CFD Online Discussion Forums

CFD Online Discussion Forums (
-   ANSYS Meshing & Geometry (
-   -   [ICEM] Auto prism layer growth to unity (

siw August 2, 2010 08:19

[ICEM] Auto prism layer growth to unity

Is there an automatic way for ICEM Tetra/Prism to keep growing a prism layer (in 3D or quad element in 2D) until the prism element aspect ratio is 1? As opposed to having to manually calculating the number of layers from the neighbouring tetra size, growth rate and first layer height, which will vary as tetra sizes vary along a surface for example.


PSYMN August 3, 2010 22:03

Max Height over Base, Prism Height Limit Factor
Yes, look up the help for two Global Prism Parameters...

Max Height over Base

Prism height Limit Factor.

These will each do this. The one adds growing layers until the specified height over base is reached and then changes the ratio to 1 and continues growing layers. The other stops layer growth once a specified height over base is reached.

siw August 23, 2010 05:25

Ok, I've read about those two and I think the Prism Height Limit Factor is the one to be used. Also watched the Boundary Layer Mesh presentation at the ANSYS demoroom and they used this as well.

But I've a question that was not covered there. Before I make the prism mesh I know the requried first layer height and growth ratio. Also, I'd like the final layer to have an aspect ratio of 1. Therefore, I don't know the number of layers required (which depends on were abouts on the model they are). So can I omit specifing the number of layers and have ICEM keep growing prisms until the final layer aspect ratio criteria is met? I've set the number of layers to 0 in the settings menu but ICEM keeps changing that to 3.


PSYMN August 23, 2010 11:33

If you know your average mesh size, initial height and the growth ratio you want, you could do a little math to figure out how many layers you need...

Or you could just set a higher number than you think you need and let ICEM CFD limit the number of layers it by the ratio.

siw August 23, 2010 14:20

Thanks Simon,

The second option sounds like the best to me as I have quite a range of surface element sizes, so I'll give that a try first with 100 layers. These will be the global prism settings:

First layer height = 0.00148
Growth rate = 1.2
Number of layers = 100
Total height = 0

Min prism quality = 1e-7 (to avoid pyramids)
Ortho weigth = 0
Fillet ratio = 0

(noticed these on the boundary layer mesh presentation at the ansys demoroom, but it didn't explain how to do what I want here)

Max prsim angle = 180
Max height over base = leave empty
Prism height over base = 1.
New volume part = PRISM

Leave the surface and volume smoothing to 0 as I'll do this afterwards globally. Max directional smoothing = 10

But with my ICEM luck I doubt it'll work:rolleyes:.

PSYMN August 23, 2010 16:40

Ortho weight set to 0 means you don't care about orthogonality... Probably set that 0.5.

Fillet Rato set to 0 may give trouble with inside corners... For most models, I wouldn't go below 0.1 on that.

Prism height Limit Factor = 1 means that the prism ratio will be reset to 1 when the layer height becomes equal to the average base side length. However, it still continues to add layers of uniform thickness after that... If this is meant to smooth transition to the volume tetras, remember that a tetra with base 1 has a lot less volume than a prism with base 1 and height 1. You may want to try a lower number like 0.7 for the smoothest transition.

It is max height over base that would stop adding layers after this ratio was reached...

I recommend doing the surface and volume smoothing in process... You can't make up for it later with global smoothing. This built in smoothing helps because it smooths between each layer and is pretty important for large numbers of layers (unless the geometry is pretty simple).

If it were me, I would set the number of layers to 7 or something like that and the ratio to 1.2... everything else would be defaults. This would give me a smooth transition to the volume. If the boundary layer thickness looked about right, great, if not, I would try again with a few more or less layers until I got it about right.

The only downside would be a lack of control for initial height. I would then go back and reset that initial height to give me the Y+ value that I wanted. I might even split some layers to give more prisms...

All times are GMT -4. The time now is 14:54.