Author Topic: CFL condition violated in HD calculation MIKE FLOOD  (Read 851 times)

simon87

  • Posts: 5
    • View Profile
CFL condition violated in HD calculation MIKE FLOOD
« on: November 13, 2019, 12:39:25 pm »
Hello all!

Has anyone else come across the 'Warning: CFL condition violated in HD calculation' error in MIKE FLOOD? I have over 200 of these in one simulation.

This error happens to me quite often and I'm not sure as to how to fix it - does anyone have any ideas?

Thanks in advance,
Simon

Christian Tomsu

  • Posts: 96
    • View Profile
    • SPEKTER GmbH
Re: CFL condition violated in HD calculation MIKE FLOOD
« Reply #1 on: December 11, 2019, 04:01:09 pm »
Hey,

since the CFL criterion depends on the timestep length divided by cell width you can do the following to reduce the CFL number:

decrease time step or increase cell size.

You can export the CFL - numbers during the simulation as model output. This gives you a good idea, where the criterion is violated.

Bests
Christian
Project management / Modelling / Advisor
SPEKTER GmbH
Herzogenaurach

Sebastian

  • Posts: 16
    • View Profile
Re: CFL condition violated in HD calculation MIKE FLOOD
« Reply #2 on: December 13, 2019, 03:08:40 pm »
Hi,

from my experience it also helps to use the "Smooth Mesh" Option in MIKE 21 to at least reduce the amount of CFL-Errors (not sure, if this really makes sense). 

Christian Tomsu

  • Posts: 96
    • View Profile
    • SPEKTER GmbH
Re: CFL condition violated in HD calculation MIKE FLOOD
« Reply #3 on: December 17, 2019, 12:41:01 pm »
Hi,

this makes sense, because this routine smoothes also the variations in element size. means: very small elements are sometimes optimized.

Bests
Christian
Project management / Modelling / Advisor
SPEKTER GmbH
Herzogenaurach

simon87

  • Posts: 5
    • View Profile
Re: CFL condition violated in HD calculation MIKE FLOOD
« Reply #4 on: December 21, 2019, 03:18:43 pm »
Thank you both for your replies.

Is it acceptable to have a model with some CFL violations? Or is it better to aim for a model with none of these warnings?

Thank you again