Connected load is not demand load
Add up the nameplate rating of everything in a building and you get a number the building will never draw, because the loads do not all run at once and most of them do not run at their rating when they do run. Sizing a service on that sum produces equipment several times larger than anything the installation will use.
Codes handle this with DEMAND FACTORS applied by load type: the first portion of a category at full value and the remainder at a reduced one, with different schedules for lighting, receptacles, cooking appliances, motors and heating. The schedules are empirical — fitted to measured behaviour of real buildings of each type — rather than derived, which is why they differ by occupancy and by country.
The most reliable input of all is measurement. Where a building already exists, codes permit sizing from its recorded MAXIMUM DEMAND over a representative period plus an allowance for the new load, and that route beats any calculation because it contains the actual diversity of the actual occupants. The calculated method exists for buildings that do not yet exist.
The direction of the error matters. An oversized service is expensive and wasteful but safe; an undersized one nuisance-trips or overheats. So the demand factors are set conservatively and a calculated result is usually above what will be measured — which means a calculated figure and a metered figure disagreeing is normal rather than evidence that one is wrong.
- P_i
- connected load of each category
- k_i
- demand factor for that category — empirical, and occupancy-specific
- V_L
- LINE voltage; the √3 converts it to the phase relationship
- pf
- power factor — the reason apparent and real power differ
Continuous loads carry a twenty-five per cent uplift
A load expected to run for three hours or more is treated differently from one that cycles. Conductors and overcurrent devices are sized at a hundred and twenty-five per cent of a continuous load's current rather than at its actual value.
The reason is thermal rather than electrical. A breaker's rating is established in a test regime that allows it to shed heat between operations; held at its rating continuously, it and the terminations around it run hotter than that test represents. The uplift restores the margin the rating assumed.
So the same current is a different design number depending on how long it lasts. Lighting in a commercial building, a heating circuit, an electric vehicle charger and a continuously-run motor all attract it; a cooker, a kettle and a socket circuit generally do not.
Vehicle charging is where this is encountered most often now, and it compounds: a charger is continuous by definition, so a nominal thirty-two-ampere charger needs a forty-ampere circuit, and a domestic board that could accommodate the first frequently cannot accommodate the second.
Three-phase: the root three, and what imbalance costs
In a balanced three-phase system, total power is the line voltage times the line current times the power factor times the SQUARE ROOT OF THREE. That factor is the single most commonly omitted term in electrical arithmetic, and omitting it under-states the current by about forty-two per cent or over-states the power by the same.
It arises because the three phases peak at different moments, so the line-to-line voltage is not simply twice a phase voltage and the currents do not simply add. It applies to balanced systems only, which is why balance is a design objective rather than a tidiness preference.
IMBALANCE has real costs. Loading one phase harder than the others means the conductor and the supply transformer are limited by the worst phase while the others are under-used, so the whole board is effectively derated to three times its heaviest phase. Unbalanced current also returns through the NEUTRAL, which in a badly balanced installation can carry more than the phases it serves.
Non-linear loads add a second mechanism that balancing cannot fix. Switch-mode power supplies, LED drivers and variable-speed drives draw current in pulses rich in the third harmonic, and third harmonics from the three phases ADD in the neutral rather than cancelling. A perfectly balanced modern office can have a neutral current far above any phase, which is why neutral conductors in such installations are not downsized and are sometimes oversized.
Motors are sized from a table, except where they are not
For conductor sizing and short-circuit protection, codes require the motor's full-load current to be taken from a PUBLISHED TABLE against its horsepower or kilowatt rating and voltage — not from the nameplate of the motor actually installed.
That reads as backwards and is deliberate. The tables are set above typical nameplate values so that a motor replaced later with a different make does not invalidate the installation, and so that the conductor is sized for the population rather than for one specimen.
The exception is OVERLOAD protection, which uses the actual nameplate current, because overload protection exists to protect that particular motor from its own heating. So a single motor circuit is designed from two different current figures, each for a different purpose, and interchanging them is a recognised error.
Starting current is the third figure and it changes the protection rather than the conductor. A direct-on-line motor draws several times its full-load current for a few seconds, which a conductor tolerates easily and a protective device must be selected not to trip on — hence the higher multiples permitted for motor short-circuit protection and the existence of soft starters and drives, which reduce the inrush rather than the running load.
A panel's spaces, poles and circuits are three different counts
A panelboard is rated for a number of SPACES, and a space is not a circuit. A single-pole breaker occupies one space and protects one circuit; a two-pole breaker occupies two spaces and protects one circuit at a higher voltage or a shared neutral pair.
Tandem or half-size breakers put two protective devices in one space, and they are only permitted where the panel is LISTED for them and only in the positions the listing allows — often a subset of the board rather than all of it. A board's maximum circuit count is therefore a separate figure from its space count, and both are marked.
