UPS Runtime Calculator

Introduction to UPS runtime and battery backup time

A UPS runtime estimate answers one question: with the utility gone, how many minutes does the battery in your uninterruptible power supply actually hold the connected equipment up? That number decides whether the UPS is a shutdown timer for a workstation, a ride-through for the ten-second gaps that trip a modem, or a genuine bridge to a standby generator. This calculator turns battery capacity, connected load, inverter efficiency and battery state of charge into a runtime in minutes, and it applies the Peukert correction that most simple estimators leave out.

UPS batteries are usually described in watt-hours (Wh), though many spec sheets list volt-amp hours (VAh) or give nominal voltage and an amp-hour rating separately. If you know the battery voltage and the amp-hour rating, multiplying them gives watt-hours, and the battery-count field scales that figure for a string of identical blocks. The load figure should reflect the equipment that will genuinely be drawing power during the outage, measured with a plug-in meter where possible, not the nameplate maximum of everything you own.

Inverter efficiency covers the energy lost while the UPS rebuilds battery DC into the AC output your devices expect. Most consumer and small-rack units fall between 0.85 and 0.95 on battery, and 0.90 is a sensible default when the manufacturer does not publish a figure. Divide the connected load by that efficiency and you get the power the inverter actually pulls out of the battery, which is always larger than the load itself.

The part that surprises people is the battery's own behaviour. A lead-acid amp-hour rating is measured at a slow, gentle discharge, normally over 20 hours. A UPS empties that same battery in ten or twenty minutes, and at that rate the chemistry simply cannot keep up: the plates polarise, the electrolyte near the active material is depleted faster than it can diffuse back, and the terminal voltage hits its cutoff while a good deal of charge is still sitting in the battery. The result is that a heavy load does not merely drain the battery proportionally faster; it also shrinks the effective capacity. That is what Peukert's law describes, and it is the single biggest reason a naive watt-hours-divided-by-watts estimate is optimistic.

Because the penalty grows with load, shedding equipment early is worth much more than shedding it late. Dropping 400 W in the first minute of an outage does not just remove 400 W of drain, it moves the whole battery onto a gentler part of its discharge curve for the entire remaining runtime. The load-shedding game further down this page is built on exactly the same equation the calculator uses, so you can feel that effect instead of reading about it.

How to use this UPS runtime calculator

1. Enter the battery capacity in watt-hours for a single battery, or leave that field blank and instead supply battery voltage and the amp-hour rating so the calculator can build the watt-hour figure from the label. Set battery count to the number of identical blocks in the string; it multiplies whichever capacity path you used.
2. Set the rating hour rate to the discharge time that the capacity label refers to. Sealed lead-acid amp-hour ratings are usually quoted at the 20-hour rate; some UPS datasheets quote a 10-hour or even 5-hour rate, and using the wrong one shifts every result.
3. Choose a Peukert exponent. Use 1.20 for an ordinary sealed lead-acid or AGM UPS battery, 1.05 to 1.15 for a new high-rate AGM, up to 1.4 or more for a tired flooded string, and about 1.03 for lithium-ion. Entering 1.00 disables the correction and reproduces the classic textbook formula.
4. Enter the battery state of charge. A UPS that has just ridden through an earlier sag, or that is still recharging, does not start the next outage at 100%.
5. Enter the connected load. Switch the unit selector to VA if your only figure is a volt-amp nameplate, then give the load's power factor so the calculator can convert VA to watts.
6. Set the inverter efficiency between 0 and 1, then read the runtime. The result updates as you type, so you can drag a load figure up and down and watch how much sooner the Peukert-corrected runtime collapses than the ideal one.

The UPS runtime formula with Peukert's law

Start with the ideal energy balance that most calculators stop at. Usable battery energy divided by the power the load demands gives runtime in hours:

That expression is only correct when the discharge is slow enough to match the rate at which the capacity was rated. Peukert's law supplies the correction. In the practical form that uses a rated discharge time, runtime is

where H is the hour rate the capacity was measured at, C_rated is that rated capacity, I is the actual discharge current and k is the Peukert exponent. A UPS runs a constant-power load rather than a constant current, but over the working part of a lead-acid discharge the terminal voltage is roughly flat, so current is proportional to battery-side power. Substituting energy for charge and power for current gives the form this calculator evaluates:

Two things fall straight out of this. First, when k equals 1 the expression collapses to the ideal formula, so the classic result is just the special case of a perfect battery. Second, the ratio between the corrected and the ideal runtime is the capacity factor

which is the fraction of the label capacity the battery will really hand over at that discharge rate. The calculator reports it directly, because it is the number that tells you whether your string is being asked to work gently or brutally.

