Your UPS Is Not for Power Cuts. It Is for the 200 Milliseconds Before the DG Wakes Up.
The costliest power event in most factories is not the two-hour outage — the DG covers that. It is the 200-millisecond dip when a grid fault passes through, or the 10-second gap while the AMF panel starts and transfers the genset. PLCs reboot, drives fault on undervoltage, servers hard-stop, a batch in process becomes scrap. The DG never even got a chance to help.
That window — milliseconds to a few minutes — is the actual job of the UPS. Sizing one starts with naming the loads that cannot ride through it.
Step 1: Separate the loads honestly
- No-break loads: PLCs and control panels, servers and network racks, safety and fire-alarm systems, critical instrumentation, that one CNC controller that loses its zero. These go on UPS.
- Short-break loads: motors, compressors, HVAC, lighting — things that restart cleanly after the DG picks up. These do not go on UPS; putting them there is how UPS budgets explode.
- The awkward middle: VFDs driving processes that cannot hiccup. Sometimes the answer is UPS on the drive's control supply only; sometimes ride-through capacitor modules on the drive; occasionally full UPS on the machine. Decide per machine, not per catalogue.
Step 2: Choose the topology for the duty, not the brochure
IEC 62040-3 classifies UPS output behaviour; in practice three families matter:
| Topology | How it behaves | Where it belongs |
|---|---|---|
| Online double-conversion (VFI) | Load always fed through rectifier→inverter; zero transfer time; full isolation from input frequency/voltage garbage | PLCs, servers, instrumentation, anything batch-critical — the industrial default |
| Line-interactive (VI) | Conditions voltage, switches to inverter on failure — a small but real transfer time | Non-critical IT, office loads; marginal on twitchy industrial electronics |
| Standby (VFD class) | Passes mains through, jumps to inverter on loss | Desktops. Not a factory device. |
In electrically noisy plants — big drives, welders, compressors sharing the bus — the double-conversion unit is also functioning as a power conditioner, which is quietly half its value.
Step 3: Size the power and the minutes separately
- kVA vs kW: modern IT loads run near unity power factor; industrial control mixes don't. Size against measured kW and kVA with the UPS's own output PF rating in view, and keep loading in the healthy band — a UPS at 95% load has no headroom for inrush, and one at 20% load wastes efficiency and capital.
- Runtime is a bridge, not a battery farm: with a reliable DG behind it, the UPS needs minutes — enough to cover start + transfer with margin, or to shut processes down gracefully if the DG fails to start. Sizing 60 minutes of battery "to be safe" triples cost for protection the DG already provides. No DG? Then runtime = graceful-shutdown time, calculated per process, honestly.
- Battery chemistry: VRLA remains the low-capex default but is temperature-fragile — its life halves fast in hot rooms, which is why "5-year" batteries die in 2 in an unconditioned Indian plant room. Lithium (LFP) costs more upfront, tolerates heat better, lasts more cycles and weighs a fraction — over ten years the comparison is closer than the purchase order suggests. Whichever you choose: condition the battery room and log its temperature.
- Redundancy: for genuinely critical plants, two smaller units in N+1 beat one big unit whose single failure is a plant event. Modular UPS makes N+1 affordable at moderate sizes.
The maintenance that decides whether it works on the day
- Quarterly: battery health check (impedance/conductance trending beats voltage-only), fan and alarm check, load reading logged.
- Annually: a real transfer test under controlled conditions — pull the input and watch it carry. A UPS that has never been tested is a hope, not a system.
- Watch the bypass: a UPS quietly sitting in bypass after a past fault protects nothing; make bypass status a monitored alarm.
Battery-runtime math is the same physics as any storage sizing — load × minutes ÷ usable capacity. Our Battery Storage calculator lets you sanity-check bank sizes and see how depth-of-discharge changes the answer; for the wider backup picture, the DG Sizing and DG Running Cost calculators complete the chain the UPS bridges into.
What we do differently
Our electrical scope designs the no-break layer from a measured load walk-down, not a nameplate list — topology per duty, runtime justified in minutes, battery room thermally designed, and the annual transfer test written into the AMC so the system is proven, on record, every year. One team owns grid, DG, solar and UPS — so the transfers between them actually work.
The three takeaways
- The UPS's job is the gap the DG cannot cover — size loads and minutes for that, nothing more.
- Online double-conversion is the industrial default; everything else needs a reason.
- Batteries die of heat and neglect — condition the room, trend the health, test the transfer annually.
Scrapped a batch to a half-second dip? Book a Free Project Blueprint & Statutory Approvals Roadmap or call +91 70099 87817 — we'll walk the loads with you and size what actually needs protecting.
More insights
Nothing Has Tripped in Years. That Is Not the Same as Safe.
Factories treat an uneventful electrical system as a healthy one. An audit measures what an absence of incidents cannot tell you — and the findings are usually the same handful of things, quietly getting worse.
ElectricalThe Building Isn't Open Yet. The EV Mandate Already Applies.
Delhi's draft EV Policy 2026 requires all new infrastructure projects to be EV-charging ready from day one — not retrofitted later, designed in from the start. Delhi has roughly 8,850 charging points against a stated need of over 36,000, and the policy is closing that gap by pushing the requirement upstream, onto every new building's electrical design. If you're breaking ground on a new project, this isn't an optional add-on anymore.