[hannes]

Hi experts,

during setup of a cavitation case, I stumbled across the fact, that there is no underrelaxation of the pressure in the interPhaseChangeFoam solver (see attached image; comparison with interFoam's pEqn.H on the right hand side).

Looks like this prevents one from using transient SIMPLE mode...

What is the reason for the removal?

Thanks for any comments...

Regards, Hannes

Auswahl_145.jpg

[massive_turbulence]

Originally Posted by hannes

Hi experts,

during setup of a cavitation case, I stumbled across the fact, that there is no underrelaxation of the pressure in the interPhaseChangeFoam solver (see attached image; comparison with interFoam's pEqn.H on the right hand side).

Looks like this prevents one from using transient SIMPLE mode...

What is the reason for the removal?

Thanks for any comments...

Regards, Hannes

Attachment 68563

"This is true only if the equations do not converge in that time step, meaning you are using an under-relaxation which is actually too aggressive and prevents the evolution of the solution in the number of prescribed iterations. Under-relaxation should not be applied at all in PISO-like algorithm for this reason, and in SIMPLE-like algorithm you must allow a sufficient number of iterations so that the solution actually is not affected by the under-relaxation." via Alberto Passalacqua

Suggested unsteady, implicit solver stable with arbitrarily large time steps

[hannes]

Dear massive_turbulence,

yes, I know that transient SIMPLE needs iterations to converge in a time step. But it also needs underrelaxation.

I'm afraid I don't understand what you mean exactly by this quote. Can you elaborate?

Thanks, Hannes

[massive_turbulence]

Originally Posted by hannes

But it also needs underrelaxation.



Thanks, Hannes

From what I could tell about implicit methods, under-relaxation isn't (always) required because larger time steps are possible as opposed to explicit methods which require smaller time steps to converge.

Non-linear equations need iterative solutions with under-relaxation for a variable that is considered at the time-step because of divergence from the solution.

This is from another post that I found.

"Hi,

In general, people agree on the following:

- There should be no relaxation on final iteration. This can be achieved by setting relaxation factor to 1 or just removing relaxation factor from dictionary (faster variant).

- There are several mentions in presentations, that for optimal performance sum of relaxation factor for U equation and for p field should be 1.

- For PISO (or PIMPLE), in general, there is no need in relaxation.

So you start from no relaxation, if your solution diverges, you add relaxation for diverging equations. BUT relaxation is not a solution for incorrectly posed BCs." by Alexey Matveichev

About interFoam

[hannes]

Dear Andrew,

thanks for hinting me to all these posts!

Hearing that SIMPLE works without relaxation is a bit surprising to me (as is for the colleague replying to Alexey's post). I guess, it might work, but probably not in general. So, for me it would thus make sense to let the relaxation in, if it is not hurting.

My impression is more that the relaxation is removed to avoid some kind of inconsistency. Though I'm not sure where is should come from.

Do you have experience with this solver?

Regards, Hannes

[massive_turbulence]

Originally Posted by hannes

Dear Andrew,

thanks for hinting me to all these posts!

Thank Google, although it still takes know how to use it, few keywords is all it takes.

Do you have experience with this solver?

The only solver I've used a lot and have tried to modify is buoyantPimpleFoam. I've used interFoam to test a LAN with the dambreak case.

I have however looked over lots of papers related to interfoam from Chalmers, but the best way to understand interfoam is to look at MULES and PISO, as interfoam is quite advanced.