Learning Forum

[bs]

Hello,

I'm curently working on evaluating the availible turbulence models for the steady state simulation of a centrifugal compressor stage (NASA HECC). For most of the models, a y+<1 is recommended for the entire domain to accurately capture the boundary layer. Naturally, this results in a relatively small first element offset. 

Using Turbogrid I am able to produce a viable mesh which satisfies the y+ target, however, the Maximum Edge Length Ratio of elements close to walls remains above the limit (1000) regardless of which mesh settings I change. Even with a mesh size of 17 million elements for the impeller alone, the Maximum Edge length Ratio is around 33000 with 15% of the cells being above 1000. 

Am I missing something? In this instance would the Maximum Edge Length Error not impact the quality of my simulation?

[C N]

Hello Brandon,

For steady state flow simulation , in terms of physics if the desired y plus settings is set then it should not impact the quality of the simulation. But definitely you can refine the mesh using mesh and topology improvement features in the latest version of 2023 r2 .I am attaching the user guide link for your reference.Depending on the complexity of the model , there are many mesh refinement options available. My suggestions include  changing the Target Maximum Expansion Rate and Cut-off Edge Split Factor to control the mesh size and the mesh quality at the boundary layer and cut-off edge and these parameters will have an impact on maximum edge length ratio.

2. New Features and Enhancements (ansys.com)

I hope this helps you in your simulation.

Thanks,

[bs]

Thank you for your response.

Firstly, the compressor's blades have rounded Leading/Trailing Edges (LE/TE), so the cut-off factors would not be applicable in this case. It is worthwhile noting that the LE/TE do not fall outside the limits of the maximum Edge Length Ratio (ELR) (as you can see in the image).

Secondly, the maximum expansion ratio is around 1.3. Manipulating the target Expansion Ratio (ER) had no effect on the maximum ELR. It did reduced the percentage of cells which were outside the given limits (only by 1 or 2%), however, the maximum ELR remained the same. I'm assuming that it increased/decreased the number of cells in the span-wise direction while the number of cells exceeding the limits mostly remained the same.

The only way I could reduce the maximum ELR was by increasing the first element offset (thereby increasing the y+) or by increasing the number of cells. If the cells exceeding the edge length limit reside in the boundary layer this makes sense, since the ER has no effect on the ELR of the elements adjacent to the wall (it might reduce the ELR for subsequent layers, although I doubt it would help much for an ELR in the order of 1e4)

From what I've gathered from various forum posts, the aspect ratio of boundary layer cells typically has little effect on the simulation quality (as you suggested). This is true in cases where the streamwise gradients are smaller than the spanwise gradients, since the elongated cells would adequately capture the streamwise/tangential gradient while the transverse gradient is captured by the boundary layer. However, in the instance here flow seperation occurs, the tangential gradient becomes significant. In this case the aspect ratio does affect the simulation quality since the point at which flow seperation originates from affects the results of the simulation. I think the only logical course is to perform a tradeoff study to find a balance between the streamwise resolution and the maximum ELR based on the degree of flow seperation experienced.