How long do solar panels last? Degradation, warranties, and the year-25 math

The short answer: output at year 10, 25, and 40

At the NREL median of 0.5 percent a year, compounded, a panel keeps about 96 percent of its first-year output at year 10, about 89 percent at year 25, and about 82 percent at year 40. Read that table twice, because it contradicts both sales directions at once: the panel is not worn out at 25 years, when it still delivers nearly nine tenths of its original energy, and it is not producing at full rating in year 25 either, which matters to any savings projection that pretends otherwise. The Department of Energy's consumer writing puts panel life in the same place from the other direction: solar panels last about 25 to 30 years, a service life, not a cliff edge.

What the warranty promises, and what it does not

A panel carries two warranties that owners routinely merge into one. The product or workmanship warranty covers defects: the panel fails, delaminates, or its junction box dies, and the maker repairs or replaces it. The performance warranty covers the fade: the maker guarantees output stays above a stated schedule, typically a higher floor in year one and a stated percentage of the original rating at year 25. The guaranteed floor sits deliberately below the expected median path, because it is written for the tail of the distribution, not the middle. A panel tracking exactly at its warranty floor is underperforming the NREL median while being fully inside its warranty.

Read your own datasheet for the two numbers that set your system's clock: the first-year degradation figure and the annual rate after it. Panel makers state these explicitly, and they differ by model, which is one honest difference between a premium panel and a cheap one. Then hold the document against the table above: if a quote's savings model shows production flat for 25 years, it has ignored the fade its own datasheet discloses, and the payback section below prices that omission.

If you ever claim under the performance warranty, expect measurement, not argument. The warranted schedule is tested under standard conditions, by the maker or an appointed installer, and compared against the floor for that year of the system's life. Your side of that conversation is the production record: monthly figures from commissioning, the commissioning report itself, and the datasheet with its stated first-year and annual rates. This is the second place the maintenance file pays for itself. A system with eight years of records can prove a fade problem; a system with none can only assert one.

The warranty term is not a death date

Treating the 25-year performance warranty as the panel's death date, in either direction: planning to replace a healthy system at year 25 because the paperwork ended, or trusting a savings model that runs year-one production flat for 25 years because the warranty period sounds like a constant-output promise. The warranty is a floor under the fade. The fade itself is about half a percent a year, and it belongs in every long projection.

The panels are not the first thing to wear out

A solar system lives on three clocks, and the panels run the slowest. The inverter is the fast clock: the inverter replacement guide, working from Department of Energy definitions and published service-life figures, puts a string inverter's working life at roughly 10 to 15 years against microinverters at about 25, with the same reporting finding that most array failures occur at the inverter level rather than in the panels. Racking, sealants, and monitoring hardware run the middle clock: they age with the roof and the weather, and they are inspected, not scheduled, which is what the maintenance routine is for.

Plan the system as the panels' 25 to 30-plus years containing one mid-life inverter event, and the lifespan question mostly answers itself. The purchase decision that matters is not whether the panels last; the field data says they do. It is whether the quote's economics survive the fade curve in the table and the inverter replacement in the middle of it. Both are knowable before you sign, from documents the installer already has.

A battery, if your system has one, runs on a fourth clock entirely: cycles and calendar together, and typically the shortest warranty in the file, with a throughput cap in megawatt-hours that can end coverage years before the printed term. Storage does not fade on the panels' schedule, and its replacement decision lands on its own arithmetic of throughput and backup value, which is why this site treats it as a separate decision in the battery guide rather than folding it into the array's lifespan. When you read a whole-system quote, line the four clocks up explicitly: panels at decades, inverter at a decade or so, battery on its own terms, roof underneath everything. Most lifespan surprises are just two of those clocks being mistaken for each other.

Why the median is a median: what moves your panel around it

The NREL review's second lesson hides in its average. The mean degradation rate across the measured systems ran higher than the median, pulled up by a long tail of systems that faded faster, which means the spread between a well-made, well-installed system and a poor one is wider than the headline rate suggests. Heat is the steady pressure: hotter operating conditions age cells and encapsulants faster, which is one reason the same panel model posts different field results in different climates. Moisture ingress, manufacturing quality, and installation care separate the tail from the middle over decades.

You cannot shop your way to a guaranteed personal rate, but you can refuse the obvious tail risks: a panel model with no published degradation schedule, an installer whose mounting and flashing work you cannot inspect in references, and a price so far below the benchmark that the savings had to come out of the equipment. The median belongs to the typical system. The tail belongs disproportionately to the careless purchase.

