Why Lumens and Lux Are Not Interchangeable

Why this matters

Somebody is going to hand you a lumen total and ask what the space will read. The temptation is to divide by the floor area, and the answer that comes out is wrong in one direction every time: too high, sometimes by a factor of three and a half. That number then goes on a proposal, and when the meter disagrees on the walkthrough the shop eats the difference in added fixtures. The gap is not a fudge factor. It is two named quantities the division silently sets to 1.0, and both of them are knowable before anyone orders anything.

The gate: when may you divide lumens by area?

The lumen method, which is the arithmetic every design tool runs underneath, says:

E (maintained, fc) = (number of luminaires x lumens each x CU x LLF) / area in ft2

CU is the coefficient of utilization, the fraction of a luminaire's output that reaches the work plane in this particular room. LLF is the light loss factor, the fraction still arriving at the end of the maintenance cycle. Dividing lumens by area is that equation with CU = 1.0 and LLF = 1.0, which would mean a room that loses no light to walls, ceiling, floor or the luminaire's own geometry, and a fixture that never ages or gets dirty. There is no such room and no such fixture, so the gate never opens. What varies is how far from 1.0 the pair sits, and that is a property of the room, not of the box.

What the room does to the flux

CU falls as the room gets tall and narrow relative to its footprint, and as its surfaces get darker. The shape variable is the room cavity ratio:

RCR = 5 x h x (L + W) / (L x W), where h is the height from the work plane to the luminaire plane.

A low RCR means a wide flat room where a bounced ray finds the work plane quickly. A high RCR means a deep well where bounced rays die in the walls. Reflectances are quoted as a ceiling/wall/floor triple, and the common tabulated case is 80/50/20 percent. Drop the walls from 50 to 30 percent and every interreflected ray that used to help now mostly does not.

The CU value itself is not derivable from the room alone. It comes out of the manufacturer's CU table for that specific luminaire, read at your RCR and your reflectance triple, and that table is generated from the photometric file. Two luminaires with identical lumen ratings and different distributions produce different CU in the same room, which is the whole reason distribution is specified at all.

What time does to the flux, and why a correction here is a re-basing

The light loss factor is a product of terms, and they fall into two families.

• Recoverable losses, which cleaning or relamping restores: lamp or LED lumen depreciation, luminaire dirt depreciation, room surface dirt depreciation, and lamp burnout where the source type has one.
• Non-recoverable losses, baked into the installation: driver or ballast factor, ambient temperature factor, supply voltage variation, and any tilt or orientation penalty away from the tested position.

Which terms a given calculation used is the definition of "maintained" in that report, which is why the report has to name them. A maintained illuminance is not an initial illuminance with something still to be subtracted; it is an initial illuminance already re-based downward by that product. Applying an LLF to a number that is already maintained applies it twice, and it is the commonest arithmetic error in the subject. On the office case below, taking the maintained 43.4 fc and multiplying by the 0.874 LLF again yields 37.9 fc, which is 5.5 fc and 12.6 percent low against a figure that was already correct.

The conventions an illuminance figure carries

An illuminance number is not a measurement until it states four things, and it states them every time it is written, including in table cells:

1. The plane. Horizontal, vertical, or normal to a named surface. The horizontal reading and the vertical reading in the same spot are different numbers describing different tasks.
2. The height above finished floor. A horizontal reading at 30 in and one on the floor in the same aisle are not comparable.
3. Initial or maintained.
4. Which loss factors the maintained figure already contains.

"48 fc" is a rumour. "43.4 fc maintained, horizontal, at 2.5 ft above finished floor, LLF 0.874 from lumen depreciation and dirt depreciation only" is a measurement someone else can check.

What the method delivers is an average, on one plane

Two limits on the answer, both of which get forgotten because the equation produces a single confident number.

It is an average over the whole area. The floor divisor in the equation is the room's area, so the result is what you would get by spreading the delivered flux evenly across it. Nobody stands at the average. A corner two feet inside the wall and a point directly under a luminaire can sit well either side of it, and the spread between them is a spacing question that a point calculation answers and this equation cannot. A space can pass on average and fail in the corner it was built for.

It is horizontal, at the height you declared. The lumen method delivers horizontal illuminance on the work plane and says nothing at all about the vertical surfaces, which is where a large share of real tasks live: a pick face, a shelf label, a panel schedule, a face across a counter. Sizing a warehouse from an average horizontal figure and then discovering the labels are unreadable is the same error as sizing it from lumens per square foot, one layer further in.

Both limits argue the same way: the lumen method is the right tool for asking whether the quantity of light is roughly correct, and the wrong tool for asking whether any particular person can see any particular thing.

