Why a Lighting Design Is Computed at End of Life

Why this matters

A retailer opened a remodeled sales floor, put a meter on the counter, and read 63 fc against a specification that said 50. Their read was that they had been over-lit by a quarter and were paying to run light they never asked for, and they asked for a row of fixtures to come out. The shop's read was that the design was exactly right. Both were describing the same installation honestly, and the disagreement is not about a measurement. It is about which day the specification refers to, because a maintained criterion is a floor the space has to clear on its worst day, and the day it is commissioned is its best.

The criterion is a floor over a period

A design target is written as a maintained value for a reason. Light output falls from the moment a system is energized, and it falls for reasons that mostly are not the fixture's fault: source flux decays, dirt lands on optics, room surfaces darken, units go out and stay out. The light loss factor card owns which terms those are and where each value comes from. What matters here is the design consequence.

If a system is sized so it produces the criterion on the day it is commissioned, then the criterion is met exactly once, and every day after that the space is under-specification. That is not a conservative design, it is a design with a lifetime of one day, and it will pass its commissioning survey and fail every survey after it. So the calculation is run backwards: pick the moment the system will be at its worst, work out what fraction of initial output survives to that moment, and size the installation so the criterion is met then.

The moment you pick is not a property of the equipment. It is the point in your maintenance cycle just before cleaning and replacement restore what is recoverable. Choose a 24-month cleaning interval and the worst moment is milder; choose 60 months and it is harsher, and the fixture count moves accordingly. The design is a maintenance-interval decision wearing a fixture count.

The day-one penalty is exactly the reciprocal

Here is the part nobody explains to the customer, and it is one line of arithmetic. If the criterion is met at the end of the interval, then on day one the space is over-lit by exactly one divided by the light loss factor. A factor of 0.78 means day one runs 1 / 0.78 = 1.28, or 28 percent above the number in the specification. A factor of 0.65 means 54 percent above it. There is no design skill involved in that figure and no padding hidden in it; it is the same factor read the other way.

That is a real cost and it is worth naming honestly rather than defending. The system draws its full input power on day one to produce light that exceeds the requirement, and it keeps doing so until depreciation catches up. Whether that is acceptable, and what to do about it, is the actual conversation.

The call: 63 fc against a 50 fc specification

Sales floor, 48 recessed luminaires in six rows of eight. Specification: 50 fc maintained, horizontal at 30 in above finished floor, light loss factor 0.78 carried in the design report.

Re-base before arguing. 50 fc maintained divided by 0.78 gives 64.1 fc initial as the value the design predicts on day one. The customer's meter read 63 fc, which is 63 / 64.1 = 1.7 percent below the predicted initial value. The installation is not over-lit. It is delivering slightly less than the design said it would deliver on day one, and it is on track to land on 50 fc at the end of the interval.

And the customer is still not wrong about what they are experiencing. They were shown a number, they are seeing 26 percent more than that number (63 / 50 = 1.26), and on a sales floor the visual result on opening day is the product. "You will lose it eventually" is arithmetically true and commercially useless.

What removing a row would have cost

Take out one row of eight from 48 and the layout keeps 40, which is 40 / 48 = 0.833 of the flux. Day one becomes 63 x 0.833 = 52.5 fc initial. Apply the same 0.78 and the end-of-interval value is 52.5 x 0.78 = 40.9 fc maintained, against a 50 fc criterion. That is (50 - 40.9) / 50 = 18.2 percent below specification at the moment the specification is written for.

The second cost is not in that arithmetic at all. Removing a row changes the spacing between the remaining rows, which changes what the distributions overlap, and that shows up as a worse maximum-to-minimum ratio rather than a lower average. The uniformity card owns why that is a separate failure with its own acceptance line; the point here is only that the average shortfall above understates the damage.

What was done instead

The fixtures were set to 78 percent output on day one. 64.1 fc initial x 0.78 = 50.0 fc delivered on day one, horizontal at 30 in above finished floor, initial condition at a 78 percent control setting, numerically equal to the maintained criterion. It contains no loss factors; the dim setting is standing in for depreciation not yet spent. The customer got the number they were promised, on the day they cared about, and the installation kept every bit of its designed capacity.

The headroom above that setting is not spare capacity and it is not padding. The dimming headroom is the light loss factor, expressed as a control setting instead of as a design allowance. Twenty-two percent of output is held back on day one and released over the interval to hold 50 fc flat, and when there is none left the interval is over, which is a far more honest end-of-life signal than a customer noticing the space got dim.

