Jon,
When you heat treat something that big with high restraint, it probably doesn't take much reduction in ductility to help start cracking.
I think you could have 2 possiblities.
1. If an flat-horizontal flux core wire was used, it most likely had bismuth added (you may want to check anyway-some manufacturers may add it to all-position wire, but not the few I've checked). Bismuth reduces ductility at high temperatures. The only time I have seen it (I think), is on repairs that have been solution annealed a second time, which makes the problem even worse, but it could happen on the first time to high temperature too. However, the cracking between the beads make some sense in this regards since the high temperature HAZ has already seen the high temperature once, and the solution anneal make
2. With the very thick sections and air cooling, if you have much ferrite at all, it can form intermetallic phases such as sigma since it takes so long to get through the embrittlement ranges. This could be one of the reasons your ferrite is cheking lower. If it transforms to sigma, it won't be magnetic anymore.
I'll send you a personal e-mail as well so you can give me a call if you want. Say hi to Angel for me.
Greg
Is it possible that an incorrect electrode was used (carbon steel?) and local repairs were made to remove the incorrect weld?
Also, you stated "We also performed RT, MP, and PT before heat treat and everything was clean." What is "MP"?
well, this one is interesting. You mentioned Having done PT and RT, but before you came along there had been repairs to the sub arc. I also gather that the problem is in the sub arc and repairs to the same. Making the assumption that RT didn't see anything, and given SAW, previous repairs for "unknown" reasons, center-line cracking under the next succeeding pass, your description of of it appearing to follow a grain boundary, I am leaning towards a dendrite problem, probably exasperated by other factors. Given that they are in fact there, and RT didn't see them, nor did PT, they were probably just below the surface. If in fact it is the problem, then you should have some more of it extending into either edge of the now visible cracks underneath the surface.
Rapid solidification of 308 can create center-line cracking problems. that same problem could have been highly stressed, especially if it also was subjected to the the stress of being in the succeeding passes heat affected zone. If in fact it was a columnar / dendritic solidification problem to begin with, they are problematic to repair. Some people will stop to earlier on the excavation of the crack, and weld over it, causing it to crack again underneath, not to be found until an attempt at annealing.
For more specifics on the testing, PT must have an open surface. I'm willing to bet that at some point in the parts life, a pt test showed a problem hence the unexplained repair.
If in fact it is the problem I've mentioned, RT sucks for finding it. Especially in thickness range you describe. Not only is the flaw orientation a problem, your going to have a problem with diffraction. At normal pipe thicknesses it's never a really an issue (< 1"). Getting into 1.5 and 2 inches and up, you run through enough of a cross section containing 9-11 percent nickel that it becomes an issue. Anything such as a fine crack in the upper surface oriented nearly parallel with the plane of energy propagation (opposite film as opposed to in contact with the film) will be muted potentially to the point that you can lose enough subject contrast to make it nearly hidden.
If the part matters enough, ship it out to or bring in a group familar with round trip tandem UT, or phased array. Either could dial it in via refracted Lwave inspections.
In short it could have been there all along, but not seen until it dropped anchor and cracked out to the surface during annealing.
Therein is my two cents worth,
Gerald