How to Calculate UPS Battery Runtime: A Worked Example

Runtime cannot be read off the amp-hour label on a battery block. The method that produces a defensible figure converts the protected load into watts per cell, reads that value against the manufacturer's constant-power discharge table, and then applies three corrections for ageing, room temperature and design margin. Done properly, a string that looks like an hour on paper usually delivers forty to fifty minutes late in its service life, and that lower number is the one a design should be built on.

Step One: Turn the Load Into Direct Current Watts

Begin with the measured load rather than the cabinet rating. A 40 kVA unit with unity output power factor running at seventy percent is supporting 28 kW. The inverter pulls more than that from the battery because conversion is not free, so dividing by an inverter efficiency of 0.95 gives about 29.5 kW leaving the direct current bus. Use metered figures, never the sum of nameplate ratings on connected equipment, which commonly overstates reality by forty percent or more. If load growth is expected within the battery's service life, add it at this stage instead of assuming it will be absorbed later.

Step Two: Divide by the Number of Cells

Discharge tables are published per cell, since that is the only figure independent of how a string happens to be wired. A 480 volt nominal string assembled from forty 12 volt blocks contains 240 cells of two volts. Dividing 29,500 watts by 240 gives 123 watts per cell. That single value is what the table is read against, and it explains why two installations with identical kilowatt loads and identical total amp-hours can behave very differently: a longer string working at a lower rate per cell always outlasts a short string worked hard.

Step Three: Read the Constant-Power Table

Manufacturers tabulate watts per cell against duration at a stated end voltage, normally 1.67 volts per cell, and a reference temperature near twenty-five degrees Celsius. A representative 100 amp-hour valve-regulated block delivers approximately 355 watts per cell for fifteen minutes, 222 for thirty, 129 for sixty and 94 for ninety. Our requirement of 123 watts per cell sits between the sixty and ninety minute columns and interpolates to roughly sixty-six minutes. Notice what just happened: a battery marketed as a hundred amp-hour unit delivered nothing resembling a hundred amp-hours, because a high-rate discharge extracts far less energy than the twenty-hour rating suggests.

Step Four: Apply the Three Corrections

Sixty-six minutes is a laboratory result for a new string in a controlled room. Three factors drag it back to reality. Ageing comes first: end of service life is conventionally defined as eighty percent of rated capacity, so multiplying by 0.8 yields fifty-three minutes. Temperature comes second, and a room actually held at twenty degrees rather than twenty-five costs a further eight to ten percent, taking the figure to about forty-eight minutes. Design margin comes third: reserving ten percent for future load leaves a defensible forty-three to forty-five minutes. Publish that number rather than sixty-six and the installation will still satisfy its specification in year five.

Why Amp-Hours Mislead at UPS Discharge Rates

The twenty-hour rating assumes a gentle trickle. Pull the same block down in fifteen minutes and usable energy collapses to somewhere near forty-five percent of nominal, an effect captured by Peukert's relationship, whose exponent for lead-acid construction falls between about 1.1 and 1.3. Dividing amp-hours by amps is therefore arithmetic without physical meaning at the rates a UPS imposes. Lithium iron phosphate behaves far more linearly, retaining well above ninety percent of rated energy at a one-hour rate, which is one reason lithium strings sized on paper more often deliver what the spreadsheet promised.

What to Do When the Answer Is Too Short

Three levers exist and they are not equivalent. Adding a parallel string halves the watts per cell, and because the discharge curve is nonlinear this more than doubles the result: sixty-one and a half watts per cell in the example above moves the string into the region of two and a half hours. Fitting larger blocks within the same string achieves something similar in less floor space but leaves a single failure path. Cutting the protected load is the cheapest lever and the one most often overlooked, since moving non-essential circuits off the critical bus costs nothing. Past roughly thirty minutes, a standby generator with a short battery bridge beats an extended string on cost per protected kilowatt-hour almost every time.

End-voltage conventions, replacement intervals and other sizing questions are answered in the technical FAQ, and discharge data for each frame we supply sits alongside the specifications in the UPS product catalogue. If interpolating a table by hand holds no appeal, send the metered load, the autonomy you need and the room temperature, and our engineers will return a string configuration with every correction factor shown in the working.

Key takeaway: convert the load to watts per cell, read the constant-power table at the correct end voltage, then derate for eighty percent end-of-life capacity, actual room temperature and planned growth. A runtime quoted straight from an amp-hour label is a number the installation will never actually see.

Frequently Asked Questions

How much autonomy does a site genuinely need?
Where a standby generator exists, five to ten minutes covers start, stabilise and transfer with margin for a failed first start. Without a generator, fifteen to thirty minutes allows an orderly shutdown of servers and process controls. Healthcare and life-safety installations are governed by local codes that set the minimum explicitly, and those override any commercial judgement.

Are parallel battery strings a problem?
Up to four strings in parallel is routine practice, subject to conditions: identical type, capacity and manufacturing date, individual string fusing, and cable runs of matched length so current shares evenly. Never add a new string to an aged one, because the older string will be driven harder and both will finish early.

Why has runtime halved after three years?
Heat is the usual culprit, followed by float voltage set too high and by repeated deep discharges. Trending impedance quarterly against the commissioning baseline identifies the weak block long before an autonomy test does, and one failed block in a series string caps the entire string's performance regardless of how healthy its neighbours are.