CFL numbers for two-phase flow

[kevin.a....@gmail.com]

I'm making good progress on ship forces in steady flow.  The computations appear to be more stable at higher CFL numbers, with 0.5 giving better results than 0.2 or 0.1.  I recently modified vof.h so that I could try CFL = 1.0, which seems to give reliable results.  I would welcome any advice or commentary regarding why the computations appear to be less stable for small CFL values.  I'd also welcome input regarding how to obtain greater stability at smaller CFL values.    

[Wojciech Aniszewski]

Hello Kevin.
On year and four days ago you were lowering the CFL from 0.5 to 0.2 seeking to improve stability, which I interpreted as an attempt
to increase stability (or shall we say repeatability as you said you couldn't reproduce results). What does the issue look like now?

If I remember you're using VOF (and even CLSVOF for some reason), so CFL>0.5 can trigger violation of mass conservation (your
mass and/or momentum inter-cell fluxes will be erroneous). Generally, vof.h will warn you about it (line 261).

General rule is implicit methods (e.g. if you have viscosity, your mgu solver) can "like" larger CFL for stability but decrease accuracy.

kindly
v

> --
> You received this message because you are subscribed to the Google Groups "basilisk-fr" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to basilisk-fr...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/basilisk-fr/3a8f2331-480c-48f1-afaf-1958fb744e1bn%40googlegroups.com.


--
/^..^\
( (••) ) Wojciech (Voitek) ANISZEWSKI
(|)_._(|)~
OrcId : https://orcid.org/0000-0002-4248-1194
Scholar : https://tinyurl.com/y28b8gfp
RG : https://www.researchgate.net/profile/Wojciech_Aniszewski
Mastodon: @w...@fediscience.org
Twitter : @echo_dancers3
GPG ID : AC66485E

[kevin.a....@gmail.com]

Hello Wojciech.  

Thanks for your reply.  

I'm getting generally good results, typically within 10 percent of experimental values.  During the past 2 weeks I've been taking a deep dive into stability and repeatability of simulations.  It appears that CFL = 0.5 is likely best.  Decreasing TOLERANCE from 0.01 to 0.001 is giving some improvements.  I think part of my problem with stability and repeatability has been due to the outlet.  I get better results when I use a longer domain (2 x longer, keeping everything else the same).  I also get better results when I change the outlet pressure condition from Dirichlet to Neumann.  Lengthening the domain or changing the outlet pressure condition to Neumann appear to reduce problems during the latter stages of a simulation.  My guess is that these problems are due to reflection from the outlet boundary.       

[Wojciech Aniszewski]

Hello Kevin,
thanks ,good to know. If you're unsure about possible reflection from the outflow bc, perhaps you could consider calculating the reflection coefficient?
(I'm assuming you have 2-phase flow with surface waves near your ship). I'm thinking of a procedure akin to the classical paper of Goda & Suzuki
(https://ascelibrary.org/doi/abs/10.1061/9780872620834.048, you wanna find something newer...). If that was the case, and you can afford an elongation of
the domain, maybe also some damping could be conceived, or one of the not-so-many ways to make a non-reflective outflow.

cheers
Voitek