open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

I have to think it over, but probably the atmospheric pressure is missing somewhere in the process, somewhere where we compute the gradient of the free surface, e.g. in the wave equation, and this results in the velocities observed. I'll try to fix this in July.

Regards,

JMH

Hello,

I could reproduce the problem. The free surface gradient and the atmospheric pressure gradient did not balance well when the free surface gradient compatibility is not 1. The modifications are not straightforward but I enclose a tentative new version of wave_equation.f that hopefully corrects the problem. I have tested that a free surface function of X**2 remains still with the adequate corresponding atmospheric pressure. I tested the cases with and without tidal flats (which require different implementations).

I hope it helps, if it gives better results I'll include it in the next version.

With best regards,

Jean-Michel Hervouet

Thank you very much!

I will test that when I'm back from holidays at the beginning of August.

Regards,
PL

Hello Jean-Michel,

I tried to use your updated version of wave_equation.f but it fails. The reason seems to be that in your new file you call t3d_stress.f with a few extra arguments than those of t3d_stress.f in v6p3r2.

Here is your call in updated wave_equation.f:
The last four variables are not defined as arguments in t3d_stress.f from v6p3r2.

Is there a new working version of this subroutine that I could use and test? Or is it a compiler related problem (I use gfortran)?

Thank you in advance for your help!

Regards,
PL

Hello,

Whoooops, sorry, I did not remember that I had changed this subroutine and did the changes in the latest version (future 7.0). You can remove the 4 arguments after MASKEL in the 2 calls to t3d_stress, so that it corresponds to t3d_stress of version 6.3.

Regards and apologies,

JMH

Hello,

I did a quick test with your updated version of wave_equation.f and I don't reach a steady state...

The only configuration that gives a steady state (still water) is with FREE SURFACE GRADIENT COMPATIBILITY = 1.0 and with wave_equation.f from v6p3r2.
When using the same case with your new version of wave_equation.f it seems the the pressure distribution is changed compared with what is defined in meteo.f. In my case, it looks like the pressure on my pier is greater than my initial condition, which results in a higher water levels on both sides of the pier, see picture below (black: result with wave_equation.f v6p3r2 and FREE SURFACE GRADIENT COMPATIBILITY = 1.0 ; colours: new version of wav_equation.f). I also have high velocities (ca 1 m/s).




Am I doing anything wrong? I even tried a similar case than yours with a parabolic water surface, but I face the same problem.

I join my files in case you want to check them.

Thank you in advance!

Regards
PL

Hello Jean-Michel,

Could you have a look at my previous post?

Thank you in advance!

Regards,
PL

Hello,

Not yet. Actually I just tested the static case where the free surface has already its final position and nothing moves. I'll do now a test similar to yours, gradually enforcing the pressure and letting the free surface move accordingly, hopefully next week.

Regards,

JMH

Hello,

OK, it is perhaps parallelism, I have now corrected a mistake in this case (the pressure gradient newly introduced lacked a parallel assembly after computation). Tell me if it is OK now (or I'll keep trying with dynamic tests).

Regards,

Jean-Michel Hervouet

Hello,

Unfortunately i don't think this will correct my problem since i made all my tests in scalar.

Furthermore all my tests were also static cases like yours, with initial water surface and pressure balancing each other, without prescribed discharge nor wind.

Maybe you can get more insight by running my simple case (see zip in previous post), and checking my fortran files (i included two versions: one for 6.3 without your new wave_equation and the same one plus your new routine with modified calls to t3d_stress, the latter showing strange results...).

I am still on holidays this week but i will try to find some time to run a few cases if needed.

Thanks in advance!

Regards
PL