open TELEMAC-MASCARET The mathematically superior suite of solvers

Dear All,
By the use of BlueKenue and Fudaa PrePro I prepared a mesh, a BC file,a liquid boundaries file, and a steering file. I used a constant elevation of 460 m as initial condition which is also the fixed water level at the downstream boundary condition. Although the first 20 time steps look sensible something seems to happen between the 20th and 21st time step. Suddenly there are enormous changes in surface elevation and thus in velocity. It looks as if something collapses in the model domain. There are huge surges as shown on the following screenshot and the simulation is cancelled.





I have no clue what is wrong? Could please someone have a look on my input and files (attached). Maybe there is a general approach completely wrong!?

Best regards
Heidi

Hello,

You can look at two things:

* Divide the time step by 2 and see what happens, it could be that the model is unstable with this time step.

* Look where exactly starts the problem : inside the domain, on a boundary condition (upstream or downstream ?). You certainly do not have these big surges appearing in 1 time step.

With best regards,

Jean-Michel Hervouet

Dear Jean-Michel,
thank you for your reply. I simulated with a time step of 15 s (nstead of 30 s) and interestingly the problems start again after 20 time steps, although the time step is only the half.

As shown on the following screenshot the surges begin at one of the liquid boundaries and in between the two embankments/abutments (where actually piers are supposed to be, but this is another problem).





One time step later the inconsistencies are already widespread.




Soon later, the simulation is canceled with the following error message:




When I prepared the simulation in Fudaa prePro a warning showed up that cartesian grid is not supprted. Could that be the problem?

Best Regards
Heidi

Hello,

some observations based on your steering file:
are your sure you want to disable the advection terms?
Complementing Jean-Michel's answer, introducing more turbulent diffusion into the system by specifying a higher eddy viscosity value helps also stabilizing the simulation (default eddy viscosity is 1.E-6 m²/s).

I don't know exactly what you want to simulate but if you want to perform a standard 2D Saint Venant hydrodynamic calculation you should consider to revisit your steering file.


Best regards,
Clemens

Dear Clemens,
thank you for your reply. I tried the simulation with advection (all of them) and unfortunately the result is the same => The simulation is cancelled.

Best regards
Heidi

Dear Heidi,

I have no idea about your mesh size, but 15s seams to be a pretty high time step..

Have you checked it?

Another idea, maybe the rise/decreas in the liquid boundary file is to rapid?

Regards,
Gabi

Hello Heidi,

I had a look at your steering file also, I wonder where you took it. You should start from a steering file of one of the test cases provided. For example you should not remove advection terms, because without them you are more sensitive to inf-sup oscillations, which is your problem. Take the steering file of test case malpasset for example.

With best regards,

Jean-Michel Hervouet