The short answer
Because on a generator rated at 0.80 power factor, 1.25 does nothing but convert kW into kVA. Since 1.25 × 0.80 = 1.00, the genset rated active output it specifies is numerically equal to the UPS kVA rating — 1000.000 kW here. Whether that is enough depends on whether the input draws more than that number, and that comparison covers rated active power only. What the rating has to cover is the UPS input: with a 1000 kVA UPS at 0.90 output power factor, 0.94 efficiency and recharge at 15% of the load-carrying input, the input draws 1101.064 kW — 101.064 kW, or 10.106%, above 1000.000 kW. That 10.1% comes from output power factor, efficiency and charging, and it is the entire reason 1.25 falls short. A conventional generator nameplated at 0.80 power factor needs at least 1376.330 kVA to cover that active power. The apparent power actually drawn at the UPS input is a different quantity and depends on the UPS input power factor, which this article does not have.
So what does 1.25× actually calculate?
Take a 1000 kVA UPS: 1000 kVA of output rating, 0.90 output power factor, 900 kW of output active power. A generator nameplate shows both kW and kVA, and at a 0.80 rated power factor the kVA number is 1.25 times the kW number. That gives at least two ways to read the same factor:
| Reading | Arithmetic | Quantity produced | Genset rated active output |
|---|---|---|---|
| On UPS output kVA | 1.25 × 1000 kVA = 1250.000 kVA | Generator nameplate | 1250.000 × 0.80 = 1000.000 kW |
| On UPS output kW | 1.25 × 900 kW = 1125.000 kW | Genset rated active output | 1125.000 kW |
Both are called 1.25×. They differ by 125.000 kW of genset rated active output. The first reading lands exactly on the UPS kVA number, because 1.25 × 0.80 happens to equal 1.00. The second lands above the 1101.064 kW the input actually needs, by 23.936 kW.
One factor, one answer that fails and one that passes — both measured in genset rated active output. Before comparing ratios, ask which quantity the factor multiplies and which quantity comes out.
What do efficiency and charging add?
The UPS input carries more than the output active power. At 900 kW of output and 0.94 efficiency:
900 ÷ 0.94 = 957.447 kW
Recharge at 15% of that figure — not of the final total — adds 0.15 × 957.447 = 143.617 kW. The share is not arbitrary: MacAllister Power Systems states that battery recharge generally runs from 0 to 25 percent of input kW, with 15 percent typical. The total input is:
1101.064 kW
Dividing by the 1000 kVA rating gives 1.101064. That ratio is fixed by three parameters and by nothing else: output power factor divided by efficiency, multiplied by one plus the recharge share, or 0.90 ÷ 0.94 × 1.15 = 1.101064.
Converting to a generator nameplate at 0.80 rated power factor:
1101.064 ÷ 0.80 = 1376.330 kVA
Two cautions belong here. First, 1376.330 kVA is a nameplate floor for a conventional 0.8-power-factor set; it is not the apparent power drawn by the UPS. That would be input active power divided by the UPS input power factor, and this article has no such figure for the model, so none is quoted and 0.8 is not reused as an input power factor. Second, 1.376330 = 1.25 × 1.101064. The whole difference between the two ratios is that 10.106%; the 1.25 itself is untouched. The 126.330 kVA gap, 9.179% of the requirement, holds only on the same 0.8 nameplate basis.
Why does a six-pulse rectifier push the ratio to 2.2?
A UPS rectifier is not a linear load. Six-pulse input current carries enough harmonic content that the voltage drop across the generator's internal impedance distorts the waveform and heats the windings, so the same kW of output calls for a larger set. Cummins makes this point in the T-030 Liquid-Cooled Generator Set Application Manual (Issue 9, 2015), and generator sizing procedures encode it.
MacAllister Power Systems, in Generator Sets and Uninterruptible Power Systems, gives a four-step procedure. For a 200 kVA/180 kW UPS with 255 kW of input, a six-pulse rectifier is handled as 1.60 × input active power = 408 kW. If other loads total 100 kW, the second arm is 1.15 × 255 + 100 = 393.25 kW (the source prints 393). Both arms are computed and the larger one governs, which here is 408 kW.
One point of wording in the source needs care. Step 2 gives 1.60 for six-pulse and 1.40 for twelve-pulse; step 3 then describes adding 1.15 times the input for a twelve-pulse UPS, while the worked example applies 1.15 to a six-pulse unit. That inconsistency belongs to the source. This article follows the example and says so rather than harmonising it. That neighbouring section is where the misreading starts: it offers one set of factors for a UPS-only set and another for a set with other loads, which reads like an either/or. The four-step procedure cited here does not work that way — step 4 compares both arms and takes the larger.
That step is also often misread as “if other loads exist, swap 1.60 for 1.15”. It is not a swap. Dropping the first arm loses it entirely: at 100 kW of input with 1 kW of other load, the arms give 160 kW and 116 kW, and 160 kW governs; keeping only the replacement arm understates the requirement by 27.5%.
The 1.60 and 1.40 factors belong to step 2: they multiply input active power and add nothing else, because other loads are step 3's business. Applying 1.60 to 1101.064 kW gives a 1761.702 kW standby genset rating, which converts to 2202.128 kVA at 0.80, or 2.202× the UPS rating. A twelve-pulse rectifier at 1.40 gives 1541.489 kW and 1926.862 kVA, or 1.927×. When the set also carries linear load, these are not the final answer: the arm must be compared with the mixed arm.
One caution belongs here. The same section of T-030 notes that past incompatibility between rectifiers and generator sets led suppliers to recommend oversizing by two to five times, and that requiring the UPS to be compatible with the set is now the more cost-effective route than oversizing. So 2.202 is arithmetic from a waveform factor under one set of conditions, not a recommended multiplier.
