Start with the requirement
Use the same watt load, exact UPS model and battery configuration on every curve. Treat published duration as a planning estimate and validate the shutdown workflow for the deployed rack.
Research based on linked manufacturer documentation. We have not physically tested this equipment.
Give every shortlisted UPS the same job
A fair UPS runtime comparison asks how long each complete candidate configuration is estimated to carry the same protected load. Use watts on the horizontal axis and minutes on the vertical axis. Do not start by comparing the biggest half-load number on a product page. Half load is a percentage of that product’s rated output, so two products can be carrying different amounts of equipment at that point.
CyberPower’s runtime tool explicitly presents total load in watts and estimated runtime in minutes, with a separate battery-module selection. It also states that its curves are approximate and based on fully charged new batteries under normal conditions. The manufacturer’s runtime calculator is the right type of evidence for a model-specific comparison.
Before opening a curve, write the intended load and required battery configuration in the worksheet. Separate the current operating estimate from the approved future planning load. A result at a lightly loaded rack today is useful, but it cannot be silently carried forward after adding servers or increasing the powered network load.
Establish what the load number includes
Measure or estimate the equipment protected by the UPS output. Include switches, storage, management appliances and the power supplied to attached devices through PoE where relevant. Keep the measurement boundary consistent across the inventory. A server’s advertised power-supply capacity is not a measurement of its normal consumption, and two redundant supplies do not automatically mean double the equipment’s consumption.
Watts and volt-amps describe different quantities. APC’s buying guide distinguishes real power in watts from apparent power in VA. Both output limits must be respected, but a runtime comparison should use the watt load requested by the manufacturer’s chart. APC’s sizing explanation supports that distinction. Resolve any chart using a different load convention before comparing it with a watt-based curve.
Do not use a wall-side reading that includes UPS losses and battery charging as though it were the UPS output load unless the manufacturer’s tool specifically asks for that quantity. Record where the reading came from and which equipment was running. For a variable workload, retain representative sustained demand and the documented peak or growth case rather than choosing whichever reading produces the longest runtime.
Half load can hide a 250W difference
The APC SMT1500RM2UC is specified at 1,000W output, while CyberPower specifies the PR1500RT2UC at 1,500W. Consequently, a watt-based half-load point represents 500W on the APC and 750W on the CyberPower. These are manufacturer output ratings, not our test results. See APC’s exact model specification and CyberPower’s exact model specification.
Suppose your rack planning load is 750W. The relevant APC curve point is 750W, which is 75 percent of its watt capacity. The relevant CyberPower point is also 750W, which is 50 percent of its watt capacity. Comparing the two products at 50 percent would give them different jobs. The percentages can help show capacity headroom, but they should not replace the common watt value.
At a 500W planning load the same rule applies in reverse: read both curves at 500W. That is half of the APC’s watt rating and approximately one third of the CyberPower’s. Save the queried watt value beside the runtime result so a reviewer can reconstruct the comparison even if a product page later changes its default slider position.
| Protected load | Share of 1,000W output rating | Share of 1,500W output rating | Curve value to compare |
|---|---|---|---|
| 300W | 30% | 20% | 300W on both models |
| 500W | 50% | About 33% | 500W on both models |
| 750W | 75% | 50% | 750W on both models |
Read the exact configuration and preserve the result
Select the full model number, including regional suffixes, and confirm the output rating shown by the tool. Check the selected battery-module count before reading the curve. Included internal batteries, one added module and several added modules are different purchase configurations. A search result can show a long-runtime family while the quote contains a base model without those batteries.
Enter or locate the agreed watt value and record the displayed estimate with its units. If the chart is a static image, use its labeled axes carefully and state that the number is approximate. Where a calculator provides an exact queried point, retain that result instead of reading an approximate intersection from a small screenshot. Record the source URL and research date in the purchase notes.
If the chosen load lies between tabulated points, avoid assuming runtime changes in a straight line. Ask the manufacturer’s tool or support team for the relevant estimate. At minimum, keep the bracketing points and identify the uncertainty. A procurement decision that barely passes only after optimistic interpolation needs stronger evidence or a different configuration.
