CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   Buoyancy in gas turbine rotor (

snusmumrick May 28, 2012 09:20

Buoyancy in gas turbine rotor
1 Attachment(s)
Good day for everyone!

I am investigate a question of heat transfer and flow through a rotor of axial compressor, like shown on a picture:

Attachment 13403

Red arrow in the left - it is an inlet into the domain.

In the right - outlet.

Arrows on the top - the heat flow from the gas in the compressor.

Black lines- supposed flow direction inside the rotor.

In my opinion I must observe strong buoyancy effect of heated from the disc air, which must transport to the axis, mix with cool air and go back to the "top" in rotating domain.

In help i've found that:

"For steady-state rotating domain simulations involving buoyancy, the gravity vector must be aligned with the axis of rotation."

But i'm confused, because did not understand which number of gravity component must be specified in the appropriate window in bouyancy domain option.

With disabled buoyancy option - there are now this effect off course.

Can anyone help me to solve this problem in CFX?
Thank you.

ghorrocks May 28, 2012 18:41

If this device is rotating then isn't the only sensible thing to do to model the whole thing? I cannot see how modelling a small slice is going to work.

snusmumrick May 29, 2012 01:31

1 Attachment(s)
Sorry. I didn't said, that i'am modelling a sector with periodic interface:

Attachment 13410

Is it enough?

ghorrocks May 29, 2012 06:38

If buoyancy is important then no, it is not enough. If buoyancy is significant then the flow will go from the bottom of the rotor to the top. This flow is not periodic. It might be symmetric - with only a single symmetry plane through the middle of the device. And in this case, guess what? The symmetry plane is parallel to the gravity vector like the original error message said.

snusmumrick May 29, 2012 06:44

But what value of gravity must be specified if i'm talking about gravity of rotating region which acceleration equals: sqr(Angular Velocity)*radius ?

ghorrocks May 29, 2012 06:48

You are not talking about gravity. You are talking about centrifugal forces. If you want to model centrifugal forces then use a rotating frame of reference, and just define a gravity vector pointing anywhere. Point it along the rotation axis to keep the solver happy.

snusmumrick May 29, 2012 07:10

I'll try it. Thank you so much

snusmumrick May 29, 2012 07:23

I'm sorry, but does it means that i must just enable rotating option in DOMAIN MOTION and Buoyancy model in additional coordinate frame?

And what about buoyancy turbulence model?

All times are GMT -4. The time now is 18:42.