Generator sizing is not the sum of nameplates plus one fixed percentage. The result must support steady operation, the largest reasonable starting event, site derating, future change and healthy engine load while avoiding excessive low-load operation.
Define required loads and recovery sequence; calculate steady kW/kVA and the maximum starting event; select the correct prime or standby rating; apply site derating; add justified expansion reserve and validate load factor. The output should be an auditable range and operating strategy, not one isolated model number.
1. Why generator sizing is not simple addition
The same 200 kW steady load can be lighting and heaters, one large pump, VFD-driven motors, or UPS and servers. Each creates different starting and transient requirements. The set must maintain critical equipment during the worst reasonable operating scenario.
Frequency, duty rating, power factor, alternator efficiency, temperature, altitude, cooling and emissions platform also affect the answer. Calculate what the loads require separately from what the set can deliver at site.
Undersizing causes failed starts, voltage/frequency dip and overload. Excess capacity raises cost and can create persistent low load, deposits or wet stacking. Professional sizing controls both risks.
2. Start with a usable load schedule—not a genset model
Record equipment, quantity, running kW or current, power factor, efficiency, starting method/current, criticality, simultaneous operation and recovery order.
- Class A: must recover first—life safety, critical process, medical or core IT.
- Class B: important but can return after seconds or minutes.
- Class C: can remain disconnected when capacity is constrained.
Remove loads outside the backup scope and calculate real operating combinations. Do not add every nameplate or invent a low diversity factor without records.
3. Calculate steady kW and kVA
kW is real power and kVA is apparent power, connected by power factor. Three-phase loads can be estimated from voltage, current and power factor, while motors also require efficiency. Equipment data, measurements and operating scenarios are the best inputs.
Total steady kW and corresponding kVA by scenario. The engine must cover real power while the alternator and current path cover kVA and current. Do not impose a generic 0.8 power factor on every load; UPS, VFD and actual power factor can differ.
Check phase balance. A moderate total with severe single-phase imbalance can still overheat one phase or move voltage outside limits.
4. Validate the largest starting event—not all starting multipliers at once
Direct-on-line, star-delta, soft-start and VFD motors have different inrush. Build a timeline: base loads already connected, the largest motor starting, permitted voltage/frequency dip and equipment that may trip.
UPS, rectifiers, compressors, lifts and transformer energisation add special events. Check engine transient response and alternator excitation together. Enough steady kW does not prove starting ability.
Staged restoration, soft starting, VFDs, unloaded starts or another parallel unit can reduce required capacity. Optimise sequence before simply increasing the set.
Continue: motor-starting load calculation.
5. Select prime or standby rating and apply site derating
Standby power covers utility outages; prime power supports longer variable-load operation. Continuous, base-load or data-centre ratings may have other definitions. Use current manufacturer data and expected hours; do not treat standby rating as daily long-term output.
Heat, altitude, restricted intake, hot-air recirculation, fuel quality and exhaust backpressure can reduce capacity. Engine and alternator derating rules may differ; the complete system follows the limiting side.
Continue: prime, standby and critical ratings and temperature, altitude and dust derating.
6. Give reserve a purpose—and check low-load risk
Reserve can cover measurement uncertainty, defined near-term expansion, environment and short transients, but no percentage fits every project. Put known future loads, dates and starting methods in the calculation. For uncertain growth, reserve distribution and paralleling interfaces rather than buying one oversized set.
Calculate typical, minimum and maximum sustained load factors. Standby sets need a testing and load-management plan; long-hour prime sets especially need protection from persistent very low load. Parallel sets, zoning or load-bank testing may create a healthier range.
Continue: capacity reserve and expansion.
7. Complete workflow from load schedule to recommended power
- Define backup scope and remove excluded loads.
- Classify criticality and recovery time.
- Build normal, worst-reasonable and expansion scenarios.
- Calculate steady kW, kVA, current and phase balance.
- Build starting sequences for the largest motor, UPS and other events.
- Select the rating that matches operating hours.
- Apply heat, altitude, ventilation, fuel and exhaust derating.
- Add purpose-based reserve and check typical/minimum load.
- Validate transients and environment against engine and alternator data.
- Accept with load tests, alarms and staged-restoration procedures.
Present a minimum technically qualified capacity, recommended capacity and expansion path, with assumptions. Users can then see how sequence or growth changes the result.
8. Information required for a reliable engineering recommendation
Provide country/city, voltage/frequency, duty, expected hours, load schedule, largest motor and starting method, UPS/VFDs, restoration sequence, temperature/altitude, expansion, acceptable outage, budget and schedule.
The final proposal should include calculation version, assumptions, recommended kW/kVA, power factor, engine and alternator model, duty rating, derating, runtime, loading/unloading sequence, protection, expansion interfaces and load-test scope.
When data is incomplete, do not disguise uncertainty with an “experience factor.” State a preliminary range, missing inputs and worst-case impact, then converge through nameplates, measurements or a site survey.