Size a UPS for your rack or room

Start with the unit mismatch

Servers are rated in watts. UPS units are sold in kVA. These are not the same thing and the gap between them is power factor.

Real power, in kW, is what the equipment consumes and converts into work and heat. Apparent power, in kVA, is what the electrical system has to physically carry. Divide kW by power factor to get kVA. At 0.95 power factor, a 100 kW load is 105.3 kVA of infrastructure.

Modern IT gear sits at 0.95 to 0.99 thanks to 80 PLUS certification, which requires 0.9 or better at half load and above. The 0.8 you may have seen in older material describes power supplies that have not been sold in decades — using it will oversize everything downstream by roughly a fifth.

N, N+1, and 2N without the jargon

N is exactly enough. If your load needs 100 kVA, you install 100 kVA. It works right up until a module fails or needs servicing, at which point the load drops. For anything you would be paged about at 2am, N is not the answer.

N+1 means one spare beyond what you need. The detail people miss is that the cost depends on how many modules you split the load across. Two modules plus a spare means installing 1.5 times your requirement. Four modules plus a spare means only 1.25 times. Same redundancy, meaningfully different bill — which is why the calculator asks how many modules you are planning.

2N is two entirely separate systems, each capable of carrying the whole load on its own. Separate feeds, separate UPS, separate distribution, ideally separate everything. It doubles your installed capacity and is what you specify when losing the room is not survivable.

One more thing that trips people: sizing to run at 100% of the UPS rating leaves you nothing. No room for the next server, no margin for the inrush when everything restarts at once, and most units are less efficient at the very top of their range anyway. Designing to about 80% is the common practice.

Why this tool will not tell you your runtime

You have probably noticed that other calculators will happily give you a runtime figure. They are guessing, and here is why.

Battery runtime depends on the specific string you install, how old it is, what temperature it lives at, and the manufacturer's discharge curve — which is not a straight line. Batteries deliver proportionally less as you draw harder. Two UPS units with identical kVA ratings and different battery configurations will give you wildly different runtimes at the same load.

What this calculator gives you instead is the kVA figure and your load percentage. Take those to a vendor along with your runtime target and ask for the discharge curve for the specific configuration they are quoting. That is a real answer; anything derived from load alone is not.

Worth knowing before you spec batteries: VRLA batteries — the sealed lead-acid type in most UPS installations — lose about half their service life for every 15°F above 77°F. A UPS room running at 92°F is on track to need replacement batteries at roughly half the rated interval. If your UPS room is an afterthought closet with poor cooling, that is a budget line you have not planned for.