Why spreadsheets fail at load calculation
Three failure modes show up again and again.
Demand factors are applied at the wrong level. A 40% factor belongs to a subtotal, not to each appliance. Spreadsheets that multiply every row by the same factor produce a number that looks plausible and is wrong by 30–40%.
The calculation and the drawing drift apart. The plan shows a 7 kW cooktop, the schedule lists 6 kW, the calculation was run last month on the 6 kW version. Nobody notices until the panel is ordered.
The method is invisible. A cell with =SUM(B2:B14)*0.4 does not tell a reviewer which code edition, which method, or which loads were included. When an AHJ asks how you arrived at 97 A, "the spreadsheet says so" is not an answer.
Software does not fix bad engineering, but it keeps the inputs, the method and the output in one place — which is most of the battle.
The three methods you actually need
Almost every residential and light-commercial job falls into one of three calculation paths. Knowing which one applies is the real skill; the arithmetic is trivial.
NEC Article 220, standard method. General lighting from Table 220.12 (3 VA/ft² for dwellings, 3.5 VA/ft² for offices), small-appliance and laundry circuits at 1,500 VA each, then the lighting demand schedule of 220.42: 100% of the first 3 kVA, 35% of the remainder up to 120 kVA. Receptacle outlets in non-dwellings get the 220.44 treatment — 100% of the first 10 kVA, 50% of the rest. Ranges and dryers use Table 220.55 and 220.54 rather than raw nameplate.
NEC 220.82, optional method for dwellings. Faster and usually smaller: 3 VA/ft² plus 1,500 VA per small-appliance circuit plus 1,500 VA laundry plus the nameplate of fixed appliances, then 100% of the first 10 kVA and 40% of everything above it. Heating or cooling is added at 100% — the larger of the two, not both (220.60 covers noncoincident loads).
IEC 60364 with national diversity tables. IEC does not publish a demand-factor table the way the NEC does. It sets the coordination rule (design current ≤ protective device rating ≤ conductor capacity) and leaves diversity to the designer or to national annexes and guides — IS 732 in India, BS 7671 guidance in the UK. This is the biggest conceptual gap between the two systems: NEC gives you a number, IEC gives you a framework and expects a justification.
Apartment, house, office: what actually changes
The same engine produces different answers because the load mix changes:
| Item | Apartment | Detached house | Small office |
|---|---|---|---|
| Dominant load | Cooking + water heating | HVAC + EV charging | Lighting + receptacles + IT |
| Lighting basis | 3 VA/ft² (NEC 220.12) | 3 VA/ft² | 3.5 VA/ft² |
| Usual service | 60–100 A single-phase | 150–200 A split-phase | 200–400 A three-phase |
| Diversity assumption | High (cooking peaks are short) | Low (HVAC runs long) | Medium, strongly time-of-day dependent |
| Typical failure | One RCD/GFCI for everything | Missing EV load in the total | Counting every receptacle at face value |
| Extra design work | Wet-area protection | Load management, generator inlet | Harmonic loads, neutral sizing |
Two notes on that table. Service sizes are typical practice, not code minimums — the calculation decides. And "typical failure" is where reviews actually catch problems, so it is a good place to start a self-check.
A worked example: 2,000 ft² house
Numbers make the method concrete. Single-family dwelling, 2,000 ft², 240/120 V single-phase, heat pump, electric range, electric water heater, no EV yet.
Using the 220.82 optional method:
- General lighting: 2,000 × 3 VA = 6,000 VA
- Two small-appliance circuits: 3,000 VA
- Laundry circuit: 1,500 VA
- Fixed appliances: range 8,000 + water heater 4,500 + dishwasher 1,200 + disposal 900 = 14,600 VA
- General subtotal: 25,100 VA
- Demand: 100% of the first 10,000 VA + 40% of 15,100 VA = 10,000 + 6,040 = 16,040 VA
- Heat pump heating 7,200 VA versus cooling 5,400 VA: add 7,200 VA (larger, not both)
- Total: 23,240 VA → 96.8 A at 240 V → 100 A service
Now the same house under the standard method: lighting 6,000 VA becomes 3,000 + (3,000 × 0.35) = 4,050 VA; small appliance and laundry stay at 100%; the range goes through Table 220.55; HVAC adds 7,200 VA. The result is meaningfully higher — around 126 A, which would push the service to 150 A.
