In using a 36” weldolet, where the branch has the same diameter than the header.
The note [12] from the B31.3 table D-300 said that the SIF is non conservative if d/D=1.
Can anybody give guide lines to choose a conservative value for that SIF?

You could ask your vendor - perhaps they have performed additional testing and can provide a "better" SIF for their fittings.

Alternatively you can use something like FE/PIPE (from Paulin Research Group) and perform an FEA run to numerically determine the SIF.