open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

I have successfully run a simple idealized channel simulation with prescribed out of phase sinusoidal boundaries for a coarse mesh resolution.
I am now trying to run the same simulation with a higher resolution mesh, however I get unwanted disturbances occurring randomly at the liquid boundaries a few days(simulated time) into the simulation, creating high velocities in the domain.

I have prescribed elevations at either end of the channel to create approx 3m/s maximum velocity in the domain with the thompson boundary condition activated.

Does anyone have any suggestions into why the simulation is falling apart with a high mesh resolution?

Any help would be greatly appreciated.

Thanks,

Scott

Hello,

Your problem is due to the fact that prescribing only the elevation is not a well posed problem when velocities are entering the domain. As velocities are free, any value is a possible solution of Shallow Water Equations. Such boundary conditions will only work if the velocities remain small (or more probably the Froude number), e.g. off shore with large depth. To overcome this problem you have to resort to Thompson boundary conditions (keyword OPTION FOR LIQUID BOUNDARIES=2;2;... as many as liquid boundaries, but you can apply it to the boundary you want, e.g. only boundary 2 will give :
OPTION FOR LIQUID BOUNDARIES=1;2).
There is a presentation on Thompson boundary conditions in the proceedings of the last user club.

With best regards,

Jean-Michel Hervouet

Hello Jean-Michel,

Thanks for your reply. I am already using the Thompson condition on both boundaries, primarily to prevent reflection of the propagating wave. Since I am still experiencing velocity fluctuations at the boundary, would it be sensible to try prescribing velocities instead of elevations at the boundaries or are more parameters required?

Thanks,

Scott

Hi

You should prescribed elevations ans Velocities and Use Thompson to relax the overconstraint problem on the boundary. Level will be imposed and velocities will be suggest ...
This solution is often use for the nested models which taking boundary condition from the large model.

Hope this helps