Practically, there is a problem where two or more customers/meters/buildings are fed from an ungrounded-secondary transfomer bank {delta, wye or 1ø.} If for instance, one place has ground-detector lights and a fault [insulation failure] develops anywhere in the system, the lights will show it. In the simplest sense, for multiple locations, it can get frustrating tracing the problem. Now, if that same 600V {or 480V} system is solidly-grounded wye, then a fault will typically operate an overcurrent device, and isolate itself.
[Warning: Rambling ”war story” follows.] I used to work in a plant that was 95% “hosedown” {very wet} having 14 [750-2500kVA] secondary-unit substations, all with [12kV primaries and] 480Y secondaries. Thirteen of those unit subs were operated ungrounded {XO bushings ‘flying in the breeze’} and one solidly grounded, so it was the only true 480Y/277V system at the plant. The only reason it was grounded was that its single load was an electrode boiler. The grounding was enforced by the boiler insurer—Factory Mutual in this case.
The other 13 had “ground lights”, which were nothing but a gasketed hinged-cover can with three external 6O-watt lamps in vaportight fixtures, and three internal 480-120V 100VA control-power transformers. When a fault showed up in any of the 13 systems, supervisors were trained like Pavlov’s dogs to run around furiously shutting down various 480V gear until local ground lights returned to their normal equally-dim state. {On a solid phase-to-ground fault, one of the three ground lamps extinguish, and the other two glow at full brightness.} The funniest incident was finding one of the ground-light sets flickering madly, only to later locate a loose 100-amp 480V pin-and-sleeve extension-cord end spitting steam from being flooded by hosedown.
