Rigid member torsion?

[gb156]

Why would a simple span, uniformly loaded (on one axis only) beam using 'rigid' member section properties have torsion?

Since RISA-3D does not have joist section properties available I used rigid members instead, and noticed that the shear and moment diagrams for the beams supporting the joists were incorrect. For some reason, the rigid members had torsion, which induced moment to the beams. Why?

I changed the joist members from rigid to a W12 shape and the torsion went away and the support beam's shear and moment were correct.

 

[JoshPlumSE]

Huh, I've never seen that happen before.

The most likely cause would be a modeling error or user error. There could be some applied load or displacement that you were unaware of. Or, some skew to the beam that makes it something other than a pure simply supported member.

Other than that, there are times where rigid members get to be too rigid. So rigid in fact, that they can exceed the internal rigidity used for boundary conditions. When that happens, you can start getting bad behavior. Older versions (before vesion 8.1.3) of RISA-3D did not have auto-detection of this issue. You would notice only when realizing that the sum of externally applied loads doesn't equal the sum of the reactions.

I should also point out that the overly rigid link issue gets exacerbated by combining multiple rigid elements together. Rigid Links directly connected to TOM offsets, diaphragms, or boundary conditions.

 

[gb156]

It's a pretty simple model. The beams aren't skewed or sloping. The load is a simple uniform distributed. There are no rigid links or diaphragms. I'm using RISA-3D version 8.1.3. The file is attached...

The original framing system used rigid members for beams and steel for the girders. I then made a copy of that framing system and changed the rigid members to steel members and the torsion in the beam goes away.

Any thoughts?

 

[JoshPlumSE]

It's related to a few things:

1) The asymmetry of the supports causes different vertical deflection and flexure/rotation of the supporting beam at each end.

2) Now, the end releases of the rigid link were set to BenPin. This releases the flexural rotations from the beam ends, but not the torsional rotation. That is still fixed. The problem would go away if one end of the beam has an "AllPin" rather than a "BenPin". Make sense?

3) The rigid links are VERY stiff in torsion. Much stiffer than the wide flange beams that modeled as the contrasting example. J = 1E6 for the rigid links vs. 0.1 for the wide flange.

This stiffness doesn't cause the problems, but it does make it much easier to see. Because of the differential deflection and rotation of the supporting member, there has to be some torsion. Though we would expect it to be relatively small. The wide flanges show some torsion, but it is extremely small (7 E-5 k-ft).

But, the rigid links are so stiff in torsion that the end connected to the more flexible beam is actually providing rotational restraint based on it's extreme torsional stiffness.

Josh