Worked example: a 720 Wh UPS carrying a 240 W rack

Take a small rack UPS with two 12 V, 30 Ah AGM blocks. That is 12 × 30 = 360 Wh each, and two of them give 720 Wh of rated energy at the 20-hour rate. The connected load is a router, a firewall, a switch and a small server that together measure 240 W on a plug-in meter. The inverter is 90% efficient on battery, the string is fully charged, and the batteries are ordinary AGM, so a Peukert exponent of 1.20 is reasonable.

• Battery-side power: 240 W ÷ 0.90 = 266.7 W drawn from the string.
• Power at the rated hour rate: 720 Wh ÷ 20 h = 36 W. The UPS is therefore discharging about 7.4 times faster than the rate its label was measured at.
• Ratio inside the bracket: 720 ÷ (20 × 266.7) = 0.135.
• Capacity factor: 0.135^0.20 = 0.670, so the battery will deliver only about 67% of its rated energy — roughly 482 Wh of the nominal 720 Wh.
• Runtime: 20 × 0.135^1.20 = 1.81 h = 109 minutes.

The ideal formula would have said 720 × 0.90 ÷ 240 = 2.70 h, or 162 minutes. The Peukert correction removes 53 minutes, a third of the answer, and that gap is not a rounding error — it is the difference between planning a controlled two-hour ride-through and finding your rack dark an hour early. Now shed the 240 W load down to 120 W: battery-side power halves to 133.3 W, the capacity factor rises to 0.770, and runtime becomes 250 minutes. Halving the load did not double the runtime, it multiplied it by 2.3. That extra 15% is the Peukert effect paying you back for shedding.

Watts, volt-amps and the power-factor trap

UPS marketing quotes volt-amps, but batteries are drained by watts. Volt-amps are apparent power, the simple product of RMS voltage and RMS current; watts are real power, the part that actually does work. The ratio between them is the power factor, and for a UPS the relationship is straightforward: watts equal volt-amps multiplied by power factor. A 1500 VA UPS with a 0.6 output power factor can only carry 900 W, and a modern unity-power-factor unit of the same VA rating carries 1500 W.

The trap runs both ways. Feed a VA nameplate into a runtime formula as though it were watts and you will overstate your load and understate your runtime, sometimes badly, because much IT equipment with active power-factor correction sits at 0.95 to 0.99 rather than the 0.6 assumed by older rules of thumb. Take the opposite shortcut, assume every device runs at 0.6, and you will underestimate a modern server rack by a third. That is why this calculator lets you enter the load in VA and supply the power factor explicitly rather than guessing on your behalf, and why the load-shedding game includes a level where the rack elevation is labelled in volt-amps.

Converting UPS specs when the battery label is incomplete

Runtime estimates still work when the battery documentation is thin, as long as you can assemble a defensible watt-hour figure. If you have nominal voltage and amp-hours, multiply them. If the datasheet quotes VAh instead of Wh, remember that VAh omits the power factor and is therefore not identical to Wh, but it will put you in the right neighbourhood. Where a UPS uses several identical blocks, the count field scales the estimate; a series string raises the bank voltage while a parallel arrangement raises the amp-hours, and either way the total watt-hours are what the runtime formula needs.

Load estimation is usually the harder half of the problem. A desktop idles gently and then spikes when a backup job starts or a GPU wakes; a network switch or modem sits nearly flat all day. Measuring the true draw with a plug-in meter produces a far better runtime estimate than trusting the maximum wattage printed on a power supply, which is a rating for the supply and not a description of the machine. When you only have a rough figure, round it up so the runtime stays conservative.

Battery chemistry, age and temperature

Chemistry sets the Peukert exponent. Most compact UPS units still use sealed lead-acid or AGM cells because they are cheap, predictable and easy to maintain, and those sit around 1.10 to 1.25. Flooded lead-acid runs higher and degrades further with age. Lithium-ion UPS batteries, increasingly common in higher-end gear, hold their capacity almost independently of discharge rate over the range a UPS uses, which is why an exponent near 1.03 is appropriate and why a lithium UPS keeps much more of its nameplate runtime under a heavy load.

Age matters just as much as chemistry. A string that once matched its label can lose a large slice of capacity after years of float charging, warm cabinets or repeated deep discharges, and the loss shows up first at high discharge rates. The practical way to represent an aged battery here is to lower the capacity and raise the Peukert exponent together. Temperature moves the answer too: lead-acid capacity falls noticeably below about 20 °C and rises slightly above it, at the cost of a much shorter service life. Heat is the reason a cramped cabinet or a pile of equipment sitting on the UPS quietly works against you.