Doubling load does not simply halve runtime
CyberPower publishes 18.2 minutes at half load and 6.5 minutes at full load for the 1,500W PR1500RT2UC. Those correspond to 750W and 1,500W. Doubling that load does not produce half of 18.2 minutes, which would be 9.1 minutes. The published runtime points illustrate why a simple inverse calculation is an unsuitable replacement for the curve.
The same caution applies to an amp-hour calculation using the battery label. A battery capacity figure alone does not describe the complete UPS output behavior across different loads. Use the manufacturer’s curve for the specified configuration instead of treating nominal stored energy as a guaranteed duration at the outlets.
Also avoid scaling runtime in direct proportion to added battery modules unless the manufacturer explicitly provides that relationship. The supported configuration, charging arrangement and runtime model matter. Preserve the module count in the comparison table and the purchase order so an impressive estimate cannot become detached from the hardware needed to achieve it.
Turn a runtime estimate into a service timeline
Use an illustrative shutdown plan to see what the minutes must cover. Suppose the service owner allocates two minutes to distinguish a sustained outage, five minutes to stop applications and four minutes to finish host and storage shutdown, then requires two minutes of planning reserve. The total is thirteen minutes. These durations are invented planning inputs, not measured shutdown results for any named system.
At a 750W constant planning load, the CyberPower published half-load point of 18.2 minutes exceeds that illustrative thirteen-minute timeline by 5.2 minutes under the manufacturer’s stated reference conditions. This is a useful initial screen, not proof that a deployed battery will deliver the same margin. Read the other candidate at 750W and compare its estimate with the same thirteen-minute requirement.
If the worksheet passes, validate the actual shutdown sequence through the manufacturer-supported procedures and the organization’s approved maintenance process. Confirm the communication path, application order and the point when storage finishes. Avoid defining success merely as the UPS still showing battery charge. The service requirement is that the dependent equipment completes its intended transition before power is exhausted.
Keep battery condition and repeat outages in the plan
Schneider Electric identifies battery age, temperature, charge state, discharge history and load characteristics as factors affecting actual runtime. It also cautions that light-load, long-runtime estimates can be less accurate. Its runtime graph guidance explains why a published curve should remain a planning reference rather than an unconditional promise.
Include the condition expected in service, not only the best case immediately after purchase. Decide who reviews battery health, replacement requirements and charging status after an outage. If a second outage arrives before recharge is complete, the original fully charged reference no longer describes the starting point. The response plan should account for that changed condition.
Keep a practical margin appropriate to the service requirement and uncertainty. There is no universal number of extra minutes that makes every rack acceptable. A remote site waiting for a generator, a lab accepting a prompt shutdown and storage serving a production application have different objectives. Document the objective instead of copying a runtime target from an unrelated installation.
Compare the finished worksheet before buying
The final comparison should show exact model, watt and VA output limits, protected load, battery configuration, manufacturer runtime estimate, required timeline and unresolved conditions. Repeat it at the future planning load if expansion is approved. Mark missing data as missing. An unverified estimate should not quietly become the winning number in the recommendation.
Use the APC versus CyberPower 1500VA comparison for the broader feature tradeoffs, then return to equal watts for the battery-duration question. The SMT1500RM2UC and PR1500RT2UC specification reviews provide the exact-model context. Headline VA similarity is a reason to inspect the details, not evidence that the battery systems are equivalent.
Choose the configuration whose documented capacity, estimated duration and management workflow meet the rack’s requirement together. Save the worksheet with the inventory and update it after significant load changes. Equal-watt reading makes the comparison fair; a recorded shutdown timeline makes the result useful to the person responsible for the service.
Sources and research basis
- CyberPower UPS runtime calculator and estimate conditionsPrimary documentation · checked 15 October 2026
- APC UPS buying guide, watts and volt-ampsPrimary documentation · checked 15 October 2026
- APC SMT1500RM2UC official product specificationsPrimary documentation · checked 15 October 2026
- CyberPower PR1500RT2UC official specificationsPrimary documentation · checked 15 October 2026
- Schneider Electric APC runtime graph guidance and estimate limitationsPrimary documentation · checked 15 October 2026
Manufacturer specifications describe the documented configuration. Buying recommendations are our interpretation of those features. No original runtime, acoustic, efficiency, switching or software test is claimed. Report a correction.