Constraints on rotating body (Siemens NX)

[Laserlassie]

Hi!
I am currently designing and simulating an electric outrunner motor (rotor on the outside). I need to evaluate the displacement of the external rotor at approximately 15000 rpm and in the software i am using (siemens nx) i apply this as a rotational load. I am fairly certain that the solver only uses this rotational speed to calculate the centrifugal forces. However, I am a bit uncertain about the constraints i have applied. In the motor mentioned, the shaft is connected to the external rotor by a slot connection and runs trough the internal stator (see figure). The shaft is supported by two bearings. I have currently modeled these bearings as infinite stiff using pinned connections (free rx,ry,rz and fixed x,y,z). However, the shaft also needs to be fixed at one location to stop rigid body rotation. I have therefore used an RBE2 (rigid connector) that is connected to face of the shaft and fixed. So my question is, does this seem like a OK way of restraining the model? Something tells me it's not.. (PS: The brown arrows in the figure are contact sets, and the spider at the shaft is the RBE2 which is fixed, the red arrows is an applied torque of 160Nm.)

Thanks for the replies!

 

[rothers]

Try posting your question here : Link

 

[Laserlassie]

Thank you!

 

[rb1957]

I have no real idea how to constrain an intentionally spinning model. I do think rigidly pinning the shaft will give you bad results, it's Much better to restrain these two points with finite stiffness (I use three orthogonal rods, and fix the ends of the rods, so the model doesn't "feel" these rigid constraints). It's possible that constraining the 6 rigid body motions is ok, but then constrain the model as …
1) three translational freedoms at one bearing (X, Y, Z) (let's call X as axial)
2) two translational freedoms at the second bearing (Y and Z, ie not axial presumably with the same line of action as the first bearing)
3) constrain RX at one bearing.

I would make use that your material properties use mass density (as opposed to weight density, eg Al is 0.1ib/in3).

another day in paradise, or is paradise one day closer ?

 

[Laserlassie]

Using finite stiff rods sounds like a reasonable approach which i will definitely look into. I have now adjusted the model to something in the line of what you suggested with the constraints, the results are similar as before but not equal so i will need to dig deeper into this. As for the mass density, the software luckily uses mass density (kg/m^3) because this is not something i had though of