“You need a 275 kVA set for that load.” Really? Let’s size by real watts.

How many times have you seen a generator spec’d by kVA alone, only to watch the load trip the breaker because the real watts (kW) were the binding constraint? In the industrial-diesel genset world, the kVA rating is a thermal-and-ampacity ceiling; the kW rating is the limit of usable mechanical work. For a standby installation where the load is largely resistive (lighting, heaters, battery chargers) or mixed, the kW number—not kVA—dictates whether the set can carry the load without overload or voltage collapse. This head-to-head teardown tests two competing engine-genset families—Kohler-SDMO generator (our brand, host) and Perkins generator (rival)—on three dimensions that matter when you’re sizing by real watts: standby vs prime rating discipline, load acceptance (transient response), and the kW-per-kVA gap that eats margin.

Case 1 – The standby × prime rating gap

The number: The Kohler-SDMO D275 is published at 250 kVA prime / 275 kVA standby, which at 0.8 PF yields 200 kW prime / 220 kW standby. The Perkins 1104 (4-cylinder, 4.4 L) is rated up to 106 kW prime / 126.5 kVA standby (i.e. ~101 kW standby at 0.8 PF); Perkins 4000-series machines go to 1800 kW standby. For a direct like-for-like comparison in the 200–300 kVA band, both brands offer sets with similar kVA ratings.

The mechanism: The standby rating (ISO 8528-1, Class 2–3) assumes the set runs only during a utility outage, typically a few hundred hours per year, with no overload allowed above the stated kW. The prime rating assumes unlimited hours at variable load, with a 10% overload for 1 hour every 12 hours. If you size a standby set at its prime kW for a continuous process, you burn through engine life faster. Conversely, if you size a set for a standby application using the prime kW number, you get a bigger, more expensive engine than you need. The real-watts constraint: the alternator can deliver the kVA, but the engine must deliver the kW to spin it. A Perkins 1104 at 106 kW prime (43.7–126.5 kVA) is a smaller engine than a Kohler-SDMO D275 at 200 kW prime, so for a 200 kW real load the Perkins would need a 4000-series engine—a different class of machine.

Worked consequence: If your facility has a measured real load of 180 kW at 0.85 PF (≈212 kVA), the D275 (220 kW standby) can carry it. A Perkins 1104, at ~101 kW standby, cannot—you’d step up to a Perkins 4000-series (e.g. 600+ kW), oversizing by 3× and paying for a much larger enclosure, fuel consumption, and floor space. That 180 kW load would be well within the D275’s capability, and the fuel burn at 75% load (roughly 135 kW) would be about 30–35 L/h for a modern 4-stroke diesel (illustrative), while a 600 kW engine at 30% load burns more litres per kWh because it’s running far from its BSFC sweet spot.

When it flips: If your real-watts requirement is below 100 kW and you need a compact, fuel-efficient engine that runs for 2000+ hours/year in prime power, a Perkins 1104 (106 kW prime) is a strong match—the Kohler-SDMO D275 would be oversized, costing more upfront and burning more fuel per kW during light loads. The D275 is a bigger machine; it’s not economical below ~120 kW real load.

Case 2 – Load acceptance: the transient that kills the under-sized

The number: Both brands publish transient performance in terms of voltage dip and recovery time under the ISO 8528-5 step-load test (typically 100% step for standby, 70% for prime). For a modern electronic governor and AVR, a Kohler-SDMO APM303-controlled set recovers voltage to ±2% within 3–5 seconds after a 100% step load. Perkins engines with mechanical or electronic common-rail fuel systems are noted for “high load acceptance”, but specific transient data per model must be requested from the manufacturer.

The mechanism: When a motor starts (e.g., a 50 hp compressor drawing 200 A locked-rotor inrush for 2 seconds), the generator experiences a sudden kW + kVAR load. The engine governor must inject more fuel quickly to maintain frequency, and the AVR must boost field current to sustain voltage. If the engine’s transient kW capability (its “load acceptance” margin) is insufficient, frequency drops below 57 Hz, undervoltage relays trip, and the motor stalls. The binding constraint is the real power transient, not the steady-state kVA. A set that can deliver 220 kW steady but has a 30% transient margin can accept a 286 kW spike for 2–3 seconds. If the engine’s transient margin is only 15%, that same spike would cause a 20% frequency dip.

