Cummins Generators

A diesel genset is a tightly-coupled system of a compression-ignition prime mover, a synchronous alternator, a control module, a fuel-and-cooling skid, and the switchgear that decides when the genset is connected to the load. Every one of those subsystems has its own failure modes, its own service intervals, and its own commissioning checklist. A specification that ignores any of them is the source of nine out of ten standby-power complaints we see in the field.

The first decision is duty class. ISO 8528 defines four duty profiles — Emergency Standby (ESP), Prime (PRP), Limited-Time Prime (LTP) and Continuous Operation (COP). A 100 kVA ESP-rated unit running continuously will derate to ~90 kVA and double its maintenance burden. Choose the rating that matches expected runtime, not the cheapest sticker price.

Sizing follows the formula kVA = Σ(running kW) ÷ pf + Σ(starting kW × inrush factor). Real installations are dominated by motor inrush — a 30 kW direct-on-line motor will pull six times running current for the first second. Soft-starters, VSDs, and step-loading sequences buy back capacity, which is why a 250 kVA genset can often replace a 400 kVA one if the controls are designed properly.

Fuel is where most field failures originate. Diesel is hygroscopic; water condenses on the inside of half-empty tanks, sulphate-reducing bacteria thrive at the water-fuel interface, and the resulting biomass blocks primary filters in months. Polishing systems, weekly water checks, and full-tank policies during long shutdowns extend injector life from 4,000 hours to 15,000 hours.

Cooling design is rarely optimised. A radiator sized for 35 °C ambient that is installed in a 45 °C engine room will overheat. Ventilation requires inlet area ≥ radiator face area × 1.5 and outlet area ≥ inlet × 1.25, with louvre free-area corrections. We have lifted three Cummins units in 2025 alone whose ducting reduced free area to under 50% of catalogue.

The alternator (the "generator end") must be selected for both kVA and short-circuit ratio. Hospitals running large MRI / CT plant need an alternator with high SCR and low subtransient reactance to ride through nuisance breaker trips during diagnostic imaging cycles. Cheap end-coupled alternators are fine for offices and useless for clinical loads.

Modern controllers — Deepsea DSE, ComAp InteliLite, Cummins PowerCommand, Woodward easYgen — implement the same core functions: AVR setpoint, governor trim, fuel theft alarm, over/under-voltage, over/under-frequency, oil pressure low, coolant high, charge-fail and emergency-stop interlocks. The controller is also where you connect the J1939 CAN bus that will give you live engine telemetry on a SCADA.

The switchgear that joins the genset to the building is the Automatic Transfer Switch. Open-transition ATSs leave a brief power gap (typically 100–500 ms once running). Closed-transition ATSs synchronise the genset with utility before transferring — required for sites that cannot tolerate even a flicker, such as broadcast and telco. Bypass-isolation ATSs let you maintain the switch without dropping load. Pick the right class — they do not retrofit cheaply.

Earthing of standby gensets is its own discipline. As a separately-derived source the alternator neutral and frame must be bonded once and only once; a second N-PE bond inside the building will draw triplen-harmonic currents through the structure. Earth resistance ≤ 5 Ω is enough for general; ≤ 1 Ω is the standard for medical, telecom and data-centre. Test by fall-of-potential or clamp method annually.

A maintenance regime is non-negotiable. Daily: visual leaks, fuel level, battery charge LED. Weekly: 5-minute no-load run, exercise the ATS. Monthly: load-bank ≥ 30% of nameplate for 30 minutes to burn off wet-stacking. 250-hour: oil and filter, coolant test, valve adjustment per OEM. 1,000-hour: injector test, cooling system flush, generator-end air-gap inspection. Skipping the load-bank is the single biggest reason "perfectly good" gensets fail to take load on the day of an outage.

Finally, the most under-spec'd item is the fuel-day-tank piping. A diesel-engine return temperature can hit 80 °C; routing the return to the bottom of a tank that already supplies the engine recirculates hot fuel and chokes capacity. Returns must enter at the top, separated from suction by a baffle, and the day-tank should be vented through a flame-arrester that itself is checked yearly.

The transfer switch deserves its own commissioning discipline, because it is the component that decides whether all the money spent on the genset ever reaches the load. Four timers govern its behaviour and all four need to be set for the site rather than left at factory defaults: the utility-failure delay, which stops a momentary dip from starting the engine; the warm-up delay, which holds transfer until voltage and frequency are inside the acceptance window; the retransfer delay, which stops the load bouncing back onto a utility supply that has not settled; and the cool-down run, which lets the engine shed heat unloaded before it stops. Sites that suffer repeated transfers during unstable-supply periods almost always have a retransfer delay set in seconds when it should be minutes. Beyond the timers, the switch itself must be exercised: an ATS that has sat in one position for a year can have stiff mechanism and oxidised contacts, and the first time it is asked to move should not be during a real outage.

Repair economics deserve to be stated honestly, because the sector has a reputation for the opposite. A well-maintained diesel genset is a long-lived asset — the engine is designed for tens of thousands of hours — so a unit failing early is usually telling you about installation, fuel or maintenance rather than about its own design life. Three pieces of evidence should exist before anyone quotes an overhaul: the running hours, the maintenance history, and a load-bank result showing what the set can actually deliver today. Those three numbers separate a set that needs an injector service from one that genuinely needs a rebuild, and they occasionally show that neither is worth doing on an old machine. Ask for them, from any contractor. A quote that arrives without that evidence base is a guess dressed up as a diagnosis, and it is the buyer who carries the cost of the guess.