open TELEMAC-MASCARET The mathematically superior suite of solvers

Dear users,

I'm running a model where I impose water levels at the offshore boundary and the velocities are free (5 4 4). For this boundary I set the Thompson boundary conditions. In the beginning of the simulation the velocities seem fine, but then after some time steps the velocities in the boundary nodes are along the boundary and don't seem realistic. I attach a screenshot of these results. Do you have an idea for the reason of this behaviour?






Thank you.
Kind regards
Maria

Hello Maria,

The only things that can be tuned in Thompson in Telemac-3D are the following:

* under-relaxation in subroutine thomps.f (library telemac2d) : there is a TETA hardcoded to 1. in thomps.f, you can try values from 0 to 1 (1 fully takes the value computed by Thompson, 0.5 takes half of it and half of the variable taken at the previous time step as guesswork). This assumes that the solution without Thompson is not too silly.

* choice of the vertical profile: in subroutine thomps_2dto3d.f there is an option IOPT hardcoded at 1 : constant velocity profile on the vertical. You can change to 2 : logarithmic profile on the vertical (keeping the same depth-averaged value). This is more complicated but certainly closer to reality.

* For those who have boundary conditions of type 5 5 5 (with Thompson it should be however 5 6 6) I enclose a safer version of bord3d.f, taken from future version 7.0 but I removed the heat exchange with atmosphere that would be incompatible with 6.3. (Anyway Maria this is not your case as you have 5 4 4).

With best regards,

Jean-Michel

Dear Jean Michel,

Thank you for your reply. I will try to change these parameters and see the differences. Anyway don't you think it is bizarre the way thompson calculates the velocities at the boundary nodes?

Thank you!
Maria