CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   ANSYS Meshing & Geometry (http://www.cfd-online.com/Forums/ansys-meshing/)
-   -   [ICEM] Aircraft unstructured mesh problem on density trailing edge (http://www.cfd-online.com/Forums/ansys-meshing/120376-aircraft-unstructured-mesh-problem-density-trailing-edge.html)

 Huo July 5, 2013 08:08

Aircraft unstructured mesh problem on density trailing edge

Hi friends,

I am doing an unstructured mesh around an aircraft. When I used the density on the trailing edge, I found there were still holes near it. So I continued to decrease the size in density function, then all of the mesh on aircraft body was disappeared. Does anybody know it? How to solve it? Many thanks!

 Huo July 5, 2013 11:06

2 Attachment(s)
Hi friends,

the first picture is the mesh I used the density bars in ICEM with size of 0.08. The second one which the body mesh disappeared is after I increased the density bar size to 0.05. If you guys know what happens, please let me know.

Thanks

 diamondx July 5, 2013 11:29

did you use the feature "define thin cut" ?? if not, that's what you should use for the trailing edge. do a google search or a forum search for define thin cut. It has been discussed before

 cfd seeker July 5, 2013 11:34

By which method you are generating the mesh? top down octree or bottom up delaunay?

 Huo July 5, 2013 14:57

Quote:
 Originally Posted by diamondx (Post 438026) did you use the feature "define thin cut" ?? if not, that's what you should use for the trailing edge. do a google search or a forum search for define thin cut. It has been discussed before
Thank you very much for such fast response! Yes, now I used the function "define thin cut". And it do work at the begining if I just defined the two parts (upper surface and down surface)for one of two wings. But I have double wings of the aircraft. Once I involved the second wing into the two parts for defining thin cut, the mesh on the aircraft body disappeared again! I was just excited to see the changes, but again the same problem comes. DO you have any idea about it? Thanks again for your help!

 Huo July 5, 2013 15:00

Quote:
 Originally Posted by cfd seeker (Post 438030) By which method you are generating the mesh? top down octree or bottom up delaunay?
Hi, I just keep the default settings, so what I am using is the bottom up delaunay. What's the difference between these two? Does this matter with the problem I mentioned above that the body mesh was diappeared?Thanks for the reminding!

Hi mate,

Quote:
 Hi, I just keep the default settings, so what I am using is the bottom up delaunay. What's the difference between these two? Does this matter with the problem I mentioned above that the body mesh was diappeared?Thanks for the reminding!
Default settings are Patch Dependent for shell meshing and Octree for volume meshing which are not compatible(unless you use "Use existing mesh" tool which is not reasonable somehow;))

Quote:
 now I used the function "define thin cut". And it do work at the begining if I just defined the two parts (upper surface and down surface)for one of two wings. But I have double wings of the aircraft. Once I involved the second wing into the two parts for defining thin cut, the mesh on the aircraft body disappeared again! I was just excited to see the changes, but again the same problem comes. DO you have any idea about it? Thanks again for your help!
Did you separate those 4 faces into 4 different parts? Also you need to separate curve between them in another new part, which seems you did it.

Try using Patch Dependent along with Delaunay algorithm, and maybe you would have no such problems.

 Far July 6, 2013 06:38

Quote:
 Default settings are Patch Dependent for shell meshing and Octree for volume meshing which are not compatible(unless you use "Use existing mesh" tool which is not reasonable somehow)
Yes you are correct. It is patch dependent quad mesh for surface meshing.

But octree method does not require surface mesh at all. So when you use octree you do not need surface mesh at all.

Quote:
 Try using Patch Dependent along with Delaunay algorithm, and maybe you would have no such problems.
Good idea and see the following link for detailed procedure for formula one meshing.
http://www.cfd-online.com/Forums/ans...rmula-one.html

But I am not in favor of this method despite of its Awesome looking mesh.

1. It is slow method and you lose all advantages of ICEM which is famous for its top down approach.
2. You need high quality and well connected geometry which is not required by octree.
3. You need to define size on each and every curve.
4. Density boxes and size functions do not work with patch dependent surface meshing.

Tetra meshing is successful, fat and accurate in ICEM (in first attempt) if you can properly define sizes and other options such as thin cut etc.

 cfd seeker July 6, 2013 11:30

I don't know why but ICEM always have problem in meshing curved leading and trailing edge surfaces of wings :confused:

If "Define Thin Cuts" is not solving your problem then I suggest another method. Split the surfaces of wings at some distance from leading edge and trailing edge both. Now you have three surfaces on the suction side and three surfaces on the pressure side. Mesh the two surfaces near leading edge(suction and pressure side) and two surfaces near trailing edge(suction and pressure side) with either Patch Dependent Method or by Autoblock method. Go to Compute Mesh and compute only the surface mesh on these 4 surfaces. Apply other settings on parts and surfaces, densities as normal. Now again go to compute mesh and calculate Volume Mesh by Octree Method by selecting " Use Existing Mesh Parts" . During the process Octree works as normal but in the end replace the mesh on these surfaces with Patch Dependent or Autoblock method. By this way you will have a nice surface mesh adapting to the curvature of the geometry.

After this you can replace the volume mesh with Delaunay and insert prisms according to your requirements.

I hope I have made myself clear but if you have doubts come back again, I will post a pic about the procedure of the method.

 All times are GMT -4. The time now is 15:23.