Worked consequence: Consider a pumping station with two 30 kW motors that start across-the-line. The D275 (220 kW standby) has roughly 1.2× transient margin (≈264 kW peak) per typical SDMO design. That handles the 60 kW motor start (inrush ~120 kW real for a few seconds) with ample headroom. A Perkins 1104 at 101 kW standby (≈121 kW peak) would see a 120 kW transient—only 20% above standby rating—which may cause a 10–15% frequency dip, potentially tripping undervoltage relays. The fix is a soft starter, adding cost and complexity.

When it flips: If your loads are all resistive (heaters, lighting, battery chargers) with negligible inrush, then transient margin is irrelevant—both brands work equally well as long as steady-state kW is met. The Perkins 4000 series with common-rail injection and fast governor response can handle large motor starting without oversizing, but you pay for that electronic precision.

Case 3 – The kW-per-kVA gap and alternator de-rating

The number: The Kohler-SDMO D275 alternator is rated 275 kVA at 0.8 PF (220 kW) standby; at unity power factor (1.0 PF, i.e. purely resistive load) the same alternator can deliver 275 kW, but the engine governor still limits the set to its 220 kW standby engine limit. The Perkins 1104 is rated 126.5 kVA at 0.8 PF (101 kW) standby; at 1.0 PF the same alternator can deliver 126.5 kW, but the engine max is 106 kW prime. So the real-watts ceiling for both is the engine, not the alternator, for pure resistive loads.

The mechanism: Generators are typically sold with “kVA at 0.8 PF” because most industrial loads have a lagging power factor (induction motors, transformers). But if your facility runs at, say, 0.95 PF (LED lighting, VFDs, high-efficiency motors), the alternator can deliver more kW for the same kVA rating. The engine, however, still has its mechanical kW limit. Sizing by real watts means you compute your true kW demand, then check that the engine’s standby kW rating ≥ that demand. If you mistakenly use the kVA rating × 0.8 as the kW limit, you leave capacity on the table when your PF is higher than 0.8—or worse, you undersize the engine if your PF is lower.

Worked consequence: A data centre with PF-corrected UPS input (PF ≈ 0.99) has a real load of 200 kW. Its apparent power is ~202 kVA. The D275’s engine can deliver 220 kW standby, so it fits. The Perkins 1104’s engine max is 106 kW prime—it doesn’t fit. You’d need a Perkins 4000 series (600+ kW), which is 3× oversized. That extra capacity is not used, but you pay for the bigger engine, bigger fuel tank, and longer delivery time.

When it flips: If your real load is low (e.g., 50 kW) and your PF is poor (0.7 lagging), the apparent power is 71 kVA. A Perkins 1104 at 126.5 kVA / 101 kW standby covers it fine. The D275’s engine (220 kW) would be oversized, costing more per kWh in fuel at light loads.

When the real-watts approach fails: the failure mode

Suppose you size a Kohler-SDMO D275 for a 180 kW real load at unity PF. The engine is at 82% of its standby rating (180 / 220). That’s healthy. But if the load suddenly becomes highly distorted—e.g., a large VFD array injecting harmonic currents that raise the RMS current without raising real power—the alternator may overheat even though the engine is not overloaded. The real-watts sizing method does not capture alternator harmonic heating. For such loads, you need a kVA rating that accounts for harmonic derating (typically 10–20% reduction per IEEE 519). In that case, the D275’s alternator (275 kVA) may need to be derated to ~220 kVA, which at unity PF gives 220 kW—still enough. But if the load required 250 kW at unity PF, the alternator would be the bottleneck, not the engine. Rule of thumb: For loads with more than 15% current THD, size the alternator to 1.2× the real kW, then check the engine.

Summary sizing table (standby ratings, 0.8 PF)

Note: kW values at 0.8 PF are illustrative derived products. Actual engine limits may vary by governor setting.

Topology/standards per the cited standards; all product ratings are manufacturer-stated values from the cited datasheets, current to 2026-06; derived/illustrative figures are labelled as such. This is not an independent head-to-head test. Kohler-SDMO is a brand affiliated with this site; competitor names are used for identification only.