[jrl4444]

Attachment 26811Bug or just the way it is? CFX Post 11.0


Problem ….User Surface moving out of location as domain rotates in a transient run.


See attached example images for quick understanding of where I am hung up. Transient run, rotating square, user surface offset inside domain, as domain rotates surface moves out of location. After investigating I find that


Vector Variable Normal on the model surfaces does not rotate with the domain if I advance steps using time step selector or run an animation.



Vector Variable Normal does update if I load any time step under file command Load Results.


This looks scary to me thinking about Post calculations / expressions that may depend on this very basic variable updating as a domain is rotating.


Clearly Post does the calculation for Normal from a surface for every node when a file is loaded, is there a way to force a recalculation as new time steps are advanced / the domain moves?


I have looked and looked for a setting or workaround:
Here is what I have considered so far.


Create my own variable MyNormal by Frozen copy of Normal Plus domain movement. ?? I don’t see an easy way to use MyNormal to place the Use Surface.


Set up a batch and load every trn file one by one outputting single image frames to stitch together later.


Load the user surface from file but then I need to transform the surface every step to move with the model. I am guessing there is a fairly simple way to do that but a call for recalculating the Variable Normal would be so much cleaner.

Thanks for your time

[ghorrocks]

Does this still happen in the current version of the software?

[jrl4444]

(Does this still happen in the current version of the software?)

I have no idea? The rotating square was a simple example to test it.

I only have 11.0 32 and 64 bit, both of these have this problem, I tried on a couple different machines different operating system all same results. I was doing a moving mesh project that included rotation. I thought this problem was do to a moving mesh issue so I set up the simple standard rotating square domain and was shocked to learn what I found. Scary because most rotating domains are round, the problem is difficult to notice on a tangent running surface.


Does it seems like a bug to you?

[ghorrocks]

It could be a bug - obviously you have to check your rotation point.

But nobody cares about bugs in old versions of software. If you find a bug in the current version then report it and hopefully get it fixed, but nothing will happen in old versions.

[jrl4444]

My Rotation point? 0,0,0 rotating about Z axis, and I did check it but true normal from a surface is Normal regardless of rotation point. Perpendicular in 3D space from a flat surface, somewhat more complicated on curved or irregular surface but still a definable vector from a node based on adjacent nodes . That is what the Variable Normal in CFX is supposed to be right?

Well newer version is not currently an option for me so digging deeper, been awhile but have done a little perl programming. CFXPostRules.ccl is my starting point but looks to be deeper than that.



Any help in a direction to go would be great, Thanks