UPS runtime calculator: how long a battery backup keeps your network up
- Load
- 96 W, 21 % of the 450 W rating
- Battery bank
- 12 V, 9 Ah, 108 Wh
- Battery current
- 9.4 A, 1.05 C
- Estimated runtime
- 27 min
- Likely range
- 20 min to 40 min
- Wh / W alone
- 57 min (too optimistic)
- At half the load
- 1 h 04 min
- At double the load
- 11 min
- For 22 min
- 7.7 Ah at 12 V (92 Wh): this bank covers it
# An estimate from the battery for new, fully charged batteries near 25 °C. Makers' charts differ by up to two times at the same load, so check yours at your watts.
Enter what your gear draws and what is inside the UPS, and the calculator estimates how long the battery carries that load, gives the range real units land in, and shows what half or double the load would do. Add the runtime you want and it works out the battery that target needs. Everything runs in your browser.
How it works
A UPS on battery turns battery current into mains power through an inverter, and the battery does not care about the watt figure on the box. It cares about current, so the calculation starts there:
bank voltage = batteries in series * volts per battery
capacity C = amp-hours per battery * strings in parallel * health
battery current = load watts / (bank voltage * inverter efficiency)
runtime (hours) = H * (C / (current * H)) ^ k Peukert's law
never more than C / current
battery for t = current * H * (t / H) ^ (1 / k)
H is the number of hours the battery's amp-hour rating is quoted over: 20 for sealed lead-acid, 5 for lithium iron phosphate. The exponent k is the part people leave out. A lead-acid battery rated 9 Ah over 20 hours (0.45 A) gives far fewer amp-hours when a UPS pulls 10 or 20 A out of it, and k describes how fast that loss grows. At k = 1 runtime would be plain watt-hours divided by watts. Real UPS figures need a lot more than 1.
The calculator uses k = 1.25 for its estimate and 1.12 to 1.35 for the likely range. Those values come from published runtime figures for 1500 VA units on two 12 V 9 Ah batteries. CyberPower lists the CP1500PFCLCD at about 10 minutes at 500 W and 2.5 minutes at 1000 W, and other makers' figures for the same battery and load differ by up to two times, so a single number would claim more precision than any chart has. Lithium iron phosphate holds its capacity much better at high current: k = 1.05, with a narrow range.
The defaults are 85 % inverter efficiency and 100 % battery health. The output also shows the straight watt-hours figure (usable Wh times efficiency, divided by the load) so you can see how far off the simple sum is, and the battery current as a C rate: 1 C drains the rated amp-hours in an hour on paper.
Worked example: a network closet
The panel opens with a typical closet: a cable modem, a router, a PoE switch with two access points, a Raspberry Pi and a small NAS, 96 W on a plug-in meter, on an 850 VA desktop unit rated 450 W with one 12 V 9 Ah battery:
battery current = 96 / (12 * 0.85) = 9.41 A (1.05 C)
runtime = 20 * (9 / (9.41 * 20)) ^ 1.25 hours
= 0.447 h = 27 min
range = k 1.35 to 1.12 = 20 to 40 min
Wh / W alone = 108 * 0.85 / 96 * 60 = 57 min
The output reads Load 96 W, 21 % of the 450 W rating; Estimated runtime 27 min; Likely range 20 min to 40 min. The straight sum promises 57 minutes, more than twice the estimate. Halve the load and runtime goes to 1 h 04 min; double it and it drops to 11 min. That asymmetry is why shutting a NAS down early buys the rest of the closet so much: at 65 W without it, the same battery gives an estimated 44 minutes.
With 22 minutes wanted, the last line reads 7.7 Ah at 12 V (92 Wh): this bank covers it.
Worked example: a bigger lab on a 1500 VA unit
Add a mini tower hypervisor and the meter shows 256 W. On a 1500 VA, 1000 W unit with two 12 V 9 Ah batteries in series (24 V):
| Case | Battery current | Estimate | Likely range |
|---|---|---|---|
| New battery | 12.5 A | 19 min | 13 to 29 min |
| Battery at 80 % health | 12.5 A | 14 min | 9.9 to 22 min |
| One external pack (2 strings) | 12.5 A | 45 min | 34 min to 1 h 03 min |
Two things stand out. An aging battery costs more than its lost capacity, because the remaining amp-hours are drained harder (1.74 C instead of 1.39 C). And a second string more than doubles the runtime, 19 to 45 minutes, because each battery now supplies half the current and wastes less of its charge.
Ask for 15 minutes with the 80 % battery and the calculator wants 9.4 Ah at 24 V, more than the 9 Ah inside, so it suggests 2 strings of 9 Ah: an external pack, or a new battery and a slimmer load.
Picking a unit from the numbers
Size the watt rating first: about twice the measured load leaves room for a startup surge and for runtime. Then use the estimate and the low end of the range to shortlist units, and check the maker's runtime chart at your watts before you buy, since that chart is the only figure that knows the inverter and cutoff voltage inside that model. Once it is installed, battery.runtime from NUT gives you the unit's own reading to compare against; monitoring a UPS with NUT covers the setup and the shutdown order.
Limits and gotchas
- The estimate assumes new, fully charged batteries near 25 °C. Cold cuts lead-acid capacity, heat shortens its life, and a battery that has not finished recharging after an outage will not give the full figure on the second one.
- UPS units stop at a low-battery cutoff and some reserve the last minutes for a clean shutdown signal. If your shutdown software starts at a low-battery flag, the usable time before that flag is shorter than the total runtime.
- Inverter efficiency changes with load. Small units run well below their rating are often less efficient than 85 %, so at very light loads try 75 to 80 % and see how much the answer moves.
- Strings in parallel means matching batteries at the same voltage, such as a maker's external battery pack. Adding a battery in series changes the voltage and only works where the UPS is built for it.
- Lithium units and power stations often limit the current their battery management allows. If a power station is rated 300 W continuous, the runtime above 300 W is zero, whatever the arithmetic says.
- PoE gear on the battery: count the switch's draw from the wall, which already includes what it sends to access points and cameras. The PoE calculator works out that budget, and the uptime and SLA calculator turns outage minutes into availability.
Questions
How do I calculate UPS runtime?
Work out the battery current: load watts divided by the bank voltage times inverter efficiency. Then apply Peukert's law, runtime = H * (C / (I * H)) ^ k, with C the amp-hours, H the rated hours (20 for lead-acid) and k about 1.25 for UPS batteries. Dividing watt-hours by watts overstates runtime by about two times.
How long will a 1500 VA UPS last?
It depends on the load and the battery inside. With two 12 V 9 Ah batteries, published figures are about 10 minutes at 500 W and 2.5 minutes at 1000 W. At 256 W the calculator estimates 19 minutes, with a likely range of 13 to 29.
Why does doubling the load more than halve the runtime?
Lead-acid batteries deliver fewer amp-hours the faster they are drained, so a higher current loses both to the extra load and to the lower usable capacity. In the closet example, 96 W runs about 27 minutes and 192 W about 11.
Does an external battery pack double the runtime?
More than double, at the same load. Two strings share the current, each battery is drained more gently, and in the 256 W example runtime rises from 19 to 45 minutes.
How much runtime does an old UPS battery lose?
At 80 % health the 256 W example drops from 19 to 14 minutes, a quarter less, not a fifth, because the smaller capacity is also drained harder. Plan with a margin of about 25 % for a battery that will age, and replace it when the runtime at a steady load keeps sliding.