Module Damage and Warranty Claim SOP

Purpose

Which warranty pays for a damaged module is decided by evidence gathered on the first visit, and most of it is unrecoverable after. This procedure guarantees a damage call produces the right route, chosen once on cause, with that route's evidence collected before anything is touched.

The routes are mutually exclusive in practice and mixing them loses both. Hail is an external cause, which mainstream module product warranties exclude, so filing it with the manufacturer burns weeks and comes back denied while the insurance clock runs. The second failure is quieter: modules with no visible damage that lost output in the same event, left off the claim because nobody screened them, paid for out of the shop's own labor a year later when they surface as a shortfall.

Scope

Covers damaged, degraded or failed modules on residential and small commercial PV, from the first report through the filed claim and the replacement, on arrays the shop installed or services.

Does not cover a shortfall with no known damage event, owned by the production shortfall investigation SOP, which may hand a case here; inverter, optimizer or microinverter failures, which run through their own manufacturer processes; the roofing damage a hail event usually brings; or modules removed for a reroof.

Roles and responsibilities

Procedure

Step 1 - Make the array safe before any diagnosis, and lead with the glass. A cracked module still generates whenever it is lit and its insulation may be compromised, so treat the frame, the mounting and any wet surface as energized: cover the damaged modules opaquely, wear cut-resistant gloves and eye protection, keep bystanders off the roof and out of the drip line below. Open the AC disconnect first, then the DC, and confirm near-zero current with a clamp meter before de-mating anything, because PV connectors are generally not load-break rated. Acceptance: damaged modules covered, disconnects open, current near zero, area below cleared. Wrong looks like a tech handling broken glass on a live string. Stop rule: red-tag and leave the array de-energized where glass is shattered across multiple modules or a conductor is exposed; do not restore it for a customer's convenience. Hazard beyond the electrical: hail that broke modules usually broke the roof, so the roof is assessed before anyone walks it, under the job's fall protection plan per 29 CFR 1926 Subpart M at 6 feet for construction-classified work.

Step 2 - Fix the event and its date from the data before you interpret anything you see. Pull module-level or string-level production either side of the reported event and find the day the step change happened. Acceptance: a dated step change in the data, or a written statement that there is none. Wrong looks like accepting the customer's date because it matches a storm they remember; the data date is the one an insurer and a manufacturer use. Hazard is evidentiary: monitoring platforms trim historical resolution over time, so export the raw interval data now rather than when the claim is written.

Step 3 - Photograph every damaged module with its serial, a scale and its position, in the same frame set. For each: a full-module shot, a close shot of the damage with a rule or coin for scale, and a legible shot of the serial label. Acceptance: three usable frames per damaged module, the serial legible enough to read aloud from the photograph. Wrong looks like a serial shot at an angle that glares out under the label's laminate; re-shoot while you are standing there. Stop rule: do not substitute the serial from the install map. Manufacturers verify the physical label, and a returned claim costs a truck roll against a two-minute task at the module. Hazard: reaching a damaged module means handling broken glass at height, so the module stays covered while photographed and nothing is lifted or flexed for a better angle.

Step 4 - Screen the entire array on data, not just the modules with visible damage. On the first clear day after the event, take the array median energy per module and flag every module at or below 70 percent of it. Acceptance: a flagged list covering the whole array with each module's energy and its percentage of median. Wrong looks like a claim listing only what the eye found; impact microcracks frequently leave the glass intact and the output down, and those modules stay invisible until next summer. Hazard is interpretive: on a string system with no module-level electronics one weak module pulls its whole series string, so a string-level screen identifies the affected string, not the module count, which has to come from a per-module test.

Step 5 - Choose one route on cause, and write the reason. External cause - hail, wind-borne debris, a fallen limb - is a property insurance matter. A materials or manufacturing defect - delamination, backsheet cracking, junction box failure, cell discoloration with no impact history - is a product warranty matter. Damage traceable to the installation, such as a clamp outside the module's published clamp zone or a fastener over-torqued into the frame, is the shop's own workmanship. Acceptance: one route named, with the cause evidence cited by date and photograph. Wrong looks like filing the same modules on two routes to see which lands first; both processes ask what else was filed, and a duplicate claim is grounds to deny. Hazard is legal: what a warranty excludes is set by the certificate for that model and purchase date, and what a signed claim form concedes is a question for your own attorney, not the warranty desk.

