open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi

I'm trying to calibrate a flood hydrograph in telemac2d. I use a cross-section in bluekenue to compare the result of the model with the measured flow hydrograph.

But I have big instability for the hydrograph curve and the flow increase from the upstream boundary.

I have tried many options but didn't manage to solve the problem.

If somebody can take a look, I have attached all simulation file and the image of the hydrograph.

Thanks for your help

Hello,

Where is this cross section, near the entrance or near the exit? If it is near the entrance the next question will be, is the flow always sub-critical during the flood? (if not we should give also the free surface elevation). If it is near the exit it would be that the stage-discharge curve triggers instabilities, then you have a relaxation parameter in function sta_dis_cur that can be tuned, it is set to 0.02, and a smaller value could be tried.

With best regards,

Jean-Michel Hervouet

Hello JM,

Thank you for your answer. In fact, the instabilities are present all over the model but are more important next to this cross section. The instabilty decrease if we go next to the entrance or next to the exit. So, It doesn't seems to come from the stage discharge rating curve. However, I set the relaxation coefficient in function sta_dis_cur to 0.01 but it doesn't change anything.

There are no particular element next to the cross section (structure, etc...)

Any suggestion to stop the instabilities ? What are the coefficients I can tune ?
First of all, I'am running a simulation with a time step of 1 second but if it works well, I would prefer not to have such a time step.

With best regards

Francis Cougny

Hello,

The smaller time steps are not always the better to avoid oscillations, so you can try higher values and work directly on the time step that give sensible results except the oscillations. Trying a quasi-bubble element for velocities would give a hint on the "inf-sup" nature of oscillations. A cheaper solution to try first is: COMPATIBILITE DU GRADIENT DE SURFACE LIBRE =0.

Diffusion should also reduce oscillations of velocities, but I see nothing on diffusion in your parameter file, try with a coefficient of 1 or 2.

Last thing, you can remove FICHIER DES PARAMETRES='cas.cas' from your steering file, it is not useful, only misleading if you change the name.

With bestregards,

Jean-Michel Hervouet