First, keep in mind that Appendix G is a nonmandatory appendix. Thus, while what it says is good practice, nothing it says is actually mandatory. I find it interesting that it only mentions the reduction of secondary stresses - I'd say that repads also reduce primary stresses.
What is mandatory is that you, as the engineer/designer, consider and evaluate all of the loadings on the vessel. Based on my experience, if I were to put a 6' x 6' top platform on a vessel and I'm attaching it with vertical legs to a 4' diameter head/shell which is 3" thick and at 400°F - I will not need repads since it is a relatively light load and transferred into the shell through shear. Take the same platform with unusually high live loads (e.g. designed to set a heavy safety valve down on it for maintenance, etc.) and put it on a 20' diameter head which is 3/4" thick and I'll have to take a close look at putting repads on since it is a relatively heavy load transferred into a large diameter thin shell through radial loading.
As vesselfab pointed out, there may be other reasons to use repads such as changes in metallurgy or to allow a bit of margin for post-PWHT changes.
There are also reasons to avoid repads, such as on vessels subject to fatigue where a fillet weld is detrimental, vessels subject to thermal shock where the differential temperature between a shell and repad could be detrimental, and my personal favorite: It is impossible to determine wall thickness on the run via UT when a repad is blocking access to the head or shell.
jt
