Maybe it's the effective section modulus if I'm not mistaken. But if it's a free end something is not right since there's is no restraint. But if the configuration of the header in both sides this branch is not flexible it is the header driving this high stress at the free end branch not the branch driving this high stress.

Modeling rigid element and using FEA SIF is a good approach since actual section modulus will be used. The D/T for 36" S10 is 115.4 so this a good candidate for using SIF's derived from FEA.

Just correct me if my interpretation of your system is correct.

Cheers,