open TELEMAC-MASCARET The mathematically superior suite of solvers

Is it realistic in this case to find a big mass increase in the final mass balance of the tracer. I used all the available schemes for the advection of tracers and I usually got an increase in the tracer mass except the SUPG which give almost no change in the mass, however the SUPG gives very high negative values.

Tracer:
Initial mass balance: 1.426
Final mass balance: 2.231
mass loss:-0.805

The mass increase is more than 50%

Hello,

No, this is certainly wrong. In the basic lock-exchange case, you should see almost machine accuracy, like 1.D-10 at every time step.

You should see that with the test-case "lock-exchange" provided and the version 6.1 or 6.2. The corrections I have sent are for version 6.2.

So first try the basic lock-exchange case in version 6.2 without my corrections and without sub-iterations. It must work with all advection schemes for tracers (except the method of characteristics which is slightly non conservative). Once this is secured you can add my modifications and see what happens, and then the iterations for non linearity, then your own personal case.

With best regards,

Jean-Michel Hervouet

Hello,

It seems a problem with time step, I have noticed that decreasing the time step two much increases the errors in the mass balance. I was using a very small time step (0.005) to keep courant number less than one (I know that telemac controls time step to insure this in case of explicit schemes).

However, it is important to say that with very high courant number in case of explicit schemes the quality of the results was also bad.