Hello everyone,

I would like to know if it's possible to define a moving axis in CFX 10 or 11?? My problem is to simulate a journal bearing with mesh deformation. As the shaft is moving with time, its rotational axis is moving too. It doesn't seem to work with a defined coordinate frame since it will stay fixed... So it would be great if you have any suggestions on how to set this rotational axis. Thanks.

I've seen journal bearings done before, but they had a fixed rotation axis for the shaft (the shaft could deform, but it's axis was fixed) while the outer walls of the bearing could deform and also move as a whole. The outer walls were also stationary, hence no problems with moving axis. Would it be possible to pose your problem in this way? If not, then the short answer is: There's no way to have a moving axis of rotation. The long answer is: Don't use rotating domain. Instead define the simulation using a stationary domain and using moving mesh to impose the rotation. Since you'll be imposing the rotation using general expressions, you're not limited to rotation about a fixed axis, but you're expressions might get kind of complicated. You'll need some way to extract the axis of rotation during the solution; to do this use areaAve(x)@RotatingWall to get the x-coord and areaAve(y)@RotatingWall to get the y-coord. Never tried this, so no guarentees!

Hi Stumpy,

It might be possible to define the problem as you said with a fixed axis and the bearing moving as a whole, I need to check that more carefully. Thanks for the suggestions, I'm working on it.

Hi Samuel,

You shouldn't use a rotating domain for this kind of problem, specifying the wall velocity should suffice. You could either do as Stumpy suggested and keep the shaft in a fixed location and move the outer walls, or you could define your shaft wall velocity in cartesian coordinates using the appropriate trigonometric expressions, but make them relative to two cartesian points which you can then move. The former may be OK for constant density, single phase flows. If you have variable dnesity you would need to add the appropriate body forces to reflect the acceleration of the reference frame (because you would essentially be moving your reference frame relative to the shaft).

To set up the alternative method, define the surface velocity as a function of x1,y1,z1 and x2,y2,z2 and make each of these a function of of you shaft dynamics. It's a little painful to think through, but easy to implement. Note that you may also need to reference the original node locations, which you can get using the following expressions:

X0 = x - Total Mesh Displacement X Y0 = y - Total Mesh Displacement Y Z0 = z - Total Mesh Displacement Z

-CycLone