Commercial Lighting Retrofit Verification

Purpose

Accept a finished retrofit against the document it was sold and permitted on, rather than against how the room looks. This is an audit, not an install: the crew that did the work is not the party that closes it out, and the fields that fail are almost never light level. They are fixture count, control function, and labeling, because those three are invisible to anyone standing in a lit room.

Labeling is the one that outlives everybody. A ballast-bypass conversion puts line voltage on lampholder pins that used to be fed from a ballast output. If the label did not go on, the next person to change a lamp in that building finds out the hard way, years after the crew, the foreman and the customer contact have all moved on.

Scope

Covers acceptance verification of a completed interior lighting retrofit in small commercial buildings, whether the work was done by our crew or another contractor, and covers the package handed to the owner and to any incentive program.

Does not cover performing the changeout, which the Lighting Retrofit Changeout Standard owns and which is the source of the pattern photograph and the pre-measurement this SOP audits against; does not cover control commissioning, which the Lighting Control and Occupancy Sensor Commissioning SOP owns and which this SOP verifies the result of; and does not cover exterior lighting.

Roles and handoffs

Procedure

1. Get all three acceptance documents in hand before walking a single area. Those are the scope count by area and type, the control sequence of operations, and the pre-retrofit light-level commitment with its measurement points. Acceptance: all three present, each dated, and each the same revision the crew worked from. Wrong looks like auditing against a proposal while the crew built to a later change order. Stop rule: a missing or conflicting document stops the audit until the office resolves which revision governs, because an audit against the wrong document produces a punch list the crew is right to reject. Hazard: none at this step, it is a document check made at a desk.

2. Reconcile installed count against scope count, area by area. Count what is installed in each area and compare it to the scope line for that area, not to the project total. Acceptance: installed equals scope for each area, or a written variance naming the area, the count and the reason. Wrong looks like reconciling only the project total, which lets a shortfall in one room hide behind an overage in another. Stop rule: any area short of its scope count goes on the punch list before the audit continues; it is not resolved verbally with the crew. Hazard: counting overhead in an occupied retail space means walking while looking up, so the count runs outside trading hours or with aisles coned, and nobody steps backward.

3. Isolate and open a sample of fixtures per type, and treat every one as line-voltage until proved otherwise. Our shop gate is 10 percent of the installed fixtures of each type, with a minimum of 3 per type. Open the branch circuit, lock and tag under 29 CFR 1910.333(b)(2), and prove dead on a known live source before and after, per NFPA 70E-2021, 120.5. Acceptance: each opened fixture matches the pattern photograph on lampholder type, conductor landing and driver mounting, with the sample size recorded per type. Wrong looks like assuming an unlabeled converted fixture is still ballast-fed, which is the exact assumption this audit exists to catch. Stop rule: a defect in any sampled fixture expands the sample to 100 percent of that type, and the expansion is recorded rather than negotiated. Hazard: this is qualified-person work under 29 CFR 1910.332 and 1910.399, and the before-and-after proof is what stops a meter with a blown fuse from reading a clean, believable zero on a lampholder that is actually at line potential.

4. Verify the conversion label on every fixture you opened and by sight on the rest. Acceptance: the kit's supplied label present, legible, stating the conversion and the line-voltage lampholder condition where that is true, plus installer and date, on 100 percent of converted fixtures. Wrong looks like labels applied inside the housing where the diffuser hides them, which satisfies nobody changing a lamp from a ladder. Stop rule: any unlabeled converted fixture puts that whole type on the punch list at 100 percent, because labeling defects cluster by crew and by day rather than scattering. Hazard: the fixture is still isolated from step 3 and stays locked until step 4 is finished on it.

5. Verify the controls against the sequence, not against whether the lights come on. Operate every switch, sensor and time-based control named in the sequence. Acceptance: each control function in the sequence observed and marked pass or fail by name, including any emergency-fed fixture confirmed to stay lit when the controlled load drops. Wrong looks like an auditor flipping the switch by the door and calling controls verified. Stop rule: an emergency or egress-fed fixture that goes dark under a control is a life-safety defect, reported to the owner in writing the same day rather than added to a punch list. Hazard: dropping lighting in an occupied space, so confirm nobody is on a ladder or a stair before any control test that darkens an area.

6. Spot-check light level by exception, at the pre-measurement points only. Re-read at a sample of the step 1 commitment's own points, same plane, same height, same daylight condition. Acceptance: each sampled point at or above the committed level, with the plane, height and daylight condition recorded beside every value. Wrong looks like new points chosen where the light happens to be good, which produces a number that cannot be compared to anything. Stop rule: any sampled point below the commitment expands the check to the full point set for that area. Hazard: none at this step beyond the floor-level walk already controlled at step 2.

