The solar eclipse came and went, and California’s power grid is still intact! While the eclipse was quite the spectacle to behold, Genscape saw fairly tame CAISO prices in the fifteen-minute market (FMM). During the hours of the eclipse, hours ending 10-12, southern California (SP-15) FMM prices averaged $19.39/MWh while day-ahead market (DAM) prices at SP-15 averaged $39.44/MWh, resulting in positive DA-FMM spreads of more than $20 on average (Figure 1). What drove these weak prices and how did our forecast of the event perform?
In our pre-eclipse blog, we used an average weekday from August 2016 as our baseline for what load would look like the day of the eclipse, with an expected peak at 39.3 GW. Instead of a normal August weekday, observed temperatures on the day of the eclipse ranged from 5-7° F below normal throughout the state. This produced the third weakest on-peak average load and fourth weakest peak hour load among weekdays this August.
During the eclipse, hours ending 10-12, load came in 3,075 MW below our average August load assumption and peak hour load came in 3,500 MW lower (Figure 2). Not only did load come in below our assumption, but day-ahead demand also cleared weaker than our assumption by 840 MW. During the eclipse, load averaged 925 MW weaker than day-ahead cleared demand, resulting in more resources being committed in the day-ahead market than were necessary in real time. A load miss of this magnitude begs the question: did the eclipse cause this directly, or were there other factors at play?
Using Locus-monitored behind the meter solar measurements, we can conclude that a reduction larger than what was forecast for rooftop solar was not the culprit behind the load miss during the eclipse. Comparing the output from Locus monitors on August 21 to Locus Energy’s forecast, we can see that, during the eclipse and into the afternoon, generation came in largely as expected (Figure 3). The most likely explanation for the load miss is a combination of cooler-than-expected temperatures (less of a need for daytime cooling) and non-weather load impacts (a lot of people took the day off to observe the eclipse).
During the eclipse, Genscape's solar forecast performed very well, with an hourly average error of only 135 MW. The bulk of the error came from solar generation missing the forecast prior to the start of the eclipse and a slower decrease in generation into the peak of the eclipse. We had expected solar to reach 6,725 MW prior to the start of the eclipse and decrease at a rate of roughly 51 MW/min until the peak. Instead, solar peaked at 6,350 MW, 475 MW lower than forecast, at the start of the eclipse and decreased at a rate closer to 40 MW/min. For comparison, the ISO forecast called for solar to reach 7,337 MW at the eclipse start while expecting a down-ramp of 70 MW/min. The other side of the eclipse saw solar return at a slower rate than expected, averaging an increase of 55 MW/min compared to our forecast for 63 MW/min and the ISO forecast for an up-ramp of 90 MW/min.
Load net solar was within 320 MW of our expectation to start the eclipse, as weaker solar was offsetting weaker load. From the start of the eclipse to its peak, we forecast load net solar to increase 7,200 MW. Instead, thanks primarily to well-below-normal temperatures, load net solar increased closer to 4,200 MW, falling short of our estimate as well as the ISO’s estimate of 6,000 MW.
The return of both utility-scale and behind the meter solar, as well as decreasing lighting load, helped drive load net solar down more than 6,200 MW following the eclipse peak. This rapid decrease in load net solar helped drive SP-15 FMM prices to below $12 for the two hours immediately following the eclipse peak. Also facilitating weaker prices during and following the eclipse was much stronger wind generation than both the ISO and Genscape’s forecast. Wind averaged 1,420 MW HE 9-13 coming in 790 MW above Genscape’s forecast and 715 MW above the ISO forecast. Had wind not come in as significantly above forecast, resulting in substantial excess generation versus the day ahead market, prices would not likely have been so weak immediately following the eclipse.