open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello everyone,

When I edit a steering file,TIME STEP and NUMBER OF TIME STEPS seems very important, change them, the cas can work well or badly.

But I don't know how to choose them, is there any criterion?

Best regards,
Li Wei

Hi Li Wei,

yes, the transport of "information" (see the CFL condition). The best TIME STEP depends on your numerical method/scheme. Most schemes perform best near CFL = 1.

For complex meshes and varying flow conditions (e.g. changing inflow) it can be very tricky to set a fixed optimum time step. Depending on the scheme, small time steps (small CFL numbers) can be diffusive.

The best way to see the influence is to make simulations with different TIME STEP and compare the results.


Best regards,

Leo

thank you, Leo, I will learn CFL condition and try it.

Hello,

The CFL (Courant-Friedrich-Levy) number is basically :

velocity * (time step) / (mesh size)

This is for advection. For shallow water équations it is :

(velocity+celerity) * (time step) / (mesh size)

celerity of waves being sqrt(g*h)


As a rule of thumb :

In the ocean take celerity

In rivers take velocity, except with finite volumes where velocity+celerity should be condidered.


With best regards,

Jean-Michel Hervouet

Hello Jean-Michel Hervouet,

Thank you, and how to decide NUMBER OF TIME STEPS? In addition, I am reading your book ...Free Surface Flows...,After I read it, It may can help me a lot to understand TELEMAC.

With best regards,
Li Wei

Hello

If you are using variable time step and a fixed Courant number, the NUMBER OF TIME STEPS is not necessary, you can give just DURATION which is the total physical simulation time. If you do not give DURATION, you can give TIME STEP and UMBER OF TIME STEPS and TELEMAC with compute DURATION as the product of these two parameters.

with my best regards
Riadh ATA