kVA, kW, and the 0.8 in between

PF = displacement factor × distortion factor = cos φ₁ × (I₁ ÷ I_rms)

Displacement is the textbook kind: the current's sine wave runs late (a motor, a transformer) or early (a capacitor) against the voltage. That's the cos φ on a motor's plate. Distortion is the other kind, and on a rig full of electronics it's often the bigger one. A load that draws its current in short, sharp gulps has an RMS current well above its useful fundamental, and the difference is harmonics: current at three, five, seven times mains frequency that does no work and still heats the copper. For a load with no phase shift, the distortion factor is 1 ÷ √(1 + THD²).

A phase-control dimmer on a tungsten lamp shows both at once, which is why it's worth working through. The lamp is a resistor, so you'd expect a perfect power factor, and at full it has one. Pull the fader down and the dimmer switches on part way through each half cycle. The chopped current is out of shape (distortion) and its fundamental lags the voltage (displacement), even though nothing inductive is involved. At half power the power factor is 1 ÷ √2, about 0.707, split almost evenly between the two.

A switch-mode supply with no power factor correction is the extreme case of distortion. Its bridge rectifier only conducts when the mains is above the voltage on the smoothing capacitor, which is a narrow slice near each peak, so all its current arrives in short spikes. Infineon put the result plainly in a PFC controller datasheet: "a power factor of around 0.5 with high total harmonic distortion."

IEC 61000-3-2 caps those harmonics for some kit. Built right up to its Class D limits (PCs and monitors between 75 and 600 W), a supply's current would have a THD of 95 % and a distortion factor of 0.726. Below 75 W the standard sets no limit for most equipment, which is why phone chargers and laptop bricks sit nearer Infineon's 0.5. In the summary I could check, professional equipment over 1 kW is exempt as well, so the standard doesn't promise PFC on big professional kit; the manufacturers below fit it anyway.

Here's what I could find on manufacturers' own pages, without guessing at the gaps. A surprising amount of kit doesn't state a power factor at all.

Kohler's 1000DSEC sheet shows the split in print. One engine, offered with different alternators, each rated at several winding temperature rises. Where the alternator is big enough, the rating stops at the engine's 910 kW prime and 1000 kW standby. Where it isn't, the alternator's heating sets the number instead, and the same engine is sold as a smaller set.

Single-phase sets are the exception. Kohler state it on their sheets: "All three-phase units are rated at 0.8 power factor. All single-phase units are rated at 1.0 power factor." The same 4R9X alternator on their 100ERESD is 98 kW three-phase and 77 kW single-phase (321 A at 240 V). On a single-phase set, kVA and kW are the same number and the power factor eats straight into it.

ISO 8528-1 defines four ratings. Read the footnote on the datasheet, because the one that matters for a show is prime, and the number in the model name is often standby.

• Datasheets don't agree with each other about the details. AGG's V110D5 sheet says prime may average 80 % over 24 hours, that 10 % overload is "available for governing purposes only", and that standby covers up to 500 hours a year. Kohler's 1000DSEC offers 10 % overload "for one hour in twelve". Go by the footnote on the set you're getting.
• Hire sets are named by prime. Power Hire's "100kVA Generator" sheet gives 100 kVA / 80 kWe prime and 110 kVA / 88 kWe standby at 415/230 V.
• Model numbers are often named by standby. AGG's V110D5 is 100 kVA prime and 110 kVA standby; the 110 is the standby. Kohler's 100ERESD is a 100 kW standby set. Standby is the mains-failure rating, and nothing about a festival is a mains failure.
• The 70 % average is the useful part. A prime-rated set can take its full rating as a peak, but over a day the average should sit at or under 70 % (or whatever the footnote says). A show that runs at its peak for hours, such as a long white look on an LED wall in a marquee with the heaters on, is closer to a continuous load than a variable one.

A capacitor across the supply draws current ahead of the voltage, and a generator handles that badly. The automatic voltage regulator (AVR) holds the output voltage by adjusting the alternator's field current. Leading current pushes the voltage up, so the AVR backs the field off, and on a lightly loaded set it can run out of room to back off. MTE's application note on generators and harmonic filters describes the end of that road: leading kvar at light load "negates the AVR field control and will force the generator to shut down on overvoltage."

