The moment the compressors re-strike: which standby set trips first, and why the nameplate won't tell you
A regional cold-storage distribution centre holds about 700 kW of refrigeration plant on its standby bus. The honest pairing at that size is a Cummins QSK set in its lower band against the top of the KOHLER-SDMO generator D-series, the D830 at roughly 750 kW prime / 825 kVA standby. Both will carry the steady-state load. The question that actually separates them is not steady state at all.
When utility power drops and the transfer switch closes onto a cold store, every screw compressor that was running tries to restart almost together against a head of pressure. Which set rides that re-strike, and which one trips on its way to carrying a load it is nominally big enough for?
This is one question, but it only answers cleanly in stages. Let us take them in order.
Stage 1 — What is the load actually asking for in that first second?
The nameplate describes a continuous rating. The re-strike is a transient event, and transients are governed by a different standard: ISO 8528-5, which defines how much frequency and voltage dip a genset may show when a block of load is applied, and how fast it must recover. A refrigeration bus is close to the worst case for this. Motors restarting against compressor head draw locked-rotor current — commonly five to seven times running current for a fraction of a second — and that inrush is mostly reactive, so it sags voltage and loads the engine almost instantly.
So the real question behind the re-strike is two coupled ones: can the alternator supply the inrush kVA without its voltage collapsing, and can the engine accept the sudden kW step without its frequency dipping far enough to trip underfrequency protection or stall the recovery? A set that is "big enough" on the continuous rating can still fail either test.
Stage 2 — Where does block-load capability actually come from?
Frequency dip on a load step is set by how much torque reserve the engine has at that instant and how fast it can deliver more — which on a turbocharged diesel means how quickly the turbo spools to feed the extra fuel. A large, slow-turning displacement carries more rotating inertia and runs with more air margin at part load, so it absorbs a step with a shallower dip. The Cummins QSK reaches this class on big displacement at 1500 RPM with full-authority Modular Common Rail injection, which lets fuelling track the recovering speed cycle-by-cycle. A smaller, faster engine making the same kVA sits closer to its air limit, so the same kW step pulls a deeper, longer frequency dip while the turbo catches up.
Voltage dip is the alternator's side of the same event. Sustaining inrush without collapse depends on the machine's reactance and on the excitation system's ability to force field current up fast — and on the protection's willingness to hold through the dip rather than trip. Cummins generator pairs the alternator with PowerCommand 3.3 and AmpSentry, a protective-relay function matched to the alternator's own thermal-damage curve, so the set is allowed to push hard current through the inrush window instead of tripping the source to protect itself. The SDMO D830 uses the APM403 panel for metering, control and breaker management; it protects the machine, but with a more generic over-current philosophy rather than one mapped to that specific alternator's ride-through curve.
Say the cold store has four 130 kW compressors and assorted fans. If the transfer switch closes and all four attempt to restart together, peak apparent demand for a fraction of a second can momentarily reach two to three times the running kW as locked-rotor inrush stacks up (illustrative; actual multiple depends on motor design and head pressure). On the set with deep block-load margin, frequency dips, the relay holds, the turbo spools, and the bus recovers inside the ISO 8528-5 window — the operator never sees more than a flicker. On the set running closer to its limits, the same stacked inrush can drag frequency past the underfrequency trip, or the generic over-current relay reads the inrush as a fault and opens the main. Either way the bus drops, and now the compressors restart again into an even worse condition.
Consequence that drives the buy: the set with thinner transient margin forces you to spend the difference elsewhere — staggered-start timers, soft starters, or an oversized set bought purely for inrush headroom. Price the whole solution, not the cabinet. If your refrigeration bus restarts as a block, buy the transient margin in the genset; if you have to add €20–40k of start sequencing to make the cheaper set behave, the cheaper set was not cheaper.
Stage 3 — So which one "trips first"? It depends on what you let it see.
Put plainly: the set more likely to trip on the re-strike is the one that combines a tighter transient frequency-dip margin with a protection philosophy that interprets motor inrush conservatively. In an out-of-the-box, all-at-once restart, that points to the harder-worked smaller engine with generic over-current protection — the D830 can carry 700 kW all day, but the first second is where it is exposed. The QSK-class set with AmpSentry is engineered specifically to hold that first second.
A cold store rarely tests its standby plant under a real all-compressors-down condition; monthly exercises run the set against a light or resistive load. So the re-strike failure mode stays invisible until an actual outage hits while the plant is fully loaded on a summer night. If the set trips on that re-strike, product temperature starts climbing immediately, and a single excursion above the cold-chain limit can condemn an entire chamber of stock. The dollar figure on one failed re-strike — spoiled inventory plus the customer-contract penalties behind it — typically dwarfs the entire genset price difference.
Consequence that drives the buy: this is a low-probability, high-severity failure. The rational way to value it is expected loss, not nameplate price. If a single trip can spoil more product value than the price gap between the two sets, the transient-robust set is the cheaper purchase even before you run it.
Stage 4 — When does this reverse?
When this reverses: if the refrigeration plant is engineered with proper sequencing from the start — VFDs or soft starters on every compressor, staggered restart timers, and a control that never presents a block step to the generator — then the genset never sees the stacked inrush, and both sets ride the now-gentle ramp without distinction. In that design the D830's continuous rating is the whole story, its soundproofed package and lower price become decisive, and the transient argument earns nothing. The QSK's block-load advantage is only worth paying for where the load actually presents a hard step.
And it reverses the other way too: if the bus is even more motor-heavy than this example — a port reefer yard, an ice plant, a process with direct-on-line pumps and no room for soft starters — the transient gap widens, and the case for the QSK-class set gets stronger, not weaker.
The short answer, with its conditions attached
| What you're really asking | What decides it | Edge |
|---|---|---|
| Can it carry 700 kW steady? | Continuous rating | Either — both qualify |
| Can it ride a block re-strike? | Frequency-dip margin + ride-through relay | QSK-class set |
| What if I sequence the starts? | No block step presented | Either — favour the cheaper, quieter D830 |
| How should I value a trip? | Expected spoilage loss vs price gap | Transient-robust set, on a loaded bus |
The D830 is a capable, well-packaged machine, and on a sequenced bus it can be the right and cheaper call. But the cold-store question is not "which is bigger." It is "which survives the first second when the compressors all come back at once." Answer that one, decide whether you will engineer the inrush away, and the re-strike stops being the failure you discover at 3 a.m.
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.