Year one deserves its own sentence because datasheets give it one. Most panel technologies settle slightly during their first period of exposure, so makers state the first-year degradation separately from the steady annual rate that follows, and the warranted schedule in a datasheet reflects that shape: a small step down early, then the long shallow slope. When you compare two panels' schedules, compare them whole: a model with a gentler first year and a steeper slope after it can cross the other model's line well inside the warranty period, and the year-25 floor, the number the warranty actually guarantees, is the honest summary of the two.

One impostor belongs in this section because it corrupts more records than any real mechanism: soiling and shade masquerading as degradation. A system whose output slides 3 percent across two dusty years has not faded faster than the median; it has got dirty, or a tree has grown, and the maintenance routine restores what degradation never took. The reverse error exists too, blaming dirt for a genuinely failing string. The distinguishing test is the shape: degradation is slow, smooth, and permanent, roughly half a percent a year, while soiling and shade arrive in steps and seasons and reverse when their cause does. Your monthly records, again, are what make the distinction visible.

The review's own histogram puts numbers on that spread. Of the 1,920 rates it assembled, 78 percent were below 1 percent a year; the mean was 0.8 percent against the 0.5 percent median, and crystalline silicon on its own showed a median of 0.5 and an average of 0.7 percent across 1,751 rates. Thin-film technologies faded faster in the same review, a median of 1.0 percent a year across 169 rates. The chart below compounds three of those rates over 40 years so you can see what a tenth of a percent a year becomes over a system's life.

Hail: the weather event the test and the warranty treat differently

Degradation is the slow clock; hail is the one weather event that can stop the panel clock in an afternoon, and it is worth understanding before you assume a 25-year life in hail country. The Department of Energy's Federal Energy Management Program, writing guidance for federal sites, opens with the reassuring part: historically, panels have survived the majority of hail events they have experienced. Then the limit: in areas that have seen very large hail, greater than 1.75 inches or 44 millimetres across, hail has caused significant damage to modules, and severe hail is the single largest contributor to insured losses on solar systems worldwide.

The certification test explains why. To be called hail certified under IEC 61215, a module only has to come through 11 strikes of 25-millimetre ice balls, about one inch, fired at 23 metres per second (51 miles per hour), with no visible damage and less than 5 percent power loss. The same standard defines optional harder classes, up to 76-millimetre balls at 39.5 metres per second, but manufacturers rarely publish those results, so a datasheet that says only hail certified has told you it passed the one-inch test. The guidance goes as far as quoting an industry white paper's verdict that all PV modules pass that basic test.

For a homeowner the guidance translates into three questions to ask before the purchase, all of them about the module rather than the brand. First, the glass: the program recommends at least 3.2 mm of front glass, notes that 4.0 mm is preferred but uncommon, and warns that glass thinner than 3.2 mm is usually heat-treated rather than tempered, and weaker. Second, the test: in hail-prone areas, ask the installer for proof the module passed a test above the IEC minimum, such as a larger-ball IEC class or a dedicated hail sequence from an independent test lab; the program names RETC's Hail Durability Test, PVEL's Hail Stress Sequence with 50 mm balls, FM Global 4478 and ASTM E1038. Third, the format: it suggests smaller modules, such as 60-cell panels, because large-format 72- and 144-cell modules may be at higher risk, and framed modules over frameless ones.

Tilt matters too, which is the one place a rooftop owner has less choice than a utility plant. Damaging hail falls closer to vertical, so panels tilted steeply take glancing blows; the program cites a tracker study in which tilting panels to 60 degrees raised the survival likelihood from 82 to 99 percent compared with flat panels. By that finding, panels on a pitched roof already sit at some angle to falling hail, while near-flat panels on a low-slope roof do not.

The part that touches this page's lifespan arithmetic is the damage you cannot see. Hail can crack solar cells at sizes and speeds below what breaks the glass, and those microcracks can spread over time, cutting output and raising hot-spot risk. That is a loss the 0.5 percent median does not describe. The program's advice after a storm is to shut down a visibly damaged system until it is inspected, test for electrical faults before repowering, and have the array scanned with infrared, ultraviolet or electroluminescence imaging within a month of the event. For an owner, the practical version is a dated production comparison and an installer inspection after any storm with hail near an inch, filed with the insurance claim; the homeowners insurance guide covers how the claim itself is read, and the panel warranties above are written for defects and fade, so read their exclusions for storm damage before assuming either one applies.