Why check genset kW and alternator kVA separately?
Multiplying input active power by 1.60 produces a standby genset kW rating — the set's rated active output, not the power its engine shaft delivers. The 2202.128 kVA figure is a second step, a nameplate conversion at rated power factor. Three quantities differ: the UPS input active load (1101.064 kW), the selected genset rating (1761.702 kW), and the engine shaft output, which needs alternator efficiency and auxiliary loads this example lacks. A larger rating does not make the UPS consume it. Cummins T-030, page 25, puts it plainly: regardless of excitation type, a generator set is a limited source in terms of both engine power (kW) and generator volt-amperes (kVA), so load changes cause voltage and frequency excursions whose magnitude and duration depend on the load characteristics and on the set's size relative to the load.
So 2.202× is not a number to quote on its own. It assumes 0.90 output power factor, 0.94 efficiency, 15% recharge, a 0.80 set rating, and that the 1.60 arm actually beats the mixed arm — with enough other load, the mixed arm can govern. Change any one and both figures move. Sets assembled with a larger alternator and a smaller engine do exist, but that is a manufacturer's configuration to be confirmed with the maker; it cannot be inferred from a waveform factor, and it does not remove the need to check the set's kW rating.
When does 1.25× actually hold?
It is not always wrong. Take a modern UPS with 0.98 efficiency, recharge at 5% of input active power, no separate waveform factor, and the same 0.90 output power factor. Input active power becomes 964.286 kW, which converts to 1205.357 kVA at 0.80 — 1.205× the UPS rating, comfortably inside 1250.000 kVA on the rated-active-power test.
Raise the output power factor to 1.00, as modern UPS units often allow, and the same efficiency and charging give 1071.429 kW, or 1339.286 kVA and 1.339×. Now 1.25 fails.
Put the active-power condition in one line: 1.25's genset kW rating is enough while input active power stays at or below the UPS kVA number. At 0.98 efficiency and 5% charging, that requires an output power factor no higher than 0.933. The familiar three conditions — high efficiency, little charging, clean waveform — are therefore incomplete; output power factor belongs on the list.
That is still only the active-power test. It does not show that input apparent power stays within the set's kVA rating: that figure is input active power divided by the UPS input power factor, which this article lacks. With the same parameters and an input power factor of 0.75, apparent power would be about 1285.714 kVA, beyond 1250.000 kVA, while the active-power test still passes. So “1.25 is enough” can only mean “enough in rated active power”, not “a compliant set”.
Does any of this cover the moment after an outage?
No, and both directions get misstated.
Start with the UPS current limit. Cummins T-030, page 30, ties the 110%–150% limit to the maximum load after an outage and explains that current limiting holds total input down by reducing the recharge rate. That sentence carries two separate facts. The load rise at the moment of transfer is a dynamic event. The input drawn while current limiting is active, and while the battery continues to recharge, can persist until the battery state changes; it cannot be dismissed as a transient on the strength of the word “peak” and dropped from the capacity check.
Then the voltage and frequency dip. When the UPS transfers from utility to generator, the rectifier loads the set and the voltage and frequency dip. If the dip leaves the UPS input window, the UPS returns to battery, and no amount of generator capacity helps while the dynamic response is inadequate. T-030 raises UPS sensitivity to voltage and frequency deviation on the same page, which places it outside what a capacity ratio can fix.
A ratio answers only whether the set can carry the steady load. It says nothing about whether the UPS will accept the set during transfer; that needs transient data, the UPS input window, and a site load test.
What is left out?
- Only steady-state power relationships are covered. Load step, voltage and frequency dip, recovery time and transfer back to battery are not.
- No UPS input power factor was obtained, so no input apparent power is quoted; 1376.330 kVA is a nameplate floor for a conventional 0.8-power-factor set.
- 1000 kVA, 0.90, 0.94, 15% and 0.80 are one set of example values. Different models, loading or recharge limits require a fresh calculation.
- The 1.60 and 1.40 factors belong to one sizing procedure and one arm of it; they are not constants for every generator and every filter arrangement.
- Altitude, ambient temperature, paralleling, fuel tank, cooling and protection coordination are outside the scope.
- Actual UPS loading is not used as a basis; the calculation follows the Cummins approach of nameplate rating plus recharge.
Open questions
Sizing a specific set still needs three sets of data: the UPS input power factor, efficiency curve, recharge limit and input voltage/frequency window; the generator transient reactance, excitation type, voltage recovery curve and altitude/temperature derating; and the real load step and battery state at transfer.
This article can explain what 1.25, 1.60 and 2.2 each measure. It cannot certify a particular set. Once the definitions are clear, the answer still comes from the two devices' measured data and a site load test.
References
- Cummins Power Generation, T-030 Liquid-Cooled Generator Set Application Manual, Issue 9, 2015, pages 25 and 30. https://www.cummins.com/sites/default/files/2024-08/t030-Liquid-cooled-gen-set-application-manual.pdf
- MacAllister Power Systems, Generator Sets and Uninterruptible Power Systems, four-step procedure with dedicated and mixed arms, accessed 2026-11-13. https://www.macallisterpowersystems.com/?p=115/
- Cummins, PowerHour FAQs: Introduction to Generator Set Sizing Software, input kVA and battery recharge. https://www.cummins.com/sites/default/files/2019-10/FAQ_Introduction%20to%20Generator%20Set%20Sizing%20Software_vF.pdf
This article is an industry observation and does not constitute procurement advice. | Updated 2026-11
