Total the running watts, survive the largest starting surge
A generator fails in the first second of a motor start, not in the steady hour after it. This calculator lists your loads with running watts and starting watts side by side, totals the running load with a diversity factor for the loads that genuinely cycle, computes the realistic peak as everything running plus the single largest starting surge (motors are assumed to start one at a time, the standard planning assumption stated openly), separates your essential loads into their own smaller scenario, and converts the answer into a recommended rating in watts and kVA with a headroom margin you control. Preset buttons insert typical published figures from manufacturer sizing guides as starting points; every number stays editable, because your nameplates outrank any guide.
| Load | Running W | Starting W | Qty | Essential |
|---|
Disclaimer: This calculator gives estimates for general information only. It is not financial, medical or other professional advice. Generator installation, transfer switching and service sizing are code-governed work for licensed electricians; this page gives an indicative rating for shopping and planning, not an installation design.
What this calculator does
It runs the two calculations that decide generator size. First the steady state: every load’s running watts times quantity, summed and tempered by the diversity factor you set, because a house where the water heater, dryer and oven genuinely never overlap needs less machine than the arithmetic sum suggests. Second the transient that actually kills undersized sets: the peak, computed as the full running total plus the largest single starting increment (starting minus running watts), on the stated assumption that motors start one at a time while everything else is already running. The recommendation applies your headroom margin to that peak and expresses it in watts and in kVA at the rating power factor, since generators are sold in both dialects. The essential-loads scenario repeats the exercise on only the loads you ticked, which is the answer for outage planning: the smaller machine (or the longer runtime) that keeps the fridge, pump and heat alive.
How to use it
- Nameplate watts beat preset watts. Presets are typical published values for orientation; the label on your appliance, or its manual’s electrical table, is the figure of record, and motors in particular vary by model and age.
- Starting watts are the motor story. Resistive loads (heaters, lights, kettles) start at their running watts. Anything with a motor or compressor (fridge, pump, air conditioner, furnace blower, power tools) draws a surge for a fraction of a second to a few seconds; if a nameplate shows locked-rotor amps (LRA), starting watts are roughly LRA times supply voltage.
- Diversity is a promise, keep it honest. Lowering diversity below 100 percent asserts loads will not coincide; on an automatically transferred whole-house set in a storm, they may. Diversity belongs to managed load lists, and the surge check here deliberately ignores it for the running base.
- Rate for the peak, run for the average. The machine must survive the surge once; it will spend its life near the running total, where fuel use and noise live. Both figures print, because both decisions (will it start my pump; what will it cost to run all night) are real.
How it is calculated
- Running total = sum over loads of (running W x quantity) x diversity factor.
- Peak = full running total (before diversity) + the largest single (starting W – running W) x quantity increment among the loads: one motor starting against everything else running.
- Recommended rating = peak x (1 + headroom) expressed in W, and in kVA as watts / (1,000 x rating PF). The essential scenario repeats all three steps on ticked loads only.
Worked examples
The outage list. Refrigerator 150 W running, 1,200 W starting; well pump 1,000 / 2,200; lights 300 / 300; furnace blower 800 / 1,600; quantity 1 each, diversity 100 percent. Running total: 2,250 W. Largest surge increment: the pump’s 1,200 W (the fridge adds 1,050, the blower 800). Peak: 2,250 + 1,200 = 3,450 W. With 20 percent headroom the recommended rating is 3,450 x 1.2 = 4,140 W, which at a 0.8 rating factor is 5.18 kVA: the 5 kVA class machine is the honest answer, and a 3 kW set, however attractive its price, stalls or trips the first time the pump starts against a running house.
Why the essential column matters. Tick only the fridge and pump as essential and the same arithmetic gives running 1,150 W, peak 1,150 + 1,200 = 2,350 W, recommended 2,820 W (3.53 kVA): a materially smaller, quieter and cheaper machine keeps water and food safe, which is the entire economics of outage planning in one comparison the results print side by side.
Reading your results
Shop by the recommended rating, then sanity-check the running total against runtime claims (manufacturers quote runtime at half load, so a machine whose rating doubles your running total will roughly meet its brochure hours). Treat a peak dominated by one huge surge (a large air conditioner’s compressor) as a design problem before a size problem: soft-start kits cut that surge dramatically and can drop the whole installation a class. And read the kVA line when comparing imports and commercial sets, whose plates are often in kVA: at 0.8 PF, kVA times 0.8 is watts, and confusing the units oversizes or undersizes by a quarter.
Limitations
- One-at-a-time starting is an assumption, stated. Two motors starting in the same instant (a pump and a compressor after power returns, both thermostats calling) produce a higher peak; transfer equipment with load management, or staggered restart delays, is how real installations enforce the assumption, and the calculator cannot enforce it for you.
- Surge figures are the softest inputs. Starting watts vary with load state (a pump starting against pressure, a compressor against head pressure) and with voltage dip during the event itself; headroom exists precisely for this softness, and 20 percent is a floor for motor-heavy lists, not a luxury.
- Altitude and temperature derate real machines. Naturally aspirated engines lose roughly 3 to 4 percent per 300 metres of altitude and sets derate in heat; a borderline rating at sea level is an undersized one in the mountains, a correction the buyer must apply from the manufacturer’s derating table.
- Sizing is not installing. Transfer switches, earthing, fuel storage, ventilation, noise rules and permits sit outside this arithmetic entirely; the brief’s own line governs: code and service sizing are a separate verified exercise.
Frequently asked questions
Why does a 3,450 W peak need a 4,140 W generator?
Because the peak figure is a best-case arithmetic moment (nameplate surges, one start at a time, a new machine at sea level on a mild day) and real life taxes every term: surges run high against pressure, engines age and derate, and an extra load always joins the list. The headroom margin converts the arithmetic peak into a rating that survives contact with those taxes; it is also why the margin is yours to set, with 20 percent prefilled as the common planning floor for motor-heavy lists, rather than a hidden fudge inside the result.
What are starting watts, and where do I find them?
They are the brief draw while a motor spins up from rest, when it behaves almost like a stalled rotor: typically two to five times the running draw for a second or less. Sources, in order of trust: the appliance manual’s electrical data, the motor nameplate (locked-rotor amps times voltage approximates them), manufacturer sizing guides (the basis of this page’s presets), and last, generic tables. Compressor loads (fridges, air conditioners, pumps) are where the number matters; for resistive loads starting equals running and the distinction vanishes.
Running watts or rated watts: which number on the box matters?
Both, in sequence. A generator’s plate carries a rated (continuous) wattage and often a larger maximum or surge wattage. Your running total must sit under the continuous rating with room (that is the headroom test here), and your computed peak should sit under the set’s surge capability, which its continuous rating plus your margin approximates conservatively. Marketing leads with the biggest number on the box; sizing reads the continuous one, and this calculator’s recommendation is expressed against the continuous rating for exactly that reason.
Can I just add up nameplate watts and buy that?
That method oversizes some houses and strands others. It oversizes where loads genuinely never coincide (diversity exists for this), and it strands any list with motors, because it ignores starting surges entirely: the pump example needs 2,250 W by that method and 3,450 W in the first second of reality. The two-figure method on this page (running with diversity, peak with the largest surge) is the standard planning shape because it prices both truths instead of averaging them away.
Related calculators
Method: running total with user diversity; peak as full running total plus the largest single starting increment; recommended rating with user headroom, expressed in W and kVA. Last reviewed: 14 October 2026.
