Maybe you are familiar with WRC's stress (SIF factors)... so you would answer a question: when you evaluate stress at nozzle-shell junction, you are interested first on weld SIF there, as "the most authentic basis for design"? You may answer to a similar question for your case.
MoverZ said "stick to simple old Kelloggs or use FE" because both of them address to the real troubles in trunnion-pipe.
Both Notes 13 in B31.3 edn 2008 and 2010 refer to Fillet welded joint or socket weld flange/fittings and mention explicitly a geometry according to Fig. 328.5.2.C/ 328.5.2.B etc. It is not your case and you cannot extrapolate them to your case because, for these configurations, the main concern is "fatigue performance of the weld" as Note 13 in 2010 edn explains.
IMO, the difference between editions is "blending the toe of the fillet weld smoothly into the pipe wall" as an improvement of the fatigue performance of weld.
Best regards.
PS. Peng's book has two sub-chapters "6.8.2 Kellogg’s Choking Model" and "6.8.3 WRC-107 Stress Evaluation" that would help you.