[lxlylzl]

Hi all

I want to mesh a cylinder with block structured mesh in the outer region and with hexahedral mesh (using cooper scheme) in the central (outlet) region. I am attaching the image, according to which i want to mesh. I have already meshed it with O-block strategy, but it generates high density mesh in the central region which I want to avoid in my study. Can anybody suggest me how to proceed?

Thanx.

[mvoss]

Afaik. with a blocking like this there is no way not to have a "denser" mesh in the central block since all blocking edges need to pass through/cut the whole blocking and the amount of elements per line is consistent for all edges along one blocking-directions (ijk). You can try to rearrange the outer most faces (where the o-grid-block exits the domain) in order to smooth the grading at that blocking face (n number of quadratic face elements with n= elem.@side1 x elem.@side2).

[lxlylzl]

Dear mvoss,

Thank you for your reply. Dense mesh in the central part will affect the results, so I need to coarsen it. Can you suggest me some other software in which the mesh of that type (the one in figure) can be created without much tussle?

[rolloblues]

Originally Posted by lxlylzl

Dear mvoss,

Thank you for your reply. Dense mesh in the central part will affect the results, so I need to coarsen it. Can you suggest me some other software in which the mesh of that type (the one in figure) can be created without much tussle?

Hi there,
if having a conformal mesh across the black line (a wall maybe?) that divides your mesh is not mandatory for your case, maybe you could use block refinements for the external part.

Playing with the refinement ratio, you could achieve a coarser mesh in the central part while keeping the benefits of a fully structured mesh

[kad]

It might be worth a try to use the unstructered all quad patch dependent method and extrude the mesh afterwards to 3D. I have applied this method successfully to create a kind of structured region in an unstructered mesh.

Use "Curve mesh setup" for both the inner and outer circle and apply the same number of nodes on them e.g. 100. Setup height, ratio and number of layers for the outer circle, you can check their meanings in the manual. Don't touch these setting for the inner curve. Now compute the surface mesh and then extrude it.

[lxlylzl]

Dear Kad and rolloblues, thank you for your replies. I will give it a try. One more option came to my mind; what if I create a 2-D mesh and revolve it by 360 degrees. It will also generate fewer mesh in the central region. The only thing will be generation of wedge shaped elements along the central axis. Secondly the problem may arise when it comes to attach the (tangential) inlet part. My design is for gas cyclone with tangential inlet (figure attached). Any idea how to do it?

[rolloblues]

Also, there is a good tutorial on YouTube about cyclone meshing with blocking technique.

Maybe you already spotted it, if not search for "Structured mesh generation using ICEM CFD"

[lxlylzl]

Originally Posted by rolloblues

Also, there is a good tutorial on YouTube about cyclone meshing with blocking technique.

Maybe you already spotted it, if not search for "Structured mesh generation using ICEM CFD"

Dear rolloblues, I have already meshed cyclone with that technique. I already mentioned that and it created dense mesh in the central region, which I want to avoid.

[rolloblues]

Yep, I knew I was going slightly off topic with my last post... but when I saw your picture of the cyclone, that video tutorial sprung to my mind and I thought that maybe you could have found it interesting more in general

[kad]

My strategy probably won't work for this geometry. I thought you were meshing a simple tube. Maybe "convert block type" works for your case.