Solar Power Systems Case Studies
Industrial5 Case StudiesUpdated Sep 2026
Industrial solar work rarely fails for lack of hardware. It fails when a design decision, a supply date, a permit condition or a shift schedule pulls against the others, and nobody has written down which one wins. The five cases below come from fictional sites and follow one habit throughout: state the business objective first, then test it against the evidence that actually exists, the limits on the site, and the people who will live with the result. Each case closes with the actions taken and what they changed.
Case Study 1
Recurring fuse holder failures in a rooftop combiner box
Calder Plastics Works, a molder of automotive housings running two shifts on one campus, with about 1,200 staff and a 2.4 megawatt rooftop array on its main building.
The situation
The array sits across the main molding hall roof and feeds three string inverters through a set of combiner boxes mounted on the north parapet. Output was designed to offset roughly a fifth of the campus daytime load. In the spring of the third operating season, the plant's energy team began logging a pattern: one or two strings dropped out on warm afternoons, the combiner box reset itself, and the strings returned by evening. Maintenance replaced two fuse holders under warranty and the trips came back within a month. The facility manager, Dana, wanted to know whether this was a supplier defect or a site problem before the next warranty window closed.
The challenge
The business question was narrower than the engineering question. Calder needed the array back at full output before the summer shutdown, when the plant runs its heaviest press schedule, and it needed the warranty claim argued on evidence rather than on a supplier's first assurance. Dana also had a second constraint: the roof membrane was under its own warranty, so any work that penetrated it required the roofing contractor on site, which added days and cost.
Analysis
The first analysis pass sorted the evidence by how much each piece could be trusted. Thermal images taken during a trip showed the fuse holder terminals running about 40 degrees hotter than the adjacent bus, which points to a contact problem rather than an overloaded string. Torque readings on the terminals were inside the manufacturer's range at installation but had not been rechecked since. The inverter event log recorded the trips as ground fault alarms, which is a symptom that can come from moisture, insulation damage, or a loose connection upstream. Nobody had yet pulled the box and inspected the gasket, so the moisture hypothesis stayed open.
Each corrective option was judged against the same objective: output restored before shutdown with little roof disruption. Replacing the fuse holders with a higher-rated model from a different supplier was the cheapest option and could be done from a lift without roofing work, but it treated the symptom if the real cause was a thermal design margin problem. Re-terminating every connection and fitting a thermal-rated combiner took longer and needed the roofing contractor, yet it addressed the mechanism the thermal images pointed to. Operations cared most about the week of the heaviest press schedule, while finance cared about whether the warranty claim would survive a second failure.
- Third season, springStrings drop on warm afternoons and recover by evening.
- First warranty swapTwo fuse holders replaced; trips returned within a month.
- Thermal imagingTerminals ran about 40 degrees hotter than the adjacent bus.
- Planned outageBox opened, terminals re-torqued, insulation and ground tested.
- Thermal alarm addedQuarterly scans now flag drift before any string trips.
Actions taken
- Pulled the combiner box during a planned outage and inspected the gasket, terminals, and busbar for heat marks, photographing each contact before cleaning and re-torquing to the manufacturer's value.
- Logged insulation resistance and ground continuity on every string before reconnecting, and set the inverter to record trip conditions at one second intervals so the next event would carry usable data.
- Replaced the remaining fuse holders with a thermal-rated model and sent the warranty claim with the thermal images, torque log, and event history, asking the supplier to confirm the failure mode in writing.
- Added a thermal scan with a 15 degree alarm above the adjacent bus to each quarterly inspection, and rechecked every termination at the next one, so drift shows up before it trips a string.
Results
The array returned to full output in the first week of the shutdown window, and the combiner box has not tripped on a warm afternoon since. The supplier accepted the claim after reviewing the thermal images and agreed to cover the replacement hardware, though the roofing contractor's invoice for the second visit was not recovered. Calder now carries the thermal threshold into its inspection checklist for every combiner on the site. The bigger lesson, Dana noted, was that the warranty window had nearly closed before anyone asked whether the fault was the part or the installation.
Questions for discussion
- Which of the three pieces of evidence would you have wanted before accepting the supplier's first explanation, and why?
- How should the warranty deadline change the order of investigation on a live array?
- What would you have needed to see in the inspection history to justify a thermal threshold on a schedule rather than on each trip?
Case Study 2
Laminator yield loss on a module assembly line
Halden Ridge Module Works, a contract manufacturer of framed photovoltaic modules with two laminating lines and about 680 staff at one plant.
