A natural experiment in solar energy

At the Wind Energy Institute of Canada (WEICan), we took yesterday’s partial solar eclipse as an opportunity to see how eclipses impact the operational data from our solar PV system.

The August 12 eclipse was a partial solar eclipse across Prince Edward Island. At our site in North Cape, Prince Edward Island, the eclipse began just before 2:00 pm, peaked at around 3:00 pm, and ended near 4:00 pm. While the eclipse was not total in PEI, a substantial portion of the sun was obscured. Atlantic Canada experienced some of the more significant partial coverage in Canada.

Our solar data provides a particularly interesting view of what that meant from an energy-production perspective.

The second graph shows Global Horizontal Irradiance (GHI), which is essentially the amount of solar radiation reaching the surface. On an otherwise smooth August afternoon, GHI would normally continue to decline gradually as the sun moves lower in the sky. Instead, we see a pronounced interruption:

• Before the eclipse, GHI was approximately 865–870 W/m².
• Around the middle of the eclipse, it dropped to approximately 525 W/m².
• That’s a reduction of roughly 40% from the pre-eclipse level.
• As the eclipse passed, irradiance recovered toward the expected daily trend.

The same pattern can be seen in the reflected irradiance measurement, which fell from approximately 135–140 W/m² to about 85 W/m².

When we look at actual PV production, we see that our array was producing approximately 90–92 kW before the eclipse-induced dip. During the deepest part of the event, production fell to approximately 62 kW. That’s a reduction of roughly 30 kW, or about one-third of the array’s output at that point in the afternoon. Then, as the Moon moved away from the Sun, production climbed back to approximately 85–86 kW before continuing its normal afternoon decline.

Not every wiggle was the eclipse

One of the interesting features of the graphs is the short-term variability before and after the main eclipse dip. Those rapid fluctuations are much sharper than the broad eclipse signature and are consistent with changing atmospheric conditions, such as passing clouds. The eclipse itself produces a much more recognizable broad, smooth reduction in available sunlight.

What can we learn from one afternoon?

For solar researchers and grid operators, an eclipse is essentially a naturally occurring experiment.

We know:

• when the reduction in sunlight should occur;
• how long the event should last;
• how much of the Sun is obscured; and
• that the change is caused by the Moon rather than a change in the solar resource itself.

That gives us an unusual opportunity to watch how a PV system responds to a predictable, temporary reduction in solar irradiance.

For a single 109 kW solar array, a roughly 30 kW reduction is relatively small. But when solar generation is deployed at much larger scales, understanding how predictable events affect thousands of megawatts of generation becomes increasingly important for forecasting, grid management and energy storage.