CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   mesh selection (http://www.cfd-online.com/Forums/fluent/28587-mesh-selection.html)

 Chetan Kadakia June 7, 2001 22:22

mesh selection

Could anyone explain what are the advantages and disadvantages of structures versus unstructured meshes? What exactly is meant by "structured" in this context? Could some of you advise as to how I should create the mesh (I am using Gambit) for the following: 1) flow over a cylinder, 2) flow over an airfoil, 3)flow through a channel, and 4) flow in a turbine. I appreciate as much detail as anyone can offer. Thanks, Chetan K.

 chiseung June 9, 2001 03:00

Re: mesh selection

As far as I know, usually "structured" grid means a grid containing hex-type cells. Generally it is known that structured grid system gives a better answer. However, there are many cases that can't be meshed with hex-type cells. Also, even if systems are meshed, a skewness problem can arise.

So, choosing a proper type of cell (which is fit to your system shape and flow direction, etc.) ,is the best way in grid generation.

Personally I don't use a automatic grid generation in gambit except simply-shaped geometry. Usually I decompose my system properly(both 2D and 3D case) and mesh it from edge to volume. This process takes lots of time but gives better skewd cells.

 Chetan Kadakia June 11, 2001 17:28

Re: mesh selection

Thanks, your answer was very helpful. Could you point me on some literature in regards to this discussion. It is best I reference everything for my work. Plus it may give me a more in depth understanding of mesh selection.

 chiseung June 12, 2001 21:01

Re: mesh selection

Sorry Mr. Chetan

I have not read specific references about grid. I just took a lecture in school and read FLUENT manual. That's what I know about mesh.

As you said,of course, mesh selection is important. However, the systems I've encountered were able to be converged only using hex-type cells by domain decomposition. So I don't have lots of experiences in choosing various types of mesh.

In my opinion, I wonder how much the results with structured grid are different from those with unstructured grid if the solutions are reasonably converged.

This is a totally my personal idea.

Good Luck!

 Chetan Kadakia June 12, 2001 22:16

Re: mesh selection

I was reading up on it lately. I guess the structured quad meshes provided better accuracy. But unstructured quad meshes are also good for accuracy. The tri grids are more for flexibility and to save on time set up. At least that is what I understood from the manual. I find the manual to be informative but lacking in detail.

 chiseung June 13, 2001 00:30

Re: mesh selection

How about spend your time on grid generation in various geometries? I think that would be a lot more helpful to you if you're not only interested in grid theory.

It is true theoretical background of mesh selection is important but you could get a feeling of mesh selection from your experiences.

For example, set the complex geomtry and apply hex,quad,tri... to your system. In this case, you can know the flexibility of tri-mesh easily.

Or, if you have experimental data which are fit to your system, compare the difference in results among various types of cell and find which one is optimal to yours.

However, practically that is a time-consuming procedure in a real project. I think a proper grid density is more important than types of grid.

Good Luck.

 chiseung June 13, 2001 00:34

Re: spend->spending in first line

Sorry.

Understand my grammar skill.

 All times are GMT -4. The time now is 18:07.