open TELEMAC-MASCARET The mathematically superior suite of solvers

Thank you so much Jean-Michele. I will use these options as well. Thank you again. In mean time I have modified the case to use 46 layers at least, for the moment with 1 sec time step. The stratification works fine on its own but when I add the tide then the stratification depends on vertical mesh scale very much, so to get a case close to reality I really need to have many layers as possible, but it comes with cost. I will try new options as suggested to that case as well for 60 sec time step for start. Thanks again. I will keep you posted.

Kind Regards!

Violeta

Hello,

To be complete, I have also set the implicitation for depth at 1. and the free surface gradient compatibility at 0. It still runs after more than 4000 iterations, which is much better than crashing at iteration 60, but I do not know if the stratification holds.

Adding planes on the vertical is not always the solution because of the aspect ratio. The higher it is, the more you may have unphysical vertical velocities, so refining on the horizontal may have a good effect also.

Regards,

JMH

Thank you very much Jean-Michele, I have amended the steering file as per your last note. Fingers crossed that stratification and tides will work fine together.

The reason I thought to add more layers as when tides come into domain the stratification is upset, what is correct, but the scale at which the vertical mixing occurs depends on vertical mesh size. Please see figures with mesh and stratification+tide, so looks like because I have big gap between 300m and 600m the stratification shoots down as soon as tide arrives. It should mix anyway, but my setting is little bit artificial, so that is why I thought by adding more layers I might get more realistic results. But I am probably wrong and there is an easier way to avoid this problem as you suggested. Thanks a mil.

Anyway, the case is running after making changes you advised, but most important it is still running for you, so it is a fantastic news:). Thank you very much for your help. I will keep you posted.

Kind Regards!

Violeta

Dear Jean-Michele,

I hope you have a better luck with my case. By using GMRES I have EXCEEDING MAX ITERATIONS, it is already set up to 500, so I wonder if I have to increase it. This message appears from nearly first step. It is still running but I suspect it will crash sooner or later. I wonder if I could use any other Solver PPE but most of them conjugate gradients. Thanks in advance.

Kind Regards!

Violeta

Hello,

I have identified in your case a problem in parallel with characteristics, probably due to elements with no volume, and like what has been described above by "jaj". I'll try to fix it as soon as possible, not very soon though... Better to use advection scheme 14 in the while. In our Berre lake application, scheme 5 also works even with flat elements, but nothing guaranteed in its theory.

Regards,

JMH

Dear Jean-Michele,

Thank you for your advice. The case had run for 6 days with big time step, ok, apart of exceeding number of iterations, but completed fine. The only problem was that with Scheme for Advection of Velocity = 14 the stratification was destroyed from the very start. I have played around before with all Schemes (however 5 gave me different error message, but not detailed, so I am not sure why it did not work) and looks like only Method of Characteristics gives me the stable stratification, unless maybe other parameters should be tuned as well, so I could use other Schemes too. So far the Method of Characteristics works with time step 1sec. It is quite slow but it is working, and that is fine with me, as case is very complicated. Thank you for your help. Let me know if there are more ideas I can try. In mean time I will be experimenting too and let you know if any progress there :).

Kind Regards!

Violeta