interFoam solution tolerances
 

Hello All,

I am working with the interFoam (v 1.7.1 and 2.0.0) solver to analyze dam spillways. To date I have primarily looked at 2D cases with fairly good success. Now as I prepare to try and run some 3D cases I am looking for ways to speed up the solution time. Until now I have been initializing (setFields) the domain with water (alpha = 1) in the reservoir but not going over the spillway. As a result there is a fair amount of time for the computation to reach a steady-state solution because the initial condition of alpha is not very close to the final solution.

I am thinking of reducing the tolerances and increasing the relTol (in fvSolutions) to help it get to steady state more quickly. Currently the fvSolutions file is the default one from the damBreak tutorial/example. My though is to “loosen” the tolerances to get to a less accurate steady-state solution faster and then run a shorter simulation with “tighter” tolerances to get to a more accurate final solution in less total time. Does this make sense? Could anyone suggest how loose I should set the tolerances?

Thanks,

Matt

I am using with the interFoam (v 1.7.1) solver to analyze dam spillways. However, I am having some trouble in 2D.
Can you help for me analyze dam spillways solver application in 2D ?
This is the my solver. However, I haven't seen any change of the free surface . So how do I do ?
I did not see liquid in that backward facing before liquid into. I use VoF method - interFoam . I want to know if there one some more file pdf with instructions which will help me learn how to set up cases for studying phase fractions using VoF.

Plz ! can you help me ? thanks !
cheers !

phuchuynh

phuchuynh,

I am happy to help much as I can; however, you haven’t provided enough information for anyone to help you. So, I will just ask you some questions.

1) Were you able to successfully run the dam break tutorial? This covers all the basics of setting up an interFoam case, and is the place you should start, if you haven’t already done so.
2) What have you done so far to set up your backwards facing step case?
3) What step of the process are you getting stuck on?

MD

Hi,
A couple suggestions. First, you will probably get more bang for your buck by trying to nudge up the Co rather than tweak solver settings. I routinely run with maxCo 1.0 and maxAlphaCo of 0.5. You could even go higher if you increase the number of nAlphaSubCycles in system/fvSolution. Sometimes I can get away with doubling both and using nAlphaSubCycles 4. This is pretty aggressive though.

BTW, you haven't said what you are interested in--transient or steady-state behavior.

Which leads to the second point, perhaps for your application, the newly introduced timestepping for steady-state VOF problems would be useful (http://www.openfoam.com/version2.0.0/steady-vof.php). I have not tried this out myself.
Hope this is helpful.
-Kent

Hi Kent,

Thanks for your suggestions. I originally posted this question a few months ago. Since then I have successfully run the model in 3D. I ultimately used a course mesh to get it going, and then progressively refined the mesh. This seemed to work fairly well, but for my next case I will try as you suggested.

Regarding the steady-state VOF, I tried it briefly, but had trouble and was already committed to using the transient solver with constant inflow. I will likely give it a shot again at some point in the future.

Matt