TheHypnotoad
Aerospace
- Jan 13, 2011
- 3
Hi everyone!
I'm working on a buckling problem (not my personal area of expertise) and I wanted to pick your collective brains on the topic of failure and Finite Element analysis.
So, we have two metallic plates, held apart by four angled L-beams. Everything is modeled as shells. (Abaqus S4R5 elements, if you want specifics). A uniform pressure is applied to the top plate, and a uniform displacement is applied on the bottom. Constraints against rigid body motion are applied, natch. The setup is symmetric, and all of these L-beams should be under an identical load in a mathematically perfect world.
If you're like me, you would expect all the beams to fail at the same time in the same way, with identical stress distributions . . . but they don't; one gives first and the top plate slides off toward a corner. Then the next buckling mode hits a few psi later and the top plate slides off toward another corner.
Why don't they fail simultaneously?
My best guess is that Abaqus has a certain amount of significant digits when assigning coordinates to nodes, and when you look closely enough, the beams are not perfectly straight. The rounding error causes minor imperfections in the beams, just enough to make one fail first, by a few pounds. Has anyone had this happen before?
On the other hand, my coworker, who has much more experience than me at this and is tremendously helpful, isn't buying my idea. What tests would you perform on this model to work out what's wrong (or prove/disprove my idea)?
Thank you!!
I'm working on a buckling problem (not my personal area of expertise) and I wanted to pick your collective brains on the topic of failure and Finite Element analysis.
So, we have two metallic plates, held apart by four angled L-beams. Everything is modeled as shells. (Abaqus S4R5 elements, if you want specifics). A uniform pressure is applied to the top plate, and a uniform displacement is applied on the bottom. Constraints against rigid body motion are applied, natch. The setup is symmetric, and all of these L-beams should be under an identical load in a mathematically perfect world.
If you're like me, you would expect all the beams to fail at the same time in the same way, with identical stress distributions . . . but they don't; one gives first and the top plate slides off toward a corner. Then the next buckling mode hits a few psi later and the top plate slides off toward another corner.
Why don't they fail simultaneously?
My best guess is that Abaqus has a certain amount of significant digits when assigning coordinates to nodes, and when you look closely enough, the beams are not perfectly straight. The rounding error causes minor imperfections in the beams, just enough to make one fail first, by a few pounds. Has anyone had this happen before?
On the other hand, my coworker, who has much more experience than me at this and is tremendously helpful, isn't buying my idea. What tests would you perform on this model to work out what's wrong (or prove/disprove my idea)?
Thank you!!