Emergency and Egress Lighting Annual Test

Purpose

Produce a defensible annual record that the egress path is still lit at the END of the required duration, not at the start of it. The normal failure of an emergency lighting system is not a unit that will not light; it is one that lights, holds for forty minutes, and quits while the building is still emptying, and every test shorter than the full duration passes that unit.

It also guarantees somebody measured the floor. A tech can watch nine heads come on, write "tested OK," and leave a stair landing too dark to descend. Illuminance at named points is the deliverable; a lit lamp only shows the test started.

Scope

Covers the annual duration test and the illuminance verification of egress lighting in small commercial and multifamily buildings, whether served by battery unit equipment, a central inverter, or a generator-backed emergency system, and covers the monthly functional test where the shop holds that contract.

Does not cover exit sign identification, replacement or self-luminous handling, owned by the Exit Sign Inspection and Replacement SOP; generator load bank testing, owned by the Generator Load Bank Test SOP; or where emergency lighting is required, which is the approved life safety plan and the authority having jurisdiction.

Roles and handoffs

Procedure

1. Fix the measurement points before any power is touched, and reuse last year's. Walk the egress path from the most remote point to the public way and mark points at each stair landing, each direction change, each door, and along corridors. Acceptance: a numbered point map with location and the measurement plane stated as the floor, matching last year's map where one exists. Wrong looks like readings taken wherever the meter happened to be, which makes this year's numbers uncomparable to any other year. Stop rule: no usable map means it is built and photographed before the test starts, never reconstructed after. Hazard: none at this step, it is a walk and a sketch made under normal lighting.

2. Confirm the building is clear for a 90-minute darkening and post the standby. Notify the owner in writing, confirm no occupants remain, and station a person at the fire alarm panel. Acceptance: written confirmation of the window, every stair and corridor on the route confirmed empty by walk-through immediately before initiation, and, because removing normal supply raises AC-loss trouble at a monitored panel and would dispatch apparatus to a darkened building, the monitoring station notified and the system placed in test for the window with the ticket number recorded, under NFPA 72 in the edition the AHJ has adopted, which binds the owner and reaches the shop through the service contract; the standby silences and records trouble signals rather than clearing them. Wrong looks like starting the test with a cleaner still on the third floor. Stop rule: any occupant unaccounted for stops the test and it reschedules. Hazard: this procedure deliberately removes normal lighting from stairs, so anyone left in the building is a fall exposure this SOP created, and the control is the walk-through immediately before initiation, not a notice sent that morning.

3. Initiate by removing normal power at the source, and start the clock at that instant. Open the normal supply to the area served, or operate the test means the system provides, and note the time to the minute. Acceptance: before the supply is opened, the panel directory plus a walk of the loads establish and record everything it feeds, with elevators recalled and confirmed empty, fire pump and life-safety loads confirmed not on that supply, and any load whose loss is itself a hazard - sump, refrigeration, medical equipment, access control that fails locked, stair pressurization - on a written list agreed with the owner; then every emergency unit and every emergency-fed head illuminated within 10 seconds of the loss of normal power, and the clock start time recorded. A load you cannot account for stops the test. Wrong looks like pressing individual test buttons unit by unit, which tests the lamp and never tests the transfer. Stop rule: a unit that does not transfer is recorded as a transfer failure on the spot and the test continues for the rest; do not stop the clock. Hazard: opening the normal supply is qualified-person work under 29 CFR 1910.332 and 1910.399, and the operator stands to the hinge side of the panel rather than square in front of the door.

4. Take the initial illuminance set within the first minute. Read the meter at floor level at each numbered point. Acceptance: per NFPA 101, Section 7.9, in the edition your authority having jurisdiction has adopted, which binds the building and its owner and reaches your shop through the inspection and the service contract, an average of not less than 1 ft-candle and a minimum of not less than 0.1 ft-candle, with a maximum-to-minimum ratio not exceeding 40 to 1; record each point's value, the average, the minimum and the ratio. Wrong looks like recording only the average, which is exactly how a dark stair landing passes. Stop rule: an initial set that already fails is recorded and the test still runs to completion, because the end-of-duration numbers are what size the correction. Hazard: reading a meter at floor level on a stair in emergency lighting only, so the reader keeps a hand on the rail and works down, never backward.

5. Hold the full duration without restoring anything. Leave the system on emergency for 90 minutes and walk the route at roughly 30-minute intervals, noting any unit that has gone out and when it was observed out. Acceptance: 90 minutes elapsed from the step 3 clock start, with a dropout observation recorded for anything that failed. Wrong looks like calling it at 60 minutes because everything still looks fine, which is the interval where these failures begin. Stop rule: a dropout does not end the test; it goes on the defect list. Hazard: where the source is a central inverter or battery cabinet, the sustained discharge and the recharge after it liberate hydrogen, so the room's ventilation stays running and nothing that sparks is operated in it; flooded cells add a sulfuric acid contact route to the same step.

