open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

I'm working on a 3D hydrodynamics model of Scheldt River. But I couldn't get good result because the velocity become very chaotic in the shallow water areas whereas it seems not bad in the main stream of the river.

I have tried two different schemes for advection of velocities (option 1 and option 14) but the problem remains.





I have no idea about this problem. Does anyone have any idea?

Another problem is that the model became unstable when salinity was added. The printout shows :

GRACJG (BIEF) : EXCEEDING MAXIMUM ITERATIONS: 200 RELATIVE PRECISION: NaN
GRACJG (BIEF) : EXCEEDING MAXIMUM ITERATIONS: 200 RELATIVE PRECISION: NaN
GRACJG (BIEF) : EXCEEDING MAXIMUM ITERATIONS: 200 RELATIVE PRECISION: NaN
GRACJG (BIEF) : EXCEEDING MAXIMUM ITERATIONS: 200 RELATIVE PRECISION: NaN
......

I guess this might be have something to do with the first problem.

Any idea will be very appreciated.
Attached are the plot and steering file used in the model.

Hi,

I'm not really involved with T3D, however, waiting for the comeback of Jean Michel, hereafter a tentative answer.

You seem to have troubles with shallow areas so naturally one can think about your vertical discretisation, how many planes are you using, is there very flat and even squeezed elements in your mesh? This can nullify the volume of some elements and then generate instabilities. The fact that instabilities increase with the salinity, could be explained also by this hypothesis.
An other point that could be investigated is the diffusion coefficients you are using in your case. maybe they are not high enough to dump reasonably the solution. I invite you to have a glance on the Berre validation case of Telemac3D to have an idea about the values of the diffusion coefficients (horizontal and vertical) to adopt. You can remark also that in this validation case, the salinity is considered.

I hope that this could helpful
Riadh

Thank you for your advice Riadh.

I have six planes in my model. Currently I'm using sigma transformation and the surface plane has a lot of flat elements. Is this the problem you meant?

For the diffusion coefficients for tracer, I'm using 1.0 for both vertical and horizontal.

I found the validation case in the documents but I didn't find it in the validation cases. Indeed this case is close to my model. Where can I find it?

Thanks!

Hello,

it seems that in your image all the velocity vectors have the same length. And so it is normal that they look chaotic.
It doesn't solve your problems with the computation but in BlueKenue you can turn off the constant length option for the arrows.

Best regards,
Clemens