open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello !

I have used T2D to simulate the breaking of dyke which protect a highly populated riverine area. The problem I faced is the correctness of the results. Here are what appeared incorrect to me:
[The water level in the river is in the order 17.15-17.30 m]
[Maximum velocity is in order of 1 m/s]
[The breach of the lateral dyke has dimensions 50 m in width, bottom elebation of the breach is 9,0 m]
[The water level in the inundation area reached 19 m].

I have attached the entire case (not including the result). It run in few minutes (0.5 or 1 hour of DURRE DU CALCUL is suifficient).

Here are the numerical parameters set in the steering file:
/

---

/
/

---

PARAMETRAGE NUMERIQUE

---

/
/

---

/
PRECONDITIONNEMENT = 2
FORME DE LA CONVECTION = 1;5
/ OPTION DE SUPG = 0;0
OPTION DE SUPG = 1;2
MAXIMUM D'ITERATIONS POUR LE SOLVEUR = 200
SOLVEUR = 7 OPTION DU SOLVEUR = 3 PRECISION DU SOLVEUR = 0.0001
STOCKAGE DES MATRICES : 3 PRODUIT MATRICE-VECTEUR : 1
IMPLICITATION POUR LA HAUTEUR = 0.6 IMPLICITATION POUR LA VITESSE = 0.6
MASS-LUMPING SUR H = 1.
MASS-LUMPING SUR LA VITESSE : 1
LOI DE FROTTEMENT SUR LE FOND = 3
COEFFICIENT DE FROTTEMENT : 30.
MODELE DE TURBULENCE = 1 COEFFICIENT DE DIFFUSION DES VITESSES = 1.
EQUATIONS : 'SAINT-VENANT EF'
BANCS DECOUVRANTS = VRAI
OPTION DE TRAITEMENT DES BANCS DECOUVRANTS : 1
CLIPPING DE H = NON
TRAITEMENT DU SYSTEME LINEAIRE : 1
/DISCRETISATIONS EN ESPACE : 11;11
DISCRETISATIONS EN ESPACE : 12;11


Thanks in advance for the time you will devote to my question.

Georges W.T.

Hello George,

I'm afraid there is no attached file.

Your result seems impossible if there is no constriction effect somewhere due to the geometry of the case ?

Anyway before seeing the case I would suggest, to go much faster and for better results, to use the wave equation option and the latest options ensuring positive depths in version 6.0, i.e.:

OPTION DE TRAITEMENT DES HAUTEURS NEGATIVES : 2
OPTION DE SUPG = 0;0
TRAITEMENT DU SYSTEME LINEAIRE : 2
SOLVEUR = 1
CORRECTION DE CONTINUITE : OUI
COMPATIBILITE DU GRADIENT DE SURFACE LIBRE : 0.9
DISCRETISATIONS EN ESPACE : 11;11

You should get a much faster run (the option solving primitive equations is now seldom or never used). We'll see then the elevation in the inundation area and see what happens.

With best regards,

Jean-Michel Hervouet

Hello Jean-Michel,

Thank you for the quick response. I should say that I thought to have attached all the necessary files for the case and an imager file,but I think I did not do all the steps for a complete attachement. I have done it again. I hope it succeed this time.

To restate my problem. The water in the reservoir is at 17,15 m elevation initially. The water in the inundate area downstream the lateral dyke reach 19 m (i.e. close to 2 m above the initial level which has only a 1 m/s initial velocity).

I need to mention that I used T2D version 5P9. The code is not upgraded for the version 6.



Best regards

Georges W.T.

Still no file. Don't know what happens, if problem you can send it to my E-mail.

You can try also my keywords in version 5.9.

Regards,

Jean-Michel

Hello Jean-Michel,

I tried the parameters you have suggested
OPTION DE TRAITEMENT DES HAUTEURS NEGATIVES : 2
OPTION DE SUPG = 0;0
TRAITEMENT DU SYSTEME LINEAIRE : 2
SOLVEUR = 1
CORRECTION DE CONTINUITE : OUI
COMPATIBILITE DU GRADIENT DE SURFACE LIBRE : 0.9
DISCRETISATIONS EN ESPACE : 11;11
The results are basically of the same sort. I have attached a picture file showing limnigraphs in the river(reservoir)and in the inundation plain. The same quuestionnable behaviour (level in the inundation plain greater than the one in the reservoir) can be seen.

I have received your acknowledgement from the case files I have sent to you by email.

I will be waiting for hearings from you.

Please, have a nice week-end.

Georges W.T.

This appeared to be a problem in parallel runs that occured only for specific numbers of sub-domains. The problem is most probably due to a user programming consisting of changing the bottom elevation of dykes to trigger a flood.

Any action consisting of counting points of a certain property must be done very carefully in parallel, as some points may be in other sub-domains. It is better to work on coordinates than node numbers, though the original global node numbers can be retrieved by array MESH%KNOLG%I(I)
where I is the number of the point in the sub-domain and MESH%KNOLG%I(I) its original global number in the mesh before partitioning.

Jean-Michel Hervouet