Serious bug in stress linearization

Victor

If you turn on Ignore through-thickness bending stress then bending stress is sometimes wrong. Consequently, membrane + bending stress (m+b) and peak stress are also wrong because they're derived from bending stress. Membrane and total stress aren't affected, nor is anything when "Ignore through-thickness bending stress" is turned off. The problem seems to only happen when the SCL is not aligned with any of the global axes.

This will be fixed in version 21 which will be released soon.

Here's an example.

The only non-zero stress component in this part is a bending stress normal to the SCL. There's no normal or shear stress in the direction of the SCL so we should expect m+b to be equal to total stress when through-thickness bending stress is ignored. Instead, m+b is 0.25 MPa while total stress is 0.5 MPa at the same location. Peak stress is also incorrect with a maximum value of 0.5 MPa instead of zero.