Regional and National Utility-Scale Solar Generation Forecasts

The MDA Solar Power Forecasting System now is used to predict generation from every utility-scale solar farm 1 MW or larger in the U.S. and aggregate these to regional, RTO, and national totals. Interesting results will be shown. A particular issue is curtailment in California, where forecasts for CAISO are much too high when curtailment is occurring. Publicly available real-time curtailment data is not available in sufficient detail to adequately correct the forecast, so a correction is being developed based on a combination of load, weather, and solar power forecasts. Another interesting case is the August 21, 2017 solar eclipse. The eclipse-day full forecasts utilizing clouds and eclipse will be discussed. Even beforehand, we ran simulations comparing eclipsed generation to full clear-sky generation at all of the roughly 2000 solar farms and summed the results, shown in the graphic below. With roughly 40% of the US total utility scale generation in CAISO, the impact on the US total is largest while the eclipse shadow is still in the western US, even though most of those solar farms are not near the region of totality. Also, it is interesting to see the comparison in responses of fixed tilt, single-axis, and dual axis trackers and solar thermal plants, each of which are modeled individually in the forecast system. Figure 1 below: Dots are individual solar farms, with 5-minute power output as a fraction of capacity shown by the size/color and fraction of solar disk covered by shadow shown by the background shading in 10% increments. Graphs on the right are percentage of capacity for clear sky conditions for eclipse day and no-eclipse day for CAISO, Duke Energy Carolina’s service area, and contiguous US for the period 10am-6pm EDT. The numbers are 5-minute generation (MW) and 5-minute ramp rate scaled to MW/hour for the time shown on the map (13:40 EDT).