Four Things Plant Engineers Get Wrong About 700 kW Cummins and Perkins Sets
An aggregates quarry needed standby for its primary crusher feed and weighbridge offices. The shortlist came down to a Cummins QSK-family set and a Perkins 4000-series set at the same ~700 kW. The shortlist also carried four folk beliefs that nearly chose the wrong machine. Each one funnels down to a single variable.
Both candidates are legitimate here — Perkins generator publishes the 4000 series at 600–1800 kW and the Cummins QSK family runs roughly 500–3010 kW — so the quarry is comparing comparable iron. The trouble is that the conversation around the table was built on four myths, and each myth hides one decisive variable. Pull that variable into the light and the myth collapses.
"The bigger-brand engine just accepts load better."
Reality — the variable is governor-plus-fuelling speed, not badge
Load acceptance is set by how fast the fuel system answers a speed dip under ISO 8528-5, and by alternator excitation holding voltage through sub-transient reactance. Cummins QSK uses Modular Common Rail under PowerCommand 3.3, one fast digital loop. Perkins offers the 4000 in both mechanical and electronically-controlled common-rail forms — and those two Perkins variants are further apart from each other than the electronic Perkins is from the Cummins generator.
Starting the crusher feed conveyor across the line is an illustrative 40–50% step on a 700 kW set. The mechanical-governor 4000 corrects after the dip; the electronic-common-rail 4000 and the QSK both pre-empt it. Buy on the variable: never specify the mechanical-governor variant for a hard motor start. Once you require electronic common rail on the Perkins, the brand gap narrows to control-system maturity — which is where the QSK's integrated AmpSentry and paralleling start to matter, not the engine block.
"A higher-rated radiator means it runs cooler in summer."
Reality — the variable is charge-air heat rejection at your ambient
Heat leaves a genset by several independent paths — jacket water, charge air, alternator losses, radiated block and exhaust heat — and they do not scale with the kW nameplate. A bigger radiator core helps only if it is matched to the charge-air rejection the engine actually produces at your altitude and ambient. Dust-laden quarry air fouling the core makes this worse, not better.
At a sustained 700 kW in a 40 °C dusty pit, the binding number is charge-air-plus-jacket heat versus core capacity and louvre airflow after derate for fouling. Buy on the variable: ask both vendors for charge-air and jacket-water heat rejection at quarry conditions, then size cooling and cleaning intervals to the larger figure. The "bigger radiator" claim is meaningless until matched to the engine's actual rejection.
"At this size, fuel burn is basically the same, so ignore it."
Reality — the variable is load factor, and it can flip the conclusion
Fuel burn ≈ load × brake-specific fuel consumption. At equal load and similar bsfc, two well-matched 700 kW sets burn comparable fuel — so far the myth holds. It breaks when the load factor is low: a genset idling at 20% of rating runs at a worse bsfc point, and the platform with better low-load fuelling behaviour wins on a duty that is mostly light.
A quarry standby set may run long light-load tests and occasional full crusher starts. If your duty sits mostly at 15–30% load, the fuel difference is a low-load-bsfc question, not a full-load one. Buy on the variable: get part-load fuel curves, not just the 100% figure, and pick on the load band you actually live in. "Same at full load" tells you nothing about a set that rarely sees full load.
"We'll never parallel, so the control platform doesn't matter."
Reality — the variable is protective-relay behaviour, which matters even single-set
Paralleling is one feature of PowerCommand 3.3; AmpSentry protection is another, and it works on a single set. AmpSentry gives the alternator a current-limiting fault characteristic that holds selectivity rather than tripping early. A Perkins 4000 package's protection depends on the controller the packager chose.
A locked-rotor fault on the crusher motor, on a single 700 kW set, is exactly where protection earns its keep: a relay that rides the fault long enough for the downstream breaker to clear keeps the office and weighbridge alive; one that trips the genset blacks out the whole site. Buy on the variable: compare the actual protective-relay scheme on each quote, single-set included. "We won't parallel" does not make control irrelevant.
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.
