What happens when you don't check your calculations ...

Engineers are always advised to "check their math" - never assume or the bridge might fall down. So, when this engineer installed a load-bearing structure (a.k.a. Ikea hanging clothes net) from an overhead, did she do any back-of-the-envelope calculations? Build a prototype? Run an FEA model first? nooooo... and look what happened. Let that be a lesson to engineers and do-it-yourself-ers everywhere.

Moral: (besides the obvious) don't use string rated as "decorative" from the dollar bins for anything other than gift wrap.

No engineers were harmed in this experiment.

Rigid Links in ProE Mechanica

rigidlinkmesh
Originally uploaded by fizz_in_space

This is a ribbed beam created in ProEngineer. The part is analysed in ProE Mechanica. One end is fixed, the other has a load applied. To prevent deformation, the load end has Rigid Links applied to the geometry. Question is, does it transfer correctly to the mesh? The model is idealized using Shell Pairs, so the original solid surfaces merge to a midsurface.

There are no Face-Face or Edge-Face Links in the AutoGEM Summary window, however it lists 1 rigid link in the Model Summary window. Adding rigid links to another collection of edges creates a second rigid link in Model Summary, confirming that each group of rigid link entities counts as 1 rigid link. But what indicates the rigid links in AutoGEM graphically? (Complete model posted to PTC Users.)

Trumpet Effect

Something mysterious happens when you apply a pure moment (twisting or torsion) to a long beam in finite element software such as ProE Mechanica. The end nodes appear to fly outward as if there is a radial force!

This effect is an artifact of the small displacement theory used to calculate the deformation. With torsion, each node must move tangentially (in a circle) as the end twists. In Cartesian (3-axis) space, this movement vector (direction) is broken down into 2 directions (because a circle lies in a plane), so this increases the error.

The calculation for each node has vector components in each direction. These calculations have a small error incurred because the theory is not the real world. As the nodes move around the circle, they also diverge outward. This effect is exagerated because the results are automatically scaled (usually) to make the deformation visible.

In this case, the appearance is false. The apparent radial motion can be safely ignored - although this does not excuse the analyst from checking the results using analytic or experimental methods. The cardinal rule of engineering applies to FEA: Never assume the pretty picture is right!

(There is no apparent literature in the software manual or online that explains this effect - thanks to the forum users at PTC Users for the clear explanation.)