Question on mfix.fixed_dt Documentation: Should it be > 0 instead of = 1?

MFIX-Exa
1

I’m reviewing the MFIX-Exa documentation for time stepping, and I noticed something a bit confusing regarding the mfix.fixed_dt parameter.

“If > 0, use a fixed time step.”
Default = 0 — meaning adaptive time stepping is used by default.

However, in the Model Options table on the Model Options page of the documentation the CFL description says it’s ignored if mfix.fixed_dt = 1. I exact CFL to be ignored when mfix.fixed_dt is set , i.e., mfix.fixed_dt > 0.

2

Ignored is maybe not the best word. There are CFL limits in the code which is the default value for both advection types. You can not exceed this limit without modifying the source code. If you set a fixed time step, e.g., mfix.fixed_dt = 1e-4 you are still limited by the CFL constraint. So your simulation may start out stepping along at dt = 1e-4 but if something causes CFL > 0.9 (or 0.5 if using MOL) for the given dt, the simulation will terminate.

3

I agree. Ignore is probably not an apt description.

Is there any reasons mfix.fixed_dt = 1 is a special case? I was wondering if that was suppose to mean that CFL is not used to constraint dt when mfix.fixed_dt is set. CFL is always considered to ensure convergence.

4

I am not aware of that, but things have been changing rapidly for the GUI. @jmusser ?

5

Asked at lunch… = 1 is no special case, should just be > 0 for fixed timestepping. @cgw is updating several things on that page and will include this. thanks for catching.

1 Like