CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   ANSYS (http://www.cfd-online.com/Forums/ansys/)
-   -   compressor shroud clearance (http://www.cfd-online.com/Forums/ansys/88482-compressor-shroud-clearance.html)

Selina_stj May 18, 2011 10:08

compressor shroud clearance
 
4 Attachment(s)
Hi everyone, for last few months, I use CFX 12 to simulate the compressor fluid without acitivating the shroud clearance by setting wall boundaries to the interface at the tip of blades. But to simulate the real flow, now I have to change the wall boundary of the shroud tip of blades to the interface with conservative flux.
Now I solve the new solver file with the results from the one with the wall boundary of shroud tip as initial results. The problem is the file refuse to run and the error information is as follows:
+--------------------------------------------------------------------+
| An error has occurred in cfx5solve: |
| |
| Indirect start method returned non-zero exit code |
+--------------------------------------------------------------------+


+--------------------------------------------------------------------+
| An error has occurred in cfx5solve: |
| |
| Indirect solver job failed |
+--------------------------------------------------------------------+
Anybody could give me some suggestions or instructions?:p

PS: the pictures of the meshes and the two kind of boundaries setting are attached

Far May 18, 2011 20:01

post pics of mesh along with boundary conditions

Selina_stj May 19, 2011 10:07

Hi Far, thank you for your reply. The pictures is added to my questions

Far May 19, 2011 13:14

why you want to create the interface between the tip and casing? is there any special requirement to do so i.e. are you going to run the casing treatment case?

Selina_stj May 19, 2011 13:23

Hi Far, the interface is generated automatically in TurboGrid.
And it's helpful for me because my meshes is too big. If I run the simulation immediately with the whole mesh. It's impossible to converge(actually, I don't know why). Then I try to close the shroud tip first by defining the wall boundaries to the GGI interfaces generated by the TurboGrid. This definition can stop the shroud tip fluid flow to other passages.
And then after this simulation converged I change the wall boundaries to interface boundary(with conservative flux). This interface boundary allow the shroud tip fluid flow to other passages which is the requirement for the real flow. And then I run the new solver file with the initial results from the simulation of the one with wall boundary to the shroud tip flow.
I hope this explonation is clear that you can see the problem.

Far May 19, 2011 13:50

this interface is perpendicular to blade tip surface?

Selina_stj May 19, 2011 14:04

yes, you are right

Far May 20, 2011 05:11

instead of using the last run as initial guess and changing the wall to interface, make the new file in CFX pre with all conditions you want to simulate.

It will surely work

Selina_stj May 20, 2011 05:18

Quote:

Originally Posted by Far (Post 308531)
instead of using the last run as initial guess and changing the wall to interface, make the new file in CFX pre with all conditions you want to simulate.

It will surely work

Hi Far, Thank you for your suggestions. I tried before as you suggested . But it could not work because my meshes is too big and too complacated. Why I set the shroud tip boundary to wall just because I want to get a first convergent result and then use is as an initial result. I'm so conused with the simulation now.

Far May 20, 2011 13:06

Quote:

because my meshes is too big .
How many no of nodes are there ?


Quote:

If I run the simulation immediately with the whole mesh..
We always with the whole mesh, no matter how big it is. If CFX is not giving any while loading it, which clearly shows that the mesh is not beyond the capability of your computer.
Can you please past the pic with the clear indiction of blade, splitter, inlet, outlet and periodic boundaries. How many chords your in and out boundaries are placed with respect to blade?

Quote:

It's impossible to converge(actually, I don't know why)
Can you put your convergence plots here please? Are you taking autotime scale, physcial time or local. By default it takes the autotime scale with 1/0.2 omega. please see the user guide for detailes.


Quote:

And then after this simulation converged I change the wall boundaries to interface boundary(with conservative flux). )
Well this is not the standard procedure.


All times are GMT -4. The time now is 12:30.