open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello

I am running a simulation in Telemac 2D but I could not obtained my result file because a mistake appeared on screen. I have followed the recommendation of the forum previously However I can not get my simulation correctly.


Thank you





Regards
Joel

Hello,

Your case is crashing, with infinite values, so to go further we would need to know when and where this happens, at the first time step or after, and we would need to see more of the listing, to know which step triggers this.

Another possibility is to have a version compiled with every check in the compiler, so that it stops whenever you have a NaN appearing, with a traceback from the compiler showing the subroutine and the line number where it occurs, this is how we debug.

If it works a few steps try to reduce the time step.

With best regards,

Jean-Michel Hervouet

Dear Jean
Thank you very much for your quick reply

Firstly, I have decreased the number of time steps from 60 to 10 (attached CAS file). As a result the model ran correctly without any mistakes. However How can I run the model for 33 days with the same time steps (10) without mistakes?

I need my results of the simulation for 33 days.

Secondly, the debug word should be at the end of my cas file.

Thank you
Regards
Joel

Hello

33 days = 33*86400 = 2851200 seconds = 285120 time steps of 10 s

with my best regards

Riadh ATA

Hello,

In a simulation for an steady case in a large domain i have the exact same issue that joel had and indeed reducing the time step (from 15 seconds to 5) solved the problem, but a new one arrised...the hugely increased computation time.

Related to this issue, there is a Keyword called VARIABLE TIME STEP that might help but i couldnt make it work because telemac only performs one step and then exits with a "My work is done" message.

Does that keyword need an additional config. that perhaps i could have overlooked?

Thank you,

José Díaz.

Hello

To use variable time step you need also to give the desired Courant number (keyword DESIRED COURANT NUMBER).
However, before using variable time step, have a glance on this topic, it will for sure help to accelerate your computation.

with my best regards

Riadh

Dear Riadh,

Thank you, i managed to do both time variant conditions and also what JM Hervouet proposed, but in the end my case still crashes with maximum iterations in the K-E turb. model using the prescribed flow-inlet condition.

As soon as i switch the inlet boundary to a thompson type, i get a rough approximation of my desired flow and fulle computes a solution, but not the expected one.

I believe is due to a meshing problem, only have 3 points where the flux should be computed, as the manual suggests there should be at least 5 (due to no-slip condition on solid boundaries i guess?), so i refined the mesh with the minimum required nodes but don't know how to "turn off" the nodes that should have zero velocity to start a new simulation from the scratch.

¿Should i post a new topic in the forum?

Thanks for your time and effort,

José Díaz

Hello

You question is not clear enough. If you have refined the mesh, you need to generate the new boundary condition file of the nex mesh. Within this file you prescribe the liquid boundary conditions you want.

If the problem is only with K-eps, try to run with other turbulence models (start with constant viscosity) and then go increasingly with more complex parameters.



with my best regards

Riadh ATA