COST OF ERROR You pick a 250 kVA SDMO D275 because the brochure says 38.1 L/hr at 75% load. That number is real — but only if you wire the generator into a system that actually lets it run at 75% load factor. If your facility cycles between 20% and 110% of prime rating three times a day, the fuel consumption curve bends so badly that the Cummins QSK60 at 2000 kW standby burns less fuel per kWh delivered. That is not a hypothetical. It is a structural consequence of how each platform handles partial-load efficiency, transient response, and cooling fan parasitics. The question is not which generator has the lower specific fuel consumption on a datasheet. The question is: under your site’s duty cycle, which generator actually keeps the efficiency it advertises?
1. Partial-load fuel economy: where the SDMO drops off the curve
The KOHLER-SDMO D275 (250 kVA prime / 275 kVA standby) is published at a fuel consumption of about 38.1 L/hr at 75% load using a mechanical fuel-injection system. That figure is respectable for a 4-cylinder diesel in this class. However, when the load drops to 25–30% — a common situation in standby applications where a facility runs only a fraction of its connected load during a utility outage — the D275’s consumption rises to about 22–24 L/hr, which is roughly 75–80% of the 75%-load rate for one-third the power output. The mechanical governor and fixed-speed injection timing cannot trim fuel delivery proportionally below about 40% load without degrading combustion efficiency. The Cummins QSK60, by contrast, uses Modular Common Rail (MCRS) fuel injection with electronic control that maps injection timing, pressure, and duration in real time. At 30% load, the QSK60’s fuel consumption drops to about 62 L/hr on a 2000 kW standby set — that is about 55% of its 75%-load rate (roughly 112 L/hr) for one-third the output. The efficiency retention ratio (actual kWh per liter at low load divided by kWh per liter at 75% load) is about 0.76 for the SDMO generator mechanical system versus 0.89 for the Cummins generator common-rail system, based on manufacturer-stated curves. Worked consequence: If your facility runs at ≤40% load for more than 30% of the generator’s operating hours, the SDMO’s gap from brochure to reality will cost you 12–18% more fuel per kWh generated compared to the Cummins — enough to justify the higher capital cost of the QSK within three to five years of prime-power usage. When this reverses: For purely standby duty with fewer than 50 hours per year and load always above 60% (e.g., a data center running full UPS+battery recharge), the mechanical injection’s simplicity and lower purchase price make the SDMO the better fit. The efficiency retention ratio barely matters if the set runs only a few hours annually.
2. Transient response: the hidden efficiency killer
One of the most common performance failures in industrial generators is the voltage dip and frequency drop when a large motor or transformer energizes. ISO 8528-5 defines Class G2 (acceptable for lighting and small motor loads) and Class G3 (tight voltage/frequency recovery for electronic loads). The SDMO D275 with its APM303 controller and mechanical governor typically recovers to ±2% voltage within 1.5–2 seconds on a 30% step load. That is within G2 but not G3. The Cummins QSK series, paired with the PowerCommand 3.3 digital control, hits G3 recovery (±1% voltage, ±0.5% frequency) within 0.5 seconds on the same step load. Why does this affect keepable efficiency? Every transient sag forces the voltage regulator to draw more field current from the generator, increasing excitation losses. On a mechanical-governor system, the frequency droop also causes the engine to momentarily run richer (more fuel per power stroke) until the governor catches up. These transient penalties accumulate. Over 200 start/stop cycles per year with heavy motor loads, the SDMO’s slower transient recovery adds roughly 3–5% more fuel burn than the Cummins, measured as excess fuel during the recovery window. Worked consequence: In a facility with multiple large motor starts (compressors, pumps, chillers), the Cummins’ faster recovery saves about 0.4–0.7 liters per start event. Over 300 events per year, that is 120–210 liters of diesel — small per event, but multiplied across a 10-year life, the efficiency gap becomes real money. When this reverses: If your load is overwhelmingly resistive (lighting, electric heating) and motor starts are rare (fewer than 10 per year), the transient penalty is negligible, and the SDMO’s simpler mechanical system avoids potential electronic failures in high-vibration environments.
3. Cooling fan parasitic load: the silent tax on net output
Every radiator fan on a diesel generator consumes engine power that never reaches the alternator. The SDMO D275 uses a belt-driven fixed-speed fan that runs at full speed whenever the engine is above 1500 RPM. At full load, this fan draws about 7–9 kW (roughly 3–4% of the prime rating). At 50% load, the fan still draws the same 7–9 kW — now 6–8% of the output. The Cummins QSK60 uses a thermostatically controlled on-demand fan (modulating hydraulic or electric drive) that reduces fan speed when coolant temperature is low. At 50% load in a temperate environment (20–25°C ambient), the QSK60’s fan draws only about 2–3 kW, which is about 0.1–0.15% of the 2000 kW rating. That difference — roughly 5–6 kW at half load — goes directly to net fuel efficiency. Worked consequence: Over a 500-hour annual run at 50% average load, the SDMO wastes about 2500–3000 kWh of fuel energy to fan drag that the Cummins delivers as usable electricity. At $0.12/kWh for diesel equivalent, that is $300–360 per year in lost efficiency. Over 10 years, the parasitic fan tax alone can approach the cost difference between the two generator sets for a midsize industrial installation. When this reverses: In hot climates (ambient >40°C) where the fan must run near full speed most of the time, the Cummins’ modulating advantage shrinks because the fan will be at high speed anyway. Also, the added complexity of the hydraulic/electric fan drive introduces a failure point that the simple belt fan avoids. For a site with extreme heat and limited maintenance budgets, the SDMO’s brute-force fan may be the more reliable choice.
4. Eligibility gate: the rule for your decision
Based on the three dimensions above, here is a threshold you can apply before signing a purchase order. If your generator’s average load factor over a year is below 45% (typical for many standby-only sites that run at 20–30% load for hours), the SDMO’s partial-load efficiency loss and parasitic fan tax will degrade its real-world efficiency by 12–18% relative to its 75%-load BSFC. That fails the eligibility gate. If your load factor is consistently above 60% and motor starts are few, the SDMO passes the gate and may be the lower total cost option. For any mission-critical application where voltage/frequency recovery must meet G3 (data centers, hospitals, precision manufacturing), the eligibility gate is tighter: the SDMO’s mechanical governor cannot deliver G3 recovery. In that case, the gate rule is binary — if G3 is required, the Cummins is the only candidate in this pair.
Failure mode / reverse case: If your site has a single large load that cycles the generator between idle and 90% load repeatedly (e.g., a rock crusher with a 400-hp motor), the Cummins’ faster transient response becomes a liability: the aggressive governor can cause fuel-rate overshoot that exceeds the SDMO’s slower but smoother ramp. In that specific scenario, the SDMO’s mechanical system may actually keep efficiency better because it avoids the fuel spike that the electronic controller introduces during rapid load changes. This is rare but real — and it is why a blanket “Cummins wins” conclusion is wrong without analyzing your load profile.
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. Cummins is a brand affiliated with this site; competitor names are used for identification only.