Degradation is already in your payback, or your payback is wrong

Here is the fade's one sharp financial edge. A production estimate and a payback model should both apply degradation from year two onward, so the kilowatt-hours shrink slightly each year. If a model instead sums year-one production across 25 years, it overstates the energy, and therefore the savings, by about 6 percent: at the median rate, the 25 yearly outputs sum to about 23.6 years' worth of first-year production, an average of about 94 percent, not 25 years' worth at 100 percent. On a long payback that gap moves the break-even date by well over a year, which is why this site's payback methodology works from degraded production.

Check your quote with one question: show me the year-10 and year-20 production figures in this model. An installer working honestly has them, because their software applies the panel's warranted schedule automatically. A model that cannot produce a year-20 figure is not a model; it is year one, photocopied. The production estimate guide gives you the rest of the audit.

Buying a house that already has panels

A second-hand system starts its life with you at an unknown point on the fade curve, and the first job is dating it. The installation date sets everything: the warranty years remaining, the expected output against the table, and the inverter's distance from its decade. If the seller cannot produce the paperwork, the Department of Energy's consumer guidance points at the same record it recommends for roof age: your local government holds the building permit, and the permit date is a reliable stand-in for the installation date. The monitoring history, if the seller will share it, is better still: real production against the modelled estimate tells you where on the curve this specific system sits.

Price what you find as remaining life, not as a new system. A ten-year-old array at the median fade still makes about 95 percent of its rating and may have fifteen or more useful years in it, with an inverter replacement likely inside your ownership rather than the seller's. Ask which warranties transfer to a new owner and get the transfer terms in writing before closing, because makers differ and a warranty that does not transfer is worth its remaining paper only. The payback if you move or sell guide works the mirror-image problem from the seller's side.

Price is the other half of that purchase. The home value guide shows that buyers in Berkeley Lab's sales study paid far less for systems about eight years old than for nearly new ones, a faster fall than the panels' own fade.

Lifespan paperwork to keep

• Get the panel's datasheet: first-year degradation and the annual rate after it, in writing.
• Ask the quote for year-10 and year-20 production; a model that cannot show them is assuming flat output.
• Shrink any flat 25-year savings claim by about 6 percent before comparing it with anything.
• Budget one inverter replacement inside the panels' life; the panels are the slow clock.
• Keep the warranty documents and the datasheet with your production records.
• At any future re-roof or repower, get the panels' removal destination named in the quote.

End of life is a decision, not an event

Because a panel at year 25 still makes about 89 percent of its rating, replacement on output grounds alone rarely pays: the new system's entire cost buys back, at first, an eighth more energy from the same roof. Systems actually come down for other reasons: a roof replacement that makes renewal convenient, a repowering project chasing higher-wattage panels on the same footprint, storm damage, or a home sale where the parties price the remaining life. When panels do come down, they are electronic equipment, not demolition waste: the industry runs recycling channels for modules through the Solar Energy Industries Association's national PV recycling programme, and your state or municipality may have its own rules for disposal (the solar panel disposal and recycling guide dates them state by state), so the removal quote should name the destination.

The practical conclusion is almost boring, which is the point. Buy panels with a published degradation schedule from a maker you expect to exist, price the payback on faded production, budget the inverter's decade, and then let the array sit there outproducing its warranty paperwork. Lifespan, for a competently bought system, is the least interesting variable in the whole purchase.

Repowering is the exception that proves the rule, and it is a footprint argument, not an age argument. Panel wattage per square foot has risen steadily across the industry, so a fixed, good roof plane can hold meaningfully more capacity today than it could fifteen years ago; where a household's usage has grown into an electric vehicle or a heat pump, replacing a healthy but small old array with a larger-capacity one on the same mounts' footprint can pencil where a like-for-like swap never would. Price it as what it is, a new purchase decision run through the cost worksheet and the sizing guide, with the old system's remaining production counted honestly as the opportunity cost.

The inverter's death, when it comes, is the one forced decision in a system's middle age, and it deserves the same cold arithmetic as the original purchase. Replacing the inverter restores the array's remaining decades at the replacement guide's benchmark-based cost; repowering replaces everything and restarts every clock at current prices. The deciding figures are the array's age and recorded health: a twelve-year-old system tracking its fade curve has most of its life ahead of it and earns its new inverter, while a system already failing on several axes may be telling you the repower year has arrived early. Either way the panels' slow fade, the subject of this guide, is rarely the deciding voice in the room.