Guidance needed regarding the Meshing issue

sidharth9426 · November 7, 2015, 01:34

Dear All,
I am working on the project of under hood thermal management. And I am facing problems after meshing.

The main model before meshing shows single surface overall, but as soon as I am meshing the model the front faces of the model has now been separated into two faces, which in combination are acting as one.

I have attached images.

So I want to know where can I made problem and what is the probable solution for it.

Thanking you,
Sidharth

Schwob77 · November 8, 2015, 15:43

Hi Sidarth,

I assume you are using the wrapper before you remesh the surface, right? If this is true then the behavior is expected. As the final volume mesh will touch both sides of the surface it's not valid for the wrapper as the volume of interest is not manifold. That's why the surfaces are inflated automatically. To avoid it you can declare the surfaces as baffle interface. In this case the wrapper won't inflate it anymore. However in reality there will be a finite thickness of the surface, that's why inflation by the real thickness in surface repair would make sense as well.

sidharth9426 · November 8, 2015, 23:42

Dear Schwob,
Thank you for the guidance.
I have selected 3 models for meshing, 2 for surface meshing(surface wrapper and surface remesher) and 1 for volume meshing (Polyhedral meshing).
So I have no idea that how does star ccm initiate the process i.e whether it does surface wrapping first or surface remeshing first.

Also creating a baffle interface is a good option, but can't there be some other way. I mean if I have complex geometry then it wont be possible for me to create baffle interface everywhere the problem arose.

So can there be other way of tackling this problem.

Thanking you,
Sidharth

Schwob77 · November 9, 2015, 04:53

Hi Sidarth,
in the end my answer is "no", there is no easier way to do that. There are alternatives, but I wouldn't lable them as easier.
First of all, the wrapper is no real surface mesher, this becomes obvious if you go for parts based meshing. There the wrapper is an operation on its own and not part of the Mesh Operation. The purpose of the wrapper is to extract a certain volume of interest out of your geometry and to deliver an error-free, water-tight and non-manifold volume. If you would get rid of all your pierced faces in the surfaces by intersecting them, you still wouldn't get a non-manifold volume due to these surfaces having no thickness. If you grow from their free edge to the inside, you will meet an edge where they are connected with another surface. This edge will be connceted with more than two faces and therefore is labeled as "non-manifold". There are only two ways for the wrapper to overcome that problem. Either the surfaces is inflated or it needs to be labeled as a baffle before by the user to show the wrapper, that you are aware of the particular problem and want to keep the surface uninflated. If you don't do it, you end up with one face that is wetted on both sides and that's not allowed, because inside/outside (=orientation of the face normal) is not clear anymore. As Baffle it has two separate faces connected with an interface. In this case inside/outside is defined well again.

I quickly want to come back to my last comment in my previous answer. In reality this surface has a real thickness. Changing it to zero can significantly change the flow field with separations in reality that you don't detect in your model without thickness. So my best guess would be to inflate the surfaces.

sidharth9426 · November 9, 2015, 12:20

Dear Schwob,
Thank you for such a good explanation. Things are getting clear for me now.

Just wanted to ask that in star ccm is there any thumb rule that a thickness below some "x" value will get inflated automatically.?
because if this information is available before hand then it will be easy to define baffle interface at all places which have value lesser than "x", and thus have a properly meshed surface.

Thanking you,
Sidharth.

Schwob77 · November 9, 2015, 12:56

Hi Siddarth,

to understand it, you shouldn't think in a limit for the thickness, but the fact if a face (=triangle) is part of the so called volume of interest. If this is true, then the wrapper will inflate it. One face always has a zero thickness. Only the part itself can have a non-zero thickness, but that's not what the wrapper sees. It only cares about the two sides of a face. So you need to avoid, that faces are "wetted" from both sides. If this is not possible for certain surfaces, then you need to inflate them manuall in surface repair or your CAD tool upstream, transfer them into a baffle or let the wrapper inflate them.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Guidance needed regarding the convergence	sidharth9426	STAR-CCM+	5	October 7, 2015 10:55
Some guidance is needed about UDF macros	highhopes	Fluent UDF and Scheme Programming	2	July 3, 2012 10:41
[ICEM] Meshing advices needed	muketa	ANSYS Meshing & Geometry	1	July 10, 2009 10:48
Help needed for meshing the following geometry...	eers	Main CFD Forum	11	June 29, 2009 16:29
[GAMBIT] Help needed with meshing!	fluentnoob	ANSYS Meshing & Geometry	24	June 6, 2009 18:41

