The biggest issue I've had with 28VLs is water or other non-vaporizing liquids in the fluid stream. These can reduce the lift-off of the seal faces to almost nothing in an instant.
We had a rash of failures on an ethane pipeline this year which was directly attributable to water contamination of only 400 ppm. The seal flush stream was filtered for debris, but there was no coalescing capability so the faces got wet.
We've since installed a revised filtration system with particulate (3 micron) and coalescing as separate stages, with liquid return to suction of the pump. This reduces the need for manual draining on the coalescer. Not the most efficient system, but the flush stream is 1 gpm out of 1200 gpm total flow.
That said, the 28VLs are quite reliable when given the proper working environment. Spend the money up front on a good support system and you won't do much except change filters.
I also have some familiarity with 53B and 53C systems, and they work well too. However, there are some drawbacks, particularly when operating high pressure systems. Barrier consumption is a major concern, so it must be compatible with your process fluid. We also have o-ring extrusion on the barrier/atmosphere side of the seal, which significantly reduces seal life. Maintaining a good pressure margin is critical in these systems, but can be difficult in a 53B if you're swapping products, changing equipment speed, or other "bad practices."
I would use the plan 53 systems when you are swapping products constantly, the product streams are heavily contaminated, or you have a light product that can change in composition quickly (say, Y-grade NGL.)