The situation
Halden Ridge Module Works assembles framed modules on two parallel lines, each feeding a laminator that presses glass, encapsulant, cells, and backsheet into a single panel. For most of the year, the lines ran at about 92 percent of their rated throughput with a first-pass yield near 97 percent. In the second month of the current quarter, throughput dropped to roughly 81 percent of rating while the rework cell filled with panels that had voids, delamination at the edge, or bubbles under the cell string. Shift leads reported that the problem appeared on some days and not others, and that the laminator recipe had not changed.
The challenge
The plant manager, Priya, had a delivery commitment to a utility-scale buyer and a contractual yield floor of 95 percent on each lot. Throughput loss cost roughly one lot a week in late shipments, and rework was consuming operator time that the line could not spare. The obvious response, adding a third shift to make up volume, would have pushed more panels through a process whose variation was not yet understood, and the quality team warned that it would simply produce more scrap faster.
Analysis
Teams grounded in Solar Power Systems usually start by drawing the process as it runs, not as the routing sheet describes it. Here the laminator was the only step where the two lines differed in a measurable way, and the difference was in the vacuum and press timing, which the controller logged but nobody had charted against defects. Encapsulant handling was a second candidate: the material sat in a cooled store, and the time between removal and layup varied with the shift. Cell alignment and glass cleaning were already stable, which left two candidates, both testable inside the existing schedule.
The priority order followed the business objective rather than the size of the defect pile. Throughput mattered for the delivery commitment, but added volume alone would raise scrap and rework together. Acceptance criteria were set before any change: first-pass yield at or above 96 percent for five consecutive production days, throughput at or above 90 percent of rating, and no lot shipped with an open void finding. Operators had no spare hours for long trials, so each test ran as a single controlled change on one line during scheduled changeovers. Quality, which signs off each lot, agreed the criteria in writing before the first test.
Actions taken
- Charted laminator vacuum and press timing against void and delamination counts for the last ninety days, split by shift and by line, to confirm whether the logged timing differed on the bad days.
- Timed encapsulant from cooler removal to layup on each shift for two weeks, then set a maximum dwell time and a visual tag system so operators could see when a sheet had exceeded it.
- Ran one controlled vacuum-profile change on Line Two during a changeover, holding all other settings fixed, and compared five days of yield against Line One on the unchanged recipe.
- Adopted the profile that met all three criteria for five consecutive days on Line Two, then rolled it to Line One after a second confirmation run and removed the extra shift from the plan.
Results
Line Two reached 96 percent first-pass yield after the vacuum profile change and held it through the following two weeks. Throughput on both lines returned to about 91 percent of rating, and the lot due to the utility buyer shipped on time with no void findings. The encapsulant dwell limit cut the tag-out problem by roughly half, though the team has not yet separated its effect from the profile change. Priya cancelled the third shift request, and the rework cell now handles about a quarter of its former volume.
Questions for discussion
- Which measurement would you have trusted least in the first ninety days of data, and what would you have done to check it?
- How would you rank the encapsulant dwell change against the profile change if both had to be justified to the buyer separately?
- What should a team do when a reasonable acceptance criterion is met for five days and then missed on the sixth?
Case Study 3
Missing arc-fault and structural evidence on a warehouse roof
Brackenridge Cold Storage, a regional operator of three refrigerated distribution centers with about 1,450 staff, planning a 1.8 megawatt rooftop array on its newest site.
The situation
Brackenridge Cold Storage planned a rooftop photovoltaic array on its newest distribution center, a steel-framed building with an insulated panel roof over ammonia refrigeration plant. The installer, a contractor named Oyelaran Electrical, submitted the design package in the third month of the project. The package included single-line diagrams, racking layouts, and a wind and snow load summary. It did not include a structural check of the purlins under the racking, a fire service access plan for the array, or evidence that the rapid shutdown system had been specified for the site. The project engineer, Tomas, flagged the gaps before the design freeze, because the permit application had already named the roof as the mounting surface.
The challenge
The review had to answer a narrow question: could the design freeze proceed, and if not, which items had to close first. Brackenridge wanted the array energized before the peak refrigeration season, because the power purchase terms and the site's demand charges both depended on it. The ammonia plant sat directly below part of the roof, so any change to the penetrations or the load path would also touch the facility's process safety file, which the site safety lead, Imani, had to sign.
Analysis
The hazards fell into three groups. The first was electrical: dc arc faults on long string runs, and the risk that firefighters would meet energized conductors on a roof with limited access points. The second was structural, because insulated metal panels carry load differently from the purlin-and-deck assumption the racking vendor used, and a snow event on a building holding a refrigerant plant changes the consequence of a sag. The third was process: any work near ammonia equipment needs its own permit, and roof penetrations can compromise the vapor barrier. For each group, the review asked what control existed, who had verified it, and what document showed it.
