CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Can anyone help me of using K-Epsilon (http://www.cfd-online.com/Forums/cfx/107678-can-anyone-help-me-using-k-epsilon.html)

 hopehope October 3, 2012 23:20

Can anyone help me of using K-Epsilon

Hi,
When I use the K-Epsilon in CFX, the result is strange. My first node is 2e-7 m from wall. After exporting the result, I find that V(y=0)=0, V(y=2e-7)=5, it's almost half of the highest one. The velocity is growing too fast.
Can anybody tell me that how can I fix that?
Thank you

Hope

 hopehope October 3, 2012 23:33

Basically my problem is the hybrid and conservative are same expect the wall velocity. And the velocity grows too fast.
Could anyone help me and tell me how to fix that?
Thank you!

Hope

 ghorrocks October 4, 2012 07:40

The parameter you want to look at is y+, the non-dimensional distance from the wall. The distance of the first node does not mean anything by itself.

And why do you say the velocity grows too fast? Do you have analytical or experimental results to show it should be slower?

 hopehope October 4, 2012 08:15

1 Attachment(s)
Quote:
 Originally Posted by ghorrocks (Post 384891) The parameter you want to look at is y+, the non-dimensional distance from the wall. The distance of the first node does not mean anything by itself. And why do you say the velocity grows too fast? Do you have analytical or experimental results to show it should be slower?
Thank you, Horrocks.

What's y+?
my width is about 0.1m, and the first node is at 2e-7 m.
Yes, I have experimental results.
my case is a wind tunnel, velocity inlet, pressure=0 at outlet, and non-slip smooth wall.
Look my attachment. The green one the experimental results, and red one is my result. at the very beginning it grows too fast. but it's better on the second node. And I don't know why.

 evcelica October 4, 2012 11:50

Like Glenn said, plot your "y plus" (not solver y plus) for the wall. I'm guessing it is below 11 and k-epsilon is using wall functions to approximate your wall improperly since the y plus is too small.
I'd recommend switching to a different turbulence model that will fully integrate the walls, like SST. Try to get your "y plus" values below 1 if you are trying to compare the boundary layers near the wall, you only use wall functions when you don't care much about the boundary layer.

This is pretty basic, CFD 101;
Your homework: Read Chapter 4 of the CFX solver modeling guide "Turbulence and Near-Wall Modeling":)

 All times are GMT -4. The time now is 09:08.