CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   rotating domain (

mach000 April 16, 2010 12:28

rotating domain
Hi to everybody,
I'm simulating a multifase rotating domain in wich the axis of rotation doesn't match with one of the global ones.

So, I defined a local Coord and specified one of them as rotation axis.

The simulation doesn't work because:

This simulation involves a gravity vector which is not aligned with the rotation axis. Because the solver currently interprets the gravity vector as being in the relative frame, the gravity vector must be made a function of time so that it is constant in the absolute frame (ie, counterrotating in the relative frame).

Could someone help me in order to undestand the problem?

I specify that I have experiences with multiphase simulation with rotating domain. I never had problems because the rotation axis was coincident with one of the global ones.

Thank you in advance

ghorrocks April 18, 2010 07:42

Looks like you might have to define gravity as a body force yourself and make it a function which makes it point in the direction you want. Then you can do whatever you like to the gravity vector - but you will have to do the maths to make it point in the correct direction, including the rotating frames of reference stuff.

mach000 April 19, 2010 10:52

the rotating velocity is "omega". Vertical global axis is Z. So, in a stationary domain the gravity vector as theese components:


Now, if I have to define a time function in order to have the g vector always alligned to Z axis, I think I have to made zero g_x and g_y defining expression such as


Is it correct?

bharath October 24, 2016 02:04

Thanks for thread.

As you told in the topic that expressions are working.

ghorrocks October 24, 2016 02:12

I suspect Ciro's equation has an error - I would not expect the x and y components to be equal in general, and for x and y I would expect one term to be a sine and one to be a cosine (ie:

Also note that I suspect there are high order terms which arise from using a moving acceleration vector like this. If your rotational speed is slow they are likely to be negligible, but as the rotational speed gets faster they will be come important. So be aware this simple approach is probably not universally applicable.

All times are GMT -4. The time now is 15:39.