CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Why Non-overlap interface occur in 1 to 1 mesh? (http://www.cfd-online.com/Forums/cfx/23315-why-non-overlap-interface-occur-1-1-mesh.html)

kake November 19, 2006 01:39

Why Non-overlap interface occur in 1 to 1 mesh?
 
As you know, in cfx, we only can specify GGI interface type to a 1 to 1 mesh when modelling rotor stator question. But one question comes out, the non overlap area fraction will occur in the physically matching 1 to 1 interface.That is to say, the non overlap will be regarded as wall defaultly. So this is very surperising. Any one know how CFX determine this ? According what? Which range we can accept? Can we cancle this determination instead of specifying 1 to 1 physically?

Robin November 20, 2006 12:51

Re: Why Non-overlap interface occur in 1 to 1 mesh
 
GGI is always used to handle frame changes, regarless of whether the mesh is 1:1 or not. This is nothing to be concerned with. The non-overlap conditions in the boundary condition specification is only used if there are non-overlap regions. If the mesh is truly 1:1, there will be no non-overlap regions.

Read the doc for details.

Regards, Robin

kake November 22, 2006 23:24

Re: Why Non-overlap interface occur in 1 to 1 mesh
 
Thanks,Robin

In fact,my mesh is truly 1:1,because the mesh is made in icemcfd with a whole block.So this can guartee the mesh node matched.But unfortunally,CFX still report the non-overlap interface region,and which occur a large fraction of the interface. So i am very surprising. My question is: According which criteria does cfx determine this? Which range can we accept?

when i check my results, the nov overlap region has been regarded as the wall.

Robin November 23, 2006 11:26

Re: Why Non-overlap interface occur in 1 to 1 mesh
 
Hi Kake,

Is the solver reporting non-overlap?

If it is a Frozen Rotor interface, the interface must be a surface of revolution about your axis of rotation. If it is not, it is possible to get this error. Check that it is and fix it as necessary, then run again.

Regards, Robin

kake November 24, 2006 05:58

Re: Why Non-overlap interface occur in 1 to 1 mesh
 
My interface type is Stage. I think it is not related to the interface type. So i'd like to know how cfx determine non overlap between two interfaces? If you want to prove my question? you also can check the tutorial case? which have the same question when you run it. it will appear in the solver *out file at the very begining.

Robin November 24, 2006 10:12

Re: Why Non-overlap interface occur in 1 to 1 mesh
 
Kake,

It can still be related to your interface type. If it is 1:1 and the edges of the regions match up, a regular GGI will report 0% overlap. That may not be the case with Stage or Frozen Rotor, so I'll ask you again, is your interface a surface of revolution? Are you domain's periodic?

Regards, Robin

kake November 26, 2006 22:03

Re: Why Non-overlap interface occur in 1 to 1 mesh
 
Thanks Robin,

in fact,i agree with your opinion.in my mesh, there are two interfaces of revolution surfaces, and the simulation domain is periodic.For the requriment of yplus,the mesh near the wall is very samll,which maybe smaller than the the geometric tollerance. So i suppose this maybe the reason.

Is it possible to override this determination of non overlap interface area in some expert mode?

Robin November 27, 2006 10:46

Re: Why Non-overlap interface occur in 1 to 1 mesh
 
If it is a frozen rotor or stage interface, you can manually specify the pitch ratio or pitch angles. Due to the finite nature of the mesh, it is sometimes possible to get some non-overlap due to discrete errors.

-Robin

HekLeR December 14, 2006 01:24

Re: Why Non-overlap interface occur in 1 to 1 mesh
 
Are you 100% absolutely certain, without any doubt, that the hub curve (lowest axial or radial position) of your interface mesh forms a perfect circle of revolution around the axis.

Are you 100% absolutely certain, without any doubt, that the hub & shroud curves on both sides of your interface mesh exactly line up, both forming the same circles of revolution around the axis.

If not, then you can get non-zero non-overlap area because the solver may not correctly figure out the pitch angles.

In this case, just set the pitch angles.


All times are GMT -4. The time now is 23:58.