The business objective was an energized array before peak season, with no permit condition that could later force a shutdown. That made the missing evidence a schedule issue as much as a safety one. Two gates were set. The design freeze required the structural check and fire access plan on file before racking was ordered. The energization gate required the rapid shutdown listing to be checked against the electrical code edition the jurisdiction had adopted, not the one the installer cited. Imani's signature carried the most weight with the insurer, and the building owner had to approve any added load in writing.
- Hazard review Group hazards by electrical, structural, and process exposure.
- Control verification Confirm each control with a named verifier and a document.
- Document audit Check listings, drawings, and sign-offs against adopted code.
- Approval gate Energize only when every condition is closed and signed.
Actions taken
- Requested a structural assessment of the purlins and deck under the proposed racking, using the insulated panel manufacturer's load tables, with snow and wind cases run at the site's design values.
- Built a fire service access plan with two roof walkways, labeled disconnects at the stair tower, and a rapid shutdown placard, then sent it to the local fire marshal for comment before the design freeze.
- Required the installer to certify the rapid shutdown listing against the adopted code edition, and to show the as-built string layout, with conductor routes marked on the roof plan and the ammonia exclusion zone drawn in.
- Issued a hot-work and roof penetration permit that required Imani's sign-off, and tied the energization gate to the process safety file so no string could be commissioned until the vapor barrier was inspected.
Results
The structural assessment found two purlin bays that needed reinforcement before the racking could carry the snow case, so the heaviest rows moved over the steel columns. The fire marshal asked for a second roof walkway, which was added without a design change. The energization gate held for eleven days while the vapor barrier was inspected and the listing paperwork reconciled. The array energized before peak season began, with every permit condition closed. Tomas estimated the review cost about two weeks of design time, far less than a permit condition that would have stopped the system after commissioning.
Questions for discussion
- Which of the three hazard groups would you have escalated first, and what evidence would have changed that order?
- Who should hold the sign-off authority on the energization gate when the insurer and the design team disagree?
- What would a review need to show before accepting a verbal assurance from an installer about the listing?
Case Study 4
Transformer delay and the array energization slip
Pelham Grange Foods, a regional food processor with three plants and about 2,300 staff, building a 3.2 megawatt ground-mounted array and battery block to cut peak demand charges.
The situation
Pelham Grange committed to energize its ground-mounted array and battery block before the start of the high-demand season, because the contract with its utility included a demand charge reduction the finance team had already booked. Construction began on schedule in the early months. The padmount transformer was the long-lead item, and the supplier confirmed a delivery date that fit the plan. Four weeks before the planned energization date, the supplier revised the date by about eleven weeks after a factory test failed on a bushing. The civil and electrical contractors were already mobilized, and the piling and fencing crews were close to finishing.
The challenge
The question for the steering group was which recovery plan to fund, and how much schedule and cost to accept in exchange. Pelham Grange's controller, Mei, had a cap on the overrun the board would approve without a new vote, and the demand charge saving depended on the battery block being live for the season's first peak. A late energization would cost roughly the full seasonal saving, while an early partial energization using a temporary connection carried its own utility approval risk.
Analysis
The evidence was better than the schedule suggested. The supplier's failure report named the bushing, and the replacement unit had passed the same test in the factory, so the revised date was credible. The utility's interconnection engineer had reviewed the single-line diagram and confirmed that a temporary feed could be approved for a limited export period. Three options were costed against the same baseline: waiting for the unit, resequencing the work so that battery commissioning could begin on a loaned transformer, and expediting the unit by air freight. Each estimate carried a range, and the team put the ranges next to one another rather than averaging them.
Each option was scored on saving retained, cap exposure, and crew idle time. Waiting kept cost lowest but moved energization past the first peak, which put roughly the full seasonal saving at risk. Resequencing recovered most of the saving with a loaned unit, with a rental and handling cost of about a fifth of air freight, but it needed a utility approval that could take several weeks. Expediting the unit was the fastest route to a firm date and the most expensive, and it left the crews idle for part of the week while the freight moved. Mei was asked to accept the resequencing cost within the cap.
Actions taken
- Obtained written confirmation from the supplier of the replacement unit's factory test report and a firm ship date, with a penalty clause tied to any further slip beyond the confirmed week.
- Secured a loaned transformer from a rental fleet for a fixed period, sized to carry the battery block and the first array block, and signed a handling agreement covering delivery, crane time, and return.
- Filed a temporary interconnection request with the utility's engineer, attaching the revised single-line diagram, the export limit, and the rental unit's test record, and asked for a decision within the approval window.
