[ICEM] 2d axisymmetric mesh

bharath · October 26, 2010, 12:32

Hi to all,
anybody help me to mesh this kind of geometry with good quality in icem cfd.this mesh am going to simulate in CFX for axisymmetric problem.

thanks in advance

BrolY · October 27, 2010, 03:14

Could you post some pictures of your issue? We can help you to solve your issues, not to create your mesh ! This is your job

You should also give us more information on what you want to do, so we could give you some advices.

bharath · October 27, 2010, 08:28

i attached a image for your understanding.this is done by regular blocking method.
sorry for, am not able bring you the exact image because it is in my office PC.
I hope you understand this.

dst · October 27, 2010, 08:43

One way is to use in the sharp corner either Y-block or Swept-Block. I would try both of them and decide depending on required mesh quality.

bharath · October 27, 2010, 08:49

hi dst,
could you explain me with some simple sketch about Y-block or Swept block in this geometry

michael_owen · October 27, 2010, 11:36

bharath,

You should "nip the tip" out of your axi-symmetric geometry. In other words, the wedge should not go all the way to the axis of symmetry, but to a small but finite radius. Make this inner boundary condition a symmetry plane or free-slip, adiabatic wall. This will prevent poorly shaped elements from being formed.

michael_owen · October 27, 2010, 12:06

PS. That advice is specifically relevent if you are modeling in CFX. Also, you probably want to make sure to run such models in double precision, as you will likely end up with high aspect ratio hex/prism elements.

PSYMN · October 27, 2010, 13:26

You could just use a wedge block (collapse the end of the block). Then you would have tri elements in the tip...

bharath · October 28, 2010, 12:19

whether collapse block gives good quality elements or i have to do some other things PSYMN ?

PSYMN · October 28, 2010, 13:53

It will give triangles (wedges) in that tight 5 degree corner. They may be fine and have high aspect ratios (depending on how many triangles there are), but the quality will always be better than with quads (hexas) shoved in there.

Simon

michael_owen · November 1, 2010, 09:13

PSYMN,

I would have to disagree. We have seen many examples of poor results in thin-wedge axisymmetric models in CFX meshed all the way to the axis of symmetry, using either quad or tet meshes. In particular, we tend to see spurious high axial velocities. These problems vanish when the axis of symmetry is nipped out at an arbitrarily small radius in the way I described. Furthermore, doing this always allows you to create high quality hex meshes. Because of the resulting aspect ratios, I always advise running in double precision for these sorts of axisymmetric models.

I will check our tech support logs to see if I can find an image to post illustrating the axial velocity problem.

michael_owen · November 1, 2010, 09:23

Can't find an image, but here is a typical support email we received illustrating this issue:

"Hi Mike

We have been continuing our evaluation of the non-flow problem with CFX.

One thing we identified is that the model representation of the XXXXX as a thin "slice" may be problematic.

The solution using the "slice" generates high velocities in the XXXXX gas at the vertex of the slice, but nowhere else.

To investigate this, I created a model that does not have the vertex. I did this by creating a model that represents half of the XXXXX (a "half model" based on a 180 degree "slice").

The solution using the half model does not include the high velocities at the center of the XXXXX gas."

I advised the "tip nipping" technique and the problem vanished.

PSYMN · November 2, 2010, 12:53

Bharath,

I am just approaching this from a meshing perspective and assuming that you had to go all the way to the axis... (I recommend a wedge instead of hexas in the tight corner because it will be higher quality).

Michael Owen seems to have more CFX solver specific knowledge about how best to deal with this situation ... His solution of simply not modeling the element right at the axis of symmetry will also produce a better mesh than trying to cram hexa corners in there...

Simon

dst · November 4, 2010, 08:45

michael_owen , PSYMN,

what are the drawbacks of the swept block approach for this kind of problems?

Concerning the mentioned problem with wedge-elements in CFX I should mention: people worked with CFX (ver.10) doing some simulations of axi-symmetric bodies got problems with velocities and pressures (non physical) in that region.

In comparison to that, the ''Star-CD (ver. 3.xx)'' team had no problems with the same mesh in that region.

Maybe, problems depend on the flow properties also and would not occur under the different conditions.

October 28, 2010, 13:53	Its all relative...	#10
PSYMN Senior Member Simon Pereira Join Date: Mar 2009 Location: Ann Arbor, MI Posts: 2,663 Blog Entries: 1 Rep Power: 47	It will give triangles (wedges) in that tight 5 degree corner. They may be fine and have high aspect ratios (depending on how many triangles there are), but the quality will always be better than with quads (hexas) shoved in there. Simon

November 2, 2010, 12:53	Michael knows what he is talking about...	#13
PSYMN Senior Member Simon Pereira Join Date: Mar 2009 Location: Ann Arbor, MI Posts: 2,663 Blog Entries: 1 Rep Power: 47	Bharath, I am just approaching this from a meshing perspective and assuming that you had to go all the way to the axis... (I recommend a wedge instead of hexas in the tight corner because it will be higher quality). Michael Owen seems to have more CFX solver specific knowledge about how best to deal with this situation ... His solution of simply not modeling the element right at the axis of symmetry will also produce a better mesh than trying to cram hexa corners in there... Simon

