The International Panel on Climate Change (IPCC) advocates for net-zero emissions by 2050, a goal adopted by numerous countries. Decarbonizing electricity grids is paramount, given their substantial contribution to global greenhouse gas (GHG) emissions. The rise of renewable energy, driven by decreasing costs and incentives, offers a pathway to emission reduction, but achieving a reliable, entirely net-zero grid remains a challenge. Researchers have explored various solutions, including firm low-carbon resources with flexible operation, transmission expansion and inter-regional coordination, and the role of energy storage. This study focuses on long-duration energy storage (LDES), defined as storage with 10 or more hours of duration, and its interaction with grid characteristics in a zero-emissions future. Previous research on LDES has explored cost sensitivities, parameter sensitivities (charging/discharging efficiencies), and the impact of LDES deployment on low- or zero-emission grids. However, these studies are often limited by modeling simplifications, such as excluding long-duration storage, multi-nodal transmission networks, and the diverse nature of renewable resources across different geographical regions. This research aims to fill these gaps by systematically analyzing how key grid characteristics affect LDES deployment and operation.

Prior studies on LDES often focused on cost sensitivity analyses or parameter sensitivity studies (e.g., charging/discharging efficiencies), aiming to understand the impact of LDES deployment on decarbonized electricity grids. However, many existing studies simplified their models by excluding long-duration storage, multi-nodal transmission networks, and diverse renewable resources. For example, Sisternes et al. (2016) studied the value of up to 30 GW of 2- or 10-h storage in reducing CO2 emissions in Texas, but didn’t model storage exceeding 10 hours, the transmission network, or diverse wind and solar resources. Dowling et al. (2020) explored a 100% renewable grid with LDES, finding that duration increased with more years of weather data; however, their model omitted transmission and averaged capacity factors across the USA. Guerra et al. (2020, 2021) are among the few studies to model transmission lines, but their approach involved separate capacity expansion and production cost modeling, preventing analysis of optimal LDES deployment and operation as a function of various grid factors. Sepulveda et al. (2021) systematically studied LDES design parameters but used simplified representations of New England and Texas. In contrast, this study addresses these limitations by using a high-resolution model to assess the impact of various grid characteristics on LDES deployment and operation.

This study uses the Switch capacity expansion model to simulate the Western Interconnect (WECC) in a 2050 zero-emissions future. The model incorporates high geographical resolution with over 8000 geolocated sites for candidate wind and solar deployment and 50 load zones connected by 126 aggregated transmission lines. The model optimizes both investment and dispatch decisions to minimize total system cost while meeting demand and considering transmission constraints. Dispatch decisions are made at 4-hour intervals, with a sensitivity analysis comparing this to a 1-hour resolution. Each load zone contains a candidate storage plant with fixed cost and performance parameters, allowing the model to optimize storage duration and size. The baseline scenario and 38 alternative scenarios were run, varying wind-vs-solar capacity shares, hydropower availability, transmission expansion costs, storage energy capacity costs, and storage mandates. The scenarios were carefully chosen to explore factors that would influence storage use or parameters that could unexpectedly change during the WECC's transition to a zero-emissions grid. A key aspect of the methodology is the model's ability to optimize both the capacity and duration of energy storage in each load zone, thus allowing the model to determine the optimal storage technology needed in various load zones under a range of possible future conditions.

The study's key findings reveal a strong interplay between LDES and various grid factors. In solar-dominant regions, 6-to-10-h duration storage is optimal, while wind-dominant regions benefit from 10-to-20-h storage. A 50% reduction in hydropower availability increases the average storage duration from 6.3 h to 23 h. Limiting transmission expansion necessitates 32% more storage energy capacity. The optimal storage duration varies widely based on the cost of storage energy capacity, ranging from 9 to 800 h (with a 75% reduction in transmission deployment). Analyzing LDES mandates, the study identifies 20 terawatt-hours (TWh) of installed storage energy capacity as the most valuable level, leading to 92% curtailment reduction, 10% reduction in total installed power capacity, 75% decrease in transmission deployment, and 70% reduction in peak-period electricity prices. In the baseline scenario, solar curtailment is high (up to 33% in a week) for much of the year, but drops to zero during peak demand in mid-summer and early winter. These peak months see significant increases in electricity prices. The southern WECC relies primarily on solar power and shorter-duration storage, while the northern WECC utilizes a mix of hydro, wind power, and longer-duration storage. The least-cost zero-emission plan highlights the importance of regional coordination but also potential reliability risks from regional dependencies during extreme weather events. This is followed by sensitivity runs based on varying the following parameters: * **Wind-vs-solar capacity shares:** LDES is more valuable in wind-dominant grids. * **Hydropower generation:** Decreases in hydropower significantly impact the need for LDES of longer durations. * **Transmission expansion:** Limited transmission expansion increases LDES value. * **Storage energy capacity costs:** Lower costs lead to longer duration storage and reduced reliance on wind and new transmission. Seasonal storage becomes cost-effective below 5$/kWh. * **LDES mandates:** A 20 TWh mandate offers the highest relative value, significantly decreasing curtailment, power capacity, transmission needs, and peak electricity prices.

The findings significantly advance our understanding of LDES's role in a zero-emissions future. The strong interaction between LDES and grid characteristics emphasizes the importance of considering regional context when planning LDES deployments. The significant cost reductions achieved through LDES mandates highlight the potential for policy interventions to optimize grid performance and enhance cost-effectiveness. The model’s demonstration that seasonal storage becomes cost-effective with lower storage costs indicates a promising path towards widespread LDES adoption. However, these cost reductions and lower price variability through LDES mandates will likely require market and tariff redesigns to ensure fair allocation of costs and benefits. The shift towards longer duration storage as hydropower availability decreases reveals the vulnerability of grid flexibility to climate change and the crucial need for resilient storage solutions. The observed reduction in transmission deployment through LDES mandates suggests a way to mitigate the challenges associated with transmission line expansion, while the lower marginal electricity prices during peak periods highlight the role of LDES in increasing grid reliability and affordability.

This study demonstrates the significant value of LDES in achieving a reliable and cost-effective zero-emissions grid, particularly when considering regional variations in resource mixes and grid characteristics. Policy interventions like LDES mandates can substantially improve grid performance, but require further market design considerations. Further research should incorporate more sophisticated demand models, examine the role of interconnection between regions, investigate the benefits of LDES in extreme weather scenarios, and consider the inclusion of other emerging technologies like carbon capture and sequestration.

The study's limitations include the 4-hour temporal resolution used in the primary simulations (though a sensitivity analysis with 1-hour resolution was conducted), the omission of interconnections with other grids, potential uncertainties in predicting future weather patterns and energy demand, the exclusion of energy reserve requirements and 'extreme' years in the simulations, and the absence of detailed modeling of carbon capture and sequestration technologies. These aspects could be explored in future research to enhance the model's accuracy and comprehensiveness.