7. Reconcile the waste record against the fixtures actually converted. Compare the disposal manifest's lamp and ballast counts to the count implied by the installed fixtures and their as-found lamp configuration. Acceptance: manifest counts equal the implied counts, or a written explanation of the difference; container labels and accumulation start dates present. Wrong looks like a manifest that is short by the contents of a box still sitting in a storeroom, which is a lamp population nobody is tracking. Stop rule: an unexplained shortfall stops the package until the missing lamps are located and manifested. Hazard: moving or counting lamps in a drum risks breaking one, and fluorescent and HID lamps contain mercury, so a break clears the area, the cleanup is stiff paper and damp wipe with ventilation to outside, never a shop vacuum, and intact lamps stay universal waste under 40 CFR 273.

8. Restore every circuit you opened, then confirm what you disturbed still works. Remove the locks and close each breaker from the hinge side rather than square in front of the door, then re-lamp-check each opened fixture and re-run the step 5 control test for the areas whose circuits were cycled. Acceptance: every opened fixture lit, every control in the sequence still passing, no breaker or driver failing on restoration. Wrong looks like a sampled fixture left with a lamp seated but not rotated home, which lights on the bench test and drops out a week later. Stop rule: any fixture or control not restored to its pre-audit state is corrected before the auditor leaves. Hazard: energy going back in is when a lampholder disturbed during the sample announces itself, so the panel door is approached from the side and the customer is kept out of the working space.

The record this produces

One acceptance package: the three step 1 documents by revision and date; the per-area count reconciliation with variances; the sample size and result per type with any expansion and its trigger; the labeling result as a percentage of converted fixtures; the control results by named function; the spot-check readings with plane, height and daylight condition; the waste reconciliation; and the punch list with the owner's dated decision per line.

The owner keeps the package because it is what an incentive program and a future buyer both ask for. The office keeps the sample expansion history, because a crew whose sample expands on three jobs running has a training problem, and that is only visible in the expansion field.

Worked pass: 96 fixtures on scope, small retail store, six areas

Step 1: scope count, control sequence and light-level commitment all present at revision B, dated, and the crew's pattern photographs attached.

Step 2: installed count is 70 of type A and 24 of type B, 94 total against a scope of 96. The two missing are both in the rear stockroom, which the crew skipped because product was stacked to the ceiling. That is a written variance on the punch list, not a verbal note.

Step 3: the sample gate is 10 percent per type with a minimum of 3. Type A is 70 fixtures, so 7. Type B is 24, so 10 percent is 2.4 and the minimum of 3 governs. Ten fixtures opened in total, which matches 10 percent of the 94 installed. Circuits 6, 8 and 11 locked and tagged in turn, each proved dead on a known live source before and after. All 7 type A fixtures match the pattern photograph.

Step 4 is where the pass fails. Of the 3 type B fixtures opened, one carries no conversion label. Step 3's stop rule and step 4's stop rule both fire, and the sample expands to 100 percent of type B. All 24 are checked: 5 are unlabeled, which is 21 percent of that type. Every one of those 5 is a bypass conversion with line voltage at the lampholder, and all 5 are on the same run, consistent with one afternoon's work by one person. The correction is 5 labels applied and verified, recorded as an expansion triggered by the sample rather than as a spot fix.

Step 5: the sequence names four control functions. All four pass, including the emergency-fed troffer at the rear door staying lit when the sales-floor sensor drops its load.

Step 6: three of the commitment's points re-read at 30 in above finished floor on the horizontal plane with the front blinds closed, as the pre-measurement was taken. All three clear the committed level, so the check is not expanded.

Step 7: the fixtures are all two-lamp, so 94 installed fixtures imply 188 lamps removed. The manifest shows 176, a shortfall of 12. Step 7's stop rule fires, and 12 lamps are found in an unlabeled carton in the stockroom, moved into a labelled closed container with an accumulation start date, and added to the manifest.

Step 8: breakers closed from the hinge side, all 10 opened fixtures lit, the four control functions re-run and still passing. The package closes with one punch item, the 2 stockroom fixtures, deferred and signed by the owner that afternoon.

References

• 29 CFR 1910.333(b)(2) with 1910.332 and 1910.399, and NFPA 70E-2021, 120.5, in the edition your employer's electrical safety program has adopted, for the isolation and live-dead-live sequence at step 3
• 40 CFR 273 for universal waste handling of the mercury-containing lamps reconciled at step 7
• UL 1598C and the specific retrofit kit's installation instructions, which supply the labels verified at step 4 and define the pattern audited at step 3
• See related: Lighting Retrofit Changeout Standard SOP, Lighting Control and Occupancy Sensor Commissioning SOP, Panel Labeling and As-Built Handover SOP