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[Other] vtk mesh or Abaqus mesh to OpenFOAM	bigphil	OpenFOAM Meshing & Mesh Conversion	27	November 23, 2015 17:31
[ICEM] Negative volume error in hybrid mesh	siw	ANSYS Meshing & Geometry	4	September 3, 2014 05:25
3D Hybrid Mesh Errors	DarrenC	ANSYS Meshing & Geometry	11	August 5, 2013 06:42
mesh missing after import in fluent	morteza08	FLUENT	0	July 23, 2010 02:22
Transform mesh in axisymmetric	safa mahjoub	FLUENT	8	December 17, 2007 06:24

October 27, 2010, 03:14		#2
BrolY Senior Member AB Join Date: Sep 2009 Location: France Posts: 323 Rep Power: 21	Could you post some pictures of your issue? We can help you to solve your issues, not to create your mesh ! This is your job You should also give us more information on what you want to do, so we could give you some advices.

October 27, 2010, 08:43		#4
dst New Member Join Date: Oct 2010 Location: Germany Posts: 25 Rep Power: 15	One way is to use in the sharp corner either Y-block or Swept-Block. I would try both of them and decide depending on required mesh quality.

October 27, 2010, 08:49		#5
bharath Senior Member Bharath kumar Join Date: Apr 2009 Posts: 169 Rep Power: 17	hi dst, could you explain me with some simple sketch about Y-block or Swept block in this geometry

October 27, 2010, 11:36		#6
michael_owen Senior Member Michael P. Owen Join Date: Mar 2009 Posts: 196 Rep Power: 17	bharath, You should "nip the tip" out of your axi-symmetric geometry. In other words, the wedge should not go all the way to the axis of symmetry, but to a small but finite radius. Make this inner boundary condition a symmetry plane or free-slip, adiabatic wall. This will prevent poorly shaped elements from being formed.

October 27, 2010, 12:06		#7
michael_owen Senior Member Michael P. Owen Join Date: Mar 2009 Posts: 196 Rep Power: 17	PS. That advice is specifically relevent if you are modeling in CFX. Also, you probably want to make sure to run such models in double precision, as you will likely end up with high aspect ratio hex/prism elements.

October 27, 2010, 13:26		#8
PSYMN Senior Member Simon Pereira Join Date: Mar 2009 Location: Ann Arbor, MI Posts: 2,663 Blog Entries: 1 Rep Power: 47	You could just use a wedge block (collapse the end of the block). Then you would have tri elements in the tip...

October 28, 2010, 12:19		#9
bharath Senior Member Bharath kumar Join Date: Apr 2009 Posts: 169 Rep Power: 17	whether collapse block gives good quality elements or i have to do some other things PSYMN ?

November 1, 2010, 09:13		#11
michael_owen Senior Member Michael P. Owen Join Date: Mar 2009 Posts: 196 Rep Power: 17	PSYMN, I would have to disagree. We have seen many examples of poor results in thin-wedge axisymmetric models in CFX meshed all the way to the axis of symmetry, using either quad or tet meshes. In particular, we tend to see spurious high axial velocities. These problems vanish when the axis of symmetry is nipped out at an arbitrarily small radius in the way I described. Furthermore, doing this always allows you to create high quality hex meshes. Because of the resulting aspect ratios, I always advise running in double precision for these sorts of axisymmetric models. I will check our tech support logs to see if I can find an image to post illustrating the axial velocity problem.

November 1, 2010, 09:23		#12
michael_owen Senior Member Michael P. Owen Join Date: Mar 2009 Posts: 196 Rep Power: 17	Can't find an image, but here is a typical support email we received illustrating this issue: "Hi Mike We have been continuing our evaluation of the non-flow problem with CFX. One thing we identified is that the model representation of the XXXXX as a thin "slice" may be problematic. The solution using the "slice" generates high velocities in the XXXXX gas at the vertex of the slice, but nowhere else. To investigate this, I created a model that does not have the vertex. I did this by creating a model that represents half of the XXXXX (a "half model" based on a 180 degree "slice"). The solution using the half model does not include the high velocities at the center of the XXXXX gas." I advised the "tip nipping" technique and the problem vanished.

November 4, 2010, 08:45		#14
dst New Member Join Date: Oct 2010 Location: Germany Posts: 25 Rep Power: 15	michael_owen , PSYMN, what are the drawbacks of the swept block approach for this kind of problems? Concerning the mentioned problem with wedge-elements in CFX I should mention: people worked with CFX (ver.10) doing some simulations of axi-symmetric bodies got problems with velocities and pressures (non physical) in that region. In comparison to that, the ''Star-CD (ver. 3.xx)'' team had no problems with the same mesh in that region. Maybe, problems depend on the flow properties also and would not occur under the different conditions.