November 8, 2015, 15:43		#2
Schwob77 Member Join Date: Nov 2015 Posts: 38 Rep Power: 11	Hi Sidarth, I assume you are using the wrapper before you remesh the surface, right? If this is true then the behavior is expected. As the final volume mesh will touch both sides of the surface it's not valid for the wrapper as the volume of interest is not manifold. That's why the surfaces are inflated automatically. To avoid it you can declare the surfaces as baffle interface. In this case the wrapper won't inflate it anymore. However in reality there will be a finite thickness of the surface, that's why inflation by the real thickness in surface repair would make sense as well.

November 8, 2015, 23:42		#3
sidharth9426 New Member Sidharth Join Date: Sep 2015 Posts: 29 Rep Power: 11	Dear Schwob, Thank you for the guidance. I have selected 3 models for meshing, 2 for surface meshing(surface wrapper and surface remesher) and 1 for volume meshing (Polyhedral meshing). So I have no idea that how does star ccm initiate the process i.e whether it does surface wrapping first or surface remeshing first. Also creating a baffle interface is a good option, but can't there be some other way. I mean if I have complex geometry then it wont be possible for me to create baffle interface everywhere the problem arose. So can there be other way of tackling this problem. Thanking you, Sidharth

November 9, 2015, 04:53		#4
Schwob77 Member Join Date: Nov 2015 Posts: 38 Rep Power: 11	Hi Sidarth, in the end my answer is "no", there is no easier way to do that. There are alternatives, but I wouldn't lable them as easier. First of all, the wrapper is no real surface mesher, this becomes obvious if you go for parts based meshing. There the wrapper is an operation on its own and not part of the Mesh Operation. The purpose of the wrapper is to extract a certain volume of interest out of your geometry and to deliver an error-free, water-tight and non-manifold volume. If you would get rid of all your pierced faces in the surfaces by intersecting them, you still wouldn't get a non-manifold volume due to these surfaces having no thickness. If you grow from their free edge to the inside, you will meet an edge where they are connected with another surface. This edge will be connceted with more than two faces and therefore is labeled as "non-manifold". There are only two ways for the wrapper to overcome that problem. Either the surfaces is inflated or it needs to be labeled as a baffle before by the user to show the wrapper, that you are aware of the particular problem and want to keep the surface uninflated. If you don't do it, you end up with one face that is wetted on both sides and that's not allowed, because inside/outside (=orientation of the face normal) is not clear anymore. As Baffle it has two separate faces connected with an interface. In this case inside/outside is defined well again. I quickly want to come back to my last comment in my previous answer. In reality this surface has a real thickness. Changing it to zero can significantly change the flow field with separations in reality that you don't detect in your model without thickness. So my best guess would be to inflate the surfaces.

November 9, 2015, 12:20		#5
sidharth9426 New Member Sidharth Join Date: Sep 2015 Posts: 29 Rep Power: 11	Dear Schwob, Thank you for such a good explanation. Things are getting clear for me now. Just wanted to ask that in star ccm is there any thumb rule that a thickness below some "x" value will get inflated automatically.? because if this information is available before hand then it will be easy to define baffle interface at all places which have value lesser than "x", and thus have a properly meshed surface. Thanking you, Sidharth.

November 9, 2015, 12:56		#6
Schwob77 Member Join Date: Nov 2015 Posts: 38 Rep Power: 11	Hi Siddarth, to understand it, you shouldn't think in a limit for the thickness, but the fact if a face (=triangle) is part of the so called volume of interest. If this is true, then the wrapper will inflate it. One face always has a zero thickness. Only the part itself can have a non-zero thickness, but that's not what the wrapper sees. It only cares about the two sides of a face. So you need to avoid, that faces are "wetted" from both sides. If this is not possible for certain surfaces, then you need to inflate them manuall in surface repair or your CAD tool upstream, transfer them into a baffle or let the wrapper inflate them.