Hard to tell from the one picture, but do the generator feet overhang the frame?
Has a vibration analysis been performed?
When you say "alignment checked ok" how was the check performed, the feeler gauge in between the generator and flywheel housing, or with a dial indicator or laser?
Are the cross braces at the rear of the generator base bolted or welded?
I have ran into this type of failure before, was related to base and mounting issues, most recent was a site with 5 units that had originally shipped with low voltage generators and had been field fitted (by dealer) with MV generators before being shipped.
Cracks in the area your picture shows are usually due the a flimsy set of base rails, or a soft foot problem. Problems with synchronizing usually shows up as excessive wear or damage to the torsional coupling. 3516C's usually have the rubber sections and the aluminum drive ring, have these been checked? I never seen load transients or oscillations cause cracking to a generator frame, and have seen some pretty severe load swings, like on crushing and drilling equipment. Severe cyclic loading usually shows us as excessive wear on the torsional coupling, misfire and poor combustion will also cause coupling wear/failure.
Do you have a copy of the CAT Application and Installation Guide? If not get the local dealer to get you one and take a good look at the section on mounting and alignment.
My personal opinion is that current base rails for MV generator sets by CAT are not as stiff as they should be, and I really dislike the base rails with the bolted cross members, no one checks them on a regular basis and if they loosen even a small amount your get flexing of the base rails. It gets worse if the vibration isolators are not properly adjusted.
I have worked on lots of units in containers over the years, up to 2.5MW standbys and 2MW prime units in long term rentals and jobs like mine sites and utility grid support. For the most part they do ok, but as noted above, they have to be level, and supported properly to allow the base rails to maintain alignment.
The old oil field bases were massive in comparison, but took huge amounts of abuse, like dragging them off transport trucks, poorly leveled foundations (usually a hard dirt patch that mostly hit both sides evenly) and "thumbnail" leveling. But they ran and ran with not many problems.
I don't use the CAT method for aligning close coupled generators, too many times have found it not to provide good operation, my preference is a laser tool. And I always try to do a soft foot check (procedure is in the A&I Guide mentioned above). I also try to regularly do vibration measurements, especially on MV machines, as my experience is that they have more problems in the field. When you make the measurements pay attention to the frequencies below first order, low frequency vibrations at higher magnitudes than first order usually indicate the frame is bouncing on the isolators. CAT specifies limits for overall and first and second orders, but seems a lot of folks overlook the lower frequency issues until something breaks.
Hope that helps, Mike L.
