Interfoam - Problem with mesh quality ?
 

I found a problem using Interfoam on a particular mesh.

I need to have a small inlet in a big domain.
I meshed the inlet with an higher refinement level than the rest of the domain. I already used this procedure without any problems but now I'm obtaining strange results.

The incoming fluid (water) is supposed to enter into the coarse domain and fall down by gravity force. The sims actually shows the fluid stucked at the fine/coarse mesh interface.

I enclose a picture of the mesh (1/4 of the total) and my setup. Here you can also find the movie: http://www.box.com/s/c71007067151cfd480e9

Any idea ? At the beginning I was thinking bad gravity sign...

Thanks,
Daniele

BTW, I have problem with alpha>1 too... I think this for sure doesn't help....

Daniele

Hi Daniele,

I've written several posts on the following thread: http://www.cfd-online.com/Forums/ope...rintstack.html - there you should find several tips learning how to use OpenFOAM by trial-and-error ;)

Best regards,
Bruno

Hi Daniele,

Is there a reason why you stopped the refinement after the inlet? Due to the coarse mesh down stream your calculated capillary forces will be very inaccurate. One of the big disadvantages of interFoam is that it needs a lot of mesh resolution to resolve the interface region. I had a case where the interface region was highly resolved in one area of the domain and coarsely resolved in another. Because the surface force was calculated differently on each region of the mesh I got strong artificial tangential marangoni type currents across the interface.

From your video it also appears that there is some "bubbles" or alpha != 1 coming in from your inlet, any idea why that would be happening?

Also, how big is this domain? Is it small enough where surface tension has a big role?

Cheers!
Kyle

@Bruno: thanks for the link, I'll study it.

@Kyle: I had to stop refinement in order to keep the number of cells reasonable.
I'm trying to simulate a sort of shower. A small inlet (400x10mm) into a 1000x1000x2000mm space. Now I better local-refined the mesh and it seems a little better, but, at the moment, I don't want to manage more than 2-3 MCell.
I'm still learning, so any advice is helpfull.

Regarding the alpha behavior I think was due to an incorrect setup of the p tollerance at the beginning of the simulation.
At first, I tried to release it (1e-4) thinking to get a faster calculation but, actually, using the "default" value of 1e-7 is better from all point of view.
This, togheter with a finer mesh, solved the problem in my current run.

Thanks.