Metrics & standards

The average passed and the desk is dark: uniformity, Uo and the measurement grid

Why an installation can hit its average illuminance and still fail, what Uo is actually the ratio of, how the area you average over decides the answer, spacing-to-height ratio, and the three clauses that make uniformity verifiable on site.

Written by Yilian Zhikong Technology (Shenzhen) Co., Ltd. · Last reviewed August 2026

An office is handed over at an average of 512 lx against a 500 lx specification. The contractor's calculation says compliant, the commissioning readings say compliant, and two people near the window wall are working at 280 lx under a fitting spacing that never suited the room. Nobody has falsified anything. The specification asked for an average and got one.

What Uo is the ratio of

Uniformity in EN 12464-1 is U_o, the ratio of the minimum illuminance to the average illuminance, and the whole content of the requirement is in the words that follow it: over the task area. Change the area you average over and you change both terms of the ratio.

This is where most disputes live. A contractor who computes the average over the whole room floor includes the space under every fitting and the corridors of light between them, and gets a comfortable number. The specification meant the area where the work happens — the desks, the bench, the aisle — which is smaller, positioned where the furniture went rather than where the grid is regular, and often sits at the unluckiest part of the layout. Two honest calculations, two different answers, because the area was never defined.

The grid decides the minimum

U_o has a minimum in it, and a minimum is only as low as your sampling finds. A calculation on a coarse grid never lands on the darkest point; a calculation on a grid whose spacing happens to align with the luminaire pitch can sample every cell at the same phase of the light pattern and report a uniformity the room does not have.

The standard handles this by defining the grid rather than leaving it to the person doing the calculation: cell size derives from the dimensions of the area being assessed, with an upper bound, and the grid is required to avoid coinciding with the luminaire spacing. Take the current rule from the edition your contract names — it has changed between editions, and it is one of the places where quoting a remembered figure gets people into trouble.

Two practical consequences. First, a lighting calculation submitted without its grid visible is not checkable, and should be returned. Second, when you verify on site, use the same grid as the calculation. Comparing a nine-point site measurement against a several-hundred-point calculated grid is comparing two different quantities, and the site figure will usually look better than the truth.

Spacing to height, and why scallops appear

Uniformity is mostly decided before any of this, by the relationship between how far apart the fittings are and how far they are above the working plane. Every luminaire distribution has a spacing-to-height ratio beyond which its beams stop overlapping sufficiently at the working plane, and it is published in the photometric data. Exceed it and you get scallops: bright patches under the fittings, dark bands between them, an average that is fine and a U_o that is not.

The ratio is a property of the distribution, not of the fitting's size or wattage, which is why swapping a fitting for a nominally equivalent one with a narrower beam during value engineering can wreck a layout that was designed correctly. This is the single most common way a retrofit that met its lumen target still fails on site — the mechanism is worked through in illuminance shortfall after retrofit.

A worked check you can do on a submitted calculation

You do not need the software to sanity-check a claim. Take the calculation's own figures:

Then ask three questions of the sheet: what area was the average taken over, what grid was used, and where in the room is E_min. If the answer to the third is "in a corner behind a column", the number may be defensible. If it is in the middle of a desk run, the layout is wrong regardless of what the average says.

What changes between calculation and building

Three things move the site figure away from the calculated one, and all three are worth anticipating rather than arguing about at handover.

1. Surface reflectances. Calculations are run with assumed ceiling, wall and floor reflectances. A dark feature wall, a glazed elevation at night or grey carpet instead of the assumed light one removes inter-reflected light, and inter- reflected light is disproportionately what fills the gaps between fittings — so reflectance error hits uniformity harder than it hits the average.
2. Furniture and partitions. High storage and screens installed after the lighting design cast exactly the shadows the calculation's empty room did not have.
3. Position tolerance. Ceiling grid modules, structural constraints and services coordination move fittings from where the calculation put them. A layout that only just met its spacing-to-height ratio has no margin for that.

Three clauses that make uniformity enforceable

1. U_o stated with the area it applies over, the reference plane height, and the requirement that task areas be located on the drawing rather than assumed to be the whole floor.
2. The calculation grid shown in any submitted lighting calculation, with the reflectances and maintenance factor used printed on the same sheet.
3. Site verification on the same grid as the calculation, at the same reference plane, with the measurement conditions recorded — meter class, stabilisation time and the treatment of daylight. Without that last clause the acceptance reading is not comparable with the thing it is supposed to verify.