UPS Runtime Explained: Why Double the Load Gives Less Than Half the Time

Most runtime surprises come from one assumption: that a battery is a fixed bucket of energy, and the load just empties it at a steady rate. A lead-acid UPS battery is not that. How much energy you get out depends on how fast you take it, how warm the battery is, how old it is, and how low the UPS lets the voltage fall before it gives up. This page walks through each factor with numbers, so you can read a runtime chart critically and predict what your own UPS will do.

Step one: how much energy is in the battery

Battery energy in watt-hours is voltage multiplied by amp-hours. A UPS with two 12 V 9 Ah blocks wired in series has a 24 V, 9 Ah battery:

That 9 Ah figure is the catch. Battery makers rate small VRLA batteries at the 20-hour rate: the current that drains the battery in 20 hours (here, 0.45 A). A UPS drains it in 5 to 60 minutes, at currents 15 to 60 times higher. At those rates, the same battery delivers only a fraction of its label capacity, often roughly 50 to 70% at a 10 to 15 minute discharge. That is why UPS-specific "high-rate" batteries also publish a watts-per-cell figure at a 15-minute rate, which is a far better guide to UPS performance than amp-hours. The UPS battery reference table lists common blocks and their nominal energy.

Step two: the inverter takes its cut

The battery's DC energy passes through the inverter, which converts it to 120 V AC. That conversion is not free. On battery, small consumer UPS units are commonly around 80 to 90% efficient, and larger business units can reach the low to mid 90s. Efficiency also drops at very light loads, where the inverter's fixed overhead (control electronics, fan, transformer magnetizing current) becomes a bigger share of the total.

A 150 W load on an 85% efficient inverter pulls about 176 W from the battery. On a 24 V battery that is roughly 7.4 A, and the current is what drives the next effect.

Step three: the Peukert effect

In 1897 Wilhelm Peukert described an empirical rule for lead-acid cells: the faster you discharge them, the less total charge they deliver. The chemistry behind it includes the speed at which sulfuric acid can diffuse into the plates and the voltage drop across the battery's internal resistance, which brings the terminal voltage down to the cutoff point sooner under heavy current.

Peukert's law, in the form most useful for runtime, is:

where H is the rated discharge time (20 h), C is the rated capacity (9 Ah), I is the actual discharge current, and k is the Peukert exponent. A perfect battery has k = 1.0. Modern AGM VRLA batteries are typically around 1.05 to 1.15; older designs and flooded batteries run higher, up to roughly 1.3.

Worked example: one UPS, three loads

Take the 2 x 12 V 9 Ah UPS above, assume 85% inverter efficiency and k = 1.15, and treat battery voltage as a constant 24 V for simplicity. Compare a naive linear estimate (216 Wh x 0.85 / load) with a Peukert estimate:

The arithmetic for the 300 W row: battery power is 300 / 0.85 = 353 W, current is 353 / 24 = 14.7 A, and the ratio C / (I x H) is 9 / (14.7 x 20) = 0.0306. Raising 0.0306 to the power 1.15 gives about 0.0181, and multiplying by 20 h gives 0.363 h, or 22 minutes.

Two things stand out. First, every doubling of load cuts runtime to about 45% of the previous value (0.5 raised to 1.15 is 0.45), not 50%. Second, the naive estimate is wildly optimistic: off by 50% or more at every load.

Step four: where the UPS decides the battery is empty

A UPS never drains its battery to zero. It shuts the inverter off at a low-voltage cutoff, typically somewhere around 1.67 to 1.75 V per cell (roughly 10.0 to 10.5 V per 12 V block) under load. Many units raise this threshold at light loads to avoid deep, damaging discharges, because a battery discharged slowly to 10 V is more deeply discharged than one that hit 10 V under a heavy current.

The cutoff also explains why runtime falls apart under very heavy load. Internal resistance drops the terminal voltage in proportion to current. At 30 A a fresh battery may already sit close to the cutoff, so the inverter gives up while plenty of chemical energy remains. An aged battery with higher internal resistance hits the cutoff even sooner, which is why old batteries often fail a self-test at full load but still run a router for an hour.

Step five: aging

Battery makers and IEEE 1188 conventionally treat a VRLA battery as at end of life when its capacity falls to 80% of rated. That is not a cliff; capacity fades gradually over years and then often declines faster near the end. The right-hand column of the table above shows the effect of an 80% battery: roughly 20% less runtime at every load.

In practice, a 3 to 5 year old consumer UPS battery is frequently below that 80% line, especially in a warm room. See UPS battery lifespan for the aging mechanisms and when to replace a UPS battery for practical tests.

Step six: temperature, both ways

Temperature affects runtime and life in opposite directions, which confuses many owners:

A UPS in a hot closet will often outperform its chart in year one and then die in year two. A UPS in an unheated garage loses runtime in winter but may last longer overall.

Why the runtime estimate on the display jumps around

The "minutes remaining" figure on an LCD or in monitoring software is a model, not a measurement. Typically, the firmware looks up the present load against a stored discharge curve, scaled by an estimate of battery health. That leads to some familiar behavior:

• The estimate swings with load. A PC that idles at 80 W and spikes to 350 W while compiling or gaming will show a runtime that jumps by a factor of four or more within seconds.
• The estimate drifts high as the battery ages. Unless the UPS has run a calibration or seen a deep discharge recently, it may still assume near-original capacity.
• The estimate collapses near the end. On battery, the remaining-time figure often drops faster than real time, because the voltage curve steepens as the battery empties.

A runtime calibration resets the model against the battery's real behavior. It is a deep discharge, so do it occasionally rather than routinely.

A bench test you can do

The only number you can fully trust is one you measured on your own equipment. A controlled test takes about an hour, most of it waiting:

1. Charge fully. Leave the UPS on utility power for at least 8 to 24 hours (check the manual for recharge time).
2. Use a predictable load. Something steady and non-critical, such as an incandescent or halogen work lamp or a fan of known wattage, avoids losing data if the battery fails early. Confirm the wattage with a meter, as described in how to measure power draw.
3. Simulate the outage at the wall. Unplug the UPS input (or switch off its circuit) and start a timer.
4. Stop at a safe point. For routine testing, stop when the UPS signals low battery rather than running to cutoff, and note the time.
5. Compare. Check the result against the manufacturer's chart at the same load. A result below about 60 to 70% of the chart figure suggests an aging battery.

What this means for buying and planning

The nonlinearity works in your favor when you oversize. Because lead-acid batteries are more efficient at lower discharge rates, buying a UPS with twice the battery energy you think you need does not merely double runtime; it typically more than doubles it. The same logic makes external battery packs unusually effective.

It also means you cannot compensate for a heavy load by assuming "a bit more battery." If your plan depends on 10 minutes at 600 W, you need a unit whose chart shows comfortably more than 10 minutes at that load with a new battery, so that the plan still holds in year three. Work out your target first with how much runtime you need, then size the battery to reach it at end of life, not at purchase.

Lithium iron phosphate batteries change some of this. They have a much smaller rate effect and a flatter voltage curve, so their runtime scales closer to linearly with load. See lithium UPS pros and cons for the trade-offs.