Solar Panels: Technical Details

Summer and Winter

The left chart below displays total kWh generated per day in July 2010. Mid-summer is almost 20 kWh per day every day. The sun always shines in this part of California.

In contrast, the right chart below shows total kWh generated per day in December 2010. It's a lot less than in summer even on a good day, and there's lots of day-to-day variation due to clouds and rain. It was very rainy from Dec 13 to Dec 25.

Daily electricity generation by solar panels in July
Daily electricity generation by solar panels in December
Seasonal Variation
Seasonal variation of electricity generated by solar panels
These curves (data recorded every 15 min) show the seasonal performance on a totally clear day in winter, spring, summer, and fall. These are displayed as local time, not daylight savings time, so that the four can be compared.

Solar noon at our house it almost exactly 12:00. The midsummer curve peaks right at 12:00. The shift to maximum generation at ≈11:00 on October 1 was at first a surprise but can be explained partially from the panels facing 15° east of south and partially by the eccentricity of the earth's orbit, which shifts the sun's position in the sky at noon.

The changing length of the day is clearly visible. Total daily generation on a sunny day on January 1 is less than half that on July 1, but that is due more to the shorter day than to the lesser peak power at noon. The fact that April meets or even exceeds July's peak is due to cooler spring temperatures boosting the panels' efficiency.
Panel Degradation
Yearly decline in solar panel efficiency
PV panels do degrade with time. This chart shows the average daily generation on the best 60% of days between May 1 and August 31. Basically days with 100% sun during a time period roughly centered on the summer solstice. The average daily generation is steadily decreasing as the panels age. The trend line is right at a 1.0% loss per year. This is pretty consistent with data reported elsewhere.

The issue seems to be the development of microscopic cracks in the panels. This occurs partly from thermal cycling – the daily expansion and contraction as the panels heat and cool – and partly from high winds flexing the panels. The reported 25-30 year life of panels is based on rather arbitrary notion that the panels have reached the end of life when the efficiency has decreased by 25%. The panels will still be fully functional, so it becomes an economic issue of whether it makes sense to replace the panels or to continue generating at a reduced level.
Panel Washing

By 2021, the 12-year-old panels had acquired a layer of sticky grime. Probably pollen adhering to the panels, then dust sticking to the pollen. Wiping a section of a panel with a paper towel after spraying with a glass cleaner left the paper towel pretty black and the panel smoother and cleaner. But cleaning 16 large panels this way was not feasible, so we hired a cleaner who used a cleaning process with hot deionized water. I was skeptical that water alone would clean them, but it did. They now feel smooth, and glass cleaner no longer cleans anything.

That said, the improvement in energy generation was very small. It's rather hard to judge because day-to-day and year-to-year fluctuations are large. By looking at the maximum daily power generation for 3 weeks before and after cleaning, and also looking at the same time span for several other years, I estimated that I should easily see a 5% increase in performance. I didn't. At best, performance increased 1 or 2 %. That makes up for 1 or 2 years losses due to panel degradation, but it doesn't in any way make up for the steady, inexorable decline. It seems a bit surprising, but the sticky grime was not really hurting performance.

The company wanted us to set up a yearly (or more often) cleaning schedule, which I declined. Considering the limited improvement and the cost of the cleaning, a cleaning more than every 5 or 6 years probably cannot be justified.

© 2021 Randy Knight