Solar Panel Degradation

Built by Jeremy Panasuk. Published October 3, 2026.

The rates are from an NREL review of published field measurements. The wattage they are applied to is this site's calculator. The straight-line compound is an illustration, because the review says linearity is not settled.

The review's summary: Jordan and Kurtz, NREL/JA-5200-51664 (June 2012), assembled nearly 2,000 degradation rates for flat-plate modules and systems. The conclusion states a mean of 0.8% per year and a median of 0.5% per year. In the histogram, 78% of the data are below 1% per year, and the count of reported rates is 1,920. Long-term rates in that histogram exclude short-term light-induced degradation. The authors say a 20% decline is typically considered a failure, and that there is no consensus on that definition.

What they would not claim

The conclusion says that linearity, and the precise impact of climate, have not been satisfactorily answered. The introduction says a higher degradation rate means less energy later, and that a wrong rate becomes financial risk. This page uses the median and the mean as published summary statistics. It does not turn them into a warranty, a climate correction, or a forecast for one brand of module.

An 800 W result with nothing left to give

Battle Born's 2,400 Wh example, at 4 sun hours, goes through sizeSystem as 2,400 ÷ 4 ÷ 0.75 = 800 W. Eight hundred is already a multiple of 100, so the round-up to the next 100 W adds nothing. The array the calculator prints is the array the day's load asked for, once the 0.75 factor is included.

Apply the median 0.5% per year as a constant compound, which is the illustration the paper's own caveat limits: the remaining fraction after 25 years is 0.995²⁵ = 0.88222. Then 800 × 0.88222 = 705.78 W. The same day's 2,400 Wh still wants 800 W at 4 sun hours. The gap from 705.78 W back to 800 W is the cushion the rounding step did not create.

The mean rate in the same paper is 0.8% per year. Held constant, 0.992²⁵ = 0.81807, and 800 × 0.81807 = 654.46 W. A 20% decline, the figure the paper says is typically called a failure, would leave 640 W. Both illustrated 25-year results are still above 640 W. They are both below the 800 W this particular load still divides out to. DOE separately says an inverter is expected to be replaced at least once in a 25-year array life. That replacement is a balance-of-system event, not the module median.

Where rounding does leave a cushion

The refrigerator example on this site is 100 W × 5 h = 500 Wh. At 4 sun hours, 500 ÷ 4 ÷ 0.75 = 166.67 W, and the calculator rounds that up to 200 W. The same 0.5% per year illustration leaves 200 × 0.88222 = 176.44 W after 25 years, which is still above 166.67 W. Held to the same constant rate, year 36 is 200 × 0.995³⁶ = 166.98 W, and year 37 is 166.14 W. Year 37 is the first year that illustration falls under the un-rounded 166.67 W. The paper's statement that linearity is unsettled applies to every one of those years. The refrigerator page keeps the 500 Wh label as an example, not a meter reading.

A cabin that sizes to winter sun hours is already buying the larger of the two lines in sizeSystem. For 2,400 Wh that winter line is 1,300 W, because 2,400 ÷ 2.5 ÷ 0.75 = 1,280 and the function rounds up to the next 100 W. Degradation and winter are different enlargements. One is a sun-hours input. The other is a literature rate applied after the array exists. The cabin walkthroughs live on off-grid cabin solar and what size solar system.

One common mistake

Subtracting 0.5% once and calling the array done. The median is a per-year rate in a skewed distribution whose mean is 0.8% per year. A single 0.5% haircut on 800 W is 796 W, which is not 0.995²⁵. The other mistake is treating the median as linear because a spreadsheet makes it linear. Jordan and Kurtz wrote that the linearity question was not settled.

Limits

These rates are a literature summary through the studies they reviewed, reported in 2012, with short-term light-induced degradation left out of the long-term histogram. They are not a measurement of a module on a roof this year. Module warranties, if you need one, are the manufacturer's document. Production at a specific address belongs in PVWatts. Electrical work belongs with a licensed electrician.

FAQ

What degradation rates do Jordan and Kurtz report?

The conclusion of NREL/JA-5200-51664 states a mean of 0.8% per year and a median of 0.5% per year. The histogram uses 1,920 reported rates, and 78% of the data are below 1% per year. Those long-term rates exclude short-term light-induced degradation. The authors say linearity and the precise impact of climate were not satisfactorily answered.

What is left of an exact 800 W calculator result after 25 years at 0.5% per year?

If the 0.5% per year median is compounded constantly, 0.995 to the 25th is 0.88222, and 800 × 0.88222 = 705.78 W. The same 2,400 Wh at 4 sun hours still calculates to 800 W. The compounding is an illustration, because the paper says linearity is not settled.

Why does the 500 Wh case still clear its un-rounded watts at year 25?

500 ÷ 4 ÷ 0.75 = 166.67 W, and the calculator rounds up to 200 W. At the same constant 0.5% per year, 200 × 0.88222 = 176.44 W after 25 years. In that illustration, year 36 is 166.98 W and year 37 is 166.14 W.

Compiled with AI assistance from the sources below. Spot an error? Email [email protected].

Sources

Jordan and Kurtz, Photovoltaic Degradation Rates — An Analytical Review, NREL/JA-5200-51664, June 2012 — accessed Oct 3, 2026.
U.S. Department of Energy, Solar Photovoltaic System Design Basics — accessed Oct 3, 2026.
Battle Born Batteries, How to Size a Deep Cycle Battery Bank, May 22, 2026 — accessed Oct 3, 2026.
SolarSizer calculator, assets/app.js (sizeSystem) — accessed Oct 3, 2026.

← Back to the solar size calculator