open TELEMAC-MASCARET The mathematically superior suite of solvers

Good morning,
I am trying to run simulation with tracers, namely dissolved Pb and Zn . I am first trying to initiate the calculation with constant upstream discharge and concentrations (attached file flperm_tr.txt). My guess is that the concentrations might be homogeneous in the riverbed after some time steps. But the concentrations of the tracers stabilized and the concentrations still decrease from upstream to downstream.
Does someone has an idea that could explain this ?
Maybe the way I parameterized the telemac steering file is wrong (see attached steering file cas_t2d_32168_trac_perm.txt) ? I also tried with DIFFUSION OF TRACERS =NON, but the problem is still there.
Or maybe my guess is wrong ?
Thanks in advance
Renaud Hostache

Hi
Diffusion of Tracers = NO or 1.E-6 for diffusion coefficient should probably similar in term of result. The numerical diffusion is higjer than 1.E-6.
But the advection should work well and some test case shows the tracer conservation is good in telemac.

You could try to use the direct solver to enhanced the precision to the machine value.
Maybe you have some tidal flats that could have an impact on tracer mass...

Hope this helps

Hi thanks for your answer. Actually there is no tide effect taken into account in the model.
I don't really understand the origine of my problem but I finally solved it by playing around with solver parameters. I know use the following parameter and the mass conservation is good:


FORME DE LA CONVECTION = 1;5;14
OPTION DE SUPG = 0;0;0
BANCS DECOUVRANTS = OUI
OPTION DE TRAITEMENT DES BANCS DECOUVRANTS : 1
TRAITEMENT DES HAUTEURS NEGATIVES = 2


COEFFICIENT DE DIFFUSION DES VITESSES = 1.E-6
COEFFICIENT DE DIFFUSION DES TRACEURS = 1.E-6
COMPATIBILITE DU GRADIENT DE SURFACE LIBRE : 0.9
CLIPPING DE H : NON
SOLVEUR : 2
PRECONDITIONNEMENT : 2
STOCKAGE DES MATRICES =3

CORRECTION DE CONTINUITE =OUI
SOLVEUR POUR LA DIFFUSION DES TRACEURS =7
MASS-LUMPING SUR LA VITESSE =1

MASS-LUMPING SUR H =1


Regards
Renaud Hostache

Hello,

I would try also to choose "PRECISION POUR LA DIFFUSION DES TRACEURS=1.E-8 or E-10.

Sometimes when the accuracy is not sufficient, the tracers just stop because the diffusion solvers do no iteration, so have an eye on the number of iterations for diffusion oftracers given in the listing. Choosing a direct solver as C. Coulet rightly suggests should have the same effect. The cost of the direct solver is about like 100 iterations of conjugate gradient, so it depends on the number of iterations you get, but more than 100 iterations for a diffusion equation is not frequent, unless you use the SUPG advection solver, which is not your case.

With best regards,

Jean-Michel Hervouet

Hello,
thanks for this new piece of information. I will actually follow your advice and set "PRECISION POUR LA DIFFUSION DES TRACEURS=1.E-10".
Have a nice day.
Renaud Hostache