open TELEMAC-MASCARET The mathematically superior suite of solvers

Dear JMH,

I did what you said and fortunately it worked.

Thanks a lot.

Amanj

Hello dear JMH,

I have done some simulation of overtopping embankment by coupling Telemac3D and Sisyphe for the attached physical lab model. I have a problem of upstream flow which is sudden release of a gate just before the crest of the dam and this in Telemac created a lot of oscillation and fluctuation which is not appear in real life. I need your advise to reduce this problem or better way to simulate. In the meanwhile, I have the result of 2D bed erosion which very different from what I got in Telemac. In Sisphy result, erosion is under estimate compare to that one in lab result.

More files

Hello,

You obviously get some inf-sup oscillations in your results, and as a matter of fact there is nothing in your parameters to prevent them (spurious oscillations may appear in Navier-Stokes equations when velocity and pressure are discretised in the same way).

try :

FREE SURFACE GRADIENT COMPATIBILITY = 0.

and see the effect.

Also change all preconditionings from 11 or 7 to 2. In 3D the most powerful preconditioning is 34 (direct solution on verticals). This does not change the results, only the computer time, so do not loose too much time on this unless you see big differences in time.


Also try :

IMPLICITATION FOR DEPTH : 1.
IMPLICITATION FOR VELOCITIES : 1.

this should stabilise and allow you a larger time-step (larger Courant numbers can also break inf-sup oscillations).

With best regards,

Jean-Michel Hervouet

Hello dear Jean-Michel Hervouet,

I updated my steering file based on your suggestions but it seems to me unstable since the calculations get longer time and as shown in the attached picture it gives over iteration in GRACJG and sometimes for few hours it show the same thing without changing the screen. The problem is happened when using preconditioning 2 while in preconditioning 11 doesn't have this problem. Would you please suggest any thing that help to solve this problem because I have the physical model for the same case and the flow is quite smoothly flowing during ovrertopping while the numerical model creates extreme osculations and fluctuations.

Thanks in advance
AMANJ

Hello,

OK if it works with preconditioning 11 and only if you change this, I am surprised it works so efficiently. You could also try 34 then. By the way preconditioning is it, I mean the full keyword so that we see which linear system is sensitive.

With best regards,

Jean-Michel Hervouet