open TELEMAC-MASCARET The mathematically superior suite of solvers

Dear all,

I have a really simple question for you!

Have you got a corresopnding simulation time with follow parameters:

Nodes Time step Time of simulation
10 000 1000 1 Hour
20 000 2000 4 Hours
50 000 2000 1 day

For TELEMAC 2D or SISYPHE model with a standard computer (32bits - 3 GHz - 2 Go Ram)

Best Regards !

Edouard D

Hello,

Very hard to answer, it depends on how you do the finer mesh with 50000 nodes, it can trigger larger courant numbers, or could it be that the cash memory of the computer is more suited for meshes up to 20000 nodes. A listing with numbers of iterations of solvers would give more information.

regards,

JMH

Hello,

Je cherche juste dans un premier temps à estimer l'impact du nombre de noeuds dans un modèle.

Je suis en train de répondre à une offre et j'essai d'anticiper les temps de simulation passés en fonction de la précision demandée.

Existe t il des courbes de comparaison? (ex: si on diminue la taille de la maille de 50% , est ce qu on augmente le temps de simulation de 80%?)



Bien à vous,

Hello,

When you do a sensitivity analysis to the mesh, you divide by 2 the mesh size, and on... In this case, every time the mesh size is divided by 2, the number of elements is multiplied by 4 in 2D, and you must divide the time step by 2 to work at constant Courant number, which gives a factor 8 in time, this is about what we observe, but as I mention, cash memory size of processors may bring surprises (e.g. a speed-up in parallelism with a single processor...).

Regards,

JMH