open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello all,

I want to run T2D simulation with finite volume method but it looks like there is a problem when I run the simulation, I get the following message (this is not an error message but it is not correct either) in the listing file, see attachment. I put as an attachment also the cas file so many you should help me.

Regards

Hello

You are using V6P3, for later versions this config is no more possible.
To use finite volumes you need to give at least the scheme and the CFL.
These informations are given by the following keywords
FINITE VOLUME SCHEME = 1
DESIRED COURANT NUMBER = 0.9

You have to increase considerably the period for listing and graphical outputs since time steps are very small for finite volumes.

with my kind regards
Riadh

Thank you Riadh for your answer.

I added the two key words you talked about and I fixed the time step at 0.01s. So now it looks like it works fine but the thing is that the computation seems very long and I have this message :

"ATTENTION: PAS DE TEMPS FIXE ET CFL NON FOURNI !"

Is there any way to make the computational time smaller ?

Sorry

You have to add the following keyword
VARIABLE TIME-STEP =Yes (PAS DE TEMPS VARIABLE =OUI)

CPU time for finite volumes is naturally much bigger than for finite volumes

with kind regards

Riadh

No problem, I forgot to precise in my previous message that I already tried with the keyword VARIABLE TIME-STEP = TRUE.

I am surprised, in this case the finite volume computational time is about 25 times longer than that of the finite element method. Is the difference between the 2 methods that huge usually ?

usually FV take much more time than FE, but a factor of 25 is a bit huge. take the same period (effective) for outputs than for FE (every 2250s for you case).
you are runng in Parallel ?
kind regards

Riadh

This is this long because the time step is very small, about 0,03 to 0,05 seconds. If you fix a greater time step it does not work correctly. So now I am currently trying with variable time step and output frequency of order 2250 time step.

What time step do you use for your FE calculations?


For finite volume you have

cfl = v*dt/dx

with
v = |v| + sqrt(h * g)

I'm not sure but for FE the water depth is not considered?

The FE schemes work also with CFL > 1. So you can estimate the differences in the possible time step size depending on your water depth.