The capacitance comes from EMI filters, harmonic filters, UPS inputs and power factor correction itself. Power Integrations point out that an active PFC supply's EMI filter capacitor (about 680 nF in their 225 W example) drags its power factor down at light load and high line. That's a real effect on one supply's figure, and a small one in absolute terms: 680 nF across 230 V at 50 Hz is 11.3 var. Hammond Power Solutions give a rule of thumb for passive filters, that capacitive kvar up to 20 % of the generator's kVA rating is tolerable. On a 100 kVA set that would take about 1,770 of those supplies switched on and idling.

So the X capacitors in a lighting rig are rarely enough on their own. The cases that go wrong involve something with a lot of capacitance on a set that's barely loaded: a UPS or a harmonic filter at light load, or PFC'd kit on a set chosen for a peak it only sees for ten minutes. If a set trips on overvoltage with the rig idling, this is the first thing to suspect. The IET's Guide to Temporary Electrical Systems (second edition, 2025) added material on leading power factor and stability with capacitive loads, which suggests it isn't a theoretical problem.

A generator can't take its full rating in one go. When load lands, the engine slows until its governor and turbo catch up, and the voltage dips until the AVR does. ISO 8528-5 grades how far and for how long into performance classes, and Kohler note that "most industrial generator sets meet G3." The class is tested with load applied in the steps the standard allows for the engine, not all at once: high output turbocharged engines take smaller steps.

• Inrush is the kit's own load step. One MAC Encore Wash CLD draws a 10.3 A half-cycle inrush at 230 V against its 2.6 A running current; the D80 limits its own to 13 A RMS. A rack of fixtures or a row of LED tiles switched on together is a step the set feels, which is why walls are powered up in sections and distros are brought up a circuit at a time.
• Motors are the big one. Kohler give starting current as typically six times full-load current, at a power factor of 0.3 to 0.5. The low power factor helps here: it asks less of the engine, and Kohler reckon a generator can supply nearly twice its rated kVA at 0.4 power factor for as long as a motor takes to get up to speed. Chain hoists start direct on line, so a bank of them starting together on a small set is worth a thought, and worth staggering.

Oversizing has its own cost. A diesel that runs lightly loaded for long periods never gets hot enough to burn its fuel completely, and the unburnt fuel ends up in the exhaust and past the piston rings. That's wet stacking. Kohler put the threshold at 30 %: "when generators run at less than 30% of capacity for extended periods of time, their engines are unable to sustain the optimal operating temperatures needed to burn fuel completely." Avtron, who make load banks, say the minimum varies by engine maker and typically falls between 30 and 50 % of the kW rating.

Event loads are lumpy, so this bites at the edges of the day: a set sized for the headliner idling through a morning of soundchecks, or overnight with only the security lights on. Two smaller sets, or a smaller set for the quiet hours, beat one big set at 10 %. Your hire company will have an opinion on their own engines, and it's the one that counts.

Ratings hold at the manufacturer's reference conditions (AGG's sheet gives 25 °C, 100 m above sea level and 30 % humidity) and come down in thin or hot air. The rule is on each datasheet and differs between them, so I've applied three published rules to one site: 300 m up, 30 °C in the shade, set in a canopy.

In the UK, temporary power for an event comes under BS 7909:2023+A1:2024, the code of practice for temporary electrical systems for entertainment, alongside BS 7671 (Section 551 covers generating sets) and BS 7430 for earthing. The generator's earthing arrangement, the RCD protection (BS 7671 asks for 30 mA RCDs on socket-outlets rated up to 32 A) and the neutral are all design decisions for a competent person, who also signs the system off before anyone plugs in. This page is for the conversation with that person, and for checking a quote. It doesn't replace them.

The method is short. List the loads at their peak, give each a power factor (published, measured, or a stated assumption), and sum the kW and the kVA. The set has to cover whichever of these is bigger: the kVA, for the alternator, and the kW ÷ 0.8, for the engine. Then add headroom for the 70 % average, for whatever nobody told you about, and for the derating. Here's a small stage worked through.

