

Thank you very much Jean-Michel for doing this. I have added it to the fortran file and see how it works. Thanks again for your help.

In terms of Coriolis force. I have run various scenarios of my case and looks like it works with tides fine but with stratificaiton it fails from the very beginning. It could be Characteristics as well but I will run it with modified fortran and see if it works too. Thank you.

I will update you on my results.

Kind Regards!

Violeta

Hello,

It seems to run even with Coriolis. However I am testing with my latest version and I had to remove your trisou.f, for some unclear reason, maybe you can now also remove it now from your Fortran file, since we do not use any specific filter.

Regards,

JMH

Dear Jean-Michel,

Thank you very much. You are right again. I removed trisou.f and it is working! I had to introduce trisou.f to the case long time ago, when we were fixing stratification as it was not stable. But we did a lot of changes around that time, so I am not sure what really fixed the problem, so maybe I do not need trisou.f after all. Fingers crossed:) I have just lunched the simulation and will see results tomorrow morning. The main problem was due to using Method of Characteristics in my stratified case and hopefully, with your great help, it has been gone. I will let you know tomorrow. Thank you very much for your help, it is much appreciated.

Kind Regards!

Violeta

Hello Jean-Michel,

Just a quick update. The simulations are still running, I set it up for one month model time, both with Coriolis and without, by using the new streamline.f. And the good news it has not crashed yet! So looks like it is working fine. As soon as it finished I will save it in bief and recompile the model, as per your advice. Thank you very much. If you do not hear from me then results are good:). Thanks.

Kind Regards!

Violet

Telemac 3D manual notes method of characteristics is not available with parallel version--Is this so ?? or has it been added

Hi Peterbell,

I am running Telemac3d in parallel with scheme of advection of velocities and k-epsilon - 1 (method of characteristics). Also I use Thomson conditions for liquid boundary condition which uses method of characteristics. It is working fine, unless there are other options you had in mind.

Kind Regards!

Violeta

Thank you Violeta for your reply--I am also doing a stratified flow problem including Coriolis.effects ..I ghuess I need to check if my version of streamline.f has the bug fix as provided by JMH

Regards

Peter B

Hi Peter,

Yes, the updated streamline worked for me fine. Firstly to check, you can add it into your fortran file, and then if it is working to replace original file by updated one. But the latest version of telemac might already have it. Here are my cas and fortran files. The stratification in domain has different values on each level, but at the open boundary I have diffrent tracers settings. I used TRA_PROF_Z for that, that is why I have Prescribed Tracers values as 0 in cas file, but it is only when tracers at the open boundary have different values that in domain. Hope it will help. Good luck!

Violeta