CFD Online Discussion Forums

CFD Online Discussion Forums (
-   ANSYS Meshing & Geometry (
-   -   [ICEM] ICEM prism layer results in Fluent NULL pointer cells (

Will Anderson February 16, 2011 13:09

ICEM prism layer results in Fluent NULL pointer cells
Good Evening.

I have a solid hollow sphere (part name=SPHERE), around which a 10 layer prism mesh is wrapped (prism volume part name=BL_SPHERE, top part name=TOP_SPHERE).

Around the sphere's boundary layer (or, surrounding the top of its prism mesh) is a normal tetrahedral mesh (part name=FLUID). Now, when I read this mesh into fluent, I received the following error:
Back in ICEM, if I only show the part TOP_SPHERE, containing the top of the prism mesh, I notice that it has holes in it, which are likely to blame for the NULL pointer error.

Any idea how i can repair this or have ICEM make a better prism mesh?

Here is the image of TOP_SPHERE
i guess i could have cropped that image....

PSYMN February 17, 2011 13:22

No idea why you got an incomplete top layer, but you can fix it with the "uncovered faces" check. (Edit Mesh => Check Mesh)

But more to the root of your problem... Should the tetras and Prisms be in the same fluid? You shouldn't need a top part or prism part at all. By default it would have put the prism in the same part as the tetras and you wouldn't have needed those shells between them at all. You can edit your way to this by right clicking on TOP_SPHERE in the parts branch and deleting it. Then right click on the FLUID part and Add to Part, then Select the second to last icon in the selection tool bar (All Volume elements).



Will Anderson February 17, 2011 15:29

I've tried to repair using the "Edit mesh -> uncovered faces check" but I've so far had no luck. I'm learning that the prism mesher (in conjunction with surrounding tetras) occasionally requires more computer than my laptop can handle, so maybe its some sort of incomplete mesh due to lack of computational resources. I will try with more horsepower and RAM and report back.

Why do I need distinct "TOP_SPHERE" and "BL_SPHERE" parts? Because these regions will be treated as rigid bodies (albeit "rigid fluid entities") by fluent in a dynamic mesh computation. The prism mesh surrounding the sphere shall retain its good boundary layer resolution, regardless of what fluent's dynamic remesher may do with the tetras throughout the domain.

Best Regards.

PSYMN February 17, 2011 17:30

Uncovered faces should always find volume elements in different parts adjacent to each other... And prism should not write out something incomplete, even if it ran out of memory.

Maybe run a cut plane thru that area where the prism top is missing and see what is going on.

Will Anderson February 19, 2011 12:44

Thanks Simon. I took a look at some cut planes which you recommended.

Maybe this is helpful for people who encounter the rare combination of elements leading to those 4 holes in the prism layer top.

I had a mesh density with 4 points positioned unfortunately close to the location of the top of the prism layer. The holes which I had in the top_sphere shell surrounded the locations of those points on which the mesh density was based. These 4 points interfered with the creation of the prism layer. The density points are now moved away from the sphere and now all is fine. The prism layer top has no more holes in it.

Thanks and Regards.

PSYMN February 19, 2011 13:17

Delete those points...
I usually delete those points after creating the density box... Otherwise nodes are forced to that location and locked. Quality suffers even if the prism layer is unaffected.

All times are GMT -4. The time now is 02:34.