The rig's power factor comes out at 0.85, better than 0.8, so it's the engine that binds: 37.7 kW ÷ 0.8 is 47.1 kVA, more than the 44.4 kVA the alternator would see. With 20 % headroom that's a 60 kVA set, which would run at 79 % of its kW rating at the peak. Between acts, with the wall dark, the amps idling and the washes at their published 64 W idle (1.5 kW for all 24), it'll be well under the 30 % that starts the wet-stacking conversation, which is a point to raise with the hire company rather than a reason to go smaller.

The kVA column adds each load's kVA as a plain sum. That's an upper bound: loads with different power factor angles, and harmonics from different supplies, partly cancel when they meet. The real figure is lower, and the meter will tell you by how much.

Put in your loads: pick a type to fill in a power factor, or type your own. It sums kW and kVA, gives the current, and picks the smallest set from a typical hire range that covers both limits with your headroom.

• Cummins, Understanding ISO 8528-1 Generator Set Ratings (PowerHour, July 2017), for the four ratings, their hours and 24-hour averages, and the 2750/2500 kW worked example.
• AGG Power, V110D5 generator specification, for prime and standby kVA and kW at 0.8, the standby current table at 415, 400 and 380 V, and footnotes that differ from Cummins' reading of the standard. Power Hire, 100kVA Bruno Iveco GX11FE data sheet, for a UK hire set named by its prime rating.
• Kohler, spec sheets 100ERESD (G4-201), 125ERESC and 1000DSEC, for alternator tables at several temperature rises, unity power factor on single-phase sets, the prime overload footnote and three derating rules. ISO 8528-5 and Generator Transient Performance (Power Systems Topics 143, 2024), reproducing Table 4 of ISO 8528-5:2022. Sizing Generators for Motor Starting (Power Systems Topics 103) for starting current and power factor. Diesel Generator Maintenance: feasibility and benefits of no-load exercising (Power Systems Topics 130) for the 30 % wet-stacking threshold.
• Avtron Power Solutions, What is wet stacking and how do load banks prevent it?, for the 30 to 50 % range across engine makers.
• Martin, MAC Encore Wash CLD specifications (martin.com), for power factor, current and inrush. Lab.gruppen, True PFC technology page. d&b audiotechnik, D80 manual, technical specifications. Robe, current product datasheets. ROE Visual, Black Pearl 2 V2 specifications. The last three are cited for what they leave out.
• Infineon, IRPLPFC1 datasheet, for "a power factor of around 0.5" without correction. IEC 61000-3-2 Class D limits (Table 3) and exemptions, from a university lecture summary of an earlier edition rather than the current standard itself.
• MTE Corporation, Generators and Matrix Filters, for leading kvar and overvoltage shutdown. Hammond Power Solutions, Passive harmonic filters: leading power factor at light load, for the 20 % rule of thumb. Power Integrations (Andrew Smith), Compensating for the impact of EMI filter X-capacitance on the power factor of active PFC, for the 680 nF example.
• BSI, BS 7909:2023+A1:2024 catalogue entry, and the IET's Wiring Matters article on the second edition of the Guide to Temporary Electrical Systems (May 2025), for the 30 mA RCD requirement and the new material on leading power factor.

Assembled 3 November 2026 with AI assistance. The current columns of three rating tables, one fixture's published watts, PF and amps, and Cummins' 70 % example are checked at build time; the dimmer, neutral and Class D figures are computed and checked against their closed-form values. Those checks catch transcription and arithmetic errors, not a wrong source, and I haven't read ISO 8528, IEC 61000-3-2 or BS 7909 themselves, only the manufacturers' and the IET's summaries of them. None of it has been measured on a real generator or a real rig. A temporary supply is designed, installed and signed off by a competent person under BS 7909, and this page doesn't change that.

Companion pages: UK power connectors for events for what the current goes through next, LED processor and power lines for panels per circuit, and cable voltage drop for what the run from the set costs you.