I agree with SAM's argument as SIF is nothing but a factor to increase bending stresses. Any attachment on a piping will result in discontinuity stresses ie local bending stresses as per shell theory,although such stresses wll die down as functions of SQRT(RT). Hence,in lieu of a more sophisticated analysis, a factor to account for additional bending stresses is something better than nothing.

Based on some comparisons that I did between FEA and Kellog method for Trunnion attachments, I have noticed that the conventional Kellog method is quite accurate in computing the local stresses( this is based on my testings only and not a published paper). Kellog method is based on a cylinder loaded with a ring of uniform bending moment.I think in lieu of SIF, if local stresses checked , it is also an acceptable method.

Regards