There have been a number of reports/studies published by EPRI on this subject. I believe they are still free. There has been much research published with regard to P91 and its manufacture and heat treatment and welding thereof.
I limited it to 2 and then required a full cut out and reweld. Excessive surface decarburization was the main reason plus over tempering. In one instance I required pipe spools to be cut out and reheat treated in vacuum furnaces after evaluating microstructure.
I would require it for backgouging of welds in heavy wall thicknesses prior to making the opposite side weld. The existing weld is under stress even if intermediate stress relief or dehy heat treating has been performed.
You are comparing apples to oranges. The manufacturers data is full size all weld metal tests performed to maximize grain refinement. Yours are subsize specmens with weld metal dilution with base material with probable lesser controlled heat inputs.
Question is too ambiguous. Some electrodes like E7018 have much reduced C today than in 1972. Typical chemistry in 1972 were 0.10% C and 1.0% Mn and in 2015 when I retired were 0.05%C and 1.0%Mn and their higher temperature tensile properties were lower.
You state that the pipe line is at a hoop stress < 20% of the SMYS. How can you truly overstress such a line unless it as hit by a truck or a 7 plus seismic event?
When accesible from the inside, like a flange to pipe weld, we would permit the repair therefrom prior to PWHT. We would have performed a dehydrogenation treatment prior to repair and initial RT..
Joint prep and excesive or thin root gap and land are common problems in field work, in which I am most familiar. Most of our clients only permitted one such repair and if unsuccessful, a complete cutout required. On all pipe < or = to 6" diameter, I required a complete cutout and reweld.
Yes the repair can be done. I have seen a number of such repairs been unsuccessful, finally requiring a complete cutout and reweld. You may have to make the repair area/length somewhat larger to effect a successful repair. I assume that you found the defect through RT before PWHT.