[nyox]

I am doing an undergraduate thesis on the turbofan and i am very new to ICEM.

I am not able to improve my mesh quality by getting rid of the highly skewed
cells, at such the mesh will not run in Fluent at all.

Can i have some suggestions as to how i can go about improving it?


This shows where the highly skewed cells between 0.96 - 1.02 are. They make up 13% of my mesh.



How i went about with my blocking;




Any help would be greatly appreciated. Thanks!

[bluebase]

Hello Shane,

check out the describtion of the Eriksson Skewness in the ICEM Help. It's ranging from 1 to 0, where 1 means good and unskewed.
So, the displayed elements (the 13%) are very fine. Try to locate the elements on the lower end (skewness lower than 0.1).

A very handy feature is called "Scan planes" its in the context menu of Premesh. It simplifies the examination of the blocking in 3d.

If you need more help, post some more screens with the real unskewed elements =)

With regards,
Sebastian

[nyox]

Hi Sebastian,

Thanks for your input. I must have read the wrong information somewhere. Skewness close to 1 is indeed good for a hexahedron element mesh.

I've got 0.05% of my mesh below 0.1 skewness and i doubt that is significant.

I am still unable to run my mesh in Fluent at the moment and I suspect it's got something to do with the mesh quality. What other qualities do you recommend to check? AR, if so what is a good gauge? Smoothness for my mesh seems fine for now.

[bluebase]

I must object your assumption, 0.05% below 0.1 might not be much. But your minimum skewness is 0.0! That means there are volume elements degenerated to a flat surface. (btw there are different definitions of skewness in most preprocessors)

The rest of the shown histogram seems good, so i assume there might only be a small area with distorted elements. Try to find them. And, determine its cause.

As a rule of thumb i usually try to get eriksson skewness > 0.3 and Derminant > 0.4.
The criteria Aspect Ratio depends on simulation settings. Sometimes its valid to have very high AR, e.g. simulating a couette flow. But often you should keep AR small where strong gradients/changes in the flow exists.
I think many studies exists which examine the influence of certain distortions.

I would suggest to use the determinant quality check and redefine its range to -1 to 0 (standard is 0 to 1) to find the negative/zero volume cells.

[nyox]

It appears i have 961 negative volume elements! Now this is highly possibly why Fluent won't run at all.



Would merging the vertices at the tip to a single point help in such a case?

[Far]

is it rounded face or cone shaped?

[nyox]

It is cone shaped.