CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   ANSYS Meshing & Geometry (http://www.cfd-online.com/Forums/ansys-meshing/)
-   -   2D Meshing - No matter how thin my model is, it is still 2 elements thick (http://www.cfd-online.com/Forums/ansys-meshing/74308-2d-meshing-no-matter-how-thin-my-model-still-2-elements-thick.html)

RossFS March 30, 2010 00:44

2D Meshing - No matter how thin my model is, it is still 2 elements thick
 
Context: Ansys Workbench, meshing for use with ANSYS CFX

With the default settings the mesh is all made of hexahedra elements and the model is 1 cell/element thick. As soon as I use inflation (which I need in the boundary layer) the elements above the inflation layers become tetrahedra elements and end up making the model 2 elements thick (as can be seen in an image below). This is definitely not a 1 cell thick model as I'm getting momentum and velocity components being calculated in the z direction during the solution.

I've even had the model 0.01mm thick and the mesh still ends up being 2 cells thick (like in the image below).

The cells above my inflation layers can be either tetrahedra or hexahedra in shape, but I have to have the model run as a 2D simulation.

Side on:

http://img710.imageshack.us/img710/3777/3003side.gif

Iso view - red indicates location of close up that follows this
http://img715.imageshack.us/img715/1636/3003iso.gif

Close up:
http://img18.imageshack.us/img18/4506/3003close.gif

PSYMN March 30, 2010 13:16

Method?
 
What method are you using? You should be using something like the sweep method...

Note, if you do use the sweep method, the inflation setup is a little odd. Start with creating the sweep method, the "inflate this method" rather than creating a separate inflation. Also, when inflating a sweep, you pick the edges on the sweep face rather than the side surfaces.

Simon

RossFS March 31, 2010 21:21

Thanks Simon.

Just to clarify the steps suggested by Simon for anyone looking at this thread from a search:

*bring up the mesh editing program/section in Workbench
*right click on mesh in the tree on the LHS of screen and click on: insert-> method
*select sweep from the 'Method' list that is in amongst the settings on the screen.
*follow the instructions for 'Inflation Controls with Sweeper' in the "Inflation Controls" help article [for meshing]

RossFS March 31, 2010 22:03

Any idea why I can't set the first layer thickness (when setting up the inflation) to be any lower than 4e-5m in height?
The mesh won't get created/an error occurs if I go any smaller than this.

Other settings for inflation are: 100 layers and a growth ratio of 1.032.

The smallest first layer thickness in a journal article I'm comparing against is something like 2.411e-7m high, with growth ratio of 1.032. Is that sort of size smaller than what ANSYS CFX can handle?

PSYMN April 1, 2010 11:57

RossFS,

This problem with the limited prism aspect ratio is fixed for 13.0... (Due out later this year)

The problem was only for the swept prisms (regular Tetra Prism did not have this limitation). Basically, there was a shape check limiting the aspect ratio for the quads used to sweep the inflation layer. This has been removed.

Sorry for the inconvenience.

RossFS April 12, 2010 03:11

So if I want to have a 2D mesh and want to inflate the boundary layer, do I have any other options than the Sweep method?

I ask as I'm wondering whether the size limitation mentioned above is the limit available on the sweep mesh method or whether it is the limit of the software. Or would I have to work in ICEM CFD to refine the mesh anymore near the bottom wall?

It is hard to tell from the help files on the different mesh methods what each of them do given my lack of understanding of CFD.

PSYMN April 14, 2010 00:13

No, sweep is for a 3D mesh (in this case, 2.5D). If you want a 2D mesh and you have a 2D model, that should work without sweep.

IF you did mean a 3D mesh, there are other options such as MultiZone and ThinSweep...


All times are GMT -4. The time now is 11:43.