Step 6 - Gather the evidence that route requires, before removing a single module. Product claims commonly require serials, install date, site photographs, production data and sometimes an electroluminescence image, which reverse-biases the module in darkness and is a lab or manufacturer test, not a ladder test. Insurance claims want the dated event, the whole-array screen and the repair scope. Acceptance: the route's own published evidence list, completed item by item, nothing marked pending. Wrong looks like an assembled claim missing one item, which restarts the submission clock. Hazard: most product warranties require the failed module to be retained and the manufacturer's authorization obtained before removal, so a module pulled early to tidy the roof can end a valid claim.

Step 7 - File, get the authorization number, and only then schedule removal. Submit through the route's process and record the claim or authorization number against every serial. Acceptance: an authorization number issued and recorded per module before any removal is scheduled. Wrong looks like a submission rejected for an unreadable serial or a missing document. Stop rule: return to site for the missing evidence rather than resubmitting from the install record, and note which module drove the return so the intake question gets fixed. Hazard is schedule, and it belongs to the customer conversation: authorization timelines are the manufacturer's or insurer's, so the customer gets that party's stated range, not a shop promise.

Step 8 - Replace with an electrically compatible module and record what changed. Confirm the replacement's current and voltage figures against the original datasheet before it goes on the roof, torque clamps within the module's published clamp zone, and re-map the new serial into the monitoring platform. Acceptance: the replaced positions report within the array's normal spread on the next clear day, and the platform map carries the new serials. Wrong looks like a mismatched replacement dropped into a series string on a string inverter, where the string runs at the weakest module's current and the shop has quietly derated the whole string to fix one module; on a module-level system the mismatch stays contained to that position, which is why the substitution is acceptable there and not here. Hazard: rehanging is roof work with live conductors upstream of the open DC disconnect, so the string stays open and locked until the last connector is seated.

The record this produces

One damage claim record on the job: the dated event and the data step change with the raw interval export attached; per-module photograph sets with serials; the whole-array screen with every module's energy and percentage of median; the flagged list separated into visible and data-only; the route chosen with its written reason; the evidence list with each item's status; claim or authorization numbers per serial; replacement model and its electrical figures against the original; and the post-replacement clear-day check.

On a questioned claim the dated screen is the whole case for the data-only modules.

Worked pass

A 26-module array on module-level optimizers, after a hail event the customer reports two days later.

Step 1 covers the four modules with visible glass fracture, opens AC then DC, confirms near-zero current and clears the area below; the roof is assessed before anyone walks it and found sound with fall protection rigged. Step 2 finds a clean step change in the array total on the storm date, matching the customer's report, and the raw interval data is exported that day. Step 3 photographs the four visible modules, three frames each.

Step 4 screens all 26 on the first clear day. Array median is 1.92 kWh per module, so the flag line at 70 percent of median is 1.34 kWh. Seven modules fall at or below it, ranging from 0.4 to 1.34 kWh: the four with visible glass damage plus three with no visible damage at all. Average energy across the seven is 0.95 kWh, so the average deficit is 0.97 kWh per module.

That is what sizes the claim. A clear-day array total would be 26 x 1.92 = 49.9 kWh. The seven flagged modules cost 7 x 0.97 = 6.8 kWh, 13.6 percent of the array. The four visible ones account for 3.9 kWh, 7.8 percent; the three invisible ones carry the remaining 2.9 kWh, 5.8 percent, and they are the ones a visual-only claim would have left on the roof.

Step 5 routes all seven to the homeowner's property insurance: the cause is external and this model's product warranty excludes external causes, with the storm date and step change cited as the reason. Step 6 assembles the insurer's evidence list.

Step 7 fails. The submission comes back with two serials rejected as unreadable, both on top-row modules where the label sat in glare. Stop rule taken: no resubmission of serials from the install map. A tech returns, re-photographs both labels shaded with a clipboard, and the claim is resubmitted and authorized. The file notes both failures were top-row glare, and the intake sheet is amended to require serial photographs be shaded.

Step 8 replaces all seven with a current model whose current and voltage figures are checked against the original datasheet. Because this array runs module-level optimizers the mismatch stays contained to each replaced position rather than pulling a series string down, stated on the record so the next tech does not read the mixed models as a defect. On the next clear day all seven replaced positions report inside the array's normal spread.

References

• The module manufacturer's warranty certificate for that model and purchase date, which sets what the product warranty covers and excludes, plus the separate power warranty terms.
• 29 CFR 1926 Subpart M and 29 CFR 1910.28 for fall protection, applied by activity classification.
• NFPA 70E-2021, 120.5 for live-dead-live verification and 29 CFR 1910.333(b)(2) for electrical work practices (29 CFR 1926.417 in construction).
• The homeowner's property insurance policy for external-cause damage, and your own attorney on what a signed claim form concedes.
• See related: the production shortfall investigation SOP, which frequently hands cases here, and the array removal and reinstall for a reroof SOP.