open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi all,

I'm trying to run a model of an estuary which has tidal flats near the mouth and which is dyked in the upstream part.
All in all I get plausible results with some exceptions.

When the waterlevel decreases and nodes are intended to get dry I get at some nodes very unplausible flow velocities and directions.

I tried different parameter sets. The latest which produced the best results is attached.

May this Problem be related to the discretization?

Best regards
Horst

Hello Horst

Your results are not bad at all, the values of velocity (<3.5m/s) are physical while they give velocity in the lower part of a dyke where gradient of bathymetry is very steep and where there are wetting/drying elements.
Numerically, it is very normal to get important (and even unphysical) velocities for partially wet/dry elements.
However, you can improve these results by the use of better advection schemes that are able to handle this phenomenon. I see that you are using option 1,5 for advection which is the method of characteristics.
For tidal flats, it is better to use the option 13 or 14.
You need furthermore to activate option for tidal flats:
TIDAL FLATS: TRUE
OPTION FOR THE TREATMENT OF TIDAL FLATS: 1 (you can test options 2 and 3)

Finally, I see that area that interests you is close to a dyke but also close the boundary which could generate an "over-constrained" flow.

I hope that this is helpful
With my best regards

Riadh ATA

Thank you very much Riadh for answering so quickly.

I'm reassured that this local phenomenon is "normal" from the numerical point of view.

I tried the different advection schemes and different options for the treatment of tidal flats but the results didn't look much better.

The phenomenon also occours further downstream. There are tidal flats close to the shipping channel (so again a part with a steep gradient of bathymetrie and here far away from the boundary).

Are there any different possibilities to deal with it?

Could a finer discretization (like max. gradient of bathymetrie within one element smaller than a certain value) be helpful?

My future aim is to simulate sediment transport and I'm afraid that these high velocities will have strong influence on it. That's why I would like to improve the results.

Best regards

Horst

Dear Horst,

For sure, a finer discretization will help to improve the results, however you need to respect the dilemma regarding the CPU time. I see that you are not using parallel option, this could be interesting to consider this option especially if you aim later to simulate sediment transport which are very demanding simulations.
An other option that you could try, is the use of finite volume kernel of Telemac. The advantage of this option, is that you don't have to tune any parameter to deal with tidal flats, it is considered automatically by the so called "hydrostatic reconstruction" technique. Schemes 1,2,5 and 6 are recommanded (2 and 6 are of 2 order in space and are more expensive in CPU time).

With my best regards

Riadh ATA