Chat
Generator Sizing

Generator Sizing Calculator

Determine the right generator kVA for your facility. Enter your running load and largest motor to account for starting surges.

How to size a generator

Generator sizing balances two requirements — steady-state running load and transient starting surge. Divide the running load (kW) by the power factor (typically 0.8 for mixed loads) to get kVA, then add a 25% safety margin for growth and ambient derating. For motor loads, the starting inrush is 4–6× the running current, so the generator must absorb that surge without excessive voltage dip. The recommended size is the larger of (running load × 1.25) or (running load + motor surge allowance).

Office and retail loads are predominantly lighting, HVAC, and electronics with mild surge. Factory loads include motors, compressors, and welding equipment with heavy surge. Data centres demand tight voltage regulation and often use parallel redundant sets (N+1). For extended outages in the UAE, diesel gensets with 500–1000L fuel tanks provide 8–24 hours of autonomy.

FAQ

Why is my generator rated in kVA and not kW?

kVA is apparent power — the vector sum of real power (kW) and reactive power (kVAR). Generators are rated in kVA because they must supply both resistive loads (lights, heaters) and inductive loads (motors, transformers). Real output in kW = kVA × power factor.

What power factor should I assume?

For typical mixed commercial loads, assume 0.8. Office buildings with lots of electronics may run at 0.85–0.9. Industrial sites with many induction motors often operate at 0.7–0.8. The generator itself is typically rated at 0.8 PF.

How much oversizing is needed for motor starts?

Direct-on-line (DOL) motor starts draw 6× full load current; star-delta starters reduce that to ~3×; soft starters and VFDs can limit surge to 1.5–2×. Size the generator so the largest motor start doesn't exceed 30% voltage dip.

Should I size for peak or average load?

Size for the maximum coincident load — the highest sum of simultaneous equipment that can realistically run together. Don't just add nameplate ratings, since not everything runs at full load at the same time. A load-demand analysis gives a far more accurate figure than worst-case summation.