Doing this touched the drivers. Where the output setting lives in a control station or a programmable interface, it is a configuration change. Where it is a physical lead or an adjustment inside the housing, the branch circuit is de-energized and locked out under 29 CFR 1910.333(b)(2) with live-dead-live proving per NFPA 70E-2021, 120.5, in the edition the employer's electrical safety program adopts, and the housing is not opened until the manufacturer's stated capacitor discharge time has elapsed and the output leads have been verified at zero, because a driver holds stored energy after the supply is gone. The units were reached from a lift rated for the height with the fall protection the equipment requires under 29 CFR 1910 Subpart D, not from the top of a stepladder.

Year three: does the plan still close

Thirty-six months into a 72-month interval, the control had stepped output to 88 percent and the same grid, plane and height read 49.2 fc.

What the system would produce with no depreciation at that setting: 64.1 fc x 0.88 = 56.4 fc. Measured 49.2 fc against that gives an in-service loss factor at year three of 49.2 / 56.4 = 0.872.

How much of the budget is spent: the design allowed a fall from 1.00 to 0.78, a budget of 0.22. Spent so far is 1.00 - 0.872 = 0.128, which is 0.128 / 0.22 = 58 percent of the budget at 50 percent of the interval. Slightly ahead, and expected: dirt accumulates quickly at first and then saturates, so the curve is steeper early than a straight line.

Does the headroom still reach: remaining depreciation is 0.78 / 0.872 = 0.894, so a further 10.6 percent will be lost. Holding 49.2 fc through it needs output raised by 1 / 0.894 = a factor of 1.119, taking 88 percent to 88 x 1.119 = 98.4 percent of full at the end of the interval, against 100 available. The plan closes with 1.6 percentage points of margin, which is thin enough to be worth saying out loud in the year-three report rather than discovering at year six.

What flips this

Two conditions break the dim-to-criterion answer, and both are visible before you commit.

A harsh loss factor. At 0.78 the headroom is 28 percent of the day-one value and it closes. At 0.60 the day-one over-lighting is 67 percent, and holding the criterion flat means running at 60 percent output for the first year, which pushes the driver into the part of its range where output regulation and flicker performance are least well characterized. Below roughly two thirds, check the manufacturer's data for the low end rather than assuming linearity, and consider whether the honest answer is a shorter cleaning interval instead.

A criterion that is not actually a floor. Some spaces want a target rather than a minimum, and a few genuinely want the extra light early. Where the customer's requirement is "not less than", dim to it. Where it is "about", the conversation is different and the arithmetic above is only informing it.

Checking your own figures

• Two different 0.78s, each applied once. The light loss factor divides once: 50 fc maintained / 0.78 = 64.1 fc initial. The control setting then multiplies once: 64.1 x 0.78 = 50.0 fc delivered on day one, initial condition. They are equal by design rather than by coincidence, and neither is applied to a figure that already contains it.
• The measured turnover value against the predicted initial. 63 fc measured against 64.1 fc predicted initial, 1.7 percent below, not 26 percent above. The 26 percent figure (63 / 50 = 1.26) is measured against the maintained criterion and is labelled as such wherever it appears.
• The removed-row arithmetic recomputes. 40 / 48 = 0.833; 63 x 0.833 = 52.5 fc initial; 52.5 x 0.78 = 40.9 fc maintained; (50 - 40.9) / 50 = 18.2 percent short.
• The day-one penalty equals the reciprocal. 1 / 0.78 = 1.28, and the day-one setting that removes it is 0.78, the same number. Stated as one relationship, not two.
• Budget spent computed against the budget, not against 1.00. 0.128 spent on a budget of 0.22, which is 58 percent, at 50 percent of a 72-month interval.
• Headroom check printed against what is available. Required end-of-interval output 98.4 percent, available 100 percent, margin 1.6 percentage points.
• Every illuminance carries plane, height and basis. 50 fc maintained, 64.1 fc initial, 63 fc initial, 52.5 fc initial, 40.9 fc maintained, 49.2 fc maintained, all horizontal at 30 in above finished floor.
• No loss factor term is derived here. Every value of 0.78 and 0.872 in this card is used as a whole-number basis; where it comes apart is the light loss factor card's subject.

References

• IES recommended practice covering maintained illuminance and design light loss factors, in the edition your specification or employer standard names, which binds through that document rather than on its own
• Manufacturer driver and control documentation for the exact ordering code, including the low-end output limit and the stated capacitor discharge time before a housing is opened
• 29 CFR 1910.333(b)(2) and NFPA 70E-2021, 120.5, for de-energizing, locking out and proving dead before adjusting a driver inside a housing; 29 CFR 1910 Subpart D for working at height in general industry and 29 CFR 1926 Subpart M in construction
• See related: What a Light Loss Factor Is Accounting For; Why Uniformity Matters More Than Average Level; What Lumen Depreciation and Lifetime Ratings Mean