The arithmetic that looks right and isn't
Stand in a finished kitchen and start adding. The range is 12 kilowatts. The microwave wants 1,500 watts, the dishwasher another 1,200, the disposal 900. Down the hall the dryer pulls 5,000, the water heater 4,500, and the air handler another few thousand on a cold morning. Add the lighting, the receptacles, the bathroom heater someone insisted on. Tally every nameplate in the house and you will land somewhere north of 60,000 watts — a 250-amp service for a three-bedroom home.
That number is wrong, and the reason it's wrong is the most important idea in service sizing. You will never run all of those at full draw at the same instant. The oven cycles. The dryer finishes. Nobody toasts bread while running the disposal while the water heater recovers while the dryer spins, all at peak, for the sustained period that actually heats a conductor. A house is not the sum of its appliances. It's a probability distribution, and the National Electrical Code sizes services to that distribution, not to the worst case that physics technically permits.
The tool for this is the demand factor, and learning to read it correctly is what separates a load calculation that passes from one that funds a service you'll never need.
Diversity is a real electrical phenomenon, not a fudge
The word the trade uses is diversity. Across many circuits, the peaks don't line up. Engineers measured this on real services decades ago: the actual coincident demand of a dwelling is a fraction of its connected load, and that fraction is remarkably stable across homes. The Code captures the measurement in a set of demand factors so that an electrician with a calculator gets the same defensible answer an engineer with load-recording equipment would.
This is why a load calculation is not a guess and not a sum. It's a procedure that applies experimentally-derived percentages to categories of load. Get the categories right and the arithmetic follows.
Walking the standard method
The standard calculation in Article 220 builds the number in layers. It's worth seeing the shape of it once, because the shape is the lesson.
General lighting and general-use receptacles start at 3 volt-amperes per square foot of living space. You don't count individual outlets — the Code rolls all the lighting and the general receptacles into one square-footage figure. A 2,000-square-foot home contributes 6,000 VA here.
The small-appliance and laundry circuits are added as flat allowances: two kitchen small-appliance circuits at 1,500 VA each, plus a laundry circuit at 1,500 VA. That's 4,500 VA, regardless of what gets plugged in.
Now comes the first demand factor, and it's the one that reframes everything. You take that combined general load — lighting plus small-appliance plus laundry — and you do not count it all. The first 3,000 VA counts at 100 percent. Everything above 3,000 (up to 120,000) counts at 35 percent. The Code is saying, in plain numbers, that most of a home's general load is statistically idle at any given moment. Our 2,000-square-foot example: 6,000 plus 4,500 is 10,500 VA connected, but after the demand factor it lands at 3,000 + (7,500 x 0.35) = 5,625 VA. Nearly half of it simply doesn't count, because nearly half of it is never simultaneously on.
The big appliances get their own logic
The heavy loads aren't folded into that 35 percent — they each carry a tailored demand factor that reflects how that appliance actually behaves.
An electric range is the cleanest example. A single household range rated up to 12 kW is not counted at 12 kW. The range demand table assigns it a demand of 8 kW. The reasoning is thermal and behavioral: an oven heats up, then cycles its elements to hold temperature, and you rarely run every burner and the oven at full simultaneously for long. The sustained, conductor-heating demand is lower than the nameplate, and the table encodes that.
An electric dryer is counted at 5,000 watts or its nameplate rating, whichever is larger. A run of four or more fastened-in-place appliances — think water heater, dishwasher, disposal, built-in microwave — can be taken at a 75 percent demand factor as a group, again because they don't all peak together.
And then there's the rule that surprises people the first time: noncoincident loads. Your electric heat and your air conditioning will, by definition, never run at full output in the same season. The Code lets you count only the larger of the two and discard the smaller entirely. Counting both would be sizing for a January afternoon that is also a July afternoon — a moment that does not exist.
Where the number finally lands
Stack those layers — the demand-adjusted general load, the range at its table value, the dryer, the appliance group at 75 percent, the larger of heat-or-cool — and the total for our example house settles comfortably under 24,000 VA. Divide by 240 volts and you're around 95 to 100 amps of calculated demand. A 100-amp service may be tight depending on the specifics, and a 150 or 200 gives headroom, but you are nowhere near the 250 the naive sum demanded. The diversity was real money: a smaller service, smaller feeder, smaller everything, all justified by a procedure rather than a hunch.
This is also why two electricians can defend different service sizes for the same house and both be right. The standard method and the optional method in the Code apply different demand factors, and the optional calculation — which lumps most loads together and takes a 40 percent factor on everything past the first 10 kVA — often yields a smaller number still. The math isn't arbitrary; it's two validated models of the same statistical reality.
Why this matters beyond passing inspection
Undersize and you nuisance-trip, overheat a feeder, and call it back. Oversize and you've sold the customer copper and gear they're paying interest on for thirty years, plus a service that may never see half its rating. The demand factor is the instrument that finds the honest middle — large enough for the coincident peak, no larger than the data supports.
The trap is that demand factors are easy to remember wrong. The 35 percent applies to the general load after the first 3,000 VA, not to everything. The range value comes from a table that changes once you exceed 12 kW or add a second range. Heating and cooling are noncoincident only when they genuinely can't run together. Miss one of these and your calculation drifts, quietly, toward a number you'll have to defend to an inspector who has the table open.
The field reality
None of this is hard math. It's bookkeeping with the right percentages applied to the right buckets, and the percentages live in tables you should never be reciting from memory on a roof or in a crawlspace. This is exactly the kind of work where a wrong constant — a 40 where a 35 belongs, a nameplate counted where a demand value was called for — survives all the way to a service that's a size off.
Voltly keeps the standard and optional load calculations on your phone, offline, with the range, dryer, and noncoincident-load logic built into the form instead of left to memory. You enter the square footage and the appliances; it applies the current demand factors and hands you the calculated amps you can stand behind. The point isn't to skip the thinking — it's to make sure the diversity you're counting on is the diversity the Code actually grants. If you'd rather check the table than trust your recall on the next service upgrade, take a look: https://voltly.lumenlabs.works