open TELEMAC-MASCARET The mathematically superior suite of solvers

Dear all,

For my thesis on numerical models at the University of Brussels I am working with Telemac-2D. The purpose of the thesis is to compare the finite element method of Telemac with the finite difference approach of another software package.

One of the things I want to evaluate is the numerical stability of the software. In this context, I set up a simple test case: the backwater curve in a straight channel with prescribed upstream discharge and downstream water level. I changed the grid sizes and time steps to evaluate effects of time and space resolutions.

At a seminary on Telemac from Sogreah, the teacher told us there is no restriction on Courant number in Telemac because the numerical schemes are implicit. However, my simulations pointed out that the maximum value for Courant number in Telemac is 1. The finite difference software has a maximum value of around 30 for the same testcase.

Could one of you explain this difference and the reason for this restriction?

Dear Edeproos
You can choose different scheme in Telemac with the keyword "TYPE OF ADVECTION".
The numerical scheme are more "semi-implicit" rather than "implicit" but with the characteristic scheme.
If you run the training test case called BUMPCRI, you will see (by changing the time step) that the simulation run well with a Courant number greater than 100.
The default value of the time step (0.03s) gives a very precise solution (no difference between the exact solution and the computed solution) but the computation could be run with other value (even 3s for the time step!).
In your case, try with other scheme (on Velocities and Water Height) to check if you conclusion still the same

Good luck

I wanted to look up the background of these schemes to fully understand everything behind it. However the user manual referred me to the formulation document. Where can I download this formulation document?

Dear Edeproos
An old version of the Formulation Document is available in the Download section (the name of the file is prin30_gb.pdf).

This document is now replace by the book of Jean-Michel Hervouet "Hydrodynamics of Free Surface Flows: Modelling with the Finite Element method" (c) 2007 John Wiley & Sons, Ldt

Regards

Thanks a lot for this answer.
Indeed, it makes a big difference when you change the type of advection.