[ICEM] 2D unstructured meshing: Prism problem

Dronzer · August 25, 2018, 09:17

Hi all,

I am quite novice in ICEM CFD. Trying to mesh a simple rectangle (flow through pipe) using unstructured meshing tools. I have,
1. created points, lines, a surface and a body called FLUID.
2. defined global maximum & minimum, patch dependent method and number of points on each side.
3. Prism mesh properties on the walls (INLET and OUTLET excluded).
4. Completed shell meshing

After these steps, I am getting some distorted cells (attachments)

. How can fix it? Please let me know.

Thanks in advance.

Dronzer · August 27, 2018, 23:47

Hi guys,

I believe it is a relatively simple problem!
Please can someone look in to it?

Gert-Jan · August 28, 2018, 02:03

I never use shell meshing, so don't know what the problem is. But could it be that the points on the corners are part of inlet and outlet instead of the walls? And what about the curves?

Btw,
- what are your trying to do? Do you want a pure 2D mesh (shells)? For fluent? Or for CFX, or something else?
- a pipe is round, and this is rectangular. Aren't you trying to model a duct? Else you should make a triangular geometry (pie).
- Have you tried a different strategy with Extrude mesh "Edit mesh>Extrude mesh"?

kad · August 28, 2018, 06:44

While setting up your mesh sizes, you have defined mesh sizes (or nodes) on the inlet and outlet curves, too. You can display them by right mouse button on curves in display tree and activate "show nodes" (or similar, haven't look it up). The patch dependent methods snap the mesh to those nodes. As you have a quite large spacing near the wall on inlet and outlet, you get a distorted mesh there.

You can try the following solutions. First, simply do not use any mesh sizes on inlet and outlet curves. Second, you can try to carefully adjust the spacing on inlet and outlet to your prism spacing by using advanced bunching laws (bigeometric). A look in the manual should help here.

Some more general hints: No need to define a body for 2d meshing. One can achieve very nice "prism like behaviour" by playing around with "curve mesh setup" (height, height ratio, numbers of layers) and using all quad patch dependent mesh. This gives you prism behaviour near walls and a nice unstructured quad mesh for interior. Of course, you might also have a look a blocking in ICEM for this kind of geometry. Should be no big deal.

August 27, 2018, 23:47	Please give me some suggestions!	#2
Dronzer Member Join Date: Apr 2016 Posts: 53 Rep Power: 10	Hi guys, I believe it is a relatively simple problem! Please can someone look in to it?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[ICEM] Prism mesh generation problem	bharath	ANSYS Meshing & Geometry	2	June 3, 2015 10:21
[ICEM] Meshing Problem ?	Leifheit	ANSYS Meshing & Geometry	3	December 4, 2014 18:09
Prism problem in ICEM CFD at inlet and outlet	benjaminhogan	ANSYS	2	November 15, 2014 11:17
[ICEM] Creating a prism layer causes problem..	earth07	ANSYS Meshing & Geometry	1	June 3, 2013 10:28
GAMBIT meshing problem	Gauthier Lambert	Main CFD Forum	1	August 3, 2000 09:22

August 28, 2018, 02:03		#3
Gert-Jan Senior Member Gert-Jan Join Date: Oct 2012 Location: Europe Posts: 1,827 Rep Power: 27	I never use shell meshing, so don't know what the problem is. But could it be that the points on the corners are part of inlet and outlet instead of the walls? And what about the curves? Btw, - what are your trying to do? Do you want a pure 2D mesh (shells)? For fluent? Or for CFX, or something else? - a pipe is round, and this is rectangular. Aren't you trying to model a duct? Else you should make a triangular geometry (pie). - Have you tried a different strategy with Extrude mesh "Edit mesh>Extrude mesh"?

August 28, 2018, 06:44		#4
kad Senior Member Join Date: Feb 2013 Location: Germany Posts: 200 Rep Power: 24	While setting up your mesh sizes, you have defined mesh sizes (or nodes) on the inlet and outlet curves, too. You can display them by right mouse button on curves in display tree and activate "show nodes" (or similar, haven't look it up). The patch dependent methods snap the mesh to those nodes. As you have a quite large spacing near the wall on inlet and outlet, you get a distorted mesh there. You can try the following solutions. First, simply do not use any mesh sizes on inlet and outlet curves. Second, you can try to carefully adjust the spacing on inlet and outlet to your prism spacing by using advanced bunching laws (bigeometric). A look in the manual should help here. Some more general hints: No need to define a body for 2d meshing. One can achieve very nice "prism like behaviour" by playing around with "curve mesh setup" (height, height ratio, numbers of layers) and using all quad patch dependent mesh. This gives you prism behaviour near walls and a nice unstructured quad mesh for interior. Of course, you might also have a look a blocking in ICEM for this kind of geometry. Should be no big deal.