Further to my previous post, I feel that to the extent an analyst has to go in for ( for example,properly computing peak stresses in FEM , incorporation of ovalization effect of supports etc) is something, an analyst has to decide based on the system in hand , his/her experience and his/her level of expertise.

For not so critical systems ( something cannot be quantified but to be assessed by an analyst), use of SIF or use of Kellog type of analysis is acceptable in my opinion.

Regarding FEPipe, my opinion is it is too tailor made with lot of limitations and can never be at par with FEA codes like ANSYS or ABAQUS and I dislike what it says " FEA MADE EASY".FEA should not be made easy as wrong results can easlily escape an analyst w/o proper training on FEA.

Regards