Learning Forum

[lugolxy]

Recently I worked on meshing by Ansys workbench (Anasys 18.1) for fluent simulation. I need very fine mesh. When I set max mesh size to be 3um (The whole structure is about 1cm in size) and all tetrahedron, I got good mesh and the element number is about 400,000. Then I tried to improve the mesh and set the max mesh size to be 2.5um. However, I met an error as

A software execution error occurred inside the mesher. The process suffered an unhandled exception or ran out of usable memory
 
I monitored the memory usage of my windows 10 desktop with 32Gb Ram. It didn't reach to a high ram percentage. I wonder if you or anyone met this problem before and do you know a solution? Thank you very much.

[raul.raghav]

Can you upload the workbench archive file so we could take a look at it?

[lugolxy]

Thanks a lot Rahul. Juan suggested me to re-scale it 1000 times. I tried it and it met the same problem. I also monitored the ram usage. It never exceeded 25%.

I made the geometry a little bit simpler and I could mesh the geometry by element size 2.5um, But it failed at 2um. 

I uploaded the file on One Drive with the following link

https://1drv.ms/f/s!AgfqIquEsf9CpyobbbIudFIV6IfJ

Thank you very much.

 

Best,

Xinyu

[raul.raghav]

Xinyu, I looked at your geometry and your meshing approach. Although the dimensions of your geometry are quite small, the geometry itself is very simple. Why do you want a very fine all tetra mesh when you can easily create a hexa mesh for your geometry? And I'm not sure what the original geometry looks like but the simplified geometry is quite straightforward to mesh. Attached is the workbench archive file with an all hexa mesh for your geometry. You can play with the sizing numbers to make it as fine as you want. Let us know if it helps you. And i took the liberty to change the named selections based on your attached file, so please make sure its consistent with your problem.

[lugolxy]

Hi Rahul,

Thank you very much. Yes the hexa mesh is very helpful. Actually the reason I use tetra mesh is to test the difference between tetra and hexa elements. People always use hexa for this kind of simple geometry. So I'd like to test how tetra mesh affects the result. Thank you!

 

Best,

Xinyu

[raul.raghav]

Xinyu, as you mentioned 2um body sizing shows the error as you mentioned, while 2.5um works. And I tried it across two computers with varying configurations and it didn't seem to make any difference. I read more about the Patch Conforming tetra mesh and it was mentioned that it was a bottom-up approach where meshing starts from the edges, surfaces and then finally the body. I tried refining it using edge and face sizings and I was able to generate a mesh with approximately 10M elements. However, even that approach eventually shows the same error below a certain size.

[lugolxy]

Got it. Really helpful. Thank you very much. Could you also attach the file or snipped images showing the setup that you refined edge and face first and then body mesh? I am still new to ansys workbench. Thanks.

 

Best,

Xinyu

[Masooma Qizilbash]

dear concerns,

i am meshing a tank  and impeller seperately. i need total number of cells 299616 when i send both meshes in the fluent solver but i am not having this numbe of cells. element size at 1mm i have 2215223 and at 1.5 mm i have 1316002. i am doing validation case and in the article they used unstructured tetrahedral mesh and no of elements 299616 so i need this much no of cells. 

when i further increase the element size it i gives constant number of cells that is 1316002 even i increase the element size uptill 500 m of impeller size and 1000 m of tank size but when i update mesh it gives the error execution error inside the mesher

what should i do to achieve number of cells 299616. please guide me 

[Rob]

Start by doing the tutorials. It's pretty much impossible to work out settings to give a cell count, instead look for the size settings used in the paper.