I've been running some complicated Finite Element Models. In most cases, the stress repartition seemed to be absolutely correct. However, on several point (complicated geometry), it appears that stress is way beyond the overall stress in that area.
The same goes for calculus of hydrostatic pressure. While all the model ranges between -150;150 kPa, there are some areas that reach the insane values of 300 or even 1MPa.
How can I be sure the singularities in those points does not alter stress repartition to a point where the model is not realistic ?