open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi Fabien,

Sorry for the mistake, I need the perl version in 64bit. Do you when it will be available?

Regards

Nicolas

Hi again,

There is not different versions for 32 ou 64 bit. You need just to compile everythings in 64 bits and use the 64-bit compiled release of metis that you can find it on the website.

My kind regards.


Riadh

Thank you Riadh.

I have make some tests in varaible time step and different desired courant numbers (3 and 5). The run with 3 succeeded but the run with 5 returned a pretty "special error" that I don't know :

================================================================================
ITERATION 53325 TEMPS : 11 H 37 MIN 58.1965 S ( 41878.1965 S)

---

ETAPE DE CONVECTION

---

ETAPE DE DIFFUSION-PROPAGATION
GRACJG (BIEF) : 10 ITERATIONS, PRECISION RELATIVE: 0.3250496E-04

---

ETAPE DE CONVECTION

---

ETAPE DE DIFFUSION-PROPAGATION
GRACJG (BIEF) : 11 ITERATIONS, PRECISION RELATIVE: 0.1471440E-04

---

BILAN DE VOLUME D'EAU
VOLUME DANS LE DOMAINE : 0.9869616E+08 M3
FLUX FRONTIERE 1 : 0.000000 M3/S ( >0 : ENTRANT <0 : SORTANT )
ERREUR RELATIVE EN VOLUME A T = 0.4188E+05 S : 0.3361472E-05
NOMBRE DE COURANT MAXIMUM : 3.901422
PAS DE TEMPS : 0.1370340
@STREAMLINE::SCARACT: THE NUMBER OF TRACEBACK INTERFACE CROSSINGS IGEN > 99



PLANTE : ARRET DU PROGRAMME APRES ERREUR
SORTIE DE PVM : APPEL DE P_EXIT

SORTIE DE MPI

RETURNING EXIT CODE: 2
Duree du calcul : 225854 secondes ( 62:44:14 ) (systeme=0.562 sec)
______________________________________________________________________________
*** RESTITUTION DES FICHIERS ***

- FICHIER DES RESULTATS : pdt_var_solv_1_tnl_2_cour_5_bat_modif.res
dans restihp (regroupe pour 4 processeurs)

________________________________________________________
Execution terminee: telemac2d.bat
________________________________________________________
No compilation/linking/file errors detected.
No execution errors detected.
Returning exit status 0

=========================================================
Telemac System 5.6 to 6.1 - Perl scripts version 6.1
=========================================================
...stopping.

Does somebody know what is it?

The .res file is well created (gretel has worked) and it seems to me that the end of the simulation has nothing special.

Regards

Nicolas

Hello,

You have this message when your velocities tend to infinity in parallel, and more than 99 interfaces between sub-domains are crossed by a characteristics pathline. So it means that the program is diverging, with large velocities...

JMH

Hello,

I'm running a simulation with Telemac 2D coupling with Tomawac (version 6.3, in parallel) and I have the same problem: "THE NUMBER OF TRACEBACK INTERFACE CROSSINGS IGEN>99". All is ok with the telemac2D-only simulation (no coupling).

Maybe my current number is too high but I think I can't use the 'VARIABLE TIME STEP' option with the coupling, can I ?

How can I see where are the nodes with higher current numbers (in order to modify the dissipation in these nodes.. maybe by increasing the friction?) ?

Thanks,

Youen

Hello,

The important thing to know is whether this happens in Telemac-2D or Tomawac, maybe you can know it by using :

DEBUGGER = 1

in both steering files. Actually there are quite large Courant numbers in Tomawac and it could be that in your case the 99 generations in parallel are "normal". Avoiding small depths in Tomawac could be a solution (there is a parameter for a minimum depth). You can also try to reduce the time step in Telemac-2D, it will reduce it also in Tomawac.

With best regards,

Jean-Michel Hervouet

Hello,

It seems to happen in Telemac-2D, during the diffusion-propagation step (cf attached file).
Is it a way to know where the higher velocity are ?
I have tried to reduce the time step in Telemac and the minimum depth in Tomawac but it does not work.
Telemac-2D only works correctly, so the fail is maybe due to the wave driven currents ? I have tried to reduce the coupling step between Telemac and Tomawac (and to reduce Tomawac time step to keep a reasonnable CPU time), but it does not work.
Do you have an idea ?
Thanks,

Youen

Hello

It seems to be a problem of convergence for the solver. If your final linear system is diagonal, the GMRES method will fail. So try to use an other solver to overcome this problem. Basically, we recommand the use of Conjugate gradient (keyword SOLVER = 1 instead of 7 for GMRES)

I hope that this can help

With my best regards

Riadh ATA