I am playing around with analyzing performance of a sailboat hull using the homogenous free surface model. I've gotten grips on most things, but one thing bothers me: The Water/Air interface is more than a meter wide, even where there is no turbulence at all. (I am using "Interface Compression = 2").

Anyone here with experience on how this influences the accuracy of my results?

/C

You are never going to do it with a grid like that. You need very fine resolution normal to the water surface, something like 10 mm or so. It won't work with a tetrahedral grid. You might make it work with prism layers, but the answer is to use a hexahedral grid.

Charles

Well I could feel that comment coming, that's why I included the mesh in the picture ;-)

The next question then is how to create such a mesh? I am using an inflated boundary around the hull regions to create the boundary mesh needed by the SST turbulence model. This boundary is very thin (resolution ~1mm). So I can't apply the same inflation to the water surface region. But it seems not to be possible to specify two different sets of Inflation parameters, only several surfacegroups using one set of parameters.

/C

Hi,

Perhaps you could use mesh adaption as in Tutorial 7 (Free Surface Flow Over a Bump)

By the way, what actually is "Interface Compression"? I think it wasn't available in CFX-5.7.1. Does it have something to do with the compressive advection scheme?

Regards,

Rui

No idea, but the manual says to set it = 2 if the fluid interface is too spread out It did improve...

To improve your mesh you could put a subdomain boundary "around" the region you expect the free surface to be located. Then inflate off of that in both directions. Inflation does not have to be off of no-slip wall boundary conditions only.

This is only really useful though if the free surface ultimately conforms to an expected shape.

Neale.