adjust model to CMM points ?

[opetrenko1]

I have several hundred CMM-points
I need to adjust the nominal (or design intent) complex but smooth surface to these CMM points (to run CFD analysis whether we can accept the part)
Does UG or third party have tools allowing me to fit a surface to these deviations?

Thank you

 

[SDETERS]

WE did something close to this. We had only 20 or so points. What we did is imported these points into th NX part file with the correct orientation. Then we used four point surface on the four most outer point that would make a nice rectangle. Then we used fit surface to fit this four pointed surface to these group of points. The points need to be grouped to use fit surface. Then we used the Synch tool and did a replace face with this new wavey type surface. This worked well for us and our real simple part.

 

[opetrenko1]

Thank you, SDETERS.
it looks like the way to go for the cases of simpler surfaces.
I have a scan of a part resembling aerodynamic scoop. Very high tolerance requirements (especially in its lip area) while rather sparse and irregular CMM points distribution that have to be even more precisely matched. Surface-fit either over-smoothes or overshoots creating deviations exceeding CMM deviations.
I almost feel like learning to program for UG, then read-in hundreds of deviations and surface-fit them in matlab, then generate thousands surface points, if not cfd grid itself. (...ten years later. lol) Thanks!

 

[BubbaK]

I'd be interested to know ... what surface fitting can you do in Matlab that you can't do in NX ??

 

[opetrenko1]

BubbaK, deviation values.

 

[aluminum2]

I used to inspect airfoils with CMM's and 100% of the time siemens, pratt, rolls royce, and GE all wanted point data as airfoil sections. If that is the case then I would use the spline tool and mesh the surface. PW Airfoil from Pratt was a great tool for doing this automatically if you have it.

 

[opetrenko1]

aluminum2, yes, we've requested the vendor to measure profiles in-plane in future. Meanwhile, I'll try moving points to be clustered in planes, fit lines and mesh. Thanks