YPlus and Aspect Ratios
 

Hey Folks,

I would like some advice regarding my mesh as I'm currently running some 3D simulations using a k-w SST turbulence model and I've been hearing some conflicting opinions on what to use for my yplus which would affect my first layer height and aspect ratio around my cube mesh for this example.

To get an accurate solution, the idea would be to have your yplus in the sub laminar region being 1 < y+ < 5 especially for SA model. However, with the k-e with enhanced wall function or k-w with SST the yplus can be higher to compensate such as having yplus at 30 < y+ < 60.

My yplus is originally set to 40 but by changing it to 5 for example the aspect ratio of the mesh near the wall is really high. Would this cause an issue with the solution as well?

Any input and advice is appreciated.

Thanks
Lee

From my experience, I think that the aspect ratio itself is not meaningful very much. It depends on the direction in which your flow will vary. For example, you have a flat plate, x is the direction of the flow, y is the perpendicular. Due to boundary layer theory, your velocity profile will vary mostly in y direction. Therefore, it is fine if you mesh has high density in y direction and it is coarse in x direction, even if aspect ratio warning appears.

If you're not using a wall function, then you need to integrate all the way to the wall boundary, and your y+ should ideally be 1; although, up to 5 is considered "acceptable" by most. If you are using a wall function, then your first cell height needs to be outside of the viscous sublayer region (logarithmic region) of your boundary layer profile.

The extent of this region varies considerably depending on your Reynolds number. For high Reynolds number flows a y+ of 100 may still be within the viscous sub-layer, and so selecting an initial cell spacing anywhere between 5 and 100 in such a case would be inappropriate. For lower Reynolds number flows, the upper limit of this region may be at a y+ 100. You have to know the type of flow you're simulating.

That being said, the "enhanced" wall function that you mention is scalable in that it determines the local y+ value and is suppose to apply the wall function if appropriate.

Best Regards,



Zach