When I at Siemens (my former employer) began building gvpi stators in the 100 mw to 300 mw size, we experienced serious vibration troubles on the end turns both locally in the series connectors and globally in the winding. This was evident in both the 50 Hz and 60 Hz fleets but was more serious in the 50 Hz fleet. This problem led to several forced outages and a few total losses. The problem was that the windings had 2-lobe resonance near the machine excitation frequencies. We were eventually able to correct this situation via a combination of series blocking and cutting away sections of the winding support structure to low tune the winding. Occasionally we would high tune, but low tuning is generally preferred. This is pretty major surgery, but it works.
On a less severe note, gvpi stators tend to show some cracking due to thermal expansion and some due to gravity induced sag. The easy way to tell is that resonant damage will tend to show dust in the cracks indicating ongoing relative motion.
In general, if I see cracking or dusting in a relatively young stator, I recommend FFA/BUMP testing of the winding to check the location of the primary and mid-basket oval modes. This will let you know how bad the problem is. It is my practice to never patch or reinforce a crack in the support structure or winding without a bump test. You might just make the problem worse.
Resonance problems are not limited to GVPI machines. The bump testing technology was developed by Westinghouse and my friend Jim to handle vibration problems with Westinghouse 4-pole machines. You'd think the huge nuclear stators would have too much mass for resonant excitation, but a 4 pole excites at 60 Hz, instead of 120. Likewise, some of the larger 2-poles had trouble in the early 80's which led to the development of the decoupled end turn support system.
Also, resonance problems occur in smaller hard-coil machines, but mostly on the phase leads or the parallel ring assemblies.
Anyway, regardless of the machine, if you see relative motion in the end turns, bump test the machine. The cost of the test is less than 20k and can save tremendous headaches. Several companies, including mine, can do the test quickly and accurately. If you make a fix without testing, you are just guessing at both the problem and the solution. I don't like guessing.