open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi all,

I have been working with a grid which have been constructed using Mercator projection and it worked fine. But, due to amplitude dumping and phase shift in my study area, which I couldn't solve by adjusting many of the parameters in Telemac2D such as those related to diffusion of velocity, turbulent, bottom friction formulations and the calibration terms that disturbed my boundary.A decision was then made to visualize the model's performance with the same grid but constructed using a different projection system.(Cartesian coordinate and longitude/latitude).The boundary condition is also adjusted in accordance to the geographic projection choices of grid's, as well as, the projection parameters in Telemac2D steering file.

But, every simulation I made with these grids gave unstable results.Also, I couldn't use ALTIMETRIE SATELLITE TPXO for the initial condition,using it terminates model's runs before starting.

One last thing, while using the primitive hydrodynamic equations with the first grid(made by Mercator projection), I did also get the same unstable results(NaN) but, the model working very fine with the wave equations.

where is my error?


Regards

Biniyam S.

Hello

Even though the disgnosis is hard to achieve with these few details, I think that unstable results are in reality due to a wrong setting of the model (boundary conditions, initial conditions etc...). You can not use TPXO option for initial conditions, so these latters are missing and this can explain all the problem.
What is the real gain to change the projection, with mercator projection every thing went well (If I have well understood)
with wave equation, final linear system is symmetric and then causes less problems to solvers, whereas for primitive equation, final system can be not symmetric which can cause troubles for some solvers.

with my best regards

Riadh

Hello,

Exactly, with primitive equations the linear system is not symmetric, so linear solvers like 3 (normal equation) 4 (minimum error) or 7 (GMRES) must be used, a simple conjugate gradient (1) cannot work. I advise to choose 7.

With best regards,

Jean-Michel Hervouet