Choosing a UPS from a runtime target

Sizing works best backwards. List the equipment that must stay up, note each device's real wattage, and decide how long each tier needs to survive: a router and modem might need to hold a site online for an hour, while a workstation only needs enough time to flush a write cache and stop cleanly. Once you know load and target duration, rearrange the formula to find the rated watt-hours you need, remembering to divide by the capacity factor rather than assuming you will get the label figure.

Most sites end up with a tiered plan. Critical network gear gets the largest UPS or a generator bridge, secondary machines get enough battery to shut down in an orderly way, and everything else is dropped at the moment of the outage. The calculator is useful here because you can swap in different load and capacity combinations and see which arrangement still clears your target once the Peukert correction is applied. It is also worth remembering that adding external battery packs raises the watt-hours without raising the inverter's rating, so the load figure remains the term with the most leverage.

Limitations and assumptions behind the estimate

These are the assumptions you are accepting when you use the result:

• Constant load. The runtime is computed for a steady draw. Real racks fluctuate, and a load that spikes for a minute costs more battery than its average suggests, precisely because of the Peukert term.
• Roughly flat terminal voltage. Converting Peukert's current-based law into the power form used here assumes battery voltage does not sag much over the working part of the discharge. That is a good approximation for lead-acid down to the usual cutoff and an excellent one for lithium, but it breaks down in the last few percent.
• A healthy string at 25 °C. No allowance is made for temperature, cell imbalance, or a battery that has not fully recharged since the previous event beyond what you enter as state of charge.
• Fixed inverter efficiency. Real inverter efficiency varies with load, usually falling at very light loads. A single figure is a simplification.
• No surge headroom. A motor, laser printer or fridge can draw several times its running wattage at startup. The UPS may trip on that surge even when the steady-state runtime looks comfortable.
• Nothing about the shutdown threshold. Most UPS units signal a shutdown well before the battery is empty. Your usable window is the runtime minus whatever margin your shutdown software claims.

The safest way to close the gap between the estimate and reality is a discharge test: simulate an outage with the real load connected, time how long the equipment survives, and compare. Repeat it annually, because the answer moves as the battery ages. Batteries also carry real hazards, so follow the manufacturer's service instructions and disconnect the unit from utility power before opening it.

Common UPS runtime mistakes to avoid

The recurring errors are easy to name. People take the label capacity at face value and forget both the conversion losses and the Peukert derating. They confuse volt-amps with watts in one direction or the other. They omit network gear, USB accessories and external drives that quietly add fifty or a hundred watts. They forget that the amp-hour rating on a lead-acid block refers to a 20-hour discharge that has nothing in common with what a UPS does. And they plan against the lowest believable load instead of the highest, which is the wrong direction for a safety margin. A careful inventory of what is really plugged in, measured rather than guessed, fixes most of these at once.

Frequently asked questions about UPS runtime

UPS runtime at a glance for common loads

This table compares the ideal energy-balance runtime with the Peukert-corrected runtime for the same battery at the same load, at 90% inverter efficiency, a 20-hour rating rate and a Peukert exponent of 1.20. The gap widens as the load rises, which is the whole point: the heavier the draw, the more of the label capacity the battery simply refuses to give you.

Note: the ideal column is , with watt-hours equal to battery voltage times amp-hours; the corrected column applies the Peukert factor described above. Both assume a constant load, a fully charged healthy string and a 25 °C ambient, so they are planning references rather than promises of battery life.

Sources and further reading

• Peukert's law, including the practical t = H(C/IH)^k form and typical exponents by battery construction — Peukert's law reference summary.
• D. Doerffel and S. A. Sharkh, "A critical review of using the Peukert equation for determining the remaining capacity of lead-acid and lithium-ion batteries", Journal of Power Sources 155 (2006) 395–400 — doi:10.1016/j.jpowsour.2005.04.030.
• IEEE Std 1184, IEEE Guide for Batteries for Uninterruptible Power Supply Systems, on sizing, selection and discharge testing of UPS battery strings — IEEE Standards Association.
• ENERGY STAR program requirements for uninterruptible power supplies, including minimum average efficiencies and the 90% minimum power factor for VI and VFI units — ENERGY STAR.
• Schneider Electric, "Watts vs. VA: what's the difference anyway?", on apparent power, real power and UPS nameplate ratings — Schneider Electric blog.
• Battery University BU-402, "What is C-rate?", on how discharge rate is expressed and why rated capacity depends on it — Battery University.