If your rack draws 600 watts, a typical 1500VA rack UPS will keep it alive for roughly 10 to 16 minutes — enough for a graceful shutdown, not an afternoon of uptime. The honest shortcut formula is: Runtime (minutes) ≈ (battery voltage × amp-hours × 0.5) ÷ load in watts × 60. That 0.5 is doing a lot of work: a lead-acid battery bank only delivers about half its nameplate watt-hours at real rack discharge rates once you account for the Peukert effect, inverter losses, and the UPS cutting off before the batteries are truly empty. The rest of this guide shows you the full calculation, step by step, with a worked example from a real homelab rack — so you can size your own backup instead of trusting the optimistic number on the box.
Key Takeaways
- UPS runtime depends on watts, not VA. VA sizes the inverter; watt-hours in the battery decide how long it runs.
- The naive formula (battery Wh ÷ load) overstates runtime by roughly 2–3x. Derate for discharge rate, inverter efficiency, and battery age.
- A 1500VA rack UPS at a 600W load gives about 10–16 minutes. That covers a graceful shutdown, not a long outage.
- Need 30 minutes or more? You need an external battery pack or a bigger battery bank — not a bigger VA rating.
- Always cross-check your maths against the manufacturer’s runtime chart before you buy.
What UPS Runtime Actually Means (and What It Doesn’t)
Runtime is the time your UPS can carry the rack on battery after the mains fails. It is not the same as the UPS’s VA rating, and the two numbers get confused constantly. A 1500VA UPS and a 3000VA UPS can have near-identical runtimes if they ship with the same battery bank — the bigger one can power a heavier load, but it does not store more energy.
Runtime is also not a fixed property of the UPS. The same unit gives 35 minutes at a 300W load and under 10 minutes at 900W. Any runtime figure without a load figure next to it is meaningless, which is why the charts in UPS datasheets always list runtime at half load and full load.
And finally, the runtime printed on the box assumes new batteries at 25°C. A three-year-old battery in a warm rack room will deliver noticeably less. If your plan depends on squeezing every advertised minute, your plan is already wrong.
The Runtime Formula: The Naive Version vs the Real One
The formula everyone quotes first is the naive energy balance:
Runtime (hours) = battery energy (Wh) ÷ load (W)
Battery energy is voltage × amp-hours. So a 24V bank of 18Ah holds 432Wh, and at a 600W load the naive answer is 432 ÷ 600 = 0.72 hours, or about 43 minutes. That number is wrong, and dangerously so — it is nearly three times what you will actually get.
The realistic version adds the losses the naive formula ignores:
Runtime (hours) = (battery Wh × discharge-rate factor × inverter efficiency × age factor) ÷ (load + UPS self-consumption)
Each factor is explained below, and the worked example puts real numbers through all of them. If you want a one-line rule of thumb, use the shortcut from the introduction: multiply battery watt-hours by 0.5 before dividing by the load. It lands within a few minutes of the full calculation for most lead-acid rack UPS units.
Step 1: Measure Your Real Rack Load in Watts
Everything downstream depends on this number, so get it right. Do not add up the wattage printed on every power supply — nameplate ratings are maximums, and a rack of nameplates will overstate your real draw by 40–60%.
The right way is to measure. A plug-in power meter on the UPS input, or the UPS’s own load readout in its management interface, gives you the true figure. Measure under normal working conditions, not idle: run a backup job or a compile and watch the peak. If you have a smart PDU, its per-outlet metering is even better — it shows you which device is eating the watts.
In my own homelab rack the numbers look like this: two 1U servers pulling about 180W each under typical load, a 24-port PoE switch at 60W, a NAS at 45W, a firewall at 25W, and rack fans plus odds and ends at roughly 110W. That totals 600W — and that is the load the worked example below is built on.
One more distinction that matters: watts, not volt-amperes. Runtime burns real power, measured in watts. VA is apparent power and only matters for sizing the inverter. Modern server power supplies run at a power factor of 0.9 or better, so the gap is small — but if you plug VA into a runtime formula, you will overstate your backup time.
Step 2: Find Your Battery’s Usable Energy
Battery energy is voltage × amp-hours, but only a fraction of it is usable. Two things eat into it before the inverter even starts.
