UPS Room Heat Load | Output kW and Efficiency

A UPS does not consume the power it delivers; it converts it. The only heat it sheds into its own room is the conversion loss plus a small contribution from batteries and controls. So the cooling load follows from efficiency, not from the full load the UPS protects. Because efficiency is output divided by input, the loss is the input minus the output, which works out to the delivered output times (1 ÷ efficiency − 1). The delivered power itself becomes heat downstream at the IT equipment, which you size separately with the server room cooling calculator.

Enter real delivered kW, not kVA or maximum nameplate rating, and use the efficiency at the actual load and operating mode. Prefer manufacturer dissipation data for the design condition where you can get it.

If the same room also houses distribution gear, add its heat too. Feeder and breaker losses are estimated the same watts-to-BTU/h way in the electrical panel heat load calculator, and a whole room of IT gear can be handled with the data center calculator.

Loss: 20 × 1,000 × (100 ÷ 94 − 1) = 1,277 W, or 1.277 kW. Total local heat: 1,277 + 200 = 1,477 W. Room heat load: 1,477 × 3.412 = 5,038 BTU/h, or about 0.42 tons. These defaults are example assumptions, not universal design conditions.

Heat to remove for common delivered loads at 94% efficiency, including 200 W of additional battery and controls heat. Notice how small the cooling requirement is compared with the load the UPS supports: you are cooling the loss, not the load.

A rough screening rule: at 94% efficiency each 100 kW of delivered output adds about 21,800 BTU/h, under 2 tons, of loss heat to the UPS room, before battery heat. A less efficient or lightly loaded unit gives more.

• Sizing cooling to the full UPS rating instead of the loss. The protected load is rejected at the IT equipment, not in the UPS room, so counting it here over-builds cooling by an order of magnitude.
• Using output × (1 − efficiency). Efficiency is output ÷ input, so that shortcut understates the loss; use output × (1 ÷ efficiency − 1).
• Confusing kVA with real kW. Enter delivered kW at the actual load, applying power factor if you only have kVA.
• Forgetting the load is also real heat somewhere. The power passed through still becomes heat at the servers; size that in the server room, but add it here if the servers share the UPS room.
• Double counting auxiliary losses. Battery and controls heat belongs in the additional watts only if it is not already in the efficiency figure or manufacturer dissipation data.
• Ignoring battery recharge. Float charging and especially post-discharge recharging add heat above the steady figure, so give recharge its own design case.
• Assuming eco-mode efficiency for worst-case heat. Eco and line-interactive modes revert to double-conversion under disturbances, so design to the lower efficiency.
• Running with no cooling resilience. UPS electronics and batteries are temperature-sensitive; ask the designers whether a single cooling unit is acceptable.

A double-conversion (online) UPS rectifies incoming AC to DC, floats the battery on that DC bus, and inverts back to clean AC. Each stage is very efficient but never perfect, so a few percent of the throughput is lost as resistive and switching heat in the rectifier, inverter, magnetics and filters. That waste obeys the same conservation of energy that underpins every load estimate in the ASHRAE Handbook - Fundamentals: the watts that do not leave as usable power leave as heat, right there in the cabinet. A unit delivering 100 kW at 96% efficiency draws about 104.2 kW and so sheds roughly 4.2 kW, about 14,200 BTU/h, into the room, while the 100 kW it delivers becomes heat at the IT load it feeds. That downstream heat is real, but it belongs to a different space and is best handled with the server room calculator rather than double-counted here.

Efficiency is not a single number. A transformer-based UPS loses more than a modern transformerless design, and every UPS is least efficient at low load fraction, where fixed losses dominate. A 100 kVA unit carrying only 20 kVA may sit well below its nameplate efficiency curve, producing proportionally more heat per delivered kW. When you read a manufacturer efficiency table, match the figure to your actual load fraction and operating mode. Eco-mode and line-interactive topologies post higher headline numbers because they bypass the converters during clean utility power, but they revert to double-conversion the instant a disturbance appears, so the worst-case heat the cooling system must absorb is the double-conversion figure, not the eco-mode one.

Batteries are the most temperature-sensitive part of the room, and they drive the design even though they contribute relatively little heat themselves. Valve-regulated lead-acid (VRLA) cells and many stationary lithium-ion batteries are characterized at a reference temperature of around 25 °C (77 °F), and that is a sensible target unless the data sheet says otherwise. A commonly cited rule of thumb, rooted in the Arrhenius relationship that governs chemical aging, is that a sustained rise of roughly 8 to 10 °C above that reference can halve the service life of a VRLA string, turning a planned ten-year battery into a five-year replacement. Lithium-ion batteries may tolerate a wider band but enforce limits through their battery management system, throttling charge or tripping offline if temperatures stray. In both cases the goal is steady temperature, not merely a low average, because short hot excursions still accumulate damage.

The heat picture changes during and after an outage. While the UPS runs on battery and then recharges, internal resistance and charge current push battery heat above the modest float-charge baseline, so a room sized only for steady-state conditions can briefly run hot exactly when reliability matters most. Flooded (vented) lead-acid cells add a second concern: they release hydrogen during charging, so dedicated battery rooms need ventilation designed to keep hydrogen well below its lower flammable limit, per the applicable fire and electrical codes. Sealed VRLA cells release far less in normal float but can vent under fault or overcharge, so the designers should address air exchange as well. Pair temperature control with humidity in a moderate band to limit terminal corrosion and static.

Treat cooling redundancy as part of the resilience the UPS exists to provide. A single cooling unit serving a tight, electronics-dense room is a single point of failure for the gear meant to deliver uptime; an N+1 arrangement keeps the space in range during a failure or routine service. The same watts-to-BTU/h math extends to any switchgear sharing the space: estimate feeder and breaker losses with the electrical panel heat load calculator and add the result to the UPS loss before selecting equipment.

Use electrical and mechanical designers for recharge, battery limits, ventilation and cooling resilience. ASHRAE Handbook - Fundamentals provides heat-transfer, air-property and load-calculation background. Example inputs here are stated scenarios, not prescribed handbook values. For residential design, ACCA Manual J, S and D address load calculation, equipment selection and duct design. This calculator does not certify a complete design or code compliance.