Utility-scale storage systems are one of the energy industry’s fastest growing segments as grids around the world strive to increase renewable generation. These storage systems not only take advantage of shifting energy but also provide valuable grid services, and are increasingly being used for capacity – deferring the need for new capacity or allowing for the retirement of existing fossil. Continuing the evolution towards high renewable grids and the replacement of fossil generation will require further buildout of these storage systems. However, most recent storage systems are battery-based and have only 2 to 8 hours of duration; thus, while these “short-duration” systems may be perfect for moving generation within a single day, they fall far short of providing the multi-day, “long-duration,” storage upon which future grids may require to operate through prolonged weather events.
Photo Credit: NASA’s Aqua satellite captured this image on January 20, 2010.
Accordingly, as the world increases the share of electricity generated by renewables over the next 10 to 15 years, it will need more than short-duration storage to maintain grid reliability—it will need storage with the ability to shift generation across longer time frames, such as multiple days, weeks, or even seasons, and to generate for multiple days without recharging. Potential solutions range from pumped-hydro storage, compressed air energy storage, "aqueous air" battery systems, to renewable hydrogen. No one size fits all solution will be able to meet all storage needs, and, as with most questions concerning reliability, we must also carefully weigh the costs and benefits of each possibility.
The California Public Utility Commission’s Integrated Resource Plan calls for nearly 45 GW of solar capacity by 2030, but what happens when the state only has short-duration storage systems to support this massive buildout of solar energy? To answer this question, we evaluated over 20 years of solar irradiance data across more than 150 sites across California. Using both the forecasted buildout of 45 GW of installed solar capacity and the forecasted load (CPUC IRP) for all 21 solar years, we determined what percentage of load in 2030 will be met by solar.
This analysis revealed multi-day periods when solar generation is at an abnormally low percentage of load. Although almost every solar year had some periods of low solar output, 2010 stood out as exceptionally bad after two historic storms hit California in January and December of that year and severely limited solar production. Indeed, these storms inflicted significant damage as far inland as Arizona and as far north as Washington, and they would also have limited some wind generation as many wind plants would have been forced to shut-in their turbines to protect them from the hurricane-force winds. These results thus demonstrate that California cannot stem effects from future storms by relying wholly on resource and geographic diversification of energy production to maintain reliability.
Fortunately, California still has a diverse resource mix, so neither of these storms resulted in reliability issues stemming from a lack of solar generation and a resource that could adequately take its place. However, this analysis clearly establishes the need for a portfolio of energy storage and flexible loads that span multiple durations. This will fill any gaps left during extended weather events.
The figure below shows the effects that the December 2010 storm would have had on California’s grid in 2030, including a six-day period (see yellow markers) when solar generation accounts for little more than 5% of load. This drop’s severity is further highlighted by comparison not only to the adjacent weeks but also to the rolling average of all 21 solar years for the same calendar days. Moreover, six consecutive days of such pitiful solar output means that the system would not be able to recharge short-duration storage facilities to counter the drop. Instead, the grid would need a facility, or combination of facilities, that could discharge over multiple days without recharging. The massive storm in January 2010, captured in NASA’s satellite image of the West Coast, also would have led to a six-day period with very low solar output and a four-day period with a severe reduction. Although rare, these types of storms are not anomalous. And as weather patterns become increasingly extreme and unpredictable, we cannot afford to ignore their potential effect on future grid reliability.
Simulating future grid operations using weather patterns from December 17-22, 2010 when there was significant storm activity across California.
Today most grids, including California’s, rely on fossil fuel-based generation to meet the generation gaps resulting from the shortcomings of short-duration storage. But as California, and many other states and countries, strive to decarbonize, fossil fuels will no longer provide a viable stopgap. Thus, if we are serious about achieving the types of aggressive clean energy goals set by California and other states, we also must seriously evaluate how to innovate and implement long-duration storage that does not involve burning fossil fuels.