Difficulty meshing - not sure where to go from here

Project Support
1

  • A description of the project and what you are trying to achieve
    I’m trying to do a conjugate thermal simulation on a 6U server rack with fans at the bottom and top and heat generating components in the middle. This is largely based off the electronics cooling box tutorial

  • An description of what is happening, screenshots or error logs are helpful
    The mesh generation keeps failing, citing that the boolean operation to prepare the model for meshing produced an unexpected result. The error highlights parts of the model for me to inspect. I’ve gone through several iterations of simplifying the model and running the mesher again, but each time it fails.

    image
    image1215×811 66.8 KB

  • [Important] Always provide a link to your project so that it is easy to find.
    6u-rack-v2 by pe5er | SimScale

2

Hey pe5er,

Thanks for posting on the forum.

Give us time to look into your project and get back to you with a possible solution for your error.

In the meanwhile, please explore (if you have not already) our documentation for possible solutions!

Best,
Satvik

1 Like
3

Hey @pe5er

I haven’t taken a look at your model, but following CFD standard practices, I suggest the following:

  • Start with a simplified section of the whole rack (a smaller section of it) to ensure that all your assumptions, boundary conditions, and meshing operations are working well.
  • Avoid using automatic meshing. Instead, use a parametric-hex based approach, which allows you to refine the mesh where thermal gradients are large (e.g., near the heat sources) and use a coarser mesh where appropriate.

I hope this helps.

Best regards,

Jairo

1 Like
4

Hi both, thanks for your help and suggestions so far.

Mostly this is a case of PEBKAC (user error) as I slowly learn the ways of CFD!

I’ve simplified my model a bit more and fixed some tiny gaps that were upsetting the mesher. I’ve also learned about “imprint”, which has got rid of all the partial contacts. The warning message about partial contacts did encourage me to do this, but I had ignored it in my quest to get a working mesh.

I’ve also read through the advanced meshing tutorials and now understand how I can optimise the mesh using local element sizing and region refinement.

The biggest barrier to getting this over the line is that the mesher takes a long time to run (~2 hours), so figuring out the correct settings is very time consuming. Does anyone know roughly how much faster the meshing might be if I had a paid licence and access to the maximum number of CPU cores?

5

Again, the answer lies on using a parametric hex-dominant mesher, where you have much more control.

6

The option to change meshing algorithm is greyed out, perhaps a limitation of my free licence?