[lejonetfrannorden]

Hello everyone,

I am working on a simple rectangular wing. Wing geometry is in the last picture. When I do the 2D T-Rex I am able to get a relatively good mesh, with skewness values between 0.5 - 0.6. However, in 3D T-Rex for this blunt TE body I am getting insanely high skewness values for just the cells that are adjacent to TE. My settings are as follows: collision buffer set to 2, boundary decay is 0.8 to keep the cell count low, no skew criteria and full layers set to 1 if I recall correctly. BCs are just whole wing set to wall with some delta-s to get me desired y+ and far-field set to match. As a side note, I do not think having more than 2 points on the TE matters, since I used 6 and 8 points and they also had very high skewness with the same settings, if I find a workaround I will refine the mesh around there.

What am I doing wrong here? Please see the pictures.

Kind Regards,
Lejonet

[RcktMan77]

What's the area ratio of all of the domains that make up the wing's surface mesh? If it's not below 4.0, then fix those areas first before using 3-D T-Rex.

If you're concerned about cell count, then you might want to look into anisotropic stretching in the spanwise direction with 2-D T-Rex.

[lejonetfrannorden]

Originally Posted by RcktMan77

What's the area ratio of all of the domains that make up the wing's surface mesh? If it's not below 4.0, then fix those areas first before using 3-D T-Rex.

If you're concerned about cell count, then you might want to look into anisotropic stretching in the spanwise direction with 2-D T-Rex.

Hello Zach,

Area ratio for the surface meshes were below 4 for all meshes for most of the part. However, there are 2-3 cells that had an area ratio around 4.1 at lower skin. Also, I have already used 2-D T-Rex. Is that unnecessary in this case? Please see the attached pictures below!

Regards,
Lejonet

[RcktMan77]

In the images you have attached, you haven't selected the tip domain. Make sure you select all of the domains on the wing including the tip and the leading edge when evaluating area ratio. You may see larger values between cells on the upper/lower surface and the blunt trailing edge. It looks like you may need to re-initialize the surface domains. It appears you have quads neighboring tris, which suggests that you ran 3-D T-Rex and used combination. You should recover your original surface mesh before running 3-D T-Rex.

[lejonetfrannorden]

Originally Posted by RcktMan77

In the images you have attached, you haven't selected the tip domain. Make sure you select all of the domains on the wing including the tip and the leading edge when evaluating area ratio. You may see larger values between cells on the upper/lower surface and the blunt trailing edge. It looks like you may need to re-initialize the surface domains. It appears you have quads neighboring tris, which suggests that you ran 3-D T-Rex and used combination. You should recover your original surface mesh before running 3-D T-Rex.

Hello Zach,

I have also examined the area ratios on tips, however did not include those in the pics. They were also below 4, mainly below 3.3 o I think that is not a problem. However, I am not sure I understand what you are saying. I ran 2D T-Rex on the upper and lower skins in order to reduce the cell count and make the LE and TE "prettier" in terms of mesh quality. That is why I have quads and tris over there. Also, the solver I am using does not mind quads vs tris I think (I am using SU2 and maybe Fluent). This is the part where I got lost in your reply, This is what you call combination, right? I recall creating very skewed tris by not combining to quads. Also, should I then just go for unstructured all the way in wing skins since you say recover your "original" surface mesh?

Regards,
Lejonet

[RcktMan77]

You need to be sure that you're checking the area ratios for the agglomeration of all of the domains (i.e. all of the domains on the upper, lower, tip, and trailing edge surfaces at once), not just individual surfaces. It isn't clear from your description whether you have looked at this or not.

You should not see quad 2-D T-Rex elements unless you're using Pointwise's quad-dominant surface meshing algorithm. The fact that all of the other cells shown in your images are tris, and not a mix of tris and quads, indicates that you're not. This also means the quads on your surface mesh have resulted from running 3-D T-Rex with combination turned on to combine the anisotropic cells marched away from the surface with T-Rex. You should not re-run 3-D T-Rex on a surface mesh that has cells which have already been combined by running an earlier iteration of 3-D T-Rex.

Given the simple geometry you're working on, if you can share your Pointwise project file, then I can highlight for you what a best-practice mesh should look like for your case.

Best Regards,


Zach

[RcktMan77]

Double Post

[lejonetfrannorden]

Originally Posted by RcktMan77

You need to be sure that you're checking the area ratios for the agglomeration of all of the domains (i.e. all of the domains on the upper, lower, tip, and trailing edge surfaces at once), not just individual surfaces. It isn't clear from your description whether you have looked at this or not.

You should not see quad 2-D T-Rex elements unless you're using Pointwise's quad-dominant surface meshing algorithm. The fact that all of the other cells shown in your images are tris, and not a mix of tris and quads, indicates that you're not. This also means the quads on your surface mesh have resulted from running 3-D T-Rex with combination turned on to combine the anisotropic cells marched away from the surface with T-Rex. You should not re-run 3-D T-Rex on a surface mesh that has cells which have already been combined by running an earlier iteration of 3-D T-Rex.

Given the simple geometry you're working on, if you can share your Pointwise project file, then I can highlight for you what a best-practice mesh should look like for your case.

Best Regards,


Zach

Hello Zach,

I am sending you the project file via PM. I believe that the problem is caused mainly by high area-ratio cells. As you pointed out I should have examined the whole wing domain. When I examine area ratio I see values up to 48 on cells that are adjecent to trailing edge.

Regards,
Lejonet