search

AdrienWehrle / diuca

GNU Lesser General Public License v2.1
0 stars 1 forks source link

Rampup viscosity through time in `icestream` FV to help convergence #23

Closed AdrienWehrle closed 6 months ago

AdrienWehrle commented 6 months ago

As suggested by @GiudGiud in https://github.com/AdrienWehrle/diuca/pull/21, here I made II_eps_min controllable, and use a RealFunctionControl together with a ParsedFunction to prescribe an exponential decay of II_eps_min from 1e-10 down to 1e-21.

AdrienWehrle commented 6 months ago

Convergence is much harder when ice weight is not applied at the calving front (above water) but let free...!

AdrienWehrle commented 6 months ago

Do you also observe this @GiudGiud ? with pressure zero or 1 atmosphere at the front (above sea level) it starts not converging around II_eps_min = 1e-15. I'll try and see if a slower exponential decay helps.

GiudGiud commented 6 months ago

I have not looked at that. I can see it being a more difficult problem to converge. The pressure has a jump at the outlet. And the real system has a 'wall' of fluid that should fall into the water

AdrienWehrle commented 6 months ago

I have not looked at that. I can see it being a more difficult problem to converge. The pressure has a jump at the outlet. And the real system has a 'wall' of fluid that should fall into the water

Yes for now I still haven't managed to reach convergence with 0 or atmospheric pressure at z > 0 on the downstream boundary. The high values of the rampup converge, but it stops converging as soon as II_eps_min isn't restrictive anymore. Investigating.