- Resequenced the crews so fencing, cable trays, and battery enclosure work closed out first, and staged the array blocks for energization on the loaned transformer, cutting idle time to about two days a week.
Results
Battery commissioning began on the loaned transformer in the third week after the revised date was known, and the array blocks followed as the utility approval came through. The energization date moved by about three weeks rather than eleven, and the demand charge saving was recovered for most of the season. The rental and handling spend came in below the estimate's midpoint because the crane was booked into a single morning. The supplier's unit arrived later and became a spare. Mei reported the outcome to the board as a schedule slip with a managed cost, which the cap allowed.
Questions for discussion
- How would the comparison change if the utility had refused the temporary feed entirely?
- Which estimate in this review had the widest range, and should the team have narrowed it before choosing?
- What penalty or incentive terms would you have asked the supplier to accept before the factory test failed?
Case Study 5
Cleaning options for a dusty carton plant array
Sable Ridge Packaging, a folding carton plant with about 540 staff and a 1.1 megawatt array over its warehouse roofs and staff parking.
The situation
Sable Ridge Packaging runs its folding carton lines, and the paper dust they generate settles on the array at a rate that monitoring showed as a steady output loss between cleanings. The array covers two warehouse roofs and a carport over the staff lot, about 1.1 megawatts in total. The site had been washing the modules with potable water from a tanker every six weeks, a method that used roughly 9,000 liters per wash, required a crew of three, and left a residue that the crew said attracted more dust. The sustainability lead, Hana, wanted the cleaning program cut in water and in the energy spent on tanker deliveries, without any drop in the output the power purchase agreement assumed.
The challenge
The team had to pick a cleaning approach that a plant with a small maintenance staff could sustain, and the choice had to be defended on lifecycle measures as well as on this season's output. Two constraints narrowed the field. The warehouse roofs carry a membrane warranty that restricts foot traffic and the use of abrasive pads, and the carport modules sit over a working lot where equipment access is limited to early mornings.
Analysis
Four options were scored against the same measures: water per cleaning, energy per cleaning including delivery and pumping, output recovered per visit, crew hours, and the disposal burden of the wash residue. Potable water from a tanker scored badly on both water and transport energy. A recycled-water wash with a neutral detergent cut the potable use sharply, but it still required a tanker for the rinse and a hose-down crew, and the residue needed filtering before disposal. A dry robotic brush used no water and little energy, though it could not remove the film that the crew had described, which affected output after a long dry spell.
The business objective was to cut water and transport energy while holding the output the power purchase agreement assumed, so the schedule had to follow soiling rather than the calendar. The soiling sensor already installed for monitoring showed that output loss flattened after rain and ran fastest during dry stretches of peak production, which made a fixed six-week interval too long in some months and too short in others. Hana cared most about the water and carbon figures, Dmitri, the crew lead, about safe access on the warehouse roofs, and finance about whether a robotic unit would pay back within its service life.
Actions taken
- Tied the cleaning schedule to the existing soiling sensor, with a wash trigger at a set output loss and a minimum interval that stops crews going out after light rain.
- Ran a six-week trial of the dry robotic brush on the carport modules, where access was easiest, and logged output recovery, charging energy, and crew hours against the tanker wash on the adjacent row.
- Replaced the tanker rinse with a closed recycled-water loop on the warehouse roofs, filtering residue to a cake that went to the plant's existing waste stream rather than the drain.
- Set acceptance limits before the trial: no more than 2 percent output shortfall against the sensor baseline in any month, and no roof access beyond the existing hatch.
Results
The sensor trigger reduced washes from one every six weeks to about one every nine weeks, and the tanker deliveries stopped once the recycled loop ran. Potable water use for cleaning fell by more than half over the period. The robotic brush stays in trial on the carport, because its purchase case depends on crew time the trial did not measure cleanly. Output shortfall stayed under the limit in every month but one, which was traced to a dust storm outside the normal season. Hana reported the results with the lifecycle figures and the unresolved question about the robot.
Questions for discussion
- Which measure would you weight most heavily if the lifecycle figures and the crew time pointed in different directions?
- Should a soiling-triggered schedule be tested against a fixed interval for a full year before it is trusted?
- What output shortfall would you accept during a dust event outside the normal season, and who should decide?
Learning Resources
Industrial CASE STUDIES
China, India, United States, Indonesia, Pakistan, Nigeria, Brazil, Bangladesh, Russia, Japan, Mexico, Philippines, Vietnam, Egypt, Turkey, Germany, United Kingdom, France, Italy, Canada, South Korea, Australia, Spain.