topic SST k-omega (Hellsten) y+ values and Meshing in CFD Forum

SST k-omega (Hellsten) y+ values and Meshing

Anonymous — Tue, 13 Oct 2015 03:40:20 GMT

Hello,

I am currently working on creating a simulation of a bicycle wheel for the purpose of finding the theoretical drag and side forces on the wheel. I have set up a simulation with the SST k-omega RC (Hellsten) turbulence model. The Guides and forum posts have been unclear as to how the hellsten correction effects the y+ calculations of the simulation. I have been operating on the assumption that the simulation should have a heavily refined boundary layer mesh and have surface values for y+ of approximatley 1 and no more than 4-5 within the viscious boundary layer. I am currently using 10 boundary layers with a layer factor of 0.45 and a gradient of 1.5. Is this the case for a simulation running with the Hellsten correction? I am thinking now that perhaps I should be using the standard SST k-omega turbulance model instead as the simulation is running in steady state with no rotation.

My mesh refinement is not too bad I think apart from maybe some skewing issues. I can't find any way of calculating skewness with this CFD package nor a way to improve it. I have inserted a picture of my current mesh refinement below. Any suggestions?

Thanks for any help that can be provided!

Hamish Knight

Re: SST k-omega (Hellsten) y+ values and Meshing

Jon.Wilde — Tue, 13 Oct 2015 15:11:26 GMT

Hi,

I am not 100% sure about your main question yet but two points I would make are:

It is easier to view the mesh on a cut plane, set to 'shaded mesh'
To understand if elements are skewed, you can plot Nodal Aspect Ratio

Re: SST k-omega (Hellsten) y+ values and Meshing

Anonymous — Tue, 13 Oct 2015 23:04:43 GMT

Hello,

I think you've answered my question about how to find the skewness in the model. You're right the cut plane is much better for showing the mesh. My main question otherwise was whether the SST k-omega RC (Hellsten) turbulance model requires a certain range of y+ values to give meaningful results. I am currently trying to run a similar set up with the standard SST k-omega turbulance model to see if that gives me the kind of results I'm looking for. To be clear I am looking to calculate the drag and side (perpendicular to drag) forces acting on the wheel volume. I have been using the wall calculator to do this. Is there anything more you need to help understand my question?

Thanks,

Hamish Knight

Re: SST k-omega (Hellsten) y+ values and Meshing

Anonymous — Wed, 14 Oct 2015 00:36:44 GMT

P.S.

I am also looking into if disabling the intelligent wall formulation (with the SST k-omega turbulence model) will help with my results.

Re: SST k-omega (Hellsten) y+ values and Meshing

LKania — Wed, 14 Oct 2015 12:55:54 GMT

As regards the meshing issue, the transition from the highly anisotropic boundary layer mesh to the isotropic region requires a layer of stretched elements as you have identified. If you are not already doing so, you might enable the "Enable enhancement blendng" checkbox in the Boundary Mesh Enhancement dialog. This will facilitate the transition between regions of the mesh but at the expense of a much larger mesh count.

Re: SST k-omega (Hellsten) y+ values and Meshing

Anonymous — Wed, 14 Oct 2015 22:54:32 GMT

Hello,

Yes I do have enhanced bleeding enabled for this reason.

Re: SST k-omega (Hellsten) y+ values and Meshing

Royce_adsk — Thu, 15 Oct 2015 02:50:31 GMT

A few comments here for you that might help:

Y+ Rules for all SST models are similar. If you want to be very accurate you want it to be less than 4-5, and theory would dictate a value of 0.3. You will probably be OK with the 4-5.
Stick with the Hellsten model. That should work fairly well for you.
Gradation at a value that large I have found to be less accurate. I would suggest sticking to 1.15 - 1.25.
Make sure you use the resid boundary force flag. This needs to be set at both the scenario and default flags list.
Lee's comments are excellent and I would always suggest checking out this option
Make a tighter surface mesh around the tire
When using many layers with gradation I tend to make my layer factor 1.0.
Confirm that you are running to beyond tight convergence. If you stop too early your forces will be probably off.

Re: SST k-omega (Hellsten) y+ values and Meshing

Anonymous — Thu, 15 Oct 2015 04:36:18 GMT

Hello,

I've set up a simulation with these settings enabled. I'll let you know if there are any additional problems.

Thanks