SimScale CAE Forum

Have I really shown that only 1 layer is needed to resolve the boundary layer on my y+=100 Hex mesh :Question:


#1

I know a lot of people don’t review new projects as they come up on the the site, so I am making this general ‘Project Support’ topic to point you to my project…

I seem to have determined that just a single layer resolves my boundary layer as well as 2, 3 or 4 layers and I want to know whether that is true or not…

Please scrutinize my project and it conclusions.

Here is the project which has a full description of what I tried to accomplish, how I did it and what I concluded


#2

Very nice and detailed description of your project Dale! Quite busy at the moment but I will have a detailed look at that ASAP!

Also inviting our SimScale community to have a look at your project and also the guys from the @CFD-SQUAD to investigate your precious work.

Cheers!

Jousef


#3

Hi Dale,

Great work. Very in depth as usual.

From my point of view and experience with layering, yes, 1 layer will def be better than no layers. Think of no layers being the same as 3 dimensional cubes representing the outline of the geometry. You would get results that are likely of significant error and unrepresentative of actual results. Having used 1 layer myself to perform quick runs with minimal headaches on very complex geometry, I agree with your conclusion and results.

As for the different between 1-4 layers. This is much more subjective. I do think some flow phenomena like down stream vortexes or any simulation involving turbulence effects down stream will be significantly influenced by the number of layers and expansion ratio. I believe that the small numerical changes associated with the varying number of layers only truly have an effect the further the influenced area is behind said flow changes that are influenced by the layers itself. Of course, there is a practical limit to it that will need to be judged depending on use case. On top of this, for a laminar or steady-state simulation, these effects would not apply.

Again, fantastic work on the layer generation. We can start recommending using a single layer which is much easier to generate than multiple layers for FSAE projects or other similar simulations.

Cheers.

Regards,
Barry


#4

Thanks Barry,

Good point on downstream vortexes, perhaps at some point I will have a look at them for my 1-4 layer meshes. However my limited experience with them says that, even if the downstream vortexes are different, the Geometry CD and CL results of my study are still valid and so are my conclusions, would you agree?

I am glad that you conclude we can start recommending 1 layer meshing in some cases.

I will be more comfortable with that if my project stands up to further scrutiny :slight_smile:


#5

Hi Dale,

Vortexes only come into play when you have a 3D wing with a finite span and a wing tip. These vortexes will help in lift generation somewhat and provide a reduction in drag if I am not wrong.

I haven’t managed to look into vortex generation significantly so I can’t make a valid judgement but I would presume that the CL and CD will not be adversely affected significantly enough to justify the increase in computing resources or time needed to generate additional layers.

In short, yes they would still be valid unless someone has done studies that says otherwise.

Cheers.


#6

Yes but my point is that I did resolve the mesh finer in the wake area so that my 1-4 layer meshes already somewhat take into account the differences in wake which the larger number of layers may introduce. This means that my 1-4 layer CL and CD results likely do take that it into account (if the wake mesh size is OK). At some point I could do a study on optimizing to mesh fineness in the wake region to look at this more closely.