open TELEMAC-MASCARET The mathematically superior suite of solvers

Dear all,
I want to simulate flash floods in catchments. For these problems, I'm using rainfall-data as input. As output, I want to use boundaries with free depth and free flowrate (BC 444). However, after some timesteps I'm getting numeric oscillations at these BC's (during some seconds the velocity and depth increases rapidly). These oscillations occurring in different catchments at different time steps. I'm using variable time steps with a desired courant number of 0.8.

If I replace the BC's 444 with 544 (prescribed depth of 0 m at all time steps), the model is running well. Nevertheless this is not a good solution.

Is it possible to use 444-conditions without numeric problems for flash flood simulations?

Thanks for your help in advantage
Thomas

Hello Thomas

Yes you can use 444 condition together with Thompson condition. The result will be more or less stable.
The best way to do in this case is to keep the 544 condition, but also to use what I call a "supercritical sponge layer" which is a layer of 2 or 3 elements on which you introduce artificially steep bathymetry gradient in order to have :
- still water when there is no ouflow
- supercritical flow when water is coming from upstream to downstream

In order to not influence the main region of the flow, try to move away this artifical boundary from your region of interest as far as possible.

with my best regards

Riadh ATA