CFD Online Discussion Forums

CFD Online Discussion Forums (
-   ANSYS Meshing & Geometry (
-   -   Prism mesh (

Mohankumarg12 July 20, 2011 06:43

Prism mesh
2 Attachment(s)
I am doing a defrosting study of windshield, in the process I encountered a problem in creating a prism mesh.

In my geometry I have three material points
1.Fluid zone(inside the cabin)
2.Solid zone(Glass of the windshield)
3.Ice zone above the solid zone.

I created a TET mesh inside my fluid zone and prism on the wall(facing the fluid zone) of the windshield, here my wind shield is not connected edge to edge to the cabin(figure-2)

The problem is I have to create a prism mesh on the windshield and for the ice layer by extruding the surface mesh which is in the bottom of the windshield.

I tried different methods
1.I created a surface mesh on all the walls and I deleted the mesh on the unwanted walls on the windshield and I created volume mesh for the cabin, while creating the prism mesh on the walls the standalone surface mesh was deleted automatically.
So I made a advanced setting in the global parameter for prism, that not to delete the standalone mesh, In this settings my standalone mesh was not deleted but I am getting a worst pyramids on the edges.

How to rectify this problem.

Thanks in advance,

PSYMN July 22, 2011 20:27

I assume you are not meshing anything out side of what you described... You have a fluid zone in the cabin, air will be circulating and transferring heat to the glass which will then pass thru and transfer heat to the ice... The glass and the ice are just solids and are fully connected to the inside of the cabin...

I would start by just meshing the cabin... You don't even need the geometry for the glass or the ice... Just the cabin as a single volume. Mesh it. Set prism settings for the inside of the cabin since your heat transfer will depend on your boundary layer capture...

When you are all done with that fluid region...

Take the shells of the window and extrude them... You probably want 3 to 5 cells thick for the glass. Extrude them in such a way that the number of layers and growth ratio add up to the thickness you wanted for the glass... During the extrude operation, you can pick the name of the volume (pick GLASS) and the name of the top (Pick GLASS_OUTER or something like that...

Then go back and repeat... Extrude "GLASS_OUTER in such a way that it extrudes by the thickness of ICE that you want and the volume is called "ICE", etc.

In the solver, give GLASS and ICE the correct solid properties and you are all set.

Mohankumarg12 July 22, 2011 23:46

prism mesh required area
1 Attachment(s)
Hi Simon,
Thanks for your reply. I understand what you are saying, but my problem is I am unable to create prism mesh all over the windshield area.

Plz see the attached picture for the problem that I am facing

Many thanks,

PSYMN July 23, 2011 13:33

No problem...
OK, no problem. If the glass is not fully connected to the volume, I guess you do need that geometry (just the one surface(s))...

I am assuming that the windshield surface(s) must be split into two main regions. The part shared by the fluid volume and the perimeter outside of that (lets call it OUTER_W).

Start as before and mesh the volume with normal tetra/prism methods. This will also mesh the inner region of the windshield surface, but not the outer edge shown in your last image.

Then go into global parameters for surface meshing => Patch Dependent. Turn on the option about "respect line elements". This will make sure that any patch dependent shell meshing you do on the OUTER_W will attach to the previous mesh along the shared boundary, but you still need to set the sizes for the outside of the OUTER_W. To that with the Mesh => Curve parameters. Set them up to be the size you want for a smooth or zero transition. Then go to Mesh => Compute Mesh => Surface meshing. make sure it is Patch Dependent and selected surfaces... Pick the "OUTER_W" surface(s). Compute.

This should generate a nice patch dependent mesh around the perimeter that uses the outer perimeter curve sizing and the line elements from the tetra prism mesh for the inner perimeter sizing... In other words, a conformal mesh across the windshield...

Then proceed with the extrusion instructions I gave before using both regions...

Note: the previous instructions assumed a uniform window and ice thickness... With a little more effort, you can also handle varying thickness.

Mohankumarg12 July 25, 2011 02:50

Meshing completed
1 Attachment(s)
Hi Simon,
I got the mesh as you suggested. Plz see the attached picture.
I really struggled to make it, but as per your idea I made it within a minute.

Many thanks,

PSYMN July 25, 2011 09:44

I am not sure this is right... I expected to see the perimeter of the volume mesh imprinted on the windshield... I don't see it... Perhaps it is just the image, but I suspect things were not done quite right yet... Is the volume mesh properly attached to this windshield?

Mohankumarg12 July 27, 2011 07:24

Mesh error rectified
1 Attachment(s)
yes Simon, You are correct, my volume elements are not proper and I re meshed the whole geometry. Now I got the correct mesh and volume mesh both. Plz see the attached picture.

Many thanks,

PSYMN July 27, 2011 16:10

Perfect... Good work. :D

All times are GMT -4. The time now is 22:34.