CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   ANSYS Meshing & Geometry (http://www.cfd-online.com/Forums/ansys-meshing/)
-   -   [ICEM] Shrinking domain while retaining mesh properties (http://www.cfd-online.com/Forums/ansys-meshing/79756-shrinking-domain-while-retaining-mesh-properties.html)

Josh September 2, 2010 16:23

Shrinking domain while retaining mesh properties
 
Hi there -

I'm completing a domain size sensitivity analysis. I already studied a 20c farfield. I want to study a 10c farfield inlet with the same 20c farfield wake and the same mesh properties (y+, expansion, etc.). To put it simplest, I would like to cookie-cut a 10c farfield grid out of my previous 20c farfield, as shown below.

http://a.imageshack.us/img820/3374/20c10c.jpg

I want the mesh to remain the same so that I can focus on the effect the inlet domain size has on the results. Can this be done simply, or should I just create a new domain with a smaller farfield but equal first layer height, expansion, and streamwise resolution?

Thanks for any help!

Josh

Josh September 7, 2010 17:03

I hate to bump topics, but I'm really struggling with this one.

BrolY September 8, 2010 03:20

I thought you could create a 2nd material point where you put the "10 to 20"c farfield elements. It would require to modify your blocking with the "10 to 20"c limit.
Then, you could delete this part and you would just have the 10c farfield.

Josh September 8, 2010 12:57

Correct me if I'm wrong, but are you suggesting the following?

Create a new body encompassing all the elements (points, curves, surfaces) I want to delete (from 10c to 20c in the farfield).

Modify my blocking so that it only applies to the region from the airfoil to the 10c farfield.

Delete the new part.

Josh September 8, 2010 18:56

I couldn't find a simpler way to do it than the one listed above and expanded upon below.

1) Create new points, curves, and surfaces defining your desired domain (in my case, the black stencil outline shown in the picture).

2) Associate the vertices and edges of the existing blocking scheme to the new geometry's points and curves, respectively.

3) Snap project the vertices.

4) Delete the parts that contain the old points, curves, and surfaces.

5) Some of the edges will be faulty because ICEM keeps the old edge length but fits it to the new geometry. Fix these edges using the "link edges" function with the unaffected edges set as the source edges and the skewed/faulty edges as the target edges.

6) Adjust the edge parameters of the faulty edges to coincide closely to the old spacing scheme. This can be done visually by enabling your new pre-mesh and the previous actual mesh, which is still preserved since you've only changed the geometry and blocking, and observing the overlap. They should overlap very nicely. If not, adjust the edge parameters accordingly.

- On that note, an easy way to assure good overlap would be to split the blocking of the original mesh at a point near to the new farfield line (for my case, at 9c or so from the airfoil). This will act as a protective "shell" around the old and new meshes. When you associate the old edges and vertices to the new curves and points, the edge parameters will maintain their original properties but they'll be in a tighter domain, so the mesh will have the same amount of elements in a smaller domain (causing denser grid). By splitting the mesh close to the new farfield boundary before performing all of the aforementioned operations, you ensure that every element within the split mesh region is unaltered after associating the old mesh to the new geometry. Only the outside elements will have require tweaking of the edge parameters.

Thanks for the help, Alex. If your method is simpler and I misunderstood it, please let me know.

BrolY September 9, 2010 03:30

This was mi idea: "On that note, an easy way to assure good overlap would be to split the blocking of the original mesh at a point near to the new farfield line (for my case, at 9c or so from the airfoil). This will act as a protective "shell" around the old and new meshes. When you associate the old edges and vertices to the new curves and points, the edge parameters will maintain their original properties but they'll be in a tighter domain, so the mesh will have the same amount of elements in a smaller domain "
If you delete the blocks of the "10c to 20c"' domain, you just have to change the number of nodes to create your new good mesh. Am I right?

Your idea of creating a new geom domain and re-associate the edges is nice too. But it means you would ahve the same edges parameters than before, so you would have the same issue of the densere grid, am I right?

Josh September 9, 2010 11:29

Hi BrolY =

I didn't realize that's what you were suggesting. It's difficult to decipher instructions for a program like ICEM.

Quote:

Originally Posted by BrolY (Post 274549)
Your idea of creating a new geom domain and re-associate the edges is nice too. But it means you would ahve the same edges parameters than before, so you would have the same issue of the densere grid, am I right?

Since I created the outer shell at the 10c farfield, all of the inner elements are protected and overlap nearly perfectly with the old mesh. The only problem elements are the ones between the shell and the 10c farfield, which should be a very small region. I just changed the edge parameters of that region.


All times are GMT -4. The time now is 02:15.