open TELEMAC-MASCARET The mathematically superior suite of solvers

Dear Jacques,
yesterday I tried to start from a completely new set-up, with less options activated. It is still running, but it seems to have a more stable behaviour, not waves propagations, homogeneous velocity field, and Qout is slowly reaching Qin value.
But... even if .cli file was unchanged, the imposed H at the outlet is not respected! So water depht varies from about 0.50 at the inlet to about 0.25 at the outlet.
(in attachment as case 2: .cas file, geometry, .cli, and actual last time step of resulting slf )

Coming back to the old .cas (it respects water depth at outle but yes, velocity field is different at each time step!)
and to your last suggestion:
- in order to use Thompson have I only to change
OPTION FOR LIQUID BOUNDARIES=2;1 (instead of 1;1) ?
or do I have to add something else?
- do I have to change .cli file, or is it still ok 544 condition for the outlet?
(in attachment as case 1: updated .cas file, .slf geometry, .cli, and last time step of the old resulting slf that could be used as the restarting point--> even if it has not homogeneous )

- Do you suggest to continue with case1 (where many advanced options are activated), or to use the new case2 (simplier , maybe I could try there Thompson)?

Best regards, Eleonora

Dear Eleonora,

For Thompson you are right, you only have to change option for liquid boundaries.
NB: with tompsom H downstream will not be strictly respected.

Sorry, I can not help you more, I have no access to a computer (until end of August) and I can not open attached files.

Best regards,
j

I'll try Thompson in the old .cas ...

hopefully by the end of august I'll have good news.

Thanks Jacques.

Hi Jacques (or anybody'd like to answer),

case1 attached few emails ago is still running. In that .cas file, I used your suggestions: TYPE OF ADVECTION = 5;5 SCHEME OPTION FOR ADVECTION OF VELOCITIES = 4 - Thompson for outlet boundary - hot start from a previous (but not good) result.
I let it goes for a long duration, hoping this could allow to reach a better result. It seems that velocity and water depth field have less variations, and Qout= 0.025+-0.002 m^3/s ... but velocity field varies at each time step, and shows vortexes, strange direction along boundaries, very variable pattern at the outlet!
in attachment an example of this situation, with the result extracted at two different time steps from the running T2DRES.

What I expect to have as result, it is a almost homogeneus slow velocity field, with flow going from inlet to outlet, and stable velocities at the outlet...

I looked again at geometry and bc files, and it seems to me they are ok
I suppose there is something wrong in my .cas file. Maybe I could simply it, deleting some options and using default ones?...
I only need to simulate a rectangular basin, with imposed Q at inlet and H at oulet, with flat bathymetry and constant roughness. The unique issue maybe is that velocities are very very small ( mm/s )

If somebody would help, you can refer to case1.zip attached few emails ago.
Thanks, Eleonora

Hi...
I'm still trying attempts with this model...

Using Jacques's suggestion received in past replies, I prepared a corrected set up (case1), but as I already wrote it doesn't reached stable results.

I tried to change the boundary conditions(new case): Fixed Q=0.025 m3/s at inlet, fixed H=0.050m and Q=0.025 m3/s at outlet (not only H). I know that's not the best...
but it worked for the flat bed. I was hopeful...then I run it on a bathymetry with transversal bed forms...it is still running...but I see from dos window that Qout is about 50 L/s instead of 25!
why?... this doesn't respect boundary condition, and the mass balance (Qin=25 L/s, Qout=50 L/s ???)
really I can't find how to make this model works for my wetlands...

I attach here original case1 (with jacques's corrections) and case_new with fixed H and Q, and two .slf files with flat and anysotropic bed forms.

My regards, Eleonora

Seems like normal cold start behaviour to me. You have a mass of water that is not moving at time zero and then you put flow just on the edges.

Either try with a flow field maybe from your flat simulation or increase discharge slowly at the boundaries.

PS, I don't see the need to specify discharge at the downstream end, if equilibrium is reached it will be the same as the upstream.

Patrick

Hi Patrick, thanks for replying.
I think there is something more than the cold start issue. Because also the flat simulation had problems (not stable velocity field) after simulating more than 4 days!
It only seems to work by imposing Qout.
But this “solution” doesn’t work for not flat bathymetry, as I previously wrote. And Qout became greater than Qin. And I also tried starting from the velocity field of the flat simulation ... with no better results....

Eleonora

Dear Eleonora,

I ran your simulations.
Basically you have a basin with a centered channel for inlet and outlet.

Case 1 : with a 0 bathymetry
=> TEATMENT OF LINEAR SYSTEM = 2 is not a good idea (faster but...)
(you can try with a file in attachments, to initialize you flow, I also add diffusion to help to remove waves)

Case 2 : with a bathymetry
I am no sure to understand, you have dry zones in this case. It is impossible to obtain a perfect steady state. You obtain unsteady behavior with vortices etc.

Best regards,
Jacques

Hi Jacques,
I'll try with your new .cas file.
Could I use it for the not flat bathymetries?...

I don't understand why with not flat cases I obtained Qout>Qin (I checked Q at several transversal sections...it passes from 25 L/s to about 50 or more!)

Do you suggest to try with a finite-volume scheme? (your .cas file is finite-element, right?).

Thanks, Eleonora

Dear Eleonora,

Yes you can use .cas for non flat bathy.
Yes you can try FV. With FV you have to hot start to avoid shocks. Be aware, some functionalities are avalable only with FE. This depens on your version.


Best regards,
J