open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi,

I work with telemac3D v6p2.
My study area is a lake with 5 rivers. On these 5 liquid boundaries, i impose a flow variation in the time.
After 17 days of simulation, my run crash because vertical and horizontal velocities near the surface become enormous for one outlet.





When i give an "academic value" of flow for each river with the function "PRESCRIBED FLOWRATES", the problem is the same.
We try to modify the bathymetric in the area but the run always crash.


I put my configuration file. Do you see a problem?
BILAN DE MASSE =OUI
LOI DE FROTTEMENT SUR LE FOND =3
COEFFICIENT DE FROTTEMENT POUR LE FOND =30
COEFFICIENT DE DIFFUSION VERTICAL DES VITESSES =1.E-6
COEFFICIENT DE DIFFUSION VERTICAL DES TRACEURS =1.E-6
COEFFICIENT DE DIFFUSION HORIZONTAL DES VITESSES =1.E-6
COEFFICIENT DE DIFFUSION HORIZONTAL DES TRACEURS =1.E-6
PRECONDITIONNEMENT POUR LA PROPAGATION =2
VALEURS IMPOSEES DES TRACEURS= 16.;38.;18.;38.;18.;0.;18.;0.;18.;0.;18.;0.
VERSION NON-HYDROSTATIQUE =OUI
MODELE DE TURBULENCE VERTICAL =3
LOI DE FROTTEMENT SUR LES PAROIS LATERALES =0
MODELE DE TURBULENCE HORIZONTAL =3
REGIME DE TURBULENCE POUR LE FOND =2
SOLVEUR POUR LA DIFFUSION DES VITESSES =1 /defaut
PRECISION POUR LA DIFFUSION DES VITESSES =1.E-6
SOLVEUR POUR LA PROPAGATION =1 / avant :2
PRECISION POUR LA PROPAGATION =1.E-6
SOLVEUR POUR LA DIFFUSION DES TRACEURS =1 /defaut
PRECISION POUR LA DIFFUSION DES TRACEURS =1.E-10
SOLVEUR POUR LA VITESSE VERTICALE =1 /defaut
PRECISION POUR LA VITESSE VERTICALE =1.E-6
SOLVEUR POUR LA DIFFUSION DU K-EPSILON =1 /defaut
PRECISION POUR LA DIFFUSION DU K-EPSILON =1.E-6
CONDITIONS INITIALES ='COTE CONSTANTE'
COTE INITIALE =-0.25
TRAITEMENT DES FLUX AUX FRONTIERES =2;2;2;2;2
OPTION POUR LES FRONTIERES LIQUIDES =2
PROFILS DE VITESSE =1
SCHEMA POUR LA CONVECTION DU K-EPSILON =1
SCHEMA POUR LA CONVECTION DES VITESSES =1
SCHEMA POUR LA CONVECTION DES TRACEURS =5
PRECONDITIONNEMENT POUR LA DIFFUSION DES TRACEURS =34
PRECONDITIONNEMENT POUR LA DIFFUSION DES VITESSES =34
PRECONDITIONNEMENT POUR PPE =34
PRECONDITIONNEMENT POUR LA DIFFUSION DU K-EPSILON =34
NOMBRE DE TRACEURS =2
VITESSE PROJETEE SUR LE FOND =OUI
VITESSE PROJETEE SUR LES PAROIS LATERALES SOLIDES =OUI
NOMBRE DE LISSAGES DU FOND =4
TRANSFORMATION DU MAILLAGE =1 / Sigma pur
CONTOURNEMENT DES VOLUMES NULS =OUI
LOI DE DENSITE =3
PAS DE TEMPS =10
DATE DE L'ORIGINE DES TEMPS =2006;01;01
VALEURS INITIALES DES TRACEURS =20.1;38
NOMBRE DE PAS DE TEMPS =518400 /180j à 30s
STOCKAGE DES MATRICES =3
COEFFICIENT D'INFLUENCE DU VENT =8.2E-6

Thanks a lot
Florian
COMPATIBILITE DU GRADIENT DE SURFACE LIBRE =0.5
NOMBRE DE PLANS HORIZONTAUX =31

Hello,

You seem to have 1 liquid boundary with a Thompson condition ? It is boundary 1 because you have :

OPTION POUR LES FRONTIERES LIQUIDES : 2

Is it really what you want or is there a misunderstanding here ?

This keyword should be given for every liquid boundary, with 5 liquid boundaries, what you do is equivalent to:

OPTION POUR LES FRONTIERES LIQUIDES : 2;1;1;1;1

If you have 5 rivers, why one would be treated differently ? Maybe it is the problem ?

With best regards,

Jean-Michel Hervouet

Hi,

This lake is connected to the sea by a canal also i have 6 boundaries: one boundary with the sea and 5 boundaries with the rivers
I put OPTION POUR LES FRONTIERES LIQUIDES : 2 to configure this maritime boundary.
I try to put OPTION POUR LES FRONTIERES LIQUIDES:2;2;2;2;2;2

But THOMPSON condition can be applied for all liquid boundaries? Even if it's a liquid boundary where we impose a flow?

Hello,

No no, if boundary 1 is the connection to the sea, leave it as it is. For the other boundaries, you should then look at the Froude number, it could be that a boundary becomes supercritical. If you have difficulties with vertical velocities at an entrance, look also at subroutine bord3d.f. At the beginning you will see that you can set the vertical velocities at the entrance to zero (the default option is that they are free).

With best regards,

Jean-Michel Hervouet

Jean-michel,

Thanks for your help.
In bord3d, i must activate these lines?

00285 ! IMPORTANT OPTION:
00286 ! VERTICAL VELOCITIES ARE SET AS HORIZONTAL VELOCITIES
00287 ! THIS IS AN OPTION, OTHERWISE LIWBOL=KSORT (SEE LIMI3D)
00288 !
00289 ! DO IPTFR = 1,NPTFR2
00290 ! IPOIN2 = NBOR2%I(IPTFR)
00291 ! DO IPLAN = 1,NPLAN
00292 ! IBORD = (IPLAN-1)*NPTFR2 + IPTFR
00293 ! LIWBOL%I(IBORD)= LIUBOL%I(IBORD)
00294 ! IF(LIWBOL%I(IBORD).EQ.KENT) WBORL%R(IBORD) = 0.D0
00295 ! ENDDO
00296 ! ENDDO

Thanks
Florian

Hello,

Yes, that's it. By default we let the vertical velocities free because if there is a slope they are not exactly zero.

Regards,

JMH

Hi Jean-Michel,

In bord3d.f, i change the parametrization of the vertical velocities at the entrance. But the run crash...
Apparently, in my mesh, i have a lot of overstressed triangles. These triangles can induce this type of velocities?

I try to use stbtel (i made a post this morning).

Florian

Hello,

It is true that some numerical schemes do not like overconstrained triangles, though, for example, the method of characteristics does not care. Think that an overconstrained triangle has only boundary consitions, so you may inadvertently prescribe velocities that force the flow to have a divergence not equal to 0, without any degree of freedom for the program to correct this situation.
For stbtel I do not know, and Yugi will be here in one week only, do the examples work ?
You can remove the line giving the name of the file stbtel.cas.

Regards,

Jean-Michel Hervouet

Hello,

i test stbtel with the examples that i took on the svn repository. But, i make the same observation. At the end of each run with stbtel, i don't have a new mesh file.
I don't understand why..