If it can fly with enough precision (say, keeping within 4" from the surface, or directly contacting the surface with an instrument; not sure what the requirement is) and the equipment can be made light enough that inspections can be completed without an egregious number of returns to base to recharge, I feel like even ultrasonic testing would be feasible.
As far as RT, it seems like it'd be a much better idea to do that with robots instead of humans who get cancer. I'm assuming that there aren't any other people within a range that would get a significant dose of radiation during this testing, but even if, it would seem to me that automated inspection could be made more precise and safer than humans could do, eventually.
For UT, the surface needs to be directly contacted with an instrument; but that's not all.
First, insulation must be cut off and disposed of (properly). It's usually tin flashing covering some kind of insulating material. If the insulation is asbestos, then it's not acceptable to have any of it blow off into the wind at all.
Next, the surface has to be polished, usually with a rasp or grinder, so that the probe has clean metal to make contact with.
Then, the probe, coated in ultrasonic-conducting jelly, gets applied to the surface, and must maintain sufficient contact for as long as it takes for the instrument to get a reading (at least a few seconds, and it's not always reliable). This step has been done in the past with robots, albeit in different access environments.
Finally, the insulation gets re-applied, covered in tin flashing, and sealed with caulk.
I'm not saying it's not possible in principle for a flying robot to carry out steps one, two, and four, but I can't forsee a technology able to do this within, say, the next ten years.
As for RT, you have to understand that a lot of refinery inspections are done in what's called turnarounds - where the refinery is turned completely off for a week or two at a time. Tons of contractors are called in to work overtime on top of one another in order to carry out planned maintenance and get the thing back online ASAP (time is money - big money). Obviously, extreme care is taken when doing RT in this kind of environment, because a mistake can be extremely dangerous.
It simply isn't safe, in any universe, with any technology, to put a radiographic source (actual radioactive elements) on a flying object and zip it around in the middle of a refinery turnaround. Period. Maybe during normal operations, but only at the cost of extreme interruption of everybody's work - eg, everybody literally has to exit the entire facility.
EDIT: on second thought, having everyone leave the refinery isn't even possible without turning the thing off. There are operators who need to be on-site at all times, monitoring and making adjustments to the various processes. And sometimes a valve really does need to be turned by hand ;)
> or directly contacting the surface with an instrument
This is more challenging than you might initially think, as conventional multirotors can't easily generate the forces you need to place a tool/sensor against a vertical structure.
As far as RT, it seems like it'd be a much better idea to do that with robots instead of humans who get cancer. I'm assuming that there aren't any other people within a range that would get a significant dose of radiation during this testing, but even if, it would seem to me that automated inspection could be made more precise and safer than humans could do, eventually.
Definitely an interesting time to be alive :)