CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   Using stage interface (

ijk October 20, 2010 10:44

Using stage interface
Dear all,

My question is regarding the use of a stage interface.

I am modelling a fan blowing air through a radiator.

Is it valid to model one fan passage out of seven and the entire radiator and to use a stage interface between the two? The radiator is square. In other words do I need to model the full fan for this case?

I am considering using the downstream velocity constraint "constant total pressure" so that the downstream velcoity can be adjusted due to the impact of the radiator.



Attesz October 21, 2010 07:50

With the stage option the fluxes will be averaged at the interfaces. The problem in your case is the high pitch ratio. You will get an 51deg periodic domain in a rotating frame of reference with the periodic side 1 and the radiator's stage will have 360 deg, so the pitch ratio will be about 7 which is high, and you can get very inaccureate results easily. I think, you should use the radiator as a periodic model, or if you can't do so, use the entire model. On the boundaries, mass flow or pressure values are better, except you have the velocity profile measured.

If you can, post some picture about the geometry.

ijk October 26, 2010 11:11

Thank you for your response.

I cannot simulate one quarter of the radiator due to its heat transfer.

I do not understand why the computational accuracy would degrade with increasing pitch change if the values of flow variables apart from pressure are simply averaged and passed downstream.

The manual says:

"The computational accuracy degrades rapidly with increasing pitch change. It is recommended that sufficient blade components be analyzed on each side of the interface to minimize pitch change. Any pitch change will result in a non-physical transient interaction: the entire 360 degree components must be solved for truly “accurate” transient interactions if the pitch ratio is non-integer. Small pitch changes will introduce relatively smaller errors, large pitch changes will introduce larger errors. "

michael_owen October 26, 2010 22:31


You are correct; the solution will not be degraded because of the pitch change, because it is an integer ratio (7).

Now the real question you should be asking is not "Can I use a stage with a 7:1 pitch change" but rather, is a stage interface appropriate at all? If the latter is true, then the pitch change is irrelevant as long as it's integer.

You need to calculate r*omega and compare it to the through velocity of the fan. If it's much greater than the through speed, then the stage approximation starts to become appropriate.

ijk October 27, 2010 07:11

Thanks Michael,

Your response helped clarify things a lot. My average through speed is about a fifth of the tip speed. Good to see you keep yourself busy with CFD whilst on the bench.:) How's the hamstring?


All times are GMT -4. The time now is 07:00.