How new battery capacity helped California avoid another blackout

Persistent heatwave conditions—which prevailed over most parts of California and the broader western U.S. from August 30 to September 6, 2022—resulted in a record peak, testing the grid’s reliability. At 51.4 GW, the actual gross peak demand reached its apex on the last day of the heatwave.

The August 2020 heatwave resulted in a surge in demand to a peak of 46.8 GW, which resulted in blackouts. Even though the September 2022 heatwave caused a higher demand of 51.4 GW, new BES contributed 3.4 GW of peak generation to help prevent the blackouts.

What made this possible? Setting aside demand response, as the exact amount of demand response data that occurred during each 5-minute block is not yet known, one key difference between 2022 and 2020 was the additional battery capacity recently brought online in California.

Forecast vs. actual peak demand

We compared the actual peak demand against CAISO’s forecasted demand. In May 2022, CAISO published its 2022 Summer Loads and Resources Assessment report. This report includes the 1-in-2 (or base case) load forecast, plus two plausible high case scenarios characterized as 1-in-5 and 1-in-10 case forecasts, which are not most likely but still have a chance to happen. The 1-in-2 forecast is used by ICF for a wide range of assessments including all CAISO base case market studies, locational marginal price (LMP) forecasting, long-term capacity expansion planning to meet the RPS targets, etc. The high scenario forecasts are used for reliability planning studies and to assess the system under stressed conditions.

CAISO operates both day-ahead (DA) and real-time (RT) markets. The DA market is a forward market that establishes the generation needed to meet the forecasted demand for the next day. On the load side, the DA market also considers the demand forecast for each time period for the next day. The RT market is a spot market in which utilities can buy power to meet the last few increments of demand not covered in their day ahead schedules. Similar to the DA market, the RT market also considers the demand forecast on 45 min to hour-ahead time intervals. The DA demand forecast, and the hour-ahead (HA) demand forecasts are shown in Table 2, for September 6, 2022. The key point to note here is that the demand forecast deviations are met by the quick ramping battery units, thereby fulfilling the grid reliability aspect.

Using the months-ahead demand forecast from the CAISO report and the DA/HA forecasts from each previous day during the heat wave week, we compared the actual demand values with reference to the forecasts. We see that towards the end of the heat wave week, the actual demand values were in the 85th and 90th percentiles. Given the extreme heatwave, higher prevailing temperatures during the late evening hours (even after 6:00 p.m.) resulted in much higher residential and commercial cooling loads. The table below shows the peak demand for each day during the heat wave, and how it compares against the 2022 report.

For example, on day 8, the observed gross peak demand was 51.425 MW. This was closer to the 1-in-10 forecast in the report (i.e., the 90th percentile forecast value). Given these sustained periods of peak demand, it is important to assess how well reserve margins held up and how blackouts were averted.

Operating Reserve Margins

Operating Reserve Margin refers to the availability of excess reserve capacity (supply), on top of the expected peak demand, to meet emergency conditions. The chart below shows the net RA capacity plus credits, plus the reserves and forced outages during the three specific five-minute time intervals on September 6, 2022. The three representative time slots were chosen such that they occur after the sun had set and the solar generation starts decreasing, on the chosen day (i.e., September 6) when the max peak demand occurred.

Following the same methodology used in the calculation of the planning reserve margins, the RA capacity credits plus reserves net of forced outages are divided by the net demand (gross demand minus wind and solar generation) to calculate the reserve margins during these three specific time slots.

Battery performance

Battery energy storage units generated about 85.4 GWh of energy from August 30 until September 6, 2022. On a five-minute interval basis, we saw that nearly all the storage units were dispatched for energy on the grid. If we consider the peak demand time slot for each day, the BES contribution to meet peak demand varies from an average of 4% to as high as 6% (calculated as the ratio of BES’s discharge MW to the demand MW in the peak time slot).

Table 4 shows the summary of key parameters, observed using the CAISO market data during the heat wave week. The data in the table indicate that other corresponding parameters when the day’s peak demand occurred (such as what was the LMP during the specific 5-min interval, etc.).

BES discharge vs. demand

The BES units’ discharge patterns were largely tracking demand, as seen in the figure below. Higher GW contributions from the BES units were concentrated during the late evening high-demand hours from 5:30 p.m. to 9:00 p.m., which coincides with the decrease in solar generation. BES generation increases exponentially with increase in demand.

It should be noted that the BES units’ max contribution was 3.36 GW, which occurred at 6:30 p.m., about 30 minutes after the peak demand for the day occurred. Nevertheless, it was still in the peak demand hours window. At this same instant, CAISO’s overall BES contributions stood at 6.9% of the peak demand.

During the evening hours from 5:00 p.m. to 9:00 p.m., when the grid was experiencing very high demand levels, the BES units discharged 1.5 GW or more continuously (reaching as high as ~3.36 GW) and contributed directly to averting blackout conditions.

Planning for the future

Climate change represents a significant challenge for California. Climate scientists have found that Southern California heat waves are becoming more frequent, more intense, and longer-lasting, as well as exhibiting higher nighttime temperatures and humidity—particularly in inland urban areas.

There are five key climate-related hazards for California: higher temperatures and extreme heat events; more severe wildfires; more frequent and intense droughts; flooding due to extreme precipitation events; and coastal flooding and erosion from sea‑level rise. Of these five hazards, periods of sustained high temperatures and extreme weather readings will have the most direct impact on the peak demand forecast, leading to more frequent scarcity condition events when electricity supply and demand becomes severely imbalanced.

Towards the end of the September 2022 heat wave, the actual peak demand values were greater than 80th percentile of the forecasted values used in the 2022 Summer Loads Assessment report. As the impacts of climate change continue to wreak havoc on long-held approaches to forecasting demand in California, ICF believes it will be important for CAISO to incorporate the occurrence of prolonged Heating Degree Days (with the daily temperature delta above 65 F) or Cooling Degree Days (with the daily temperature delta below 65 F) into its weather simulation model.

By properly maintaining the state of charge, battery units can serve unplanned load deviations for durations as observed during the California heat wave. The strong performance of battery storage in helping CAISO prevent rotating blackouts recently is a positive sign for grids across the US, particularly as the incentives in the Inflation Reduction Act lead to increased battery deployment as well as (on the demand side) electrification of vehicles, heat pumps, and other technologies. Events such as winter storm Uri or heat waves resulting in prolonged high temperatures will, however, also require longer duration batteries, along with other dispatchable capacity and demand response events in order to maintain grid reliability.