[krombacher]

What are the best software to simulate a 3d, free surface, sloshing problem for an engine oil pan, on a race track (addressing braking, turning, and accelerating forces)? We are having oil starvation issues and have a number of baffle designs to try and optimize.

Here is a video of the kind of simulation I want to do:

https://youtu.be/-Iyh-S-LGN8

Or

http://mdx2.plm.automation.siemens.c...-tank-sloshing

I've tried ansys fluent and it seems to very slow even with a simple 2.5 D tutorial problems let along a true 3D one.

I have a mesh .stl file of the oil sump surfaces.

[Ahmed Alkaisi]

Openfoam

Sent from my SM-G900I using CFD Online Forum mobile app

[krombacher]

Thank you Ahmed!

[arjun]

"I've tried ansys fluent and it seems to very slow even with a simple 2.5 D tutorial problems let along a true 3D one."

Most VOF simulations of these kind are slow. If you find fluent slow then you will be surprised with openFOAM.

Looking at the animation it seems like the simulation has good amount of body force applied to it. For VOF simulations Fluent offers something called body force weighted scheme so if I have to guess Fluent would be much better to use compared to openfoam here.

PS: FVUS/wildkatze also has this body force weighted scheme and have same discretization procedure, so what fluent can do it can too. Only difference is FVUS offers thinc scheme for VOF which shall be slightly better than VOF scheme in fluent as far as sharpness goes.

[JBeilke]

Are you sure that the oil/air surface remains sharp in reality? Do you expect any surface breakup or droplet/spray generation?

[akidess]

Originally Posted by arjun

Most VOF simulations of these kind are slow. If you find fluent slow then you will be surprised with openFOAM.

The difference being if you put the $50'000 you save on the Fluent license into hundreds of cores, you may get the results you are looking for quite a bit faster

[arjun]

Originally Posted by akidess

The difference being if you put the $50'000 you save on the Fluent license into hundreds of cores, you may get the results you are looking for quite a bit faster

In this case it is still no guarantee though, given the fact that how courant number constraint drops the timestep size not to mention the problems in stability due to huge body force on fuel.
Anyway if one does with openFOAM then we will come to know how fast or easy it is with openFOAM.

If saving 50000 dollars is main thing, FVUS is closest to fluent in terms of solver as it has same method with very small differences (I followed what fluent does because they lead this market).. So if fluent can do you should get it with FVUS too. (openFOAM is different algorithm minus body force weighted scheme).


PS: Actually i believe FVUS hybrid solver (fractional plus implicit) shall be most efficient as it drops to implicit scheme below certain timestep size (that user specifies).

[krombacher]

Originally Posted by JBeilke

Are you sure that the oil/air surface remains sharp in reality?

No - but how does my answer effect things?

Originally Posted by JBeilke

Do you expect any surface breakup or droplet/spray generation?

Yes, probably.

[krombacher]

So fluent may be the fastest, follows FVUS, followed by open foam(?)

Does the SPH method help speed things up? I don't need high levels of accuracy here. It's for visualization.

[arjun]

Originally Posted by krombacher

So fluent may be the fastest, follows FVUS, followed by open foam(?)

Does the SPH method help speed things up? I don't need high levels of accuracy here. It's for visualization.

In this case, check the stability of the solver first. I spent a lot of time with VOF and comparing with Fluent etc and I think the body force dominated calculations are difficult to do.

More than speed this shall be your criteria for it as of now, because the animation does look good but you don't know the problems they faced developing it.

Speed of calculation as to do with this courant criteria related time stepping. Implicit scheme as provided by Fluent starccm defuse the interface too much and is not a desirable thing. So explicit scheme is what is suggested. Now with explicit scheme the courant will decide the speed of calculation.

Fluent, openfoam, starccm provide PISO for this purpose. In FVUS I put hybrid solver that is it is fractional plus implicit type time stepping. So typically you can do say 100 fractional step and then 1 implicit type step. (That removes the error build up due to fractional steps and thus more stable than normal fractional solvers).

PS: This body force was the reason, I reversed engineered Fluent's body force weighted scheme in FVUS.

[gbacker]

WRAFTS works well on oil slosh problems.

[krombacher]

Originally Posted by gbacker

WRAFTS works well on oil slosh problems.

I cannot seem to find a copy of it. Where can I try a demo?