Ask a facility manager which backup power source they need and the answer usually starts with budget, when it should start with a different question entirely: how does this load behave during an outage? A server rack that cannot tolerate even a 10-millisecond gap needs a fundamentally different backup architecture than a set of bore-well pumps that can wait thirty seconds for a diesel generator to spin up. Treating "backup power" as one category is where most oversized or undersized systems start.
A UPS sits electrically between the load and the mains supply, so when mains fails, the battery takes over with no perceptible switchover gap — this is why servers, medical equipment and process controllers that cannot accept any interruption sit on UPS, never directly on a generator feed. The trade-off is duration: a UPS is sized in minutes to a couple of hours of runtime, not the hours a genuine grid outage can run to.Sizing a UPS correctly means calculating actual VA draw of the connected load, not nameplate wattage, and building in headroom for startup surge on equipment with motors or compressors — the exact calculation is laid out step by step in a dedicated UPS VA rating sizing guide. A common mistake worth flagging here: nameplate wattage on IT equipment and most electronics is a power factor away from VA, and undersizing a UPS by ignoring that gap is one of the more frequent reasons a "correctly rated" UPS trips under load it was supposedly specified to handle.
A diesel generator takes anywhere from 10-30 seconds to start, stabilise and take over the load through an automatic transfer switch — a gap that's irrelevant for pumps, lighting, and most industrial machinery, but fatal for anything running on a UPS already covered above. What a DG set offers in exchange is duration limited only by fuel supply, which is why it remains the default for facilities needing to ride through multi-hour grid outages rather than brief ones.Sizing a DG set against the full connected load, including the harder-to-predict starting current of motors and compressors rather than just their running current, is the single most common sizing error — reviewed in more detail against a facility's actual DG set specification before committing to a capacity. A motor's starting current can run several times its running current for a brief period, and a DG set sized only against the sum of running currents across a facility can stall or trip on the very first large motor start after an outage begins, even though its nameplate capacity looks comfortably larger than the "steady state" load it's meant to carry.
Solar paired with battery storage doesn't fit neatly into the "backup" framing above — it's less a standby system that activates on failure and more a parallel supply that reduces reliance on the grid in the first place, with the battery bridging the gap when both solar generation and grid are unavailable (typically overnight outages). Its economics depend on daily usage pattern and local tariff structure far more than DG or UPS sizing does, which is why it tends to make sense alongside, not instead of, a DG set for facilities with genuinely long or unpredictable outage patterns.A facility with a daytime-heavy load profile — offices, retail, most commercial operations — captures the most value from solar generation offsetting grid draw directly during the hours the panels produce, with the battery sized more for short evening or early-morning bridging than for carrying the full facility through a multi-hour outage. A facility running significant night-shift load gets proportionally less daytime offset value and needs the battery and DG combination to do more of the actual backup work, which changes the sizing conversation for the battery considerably compared to a daytime-only operation.
Most commercial and light-industrial sites in Telangana end up with a layered answer rather than a single technology: UPS for the loads that cannot tolerate any gap, DG set sized for the loads that can wait a switchover but need hours of runtime, and increasingly solar+battery layered in to offset daytime grid draw. Getting the layering wrong — undersizing the UPS because "the DG will cover it" — is how a facility ends up with equipment resets every time mains drops, DG switchover gap included.A practical way to build the stack correctly is to sort every load into one of three buckets before sizing anything: loads that cannot accept any interruption (UPS-only), loads that can accept a 10-30 second gap but need hours of duration (DG-fed), and loads that are flexible on both timing and duration where solar offset makes the most economic sense. Loads that get mis-sorted — typically a load assumed to tolerate a gap that actually can't, like a networked access control system or a BMS controller — are where facilities discover gaps in the design months after commissioning rather than at design review.
A DG set's rated runtime is only as real as the fuel actually available on site and the generator's demonstrated ability to start and carry load on demand. Fuel storage capacity needs sizing against the facility's realistic worst-case outage duration, not an optimistic assumption of short outages — and in practice, fuel storage sizing is one of the more commonly under-specified parts of a DG installation, since it's easy to size the generator itself correctly while leaving the fuel tank sized for convenience rather than for the duration the generator is actually meant to cover.Equally important and more commonly skipped is routine load testing — starting the DG set under realistic load on a scheduled basis, not just running it unloaded to confirm it starts. A generator that starts cleanly unloaded can still fail to pick up and carry a real facility load if a problem has developed in the alternator, the governor, or the automatic transfer switch's load-sensing logic — faults that an unloaded start test simply doesn't exercise. Facilities that only discover a DG set problem during an actual outage, rather than during a scheduled load test, are discovering it at the worst possible time.
