[K.C.]

Dear Foamers,

during my research I noticed, that there is a time step restriction for the momentum predictor. I tested different meshsizes and timesteps in the cavity case of icoFoam found then following restrictions:

1.) Co = u_wall*dt/dx < 1 (as usuall)
2.) nu*dt/dx² < 0.25

The last restriction seems strange, because icoFoam uses an implicit scheme.

Why do we have this restriction, where van-Neumann-stability-analysis predicts unconditional stabilyty for the implicit diffusive term in momentum equation? Do you observe the same restriction? Do you know how to overcome this problem?

Thanks for your help,
K.C.

[K.C.]

Does anybody have an idea?

By the way an increase in corrector loops seems to soften the last restriction, but nevertheless a reson for this restriction is missing.

[K.C.]

Please,
could anybody just confirm my observation and try the icoFoam-cavity case with nu*dt/dx² > 0.25?

Every literature tells me that implicit treatment of viscous stresses should avoid this time step restriction, but my icoFoam has this restriction.

Thanks to you.

[K.C.]

This restriction is a huge problem for me and I dont understand, that every literature (e.g. Ferzinger) writes that solving the momentum predictor should be unconditional stable.

[zzzlllhh]

Dear K.C.

Did you find the answer to this problem? I tested with icoFoam and found the same restriction on time step. Could you please tell me what produces this restriction if you already got the answer? Thank you very much!

Zhang