First, the discharge rate. Lead-acid capacity is rated at a gentle 20-hour discharge. A rack UPS pulling 25 amps from an 18Ah bank is discharging at well over 1C, and at that rate the chemistry simply cannot deliver its full capacity before voltage collapses. This is the Peukert effect, and it is the single biggest reason naive runtime maths fails. At a 300W load the bank might deliver 65% of its rated energy; at 900W, barely 47%.
Second, depth of discharge. The UPS will cut off at a set low-voltage point to protect the batteries, so the last 10–20% of the bank’s energy is never delivered to your load. Lithium batteries suffer far less from both effects, which is why lithium UPS units quote longer runtimes from the same watt-hour rating — but lead-acid is still what ships in most rack UPS units, and this guide assumes lead-acid unless stated otherwise.
So for a typical 1500VA rack UPS with a 24V bank made of two 12V 18Ah batteries: nominal energy is 24 × 18 = 432Wh, but the usable energy at a 600W load is roughly half that once the discharge rate is accounted for.
Step 3: Apply the Derating Factors
Three more losses sit between the battery and your servers.
Inverter efficiency: converting DC back to AC wastes roughly 10–12% as heat. Budget 88% efficiency for a decent line-interactive or online UPS.
Battery age: a lead-acid battery at the end of its typical 3–5 year life delivers about 80% of its original capacity. If your batteries are three years old, knock 20% off. This is also why runtime slowly shrinks year after year even though nothing else changed.
UPS self-consumption: the UPS itself draws 20–30W to run its own electronics, fans, and monitoring. Add that to your load before dividing.
Temperature matters too. Capacity drops in a cold room and battery life drops in a hot one; 20–25°C is the sweet spot manufacturers assume. A rack room running at 30°C will age batteries roughly twice as fast.
Worked Example: A 600W Homelab Rack on a 1500VA UPS
Here is the full calculation for the rack from Step 1 — a 600W measured load on a 1500VA rack UPS with a 24V, 2 × 12V 18Ah battery bank (432Wh nominal).
Step 1 — Nominal battery energy: 24V × 18Ah = 432Wh.
Step 2 — Discharge-rate factor: at 600W the bank supplies 600 ÷ 24 = 25 amps, about 1.4C for an 18Ah bank. Effective capacity is roughly 55% of rated: 432 × 0.55 = 238Wh.
Step 3 — Inverter efficiency at 88%: 238 × 0.88 = 209Wh delivered to the outlets.
Step 4 — Age factor for three-year-old batteries at 80%: 209 × 0.80 = 167Wh of genuinely available energy.
Step 5 — Effective load including 25W of UPS self-consumption: 600 + 25 = 625W.
Step 6 — Runtime: 167Wh ÷ 625W = 0.267 hours, or about 16 minutes.
Compare that with the naive formula’s 43 minutes. The naive number would have told you that you had time to finish a backup and shut down calmly. The real number says you have about a quarter of an hour — still plenty for a graceful shutdown script, which typically needs 5–10 minutes, but nowhere near enough to ride out a 30-minute outage. If you need 30 minutes at this load, you need roughly double the battery — an external battery pack — not a bigger VA rating.
Runtime vs Load: Same UPS, Very Different Answers
Run the same calculation at different loads and the curve is steep. These are calculated estimates for the same 1500VA UPS and 432Wh bank, using the derating method above — always verify against the manufacturer’s runtime chart before buying.
| Rack load | Discharge current | Calculated runtime |
|---|---|---|
| 300W | 12.5A | ~37 minutes |
| 450W | 18.75A | ~22 minutes |
| 600W | 25A | ~16 minutes |
| 750W | 31A | ~12 minutes |
| 900W | 37.5A | ~9 minutes |
Two things jump out. First, runtime falls faster than load rises, because the Peukert penalty grows with current. Second, halving your load more than doubles your runtime — shedding non-essential gear (a second NAS, lab VMs, PoE cameras) during an outage is the cheapest runtime upgrade there is. Many smart PDUs and UPS management tools can do this automatically with load-shedding rules.
How Much Runtime Do You Actually Need?
