[wjy-c]

In CFX mesh quality, I do not quite understand the meaning of expansion factor.

i read the CFX help document and regard expansion factor as an ratio between one gird's area or volume and the other gird's area or volume which next to that one. am i right?

in my model, i read the CFX solve document and notice that all my expansion factor in every domain is “!”, so i remeshed my model and make sure all the girds are nearly the same size. but the expansion factor is still "!".

I really do not what is wrong, does anyone have some suggestions for me? Thank you!

[dbagtbag]

Mesh expansion factor is the ratio of the volume of the larges and smallest sector volumes in your domain. Its a common measure of mesh quality but as usual take it with a grain of salt as its pretty case dependent. I'd look at your simulation again and see if it justifies a large expansion ratio where you're getting them.

[brunoc]

Expansion factor is the change in volume between two neighboring cells, more specifically, between two sub-control volumes on neighboring cells.

Expansion factors are controlled by mesh expansion ratios. But even with small growth ratios (< 1.1) you can have high expansion ratios on tetra meshes depending on your geometry since fine-control on tetra meshes in not something easy to do.

High values are also expected when boundary layers elements (usually prisms) are not properly created, resulting in a big volume change when the mesh switches from prims to tetra layers.

You should open your .def file in Post and check where the regions with high expansion ratio values are, then go back to your mesh and try to see if anything can be done do correct. If it's caused by geometry, it's on just a small number of cells and also on a not-so-important region, just leave it alone. If its caused by prism layers, definitely give it attention.

[ghorrocks]

Yu Omori's explanation is incorrect. Bruno's explanation is correct.

The significance of expansion factor is problem dependent. The OK/ok/! in the output file is a just a guide and your simulation may be more or less sensitive to this. For instance, for free surface modelling with surface tension it is highly sensitive to this parameter. If it goes over 1.1 on hex meshes you are going to get significant errors. But general single phase CFD is much more tolerant of this parameter.

[dbagtbag]

I stand corrected!

[wjy-c]

Hi, brunoc! Thank you very much, now I understand why my expansion factor is not acceptable.

In my model, all my cells are tetra elements, since you said that "But even with small growth ratios (< 1.1) you can have high expansion ratios on tetra meshes depending on your geometry since fine-control on tetra meshes in not something easy to do", I think my tetra elements are the problem.

However, my model has a complicated structure, it is so hard to have it meshed by Hexahedral elements, and the tetra element is the best choice for me. The ".def file" in Post" you mentioned only told me which domain has the unacceptable expansion factor, I do not know how to locate to some special region where the problem occurred. So, did I missed some important message in ".def file"? if not, how do I figure out the expansion factor problem, do i need to remesh the model to hexahedral elements?

Thank you

[ghorrocks]

Do a sensitivity check. Generate meshes with different expansion factors and see if it makes any difference to the results you care about.

[brunoc]

Originally Posted by wjy-c

The ".def file" in Post" you mentioned only told me which domain has the unacceptable expansion factor, I do not know how to locate to some special region where the problem occurred.

On Post, first go to 'Tools > Mesh Calculator' and use the 'Element Volume Ratio' function. You'll now have a new volume ratio variable. Use it to create an isovolume ('Insert > Location > Volume') with a value equivalent to those you intend to find. You'll now have a volume showing you where your high volume ratios are.

Evaluate if these are important regions. If they are, go back to your mesh (can be tetra, no need to switch to hexa on complex geometries) and use local controls to try and get a smooth volume transition on the regions you found.

[wjy-c]

I got it, Thank you!