Two methods, different use cases
Two industry methods exist for estimating UPS runtime, and they answer slightly different questions.
Watts-Per-Cell (WPC) — the dealer's tool
Watts-per-cell is what every UPS OEM sizing tool uses internally — Riello GUS, Schneider Battery Selector, Eaton 9PX/9SX sizing, Vertiv Liebert configurator. Each battery model has a published discharge table that lists, for that exact part number, the watts available per 2 V cell at different runtime durations (typically 5, 10, 15, 30, 60, 120, 240, 480 minutes). To use it: take your load, divide by the number of cells in the bank, look up the runtime in the table for that WPC value.
This calculator uses a typical WPC curve calibrated against Yuasa NPL, Panasonic LC-P, CSB GP/HRL, and Eaton 9SX datasheet medians for VRLA AGM, and against CATL / EVE / BYD commercial datasheets for LiFePO4. Capacity scaling is linear: a 200 Ah block has twice the watts-per-cell of a 100 Ah block at the same runtime. For mission-critical sizing, always cross-check against the actual datasheet of the battery you're specifying.
Peukert — the academic tool
The Peukert equation models effective Ah capacity as a function of discharge current: Ah_effective = Ah_rated × (I_rated / I_actual)^(k − 1). With I_rated at the C/20 reference. Typical Peukert exponents: 1.10 for VRLA AGM, 1.15 for Gel, 1.30 for flooded, 1.03 for LiFePO4 (essentially flat). The model is accurate at low-to-moderate C-rates and gets optimistic above 1C because it doesn't account for terminal-voltage sag or end-of-discharge cutoff. Use it for academic comparison or when you need to model a non-standard cell.
When the two methods diverge
For typical UPS sizing scenarios (C/3 or slower discharge) the two methods agree within ~20%. For aggressive discharges above 1C, Peukert can over-predict runtime by 2–5×. WPC stays closer to real-world measurements because it's derived from actual datasheet pull tests at each discharge rate. If your scenario is at C/1 or faster (a small battery hit hard), trust the WPC method and treat Peukert as a sanity check.
Why dealers care about runtime
Customers don't buy “watt-hours” — they buy minutes. They specify the runtime they need (15 minutes to gracefully shut down servers; 30 minutes for the generator to start; 4 hours for a small office UPS) and expect you to deliver the battery sizing that achieves it. Sizing too generously wastes their money and your margin; sizing too tight risks a callback when the UPS drops the load before the customer expects. This calculator gives you a defensible starting point. Power Stack's installed-base tracker then monitors aging across your fleet so you know when each customer's real runtime starts dropping below their spec.