The right target depends on what the UPS is for. Most racks fall into one of four cases:
| Scenario | Target runtime | What it covers | Typical setup |
|---|---|---|---|
| Graceful shutdown only | 5–10 minutes | Save work, shut down cleanly, avoid data corruption | Standard 1500VA UPS, no extras |
| Generator bridge | 10–15 minutes | Hold the load while a generator starts and stabilises | Standard UPS, tested auto-start generator |
| Short outage ride-through | 30 minutes | Stay up through brief grid failures | UPS + one external battery pack |
| Extended uptime | 60+ minutes | Keep services running through long outages | UPS + multiple battery packs, or lithium |
For a homelab, the graceful-shutdown case is the honest answer for most people. Size for 10 minutes at your real load with aged batteries, automate the shutdown, and spend the money you saved on a second UPS or better monitoring instead of a battery bank you will never fully use.
For a small business rack, the generator-bridge case is the common spec — but only if the generator is actually tested. An untested generator plus a 10-minute UPS is just a 10-minute UPS with extra steps.
Common Mistakes That Make Your Runtime Math Wrong
Using nameplate wattage instead of measured load. Nameplates are maximums. Your real draw is usually half the sum of the nameplates, and using the bigger number makes you buy more UPS than you need — or worse, makes you think a small UPS is overloaded when it is fine.
Trusting VA instead of watts. A “1500VA” UPS is not a 1500W UPS. Most 1500VA rack units deliver 900–1000W, and some budget models less. Check the watt rating on the spec sheet, not the headline VA number.
Forgetting the load grows. Racks fill up. If you size for today’s 600W and add a server next year, your 16 minutes becomes 12. Leave 20–25% headroom in the UPS watt rating for growth.
Ignoring battery age. Runtime is a moving target. The 16 minutes you measured on day one will be closer to 12 by year three. Test annually and replace batteries on schedule, not when they fail.
Skipping the manufacturer’s runtime chart. Your calculation is a planning tool; the chart is the verdict. If the chart disagrees with your maths by a wide margin, trust the chart — it was measured on the real hardware, and its pessimism is usually the Peukert effect being worse than the textbook factor.
How to Test Your Real Runtime Safely
Do not test by pulling the plug on a live rack and watching the clock — that is how you discover your shutdown script was broken. Test in stages.
First, read the UPS management interface. Most modern rack UPS units report estimated runtime at the current load, calculated from actual battery voltage. It is the single most useful number you own, and it updates as batteries age.
Second, do a controlled test during a maintenance window. Shut down non-essential VMs, confirm the graceful-shutdown script is armed, then pull the mains and time how long the UPS holds the remaining load. Compare with your calculation. If reality is more than 25% shorter than the maths, your batteries are older than you think.
Third, repeat the test yearly. A runtime that shrinks 20% in a year means the batteries are near end of life. Replace them before they strand you — a battery that fails under load usually fails without warning.
Frequently Asked Questions
How long will a 1500VA UPS run a server rack?
It depends entirely on the load in watts. At 300W you can expect roughly 30–37 minutes; at 600W about 10–16 minutes; at 900W under 10 minutes. The VA rating does not set runtime — the battery bank does.
Is UPS runtime based on VA or watts?
Watts. Runtime burns real energy, and watts are the real-power measure. VA only matters for sizing the inverter so it is not overloaded. Use measured watts from a meter or the UPS display, never the sum of nameplate ratings.
Can I add batteries to increase UPS runtime?
Yes, if your UPS supports external battery packs (EBMs). They connect in parallel with the internal bank and can double or triple runtime. Check the UPS manual first — some budget units have no EBM connector, and daisy-chaining too many packs can exceed the charger’s ability to recharge them in a reasonable time.
Does a bigger VA UPS automatically mean longer runtime?
No. A bigger VA rating means it can carry a heavier load, but if it ships with the same battery bank, runtime at the same load is nearly identical. For longer runtime, compare battery watt-hours and EBM support — not VA.
How do I test UPS runtime without crashing my servers?
Use the UPS management interface’s runtime estimate first, then run a controlled test in a maintenance window with non-essential loads shut down and the graceful-shutdown script armed. Time the real hold-up, compare it with your calculation, and repeat the test once a year.
Bottom Line
UPS runtime is a battery-energy question, not a VA question — and the naive formula overstates it by two to three times. Measure your real load in watts, work out the battery’s usable energy after discharge-rate, inverter, and age losses, and cross-check against the manufacturer’s runtime chart. For most homelab and small-business racks, a 1500VA UPS giving 10–16 minutes at a 600W load is exactly right: enough for a clean, automated shutdown, with money left over for the monitoring and battery replacements that actually keep you safe.
