Jackson306;
to add to Meteng's list, some interpretations of the ASME rules on preheat and PWHT allow the time to be based on the thinner walled part, but in the case of P91 , I would suggest the time be based on the thicker walled part. For piping attached to a thick valve, there can be a large difference in thickness, and the rate at which the parts heat up is based on the square of the wall thickness. Place the primary monitoring thermocouple on the thicker part, plus add'l thermocouples under the heating pads as per AWS guidelines, to avoid overheat / overtempering under the pads.
A related cracking issue can occur during cyclic operation at the weld between a thin pipe and a very thick cast valve. Again, the rate of heating and cooling during operations will be related to the square of the wall thickness, and fast startups / shutdowns can generate temperature differences at the butt weld between these 2 parts in excess of 100 C. Cracking of the butt weld ( within 12000 hrs) has occurred in P91 cases where the weld detail did not include the use of a forged F91 transition piece and also did not provide the code-required max wall thickness transition angle at the butt weld of <30 degrees ( max of 15 degrees is required in EU codes, see EN 12952-3 table B.1 , weld notch group K1).
"Whom the gods would destroy, they first make mad "