This is why a board that looks half empty can be full. Adding a circuit that needs two poles to a board with two scattered single spaces does not work; the spaces must be adjacent, and getting them adjacent means moving existing circuits — which is work, and which changes the phase each existing circuit sits on.
That last point links back to balance. In a board with alternating phase assignment, moving a breaker one position moves its load to a different phase, so a re-arrangement done for space is also a re-arrangement of the balance, and it is worth doing both deliberately at once rather than discovering the second afterwards.
The losses have to go somewhere, and it is usually a small room
Every transformer, drive, breaker and conductor converts a fraction of what passes through it into heat. Individually the percentages are small; collected into a switchroom they are a continuous thermal load that has to be removed, and it is a genuine input to the building's cooling rather than a detail.
The load is not constant. Transformer losses split into a no-load component that is there whenever the unit is energised, and a load component that grows with the SQUARE of the current — so a transformer at full load produces four times the copper loss it produces at half load, and an electrical room's heat output is strongly dependent on how loaded the building is.
There is a feedback that makes it worse. Equipment ratings are established at a reference ambient, and a room running above that ambient derates the equipment in it; a derated transformer carrying the same load runs hotter and dissipates more. A switchroom whose cooling has failed does not simply get warm — it moves down its own derating curve.
Which is why an electrical room's ventilation or cooling is sized from the equipment's published losses at the expected load rather than from the room's volume, and why the calculation belongs in the mechanical design as much as in the electrical one.
Lighting: lumens are not lux, and the room decides the difference
A lamp emits LUMENS; a surface receives LUX. Dividing the total lumens installed by the floor area gives a number that is always higher than what will be measured, because not all of the light reaches the working plane.
The correction is the COEFFICIENT OF UTILISATION, and it depends on two things: the shape of the room, expressed as a cavity ratio relating its height to its plan, and the REFLECTANCES of the ceiling, walls and floor. A tall narrow room loses light to its walls; a room with dark surfaces absorbs what a pale one would bounce back down.
The reflectance term is larger than people expect. Redecorating a space in dark colours can require a substantially higher installed load to hold the same illuminance, and a lighting design produced against pale assumed surfaces will under-perform in a room finished dark. Reflectance is a lighting decision made by an interior designer.
The second correction is the LIGHT LOSS FACTOR, which is not about the room but about time: lamps depreciate, luminaires and surfaces collect dirt, and lamps fail between maintenance visits. A design is done to the MAINTAINED illuminance — the value at the worst point of the cycle — so a newly commissioned installation should measure comfortably above its design figure. One that measures exactly at target on day one is already destined to fall below it.
Emergency lighting is an evenness problem, not an average one
Escape lighting is not specified as an average level over an area, and treating it that way produces a compliant average with unusable dark patches. The requirements are a MINIMUM along the path of egress and a limit on the ratio between the brightest and darkest points on it.
The uniformity limit exists because of how eyes work. Vision adapts to the brightest thing in view, so a bright pool next to a dark stretch makes the dark stretch effectively invisible — worse, for someone moving through it, than a uniformly dimmer corridor. A high maximum-to-minimum ratio is a hazard even when the average is generous.
That inverts the usual spacing logic. Ordinary lighting is spaced to achieve an average efficiently, which favours fewer brighter fittings; emergency lighting is spaced to hold a ratio, which favours more, dimmer, closer ones. The two layouts are not the same and a fitting count derived from the first will fail the second.
Everything on this page is a screening calculation. A real distribution design is a load study with the actual schedules and diversities, a real lighting design is a point-by-point computation over the actual room with the actual luminaire photometry, and both are checked by measurement after installation. These pages place a design in the right region and say which standard's assumptions they used to do it.
Calculators that use this method
Basis
- NFPA 70 (National Electrical Code) Article 220 — branch circuit, feeder and service load calculations, the demand factor schedules, and 220.87 for sizing from measured maximum demand in an existing building.
- NFPA 70 Articles 210.19 and 210.20 for the 125 per cent continuous load provision, and Article 625 for electric vehicle supply equipment as a continuous load.
- NFPA 70 Article 430 — motor full-load current tables for conductor and short-circuit protection, and the nameplate current used for overload protection.
- NFPA 70 Article 408 and UL 67 for panelboards: space counts, listed positions for tandem breakers, and maximum circuit counts.
- IEC 60364-5-52 and BS 7671 Appendix 4, for the equivalent European treatment of diversity, grouping and neutral conductor sizing with harmonic currents.
- IEEE Std 519 on harmonics, and the triplen harmonic addition in the neutral of a three-phase four-wire system.
- IES Lighting Handbook and the zonal cavity method — coefficient of utilisation from room cavity ratio and surface reflectances, and the light loss factor behind maintained illuminance.
- NFPA 101 Life Safety Code and IBC 1008, emergency egress illumination: the minimum along the path and the maximum-to-minimum uniformity ratio.
- Manufacturers' published no-load and load losses for distribution transformers, and IEEE C57.12 for the ambient-derating relationship described here.