Outcome one: the office bay

Twelve luminaires at 4,000 lumens each, so 48,000 lumens installed, over a 20 ft by 30 ft room, 600 ft2. Ceiling 9.0 ft, work plane 2.5 ft, so h = 6.5 ft. Reflectances 80/50/20. Clean environment, 24-month cleaning interval.

RCR = 5 x 6.5 x (20 + 30) / 600 = 32.5 x 0.0833 = 2.71.

CU read from the manufacturer's table at RCR 2.7 and 80/50/20: 0.62 for this walkthrough. Treat that as illustrative; the real one comes off the table for the luminaire you are actually quoting.

LLF = lumen depreciation 0.95 x dirt depreciation 0.92 = 0.874. Driver factor 1.00 and ambient factor 1.00 at an office ceiling, so neither moves it.

E = 48,000 x 0.62 x 0.874 / 600 = 26,010 / 600 = 43.4 fc maintained, horizontal, at 2.5 ft above finished floor.

The naive division gives 48,000 / 600 = 80 fc, which overstates by a factor of 1.84.

Outcome two: the shop bay

Same twelve luminaires, same 48,000 lumens, same 600 ft2 footprint. Ceiling 16.0 ft, bench work plane 3.0 ft, so h = 13.0 ft. Reflectances 50/30/20 with exposed structure and dark walls. Dirty environment, same 24-month interval. Ambient temperature at the ceiling elevated enough that the manufacturer's temperature curve derates output.

RCR = 5 x 13.0 x 50 / 600 = 65 x 0.0833 = 5.42.

CU at RCR 5.4 and 50/30/20: 0.38, again illustrative and again owned by the table.

LLF = 0.95 lumen depreciation x 0.78 dirt depreciation x 0.95 ambient temperature factor = 0.704.

E = 48,000 x 0.38 x 0.704 / 600 = 12,841 / 600 = 21.4 fc maintained, horizontal, at 3.0 ft above finished floor.

The naive division gives the same 80 fc, overstating by a factor of 3.74.

What the pair proves

Identical installed flux and identical floor area, and the delivered maintained illuminance differs by 43.4 / 21.4 = a factor of 2.03. Note also that the two figures sit on different planes, 2.5 ft and 3.0 ft above finished floor, which is part of the point: they were never the same measurement even before the numbers diverged.

Split the factor of 2.03 into its causes and you get the design lever. The CU ratio is 0.62 / 0.38 = 1.63, and the LLF ratio is 0.874 / 0.704 = 1.24, and 1.63 x 1.24 = 2.02, which closes to the 2.03 within rounding. So roughly two thirds of the loss is room geometry and surface reflectance, and one third is maintenance and thermal environment. In the shop bay, painting the walls and picking a distribution suited to a high cavity ratio moves more than any lamp change will, and a shorter cleaning interval moves the rest.

What would flip this: if the shop bay had light surfaces and a clean environment, its CU and LLF would climb toward the office case and the naive division would still be wrong, just less wrong. The gate does not open, it only narrows.

Checking your own figures

• Gate stated and tested. The division was allowed only at CU = 1.0 and LLF = 1.0. Office actual: CU 0.62, LLF 0.874. Shop actual: CU 0.38, LLF 0.704. Neither pair is 1.0, so the naive division was refused in both.
• No double application. The 43.4 fc and 21.4 fc each contain their LLF once. Re-applying 0.874 to the office figure gives 37.9 fc, a 12.6 percent understatement, and it was not done.
• The factors decompose to the answer. 1.63 x 1.24 = 2.02 against the directly computed 2.03, closing within rounding rather than being asserted.
• Conventions printed on every value. Both results carry plane (horizontal), height (2.5 ft and 3.0 ft above finished floor), basis (maintained), and the named loss terms.
• Nothing new appeared in the examples. Every term used, CU, RCR, the two LLF families and the temperature factor, was defined above before it was applied. The 0.95 ambient factor in the shop case is the only term absent from the office case, and it is absent there because an office ceiling sits inside the tested temperature range.
• Both results labeled as averages, not as point values. 43.4 fc and 21.4 fc are area averages over 600 ft2 each, and neither was quoted as what a meter reads at any named spot.
• Illustrative values labeled. The two CU figures, 0.62 and 0.38, are marked as stand-ins for the manufacturer's table, not as values for any real product.

References

• IES recommended practice covering the lumen method and light loss factors, in the edition your specification or employer standard names, which binds through that document rather than on its own
• Manufacturer coefficient of utilization table and photometric report for the specific luminaire, which owns CU for your room
• Manufacturer ambient temperature derating curve for the driver and LED module, which owns the temperature factor
• 29 CFR 1926.56(a) and Table D-3, construction illumination minimums, which are floors rather than design targets
• See related: What Light Is Measured In, and Why the Units Confuse People; What a Photometric File Actually Describes