open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi

Is the key word FREE SURFACE GRADIENT COMPATIBILITY only relevant when you choose to use the wave equation, i.e. when TREATMENT OF THE LINEAR SYSTEM = 2?

Thanks

Fay

Hello Fay

Theoritically, this option is relevant with both options (1: primitive equations and 2: wave equation). The key idea behind, is to break the rigorous compatibility between water depth and velocity components which verify continuity equation in order to damp spurious oscillations that can appear in areas with high gradient of topography.
For Telemac-2D, this option is only implemented with wave equation. This is logical while this option is more fast and stable than option 1 and thus is oscillations appears with this option they will appear and will grow more and more with option 1.

I hope that this will help

With my best regards,

Riadh ATA

Hello Everyone,

I was searching the forum for a question related to this topic and I found this answer very helpful.

I have a question related to it and did not wish to create a new topic.
I find that in all examples I've seen this key-work is used with the value
FREE SURFACE GRADIENT COMPATIBILITY = 0.9
but I still get a lot of oscillations so I try to decrease it even more (to 0.6) and the results were smooth, now my question, is this a valid think to do? Or the result is completely artificial? Should I try something different instead?

Hope you can help me.

Hello,

Smoothing the solution using such a low value FSGC seems to be pretty artificial to me.

Before going to such lenghts, try refining (and checking) your mesh, adjusting time step accordantly (or setting a variable time step with a desired courant around 0.9) or even trying the finite volume scheme, along with some other less agressive numerical tweaks.

Hope you find somehting useful along the lines. Regards,

José D.

Jose, Thank you for your answer. That is what I thought.

I manage to get a smooth solution with a FREE SURFACE GRADIENT COMPATIBILITY = 0.9 by changing the OPTION FOR THE DIFFUSION OF VELOCITIES from 2 to 1 and also adding PSI SCHEME OPTION = 2. My simulation is still a little off with regards of the real data and I am still trying to make it better.

Hello,

Just a personal remark. In most of the cases I've found much better results using K-E (or even mixing-length) turbulence model over constant viscosity when comparing with measured info. I usually deal with flood hazard mapping though.

Out of curiosity, are you trying to match a Lab test case?

Regards,

José D.

Thank you for the remark, I haven't try K-E but I will give it a try.

I'm trying to match real data taken from a measurement station in order to be able to simulated sediment transport in the area.