CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   2d axisymmetry (http://www.cfd-online.com/Forums/cfx/23514-2d-axisymmetry.html)

Cote January 11, 2007 05:08

2d axisymmetry
 
Hi, Im beginner to CFX and I need some help regarding how to handle 2D axisymmetry problem in CFX (ANSYS CFX 10.0)

The problem is quite simple: a fluid in a cilyndrical boundary is heated at the top with a heat flux of cylindrical symmetry

Following the ansys cfx tutorial i have created a XZ rectangle and extruded with a small rotation angle around Z(1 degree) and mesh it with 1 element depth. "Symmetry boundary" condition is applied to the high and low theta planes (no swirl is forseen) . The heat flux is applied according to its xy distribution.

However im a little bit confused on how CFX solves this kind of axisymmetry (other codes preview special elements and procedures for axisymmetric problems wich a i have not found here): 1) Is there some restriction on the choice of rotation angle? shuld it be X? Z? .... 2) Should cylindrical coordinates be used? 3) What kind of boundary conditions should be applied in the center axis? 4) Finally, and to summarize, is axisymmetry correctly handled in CFX?

Thanks a lot...

Fusion January 11, 2007 09:50

Re: 2d axisymmetry
 
First of all, CFX don't account for special procedure to solve axisymmetric problems.

You cannot model the axis, you should provide a little empty space between the axis and the first layer of cells (no "corner" elements).

Finally, I found some problems depending on the ratio between the first elements layer size and the biggest one.

Good luck

Joe January 11, 2007 10:04

Re: 2d axisymmetry
 
1) Is there some restriction on the choice of rotation angle? No but keep it small.

shuld it be X? Z? .... No

2) Should cylindrical coordinates be used? No

3) What kind of boundary conditions should be applied in the center axis? None

4) Finally, and to summarize, is axisymmetry correctly handled in CFX? Yes.

Robin January 11, 2007 10:36

Re: 2d axisymmetry
 
CFX always solves the full 3D cartesian equations. You acheive axisymmetry by virtue of your setup. It doesn't matter whether your axis of symmetry is aligned with the X, Y, or Z axis, it can be completely arbitrary, although you are unlikely to create a model in this manner. All that matters is that your geometry is axisymmetric.

About the only restriction is that your symmetry planes must be planar. This restriction does not exist if you use periodic boundary conditions.

If your wedge goes right down to the axis of rotation, there is nothing special you need to do since the symmetry planes will come together at this point. What I would avoid is refining the mesh towards the axis. Refinement like this is not really necessary, as there is really no physical significace to the axis.

If you are using CFX-Mesh, there is no need to worry about what angle to revolve your mesh. Under Mesh>Options set your Meshing Strategy to "Extruded 2D Mesh" and the 2D Extrusion Option to "Partial". The Partial option will extrude or revolve the mesh between your "Extruded Periodic Pair" by one element and automatically set the angle or thickness to an appropriate value.

Regards, Robin

Paul January 12, 2007 07:20

Re: 2d axisymmetry (To Robin)
 
Dear Robin,

You said "If you are using CFX-Mesh, there is no need to worry about what angle to revolve your mesh". While using ICEM CFD, whats the maximum angle beyond which it is not suggested to revolve to obtain a good solution in CFX? I remember some line in the manual which says that the angle of revolution should be kept below 5 degrees for axisymmetric case. What do you say?

Regards,

Julia


Robin January 12, 2007 10:31

Re: 2d axisymmetry (To Robin)
 
I would suggest between 2.5 and 5 degrees.

Manu January 12, 2007 11:08

Re: 3d axisymmetry (To Robin)
 
Wht will happen if you make a 3D axisymmetric simulation by revolving a 2D mesh by 360 degree and many no of layers.Would it be a replacement of o grid??In this case you will ge a point at centre and in o grid a square.

Robin January 12, 2007 11:41

Re: 3d axisymmetry (To Robin)
 
That's fine too. The discussion here was regarding 2D cases, for which you need a single layer. If you have a 3D axisymmetric case you can create a grid by revolving it.

That said, you can only revolve a mesh in CFX-Mesh if it has two periodic sides. If you need a full 360, I would suggest creating a 90 degree wedge to generate the mesh, then copy and rotate the wedge in Pre to get the full 360.

Regards, Robin


All times are GMT -4. The time now is 00:50.