open TELEMAC-MASCARET The mathematically superior suite of solvers

Dear Users,

I hope somebody could point me into right direction. I used to run simulation on stratification in experimental flat tank, and it worked fine. But now I am trying to introduce the real bathymetry, that is very steep, and it does not work. I am using nearly same steering file as I did for tank experiment but all surface level gets distorted immediately. Please see attached the pics of simulation runs. I have also attached the cas, fortran and result files. I have tried to set up the Mesh Transformation=3 in fortran file, as presented on page 30 of User Manual - Telemac Modelling Sytem, figure 4, but it did not work either, so I am doing something wrong. I have similar case for same bathymetry run for tidal model and results look ok, but when I try to use it for stratification case (not tides, no velocities) then it does not. Please advise. Thanks in advance.

Kind Regards!

Violet

Hi Violeta

Even Telemac doesn't like steep bathymetry, I think you should first try to enhance the number of planes.
Only 3 planes for such water depth is not sufficient. You have only 2 vertical cells.
Depending on what you really want to focus on, and after finding a quiet good vertical resolution you could try to use different mesh transformation in order to limit the total number of planes (just refining near the salinity gradient or near the thermic gradient ...). The method 5 (AMR method) couls also help you do automatically adapt the vertical refinement.

Hope this helps

Great, thank you very much Christophe for advice. I will try to use Mesh Transformation = 5 and see how it works. I had tried already to use 10-15 vertical layers and got same results, the huge surface distortion and hence velocities are huge. I also will try to refine the vertical mesh at mixing layer level. Thanks again.

Kind Regards!

Violeta

Hello,

To clear up problems, you should try to do a run without salinity or temperature gradient. Only when it works add temperature and salinity, and you can also try the keyword:

HYDROSTATIC INCONSISTENCY FILTER = YES

to cancel temperature and salinity gradients in highly distorted areas of the mesh.

The gradients given by the initial temperature and salinity are also very important, if you start from very artificial conditions it will trigger important currents.

With best regards,

Jean-Michel Hervouet

Thank you very much Jean-Michel. I will do so and hope it will work. Thanks again.

Kind Regards!

Violeta

Dear Jean-Michel/Christophe,

Thank you for your advice on this case. Unfortunately it did not fix the problem, the surface level is still distorted straight from the beginning of the run. I run the same bathymetry without stratification but with tides and it worked fine, the minute I added the stratification but without tide the problems begin. I also experimented little bit with this case and closed the open boundary to see if steep bathymetry still caused the problems, and it did not. It worked ok, the surface level was not distorted at all. So maybe the boundary conditions at the open boundary are wrong? I have it as 544 and 4 (Open boundary with prescribed H and free tracer). Since I am going to add tide to the case, when stratification is fixed, I think I will need to have open boundary as 544 or 566 and that works fine for tidal modelling for my case, but it does not for stratification. Could you please advise if I can use something else instead that suits this scenario, or maybe my problem is not on boundary conditions at all. I attache my output and steering files, just in case. Thanks in advance.

Kind Regards!

Violeta

Hi Violeta

Yes, you're probably right about the open boundary and the tracer. When you impose only H for the tides, the problem is ill-posed as the velocities are free when the water filling the model. I suppose you activate the Thomson method but it's also a problem for the tracer as you declare it free. You should probably try with a fixed value which will be only valid when the water entering into your model as telemac automatically switch the condition of the tracer according to the direction of flow.

Another point you should explore is the way you move from a solution without stratification (which works well according to your post) to the solution with stratification?
Hot restart?

Thanks a mil Christophe. Looks like the problem was at open boundary. I opened it again and prescribed UV only, at the open boundary and surface level behaves very well! The layers are not distorted anymore. I also use the mesh transformation = 5. However, I have run it for small period of time only, but I am waiting for long period results at the moment. The only negative thing I have noticed that in my output file the Mass exiting (Boundaries or source) for Tracer 1 and 2 - the error are huge. I will try your suggestions to see if it improves the result. I am little bit worried when I am going to add the stratification to the tidal case as I need to have prescribed H at open boundary there. I will follow your advice. Thank you.

Kind Regards!

Violeta

Hi Christophe,

Thank you so much for your advice. The Thompson method solved the problems at open boundary. By activating it, I was able to go back to my original open boundary settings 544 and even 566, and surface behaves very well. I managed to add the stratification to the tidal case and both work fine.
My only concern is that in output file (it is running at the moment for 6 days model time) I can already see that on Tracer1 and Tracer2 the error on the mass during each time step is very high. You mentioned the prescribed tracers value to be specified at open boundary. I used it for my old box stratification experiment with one open boundary before and it did not work well, as I had two stratified layers there. When I prescribed the tracers values in steering file it gave me the wall of warm and saline water entering into domain. Could I actually prescribe tracers values as two layers, as in steering file I could only enter one value for temperature and one value for Salinity. But in domain I have one layer with some temperature and salinity values and another layer with another temp and salinity. So really I have to prescribe 4 values in total. So I that is why I have free tracers values now. But I think it causes tracers errors in this case. Please advise. Thanks again for your help.

Kind Regards!

Violeta

Violeta

In such case, you have to program the vertical profile of tracer on your boundary if you want stratified values.
Have a look into TRA_PROF_Z subroutine

Good luck