open TELEMAC-MASCARET The mathematically superior suite of solvers

I am modelling a generic tracer in a river with telemac 2d. I have found some weird results at the plume front under typical flow condition (see attached). The front under max flow condition does look fine. The .cas file is also attached.

Does this result suggest something wrong in my model setup? Thanks

Hailiang

Hello,

You can check first on the listing that the solver for the diffusion of tracers always does at least 1 iteration, if not increase the accuracy, but it is already 1.D-12, so probably enough. A most probable reason is the OPTION FOR THE DIFFUSION OF TRACERS which perhaps causes trouble or differences of propagation on shallow zones, try 1 instead of 2 and see what happens. It is true that in theory we should not simplify the depth across the derivative, but in practice the diffusion in a depth-averaged context also contains the dispersion due to the heterogeneity of velocity on a vertical, which sometimes is even not treated as a diffusion but as an advection, so we have a lot of uncertainty here.

With best regards,

Jean-Michel Hervouet

Thanks Jean-Michel for the reply. After i changed OPTION FOR THE DIFFUSION OF TRACERS to 1, I still observe the same pattern at the plume front.Does this mean i may need a 3D model to get a more reasonable pattern?

My second question is: with the current case file configuration, i get viscosity around 0.2 m2/s. Is it a reasonable number (or how can i tell whether it is reasonable)? If i set the turbulent scheme to constant viscosity 0.2, should i observe similar velocity and depth? I guess i am not quite sure when should I resort to k-epsilon scheme.

Thanks

Hailiang

Hello,

This is strange, could you test without friction on the lateral boundaries?

LAW OF FRICTION ON LATERAL BOUNDARIES : 0

I find strange that with a large friction on the boundaries you get a tracer which is faster on the boundaries. Maybe it is explained by water depth and the velocity pattern?

Otherwise yes, if you put a constant diffusion of 0.2 m2/s it will give about the same result as k-epsilon.

Last thing, NAMES OF TRACERS should be given in 32 characters, 16 for the name, 16 for the unit.

With best regards,

Jean-Michel Hervouet