open TELEMAC-MASCARET The mathematically superior suite of solvers

I've try to the tide test case given there : www.opentelemac.org/index.php?option=com...4833&Itemid=100#4851

The cas1 works very well. I've tried to make my own simulation, I've just changed the bathymetry file, the CL file, and GEOSYST in princi1.f
After few seconds of run I get very high velocities, upper than 1000m/s... I should miss something but I don't understand what...
Thank you for your help

Hello Francois,

You have to change more things to adapt the test case to yours. In particular, I think of:
- in the steering file:
- the initial conditions,
- your time step, depending on your case (mesh size, velocity on the area),
- in the Fortran file:
- in TIDAL_MODEL_T2D, if you change GEOSYST, you may have to change NUMZONE also (see the comments), the arguments CTIDE and MSL are used for calibration and depend on your case, and XSHIFT and YSHIFT values are very specific to your case. Normally, they should be equal to 0.D0, but they were changed for this example,
- in corfon, the modifications are really dependent on the mesh numbering, in particular on the boundaries. That treatment prevents on problems for imposing boundary conditions of tide for nodes that are not wet. This treatment works in parallel,
- you can delete bord_tidal_bc in you Fortran file, because it was just changed for the value of BETA (by default, it is 0.D0) for this test case.

You should also check your harmonic constituents file ('FICHIER DES CONSTANTES HARMONIQUES'), just edit it and see if there are no strange values. If there are, that means that some parameters previously told above should be changed.

Hope this helps,

Chi-Tuan
-

Hi Chi-Tuan,
Thanks a lot,
I've done the changes but my problem is still there. I don't know how is done the harmonic constituents file (maree file in your case right?) but I've very high values compared to your test case.

Here is the beginning of the file:
1
210 273
210 -32154.87
2.619 4.1 0.670 166.7 0.488 63.1
1.041 32.1 0.289 201.2 0.300 96.7
0.628 350.5 0.166 169.4 0.092 57.8
0.000 356.0 0.000 356.0 0.000 356.0
211 -32154.87
2.619 4.1 0.670 166.7 0.488 63.1
1.041 32.1 0.289 201.2 0.300 96.7
0.628 350.5 0.166 169.4 0.092 57.8
0.000 352.0 0.000 352.0 0.000 352.0
212 -32154.87
2.619 4.1 0.670 166.7 0.488 63.1
1.041 32.1 0.289 201.2 0.300 96.7
0.628 350.5 0.166 169.4 0.092 57.8
0.000 348.0 0.000 348.0 0.000 348.0
213 -32154.87
2.619 4.1 0.670 166.7 0.488 63.1
1.041 32.1 0.289 201.2 0.300 96.7
0.628 350.5 0.166 169.4 0.092 57.8
0.000 344.0 0.000 344.0 0.000 344.0

The values -32*** are not good right?

Hello Francois,

You are writhe, the harmonic constituents file in the test case is called "maree"

Although these values -32*** are not used after for computation, they are very strange. These values correspond to the interpolation of the bathymetry at the open boundary of you local domain, in the "geofin" model. The problem is that the minimum value of bathymetry in this last model is around -5000 m. I think that all information is not well filled.

Moreover for your 4 first nodes, the harmonic constituents do not vary for the 3 first waves.

It seems that the interpolation is not good.

Can you send me your case (geometry + boundary conditions files) please?

Regards,

Chi-Tuan

I've put all the files in the zip file,
Thanks a lot for your help
François

Hello again,

Thanks to submit this case. That has showed me a bug in BORD_TIDAL_BC. Please find a correction for your Fortran file. There was a bug when the coordinates are in longitude/latitude, sorry. I have not checked this possibility. It will be fixed in next version.

Nevertheless, you will face a problem with Telemac. During the computation, Telemac requires a mesh with coordinates in plane projection (eg UTM, Lambert, Mercator compatible with Telemac). You have two solutions:
- to convert your coordinates in CORRXY if you want to keep your initial mesh in longitude/latitude (I let you do it by yourself, good luck, you may need Jean-Michel Hervouet's book),
- to rebuild your mesh with coordinates in a plane projection (UTM or Lambert). I recommend this last suggestion if your domain is not too wide (like yours).

Best regards,

Chi-Tuan

I have convert the lon/lat into Lambert (with Bluekenue, very usefull!), I'm not sure of the parameters for the conversion, I've to check that.
And it works! sorry for my mistake...
Thanks a lot !!!!

Francois,

Be careful with your conversion formulae if you want to use the subroutine bord_tidal_bc that allows to generate the harmonic constants file. If you want to use standard Lambert projections for France (1, 2 etendu for your particular case), you have to take the same constants in the formulae as in IGN algorithms. In your video, you have x that can be equal to 0. and that does not correspond to "classical" coordinates with IGN algorithms. Otherwise, you have to implement your "own" Lambert projection with GEOSYST = 0.

Anyway, your harmonic constants file should not be so different from the one in my previous archive.

Regards,

Chi-Tuan

Ok, I confess I've done it quickly to see if the model was running. I found the constants there if someone else is looking for them: support.esrifrance.fr/documents/generali...ctionsfrancaises.htm
I'll check the harmonic constants also,
Once again, thank you!
Regards,
François

I converted the lon/lat in lambert 2 etendu, I then changed GEOSYST and NUMZONE.
At the beginning of the run I have the error:

ERROR WHEN INTERPOLATION, K= 1 DIS= -2.1600685494339502
ERROR WHEN INTERPOLATION, K= 268 DIS= -0.12429054761042899
ERROR WHEN INTERPOLATION, K= 269 DIS= -0.36608741463576028
ERROR WHEN INTERPOLATION, K= 270 DIS= -0.87967730735258998
ERROR WHEN INTERPOLATION, K= 271 DIS= -1.2151910827308545
ERROR WHEN INTERPOLATION, K= 272 DIS= -1.5507924772923514
ERROR WHEN INTERPOLATION, K= 273 DIS= -1.8862037779934635

I tried to take the file princi1modif.f before your modifications and i have not the error anymore.

However I have some strong velocities in the south-west corner (up to 65m/s):

I think I will remove the corner from the mesh to avoid the problem...