6. Take the end-of-duration illuminance set BEFORE restoring normal power. At minute 90, repeat the step 4 readings at the same numbered points, same floor plane. Acceptance: per the same NFPA 101 section and adopted edition, an average of not less than 0.6 ft-candle and a minimum of not less than 0.06 ft-candle at the floor, with the maximum-to-minimum ratio still not exceeding 40 to 1, and every value recorded per point. Wrong looks like restoring power first and reading afterward, which produces a set of numbers that means nothing. Stop rule: any point below the minimum, or a ratio over the limit, is a failure recorded per point and reported to the owner that day even if the average passes. Hazard: same as step 4, and by minute 90 the levels are at their lowest, so the reader carries a hand light for footing and does not use it near the meter's sensor.

7. Restore normal power, and verify the system recharges and re-transfers. Close the supply from the hinge side of the panel, confirm every emergency head returns to its normal state, and confirm each unit's charge indicator is charging rather than faulted. Acceptance: all units back on normal, charging indication present at each, no breaker or fuse operated during restoration, and the fire alarm system taken out of test with the station confirming normal supervision restored, which is what closes this step. Wrong looks like a unit whose charger cooked during the deep discharge and now shows a fault lamp nobody looked at, which guarantees a failure at next year's test. Stop rule: a unit not charging is a defect on the same list as a duration failure. Hazard: restoration is when a recharge inrush finds a marginal charger or a damaged battery, and the battery is the thing that vents or gets hot, so nobody stands over an open cabinet during the first minutes of recharge.

8. Close the defect list against the owner before demobilizing. Acceptance: a written list where every failure carries the unit or point number, the measured value, the criterion it missed, and either a correction that day or a dated owner deferral. Wrong looks like a verbal mention on the way out. Stop rule: a life-safety failure left with no dated decision goes to the office the same day. Hazard: none at this step, it is a conversation held at a desk.

When the building will not cooperate

The owner will not release a 90-minute window. The monthly 30-second functional test is not a substitute, and saying so plainly is part of the job. Offer a night or weekend window; if it is refused, the refusal goes in writing with the date, so the record shows the test was offered and declined.

The units are self-testing. They run their own duration test and report by indicator, which NFPA 101, Section 7.9, in the adopted edition, treats differently from a manual test. Read and record the indicators per unit, and confirm what the AHJ accepts before dropping the manual test from your contract.

The record this produces

One test record per building: the numbered point map and its plane; the clock start; both illuminance sets per point with average, minimum and ratio; transfer and recharge results per unit; dropout times; the monitoring ticket number; and the defect list with the owner's dated decision on each line.

Worked pass: two-story office suite, 14 points, 11 units

Step 1: last year's map is reused, 14 points, all at floor level, points 8 through 10 on the interior stair.

Step 3: normal supply opened at 6:12 p.m. Ten of the 11 units transfer within 10 seconds. Unit 5 over the rear corridor does not illuminate at all, recorded as a transfer failure; the clock keeps running.

Step 4, taken by 6:14 p.m.: the 14 points average 1.4 ft-candles, the minimum is 0.3 ft-candle at point 9 on the stair landing, and the maximum is 6.0 ft-candles at point 2 by the lobby door. The ratio is 6.0 divided by 0.3, which is 20 to 1, inside the 40 to 1 limit, so the initial set passes all three criteria.

Step 5: at the 30-minute walk everything except unit 5 is still lit. At the 60-minute walk unit 6, also on the rear corridor, has gone out, so the record reads "out at the 60-minute walk, last seen lit at 30" - at best 30 minutes short of the required 90 and possibly 60, because it cannot claim a minute nobody observed.

Step 6, at 7:42 p.m., 90 minutes from the 6:12 start: the same 14 points average 0.7 ft-candle, exactly half the initial average, against a criterion of 0.6, so the average passes. The minimum is 0.04 ft-candle at point 9, against a minimum of 0.06, so it fails. Maximum is 3.2 ft-candles at point 2, and 3.2 divided by 0.04 is 80 to 1, which fails the 40 to 1 limit as well, so point 9 fails on two criteria while the average passes.

Step 7: supply closed from the hinge side at 7:45 p.m. Nine units return to charging indication. Unit 5 shows a charger fault, consistent with its transfer failure, and unit 6 shows charging.

Step 8: three items to the owner. Unit 5, transfer failure plus charger fault, quoted for replacement. Unit 6, out by the 60-minute walk against a 90-minute requirement, battery. Point 9, 0.04 ft-candle and 80 to 1 at 90 minutes, which needs an added head on the stair landing rather than a battery. The owner authorizes units 5 and 6 that evening and defers the stair head, signed and dated.

References

• NFPA 101, Section 7.9, in the edition your authority having jurisdiction has adopted, for emergency illumination levels, the maximum-to-minimum ratio, the 90-minute duration and the testing intervals; pull the adopted edition, since these provisions renumber between cycles
• NEC Article 700 in the edition your jurisdiction has adopted, for emergency system wiring and unit equipment connection
• NFPA 111, in the edition adopted locally, where the source is a stored electrical energy system rather than unit equipment
• 29 CFR 1910.332 and 1910.399 for the qualified person operating the supply at step 3, with 29 CFR 1910.333(b)(2) for the associated work practices
• See related: Exit Sign Inspection and Replacement SOP, Generator Load Bank Test SOP