CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   high aspect ratio elements (http://www.cfd-online.com/Forums/cfx/19555-high-aspect-ratio-elements.html)

Olivier April 2, 2003 11:06

high aspect ratio elements
 
Hello everybody

I hope you will not mind answering yet another question. I'm trying to model the flow around a heated cylinder placed in a rectangular chamber. I model only one quarter of the whole geometry. I put surface mesh controls on the three surfaces of the cylinder (top, front, and back) and an inflation layer to get y+ below 5 (SST model). The problem is that this thing creates high aspect ratio elements near the edges of the cylinder (where the lateral surface meets the front and back surfaces) and it doesn't converge. I have tried to change the parameters of the surface mesh controls but it does no good.

Does someone have an idea ? I read the comments of some people who say that CFX-Build makes lousy meshes, but we don't have another mesh generator so I'm pretty stucked there.

Many thanks

Cheers

Olivier

Pascale Fonteijn April 2, 2003 11:58

Re: high aspect ratio elements
 
Some hints:

- Switch on edge proximity. - use AFI instead of the default Delauney surface mesher. - do not use inflation on the symmetry planes, periodic boundaries, inlets and outlet.

Pascale


Robin April 2, 2003 16:36

Re: high aspect ratio elements
 
Hi Oliver,

Are you certain the high aspect ratio is the problem? It is normal to have high aspect ratio's in the first few prism layers. Generally, these are not problematic because they are aligned with the walls.

Robin

Olivier Macchion April 3, 2003 02:57

Re: high aspect ratio elements
 
Thank you for your answers Pascale and Robin.

Pascale, I didn't put inflation layers on the symmetry planes, inlet and outlet. And I think that I had turn on Edge proximity. But I will look thank you.

Robin, the solver says to me that there is 12 flat tetrahedra at a given location. I looked at it in Post and it is very close to the edge of the cylinder. I looked in the journal file of the database, and it appears that when it makes the inflation layer, it takes away many prisms. I guess that is at the edge of the cylinder where the inflation layers of the lateral surface and of the front surface meet. But I'm not sure.

A solution would be maybe to put a triangle mesh control in front of the cylinder to see if it helps to get a more homogeneous mesh. What do you think ?

Thank you

Cheers

Olivier

Helper April 3, 2003 05:31

Re: high aspect ratio elements
 
The high aspect ratio elements are created to capture flow that has high gradients normal to the element, as I'm sure you are aware. The high aspect ratio elements on the main body of your cylinder are fine, however, near the ends of the cylinder the normals used to create the prisms are smoothed and on the edge share between the cylinder and the end will be at 45 degrees. The limitation that only one normal is present at each surface node means that the elements in this region will be skewed, this skewing may result in poor quality prisms that will be stripped out (You can reduce the minimum internal angle on the set->inflation params form to force more prisms to be retained, and this may reduce the flat tetrahedra problem). However, the prisms may not be the best element in this region as the flow may not be so well aligned with the elements.

I suggest that you place a series of line mesh controls around the edge at each end of the cylinder to reduce the local length scale and capture the flow better, you will still get some inflated mesh and this will be combined with tetrahedral mesh away from the edge.

Hope this helps.

Olivier Macchion April 3, 2003 05:55

Re: high aspect ratio elements
 
Thank you very much

That was help

Hälsningar

Olivier


All times are GMT -4. The time now is 11:09.