[ICEM] hybrid mesh

vmlxb6 · March 5, 2011, 02:21

There was a thread which talks about hybrid mesh (hexa + tri) but I didnt follow it.
I an simulating a 2-D one way FSI for a flow over a cylinder and want a hexa mesh around the cylinder and tri or quad mesh in the rest of the region. A complete hexa mesh would also do. Since my mesh (cylinder) is moving, there is stretching and expansion of the boundary. I guess having a good hexa mesh should work. I created an o-grid mesh but the corner elements got distorted. I have attached 3 pics. Mesh 3 shows where the distorted elements are, mesh 1 shows the mesh around the cylinder.

http://img717.imageshack.us/i/mesh1.png/

http://img195.imageshack.us/i/mesh2q.png/

http://img694.imageshack.us/i/mesh3.png/

I am looking at something like this:

http://img233.imageshack.us/i/requiredmesh.png/

Thank you.

PSYMN · March 6, 2011, 16:45

I guess you are using the hexahedral mesh smoother set to Orthogonality.

We are working on improving this smoother so it can relax its orthogonality in a better way, but right now it is trying to apply orthogonality (make the mesh right angles) to the walls and this causes trouble with the 45 degree angles in the corners.

The solution is to release the orthogonality for the FF walls. You will find a control for this where you setup your smoother settings.

At the same time (and with the same option), I suggest you set the initial cell height on the circle so the ortho smoother will control that better and not cause drifting. You can set a specific height if it is constant, but the non-sensical default of -1 means "keep existing height", and is useful if that initial height varies.

The Ogrid method is much more efficient than the HGrid method and much better able to capture boundary layer effects.

PSYMN · March 6, 2011, 16:49

One other quick thought... You could have a circular far field. Then you would have a perfectly radial model that could be orthogonal everywhere without any pesky corners.

vmlxb6 · March 6, 2011, 16:58

for a circular far field how do i set the BC ??????? Will it just be inlet ???

I made another mesh which is O-grid at the center near the cylinder and H-grid else where. Could you please have a look at it. Can I can email you the msh file.

I am studying a moving cylinder problem (1 way FSI). And I cannot decide as to what the problem is. I presume the problem may be the mesh deformation at the boundaries.

any idea what should the mess stiffness be ???????

vmlxb6 · March 6, 2011, 17:25

Am attaching some pics of the mesh. Can you tell me if its appropriate for a one way moving cylinder problem ???

Thank you very much.

PSYMN · March 9, 2011, 14:44

I would probably have fanned out the Hgrid a bit so the perimeter would be uniform spacing, but that isn't necessary... I think it looks good.

I have no idea what to use for stiffness... What are you using to Morph? Does your morpher allow you to move the entire region around the circle so that the boundary layers would be moved as a whole and the interpolation could happen somewhere else?

March 5, 2011, 02:21	hybrid mesh	#1
vmlxb6 Senior Member Ugly Kid Joe Join Date: Aug 2010 Posts: 193 Rep Power: 16	There was a thread which talks about hybrid mesh (hexa + tri) but I didnt follow it. I an simulating a 2-D one way FSI for a flow over a cylinder and want a hexa mesh around the cylinder and tri or quad mesh in the rest of the region. A complete hexa mesh would also do. Since my mesh (cylinder) is moving, there is stretching and expansion of the boundary. I guess having a good hexa mesh should work. I created an o-grid mesh but the corner elements got distorted. I have attached 3 pics. Mesh 3 shows where the distorted elements are, mesh 1 shows the mesh around the cylinder. http://img717.imageshack.us/i/mesh1.png/ http://img195.imageshack.us/i/mesh2q.png/ http://img694.imageshack.us/i/mesh3.png/ I am looking at something like this: http://img233.imageshack.us/i/requiredmesh.png/ Thank you. Last edited by vmlxb6; March 5, 2011 at 03:40.

March 6, 2011, 16:45	Release Orthogonality	#2
PSYMN Senior Member Simon Pereira Join Date: Mar 2009 Location: Ann Arbor, MI Posts: 2,663 Blog Entries: 1 Rep Power: 47	I guess you are using the hexahedral mesh smoother set to Orthogonality. We are working on improving this smoother so it can relax its orthogonality in a better way, but right now it is trying to apply orthogonality (make the mesh right angles) to the walls and this causes trouble with the 45 degree angles in the corners. The solution is to release the orthogonality for the FF walls. You will find a control for this where you setup your smoother settings. At the same time (and with the same option), I suggest you set the initial cell height on the circle so the ortho smoother will control that better and not cause drifting. You can set a specific height if it is constant, but the non-sensical default of -1 means "keep existing height", and is useful if that initial height varies. The Ogrid method is much more efficient than the HGrid method and much better able to capture boundary layer effects.

March 6, 2011, 16:49	Circular Far Field...	#3
PSYMN Senior Member Simon Pereira Join Date: Mar 2009 Location: Ann Arbor, MI Posts: 2,663 Blog Entries: 1 Rep Power: 47	One other quick thought... You could have a circular far field. Then you would have a perfectly radial model that could be orthogonal everywhere without any pesky corners.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Gambit problems	Althea	FLUENT	22	January 4, 2017 04:19
[ICEM] surface mesh merging problem	everest	ANSYS Meshing & Geometry	44	April 14, 2016 07:41
[ICEM] hole in hybrid mesh	malkovitch	ANSYS Meshing & Geometry	1	June 11, 2010 09:21
Tgrid Hybrid mesh	JSM	FLUENT	2	December 5, 2007 00:03
hybrid mesh	SC Hu	FLUENT	1	December 23, 2000 16:40

March 6, 2011, 16:58		#4
vmlxb6 Senior Member Ugly Kid Joe Join Date: Aug 2010 Posts: 193 Rep Power: 16	for a circular far field how do i set the BC ??????? Will it just be inlet ??? I made another mesh which is O-grid at the center near the cylinder and H-grid else where. Could you please have a look at it. Can I can email you the msh file. I am studying a moving cylinder problem (1 way FSI). And I cannot decide as to what the problem is. I presume the problem may be the mesh deformation at the boundaries. any idea what should the mess stiffness be ???????

March 9, 2011, 14:44		#6
PSYMN Senior Member Simon Pereira Join Date: Mar 2009 Location: Ann Arbor, MI Posts: 2,663 Blog Entries: 1 Rep Power: 47	I would probably have fanned out the Hgrid a bit so the perimeter would be uniform spacing, but that isn't necessary... I think it looks good. I have no idea what to use for stiffness... What are you using to Morph? Does your morpher allow you to move the entire region around the circle so that the boundary layers would be moved as a whole and the interpolation could happen somewhere else?