Issues with 3D results for some problems about pynite HOT 5 CLOSED

Actually, I think this may be an issue with my statics check code, not the main program. I'll look into this and resolve this issue. For now, use caution and be advised that this may or may not be an issue.

from pynite.

Just ran some more tests. It's definitely a problem with the main program. 2D models in the XY plane don't seem to be affected by this issue. More updates to come.

from pynite.

Update: Still working on this. It doesn't seem to affect every 3D problem. It's definitely affecting more than just reactions though. I think it may be a problem with how the 3D transformation matrix is being formulated. Tricky problem to fix.

from pynite.

Still working on this. I found some major math errors in the textbook I was using to develop the transformation matrix. I've corrected those errors, but still seem to be getting bad results. I've got a different text arriving next week, and will check the transformation matrix against that one.

from pynite.

I think I've got it fixed! I'm going to run some more test routines on it and make sure I'm getting results that are correct.

The textbook I was using had some major flaws in the transformation matrix. I ended up deriving a new transformation matrix myself. The new one is "smarter" as it keeps the local z-axis horizontal in all circumstances. The previous transformation matrix kept the local y-axis perpendicular to the global Z-axis, which led to unpredictable behavior in 3D, since this constraint caused beams to be rotated about their longitudinal axis in some cases. Keeping the local z-axis horizontal simplifies working with structures where the global Y-axis is vertical. It also models real-world problems better.

The new transformation matrix also keeps the local y-axis in the global +/-X direction for vertical members. This convention makes it natural to model 2D frames in the XY plane by making the z-axis the flexural axis for all members in 2D problems modeled in the XY plane.

from pynite.