The ground near Florianópolis was flat and the panels were brand new.
Regular inspections started almost immediately after commissioning.
Within weeks, something was wrong.
Technicians began finding broken glass on modules, first a handful, then a pattern.
Month after month they logged an average of 14 cracked panels, and by the nine month mark more than half of the installed modules had broken glass. Nothing had hit them.
Why panels this thin can crack with nothing hitting them
The site ran large format glass on glass modules, the double sided design that swaps the polymer backsheet for a second sheet of glass, here about 0.08 inch thick. Most of the array sat on single axis tracking rows that rotate through the day to follow the sun. That combination, big panels with thin glass on moving frames, has spread quickly across the industry.
No single cause has been established. A survey by US national laboratory researchers lists several candidates that may act together: less thermal strengthening in thinner glass, microscopic flaws on glass edges or surfaces, lamination stresses such as edge pinch, and contact between the glass and the frame or debris trapped in that gap.
The testing lab that published the Brazilian case points to tracker compatibility with large format modules as a critical variable. Weather monitoring at the site showed no temperature or wind anomalies over the period. In the majority of cracked modules, 59 percent, it was the rear glass that broke.
What the modules looked like up close
The fractures are not what a hailstorm leaves, with its round central impact dimple and radiating lines. Federal researchers describe a pattern in which a few large cracks form rather than hundreds of fragments, often with no clear origin and no link to severe weather or an impact event.
On a glass on glass module both faces are structural, so damage is not always confined to the side where it starts. A cracked panel does not stop generating power immediately. But moisture can enter through the fracture line and reach the cell contacts underneath, and losses compound from there.
The lab that documented the Brazilian case says it has been contacted multiple times in the past two years with reports of broken glass at PV sites worldwide.
What the inspections at the Brazilian test site recorded
The array was designed as a 100 kWp research testbed to evaluate bifacial module performance in Brazilian conditions. It used large format bifacial glass on glass modules with 2 mm heat strengthened glass, installed across five single axis tracking rows and one fixed tilt system. The team ended up studying glass breakage instead.
Regular inspections recorded 14 broken modules per month over a seven month period, and within nine months of commissioning over 50 percent of installed modules had cracked glass, as documented in the published case. Front glass broke in 28 percent of cracked modules and both faces in 13 percent.
Work at a German research institute describes the same broad picture from field investigations: a marked rise in breakage of glass on glass modules with thin glass and large areas, often within months of deployment, with severe weather and faulty installation largely excluded. Modules that passed required certification tests were still failing in the field.
What the wider inspection data shows, and the numbers behind it
One major testing lab describes spontaneous glass breakage in glass on glass modules as the most significant reliability issue affecting modules today, and reports seeing it in multiple countries, module types and mounting systems. Its mechanical stress sequence testing hit a high in the second quarter of last year, with roughly a third of modules breaking, easing to about a quarter later.
Its vice president told pv magazine that the industry has eroded safety margins, so that edge defects, misplaced silicone, edge pinch and busbar pressure can now be enough to break a module. On the factory side, a major inspection body graded more than 70 percent of audited PV factories in the bottom two tiers last year, with no plant earning its top rating, across more than 80,000 inspections at over 340 sites.
Mounting geometry matters too. Operators in Texas who stowed their panels at a steep tilt before a major hailstorm showed a related principle: the angle at which a panel sits when stress arrives can decide whether the glass holds, a case covered in the Texas farms report.
What happens to a cracked panel, and what the industry is doing about it
A cracked module is rarely swapped out the day the crack appears. It still generates power, and a ground level visual pass can miss a hairline fracture on the rear glass entirely. An electroluminescence image can reveal cell damage spreading inward that the eye cannot catch.
The panels that do get pulled early feed a growing stream of retired glass. Recycled glass from retired panels has been tested in new modules in Georgia, where it matched material made from virgin sand. That matters more if early retirements keep arriving well ahead of the 25 to 30 year schedule most sites assume.
The fixes now being pushed are measurement based: non destructive checks of glass surface stress on finished panels, tighter process control in lamination and framing, and testing specific module and tracker combinations rather than assuming a certification pass transfers to the field. The Florianópolis site remains the clearest published record of how fast a new array can go from commissioning to more than half its modules cracked.
Read the whole thing?
Get the week's signal, not the noise
Our sharpest reporting on energy, climate and nature — free, once a week.