Not every facility sizes its backup power for 100% of connected load — doing so is sometimes simply not economical, particularly for a facility with occasional very high peak loads that don't coincide with typical outage timing. A load-shedding or priority scheme, where non-critical loads (certain HVAC zones, decorative lighting, non-essential equipment) are automatically or manually shed during an outage to keep the generator or battery within its sized capacity for the loads that actually matter, is a legitimate design approach rather than a compromise to be embarrassed about.Getting this right means the priority list has to be decided and documented in advance, with the switching logic (automatic load-shedding contactors, or a documented manual procedure) actually built and tested — a priority scheme that exists only as an assumption in someone's head, with no actual mechanism to shed the load when an outage hits, isn't a real mitigation. This is also where interval load data from a facility's digital metering becomes useful input, since knowing the actual historical peak demand and its timing is what makes a realistic load-shedding threshold possible to set in the first place.
| Factor | UPS | DG Set | Solar + Battery |
|---|---|---|---|
| Switchover gap | None (online/double-conversion types) | 10-30 seconds (start + ATS transfer) | None for battery-covered loads; depends on inverter config |
| Typical duration | Minutes to a few hours | Hours, limited by fuel supply | Hours, limited by battery capacity and daily solar yield |
| Best suited for | Servers, medical equipment, process controllers | Pumps, lighting, general industrial machinery, multi-hour outages | Daytime grid-draw offset; overnight bridging with the right sizing |
| Main sizing error to avoid | Using nameplate watts instead of real VA draw | Sizing against running current, not motor starting current | Underestimating overnight/no-sun bridging need |
The automatic transfer switch (ATS) is the component that physically decides when to hand load from mains to generator and back, and its own condition and settings matter as much as the generator's rated capacity. An ATS with a transfer time slower than specified, contacts worn from repeated switching over years of service, or sensing thresholds drifted from their original calibration can introduce exactly the kind of switchover delay that a UPS-dependent load can't tolerate — even on an installation where the generator itself is perfectly healthy and correctly sized. Because the ATS only operates during an actual outage or a scheduled test, a developing problem with it can go unnoticed for a long time between genuine triggering events, which is one more reason scheduled load testing (discussed above) needs to exercise the ATS itself, not just confirm the generator starts and runs.
Can a single large UPS replace a DG set for a facility that wants to avoid running diesel generation altogether?
Only for short outages — a UPS sized for multi-hour runtime at meaningful load becomes prohibitively large and expensive in battery capacity compared to a DG set covering the same duration, which is why UPS and DG remain complementary rather than substitutes for most facilities, rather than a straight either-or choice.Does solar+battery reduce the required DG set size, or does the DG still need to be sized for the full load regardless?
In most practical designs the DG set is still sized for the full critical load, since solar generation isn't guaranteed to be available during every outage (a grid failure at night, or during heavy cloud cover, gets no help from solar) — the DG remains the full-coverage layer, with solar+battery reducing average grid draw and fuel consumption rather than reducing the DG's required capacity.How often should a DG set actually be load-tested, and what counts as a meaningful test versus a token one?
Monthly load testing under a realistic percentage of actual facility load — not just an unloaded start — is a common and reasonable baseline for commercial and light-industrial installations, with more frequent testing for critical facilities. A token test that only confirms the engine starts and idles, without the automatic transfer switch actually picking up real load, misses most of the failure modes that matter and shouldn't be treated as equivalent to a genuine load test.Is a load-shedding scheme only relevant for very large facilities, or does it make sense for a smaller commercial site too?
It scales down perfectly well — even a small commercial site with a modest DG set or UPS can benefit from deciding in advance which loads get power during an outage and which don't, rather than discovering during the first real outage that the backup capacity is split unevenly across critical and non-critical loads with no way to prioritise between them in the moment.How much fuel storage is actually enough for a facility that regularly sees multi-hour outages?
This depends entirely on the facility's historical outage pattern and the DG set's fuel consumption rate at expected load, not a generic rule of thumb — reviewing at least a year of actual outage duration data for the site, where available, gives a far more realistic sizing basis than assuming a fixed number of hours of storage is universally adequate.The Central Electricity Authority's guidelines on captive and standby generation, available via the CEA, are a useful independent reference when a facility's backup design needs to be justified against regulatory expectation rather than a single vendor's recommendation.
The question isn't which backup technology is "best" — it's which load tolerates which gap, for how long. Map that first, size each layer to the load it's actually covering, account for starting current and real VA draw rather than nameplate numbers, and the backup stack stops being guesswork.