The California ISO recently had a quarterly market performance and planning call during which they addressed the upcoming eclipse on August 21. In the meeting, they outlined their assumptions for load, the rate of decline in solar production heading into the eclipse, and the rate of increase in solar production following the eclipse peak. After reviewing the data they provided, certain underlying assumptions raised several questions. Using historical data, Genscape’s in-house meteorologists and power market analysts examined growth in utility-scale solar to forecast load and solar generation for the eclipse and compared it to ISO expectations.
The ISO displayed its forecast load profile for the upcoming eclipse alongside a load profile from August 22, 2016. According to their plot (Image 1) peak load for the day was approximately 35,000 MW, but referencing data from the ISO’s OASIS website shows actual load peaked at 37,805 MW – close to 2,800 MW higher than indicated!
Another question arising from the CAISO load forecast is: what makes Monday August 22, 2016, a good starting point for load on August 21, 2017? The meteorological and climatological answer is: nothing. Consider July 26, 2017 compared to July 27, 2016: load peaked at 40,871 MW in 2017, while exactly one year earlier, in 2016, it peaked at 45,761 MW, nearly a 5,000 MW difference! A more logical approach is to average the load on all 2016 August weekdays as a starting point and determine if this August is likely to see similar, weaker, or stronger load, on average, than last year.
Genscape’s own meteorologists are forecasting above-average temperatures this August for California, with northern reaches of the state experiencing slightly warmer conditions, 1-2° F on average, and southern regions seeing moderately warmer conditions, 2-4° F on average. In looking back at last August, this forecast compares favorably, as the state saw temperatures average 2.5° F warmer than normal. Assuming similar load to last August yields an on-peak average of 34,568 MW and a peak hourly value of 39,339 MW.
Capacity for utility-scale solar has increased year-over-year with current capacity at approximately 10,400 MW, making a direct relation to last August difficult, as capacity was closer to 8,500 MW. Instead, one can apply the observed hourly change in solar generation from last July to August on a percent basis and apply it to observed solar generation from July so far. This exercise shows that solar from July to August does not change drastically across the strongest hours, roughly 9 AM to 5 PM, but there is a reduction during the ramping hours (Image 2).
Constructing the forecast load net solar curve for August 21, 2017
In the previously mentioned California ISO presentation, they provided a more granular look at forecast solar the day of the eclipse and how low it will drop (Image 3). Using their expectation of where solar generation will be to start the eclipse, the peak of the eclipse, and the end of the eclipse, we can infer average rates of increase and decrease in solar generation for each minute. Doing this calculation, we find that utility-scale solar is likely to decrease at a rate of 51.1 MW/min headed into the peak of the eclipse, and increase at a rate of 62.8 MW following the peak of the eclipse.
Accounting for behind-the-meter solar
Using data from nearly 50,000 Locus Energy meters in the state of California that collectively account for nearly 500 MW of solar capacity, we can produce reliable estimates of clear-sky, behind-the-meter solar production across the state. From there, we can then model the impact that the eclipse will have across the state. Because California is such a geographically large area, the impact of the eclipse will vary dramatically from North to South. Peak obscurity during the eclipse will vary from roughly 60-80 percent, depending on location. As you can see in the graphic below, solar production in the Pacific Gas and Electric (PG&E) service territory will show the largest dip in production, as it covers most of the northern half of the state.
Unlike utility-scale solar, which is heavily concentrated in the southern part of the state, behind-the-meter solar is highly dispersed. According to state data, 2.8 of the roughly 5.5 GW of behind-the meter solar in the state are located in the PG&E territory. The map below further shows how the capacity is distributed across the state, with the heaviest concentrations located around the major metropolitan regions of San Francisco, Los Angeles, and San Diego. Since so much of this capacity is in the northern half of the state, the relative effect of the eclipse will be more significant on distributed solar than utility-scale solar.
By combining the distributed solar capacity data from the state with the modeled capacity factors mentioned above and shown in the graph below
), we can estimate the total MW impact of the eclipse on August 21. At its peak, the eclipse will cause behind-the-meter solar production to be a little over 2,000 MW below what it would normally be on a clear August day. From the CAISO perspective, however, this production drop amounts to what is essentially a 2,000 MW bump in the load curve for that day.
Combined with the 5,800 MW from the effect on utility-scale solar, the overall effect of the eclipse at its peak will require that non-solar generators will have to make up for an additional 7,800 MW more than they would on a clear day.
In comparing forecasts, it is evident we anticipate load net solar on August 21 will be stronger than CAISO is currently expecting. We project load net solar will be roughly 30.5 GW at the eclipse peak while the ISO expects more than 2.5 GW less given their forecast for load to be under 31 GW and utility scale solar production to be over 3 GW. Should the ISO move forward with the assumptions they’ve outlined, an average August load day would drive a considerable under commitment of resources and increase the likelihood of price volatility.
Genscape and Locus Energy will be closely monitoring the solar eclipse event on August 21. Using proprietary power market data and behind-the-meter solar data, our power market analysts will be assessing both generation and production impacts. At the end of the month we will be sharing a follow-up blog on the event on our observations.