Reply To: Discrepancy between LSPrepost and hand calculations -Shell element stress output

[Armin]

Thank you for sharing these observations.

I would suggest using the maximum value since it is safer from a design perspective.
As a side note, I know that for some fracture models, you can specify the percentage of integration points that must reach failure criterion before the element is considered fully failed. For example, you can specify that 60% of integration points of a shell element should reach failure condition so that the whole element is considered as failed and be removed (eroded) from the model.