That gap is not a bug. It is the reason the optional method exists, and it is the first thing a reviewer will ask about. A tool that shows the demand breakdown line by line lets you answer in seconds.
What good software does that a spreadsheet cannot
- Keeps the method explicit. Each subtotal carries the rule it came from, so the output reads like a calculation sheet rather than a guess.
- Pulls loads from the drawing. Circuits, devices and nameplates come from the same model as the plan, so the "plan says 7 kW, calc says 6 kW" drift disappears.
- Recomputes downstream. Change the heat pump and the service, feeder, breaker and panel schedule update together.
- Flags the impossible. A 96.8 A load on a 100 A service is legal but leaves no margin; a tool should mark that.
- Exports something a reviewer accepts. A load summary in the permit package beats a spreadsheet screenshot, and it feeds the panel schedule directly.
Formats, limits and calculation outputs are detailed in the documentation.
If you want to see the calculation sitting next to the plan instead of in a separate file, open the interactive demo and run a small dwelling through it.
Where the numbers stop being reliable
Honest limits, in order of how often they bite:
Nameplate data is assumed, not measured. If the actual heat pump model is not selected yet, someone typed a placeholder, and every downstream number inherits that guess.
Diversity is a judgement, not a formula. IEC-style calculations especially depend on assumptions about simultaneity. Software can apply your factor; it cannot know whether the factor is right for a 40-person office with a kitchenette.
Code editions differ. NEC 2020, 2023 and 2026 are not identical, and local amendments sit on top. Verify every rule against the edition your AHJ enforces.
Load management is outside the arithmetic. If the design relies on an EV charger interlocked with a dryer, the calculation must reflect the control scheme, not the sum of nameplates. And the client who adds a hot tub next year is not in your model — document the spare capacity you assumed.
A workflow that survives review
- Collect nameplates first. Model numbers, not guesses. Mark unknown values as provisional.
- Pick the method and write it down. Optional method for a dwelling, standard method where the AHJ expects it, IEC/national diversity elsewhere.
- Build the load list from the model. Circuits and devices already carry ratings.
- Review subtotals, not the total. Errors cluster in appliance lists and HVAC assumptions.
- Compare against the service and feeder. Utilization above roughly 90% deserves a second look.
- Export the summary with the project. It is part of the deliverable, not a private note.
For the calculations that follow — conductor selection and voltage drop — the guides on cable sizing and voltage drop cover the formulas and derating factors.
Ready to try it on a real plan? Start with the free tier and demo, then check pricing when you hit the project limit.
Sources
- NFPA 70, National Electrical Code — Article 220 (branch-circuit, feeder and service calculations), Table 220.12, 220.42, 220.44, 220.55, 220.82: nfpa.org/70
- IEC 60364 series, Low-voltage electrical installations — IEC Webstore
- Schneider Electric, Electrical Installation Guide (design rules and diversity practice): electrical-installation.org
- Bureau of Indian Standards — IS 732 (wiring installations) and IS 3043 (earthing): bis.gov.in
Code references are for orientation. Editions and local amendments vary — verify against the text your authority having jurisdiction enforces. This article is not legal or engineering advice.
Software output is accepted when it shows the method, the inputs and the arithmetic — essentially a calculation sheet. The AHJ reviews the engineering, not the brand of the tool. Attach the code edition and the assumptions used.
Yes. The EV charger is a fixed load and belongs in the general load list under the 100%/40% treatment, unless a load-management scheme justifies a lower value. Check 220.82 and any local amendment before relying on the reduction.
IEC 60364 sets coordination rules and leaves diversity factors to national annexes and engineering judgement. The NEC prescribes demand factors directly in Article 220. The physics is the same; the audit trail is different.
It converts the calculated load to amps and suggests a standard rating, but the final choice involves conductor ampacity, continuous-load rules and utility requirements. Treat it as a starting point and confirm it against your code.
Balanced three-phase loads divide by √3 × line voltage. Neutral sizing needs separate attention with harmonic-rich loads (IT equipment, LED drivers, VFDs), because triplen harmonics add in the neutral.
Yes, but model dwelling units and house loads separately, then apply the appropriate demand rules to each. A single diversity factor for a whole building is the most common multifamily error.
Design it in the browser
Turn the plan into circuits, a panel schedule, a single-line diagram and a calculated bill of materials. Start free or open the demo.
Open demo See pricing