The value of long-duration energy storage under various grid conditions in a zero-emissions future

Decarbonizing electricity is central to achieving global net-zero goals by 2050 as the power sector accounts for roughly a third of greenhouse gas emissions. While rapidly declining costs have expanded renewable deployment, achieving a fully net-zero and reliable grid remains challenging. Prior research has explored solutions including firm low-carbon resources, expanded transmission, and energy storage. Following emerging practice, the paper defines LDES as storage with duration of 10 hours or more (seasonal storage operates over months). The research gap addressed here is how grid characteristics—generation mix (solar vs wind dominance), hydropower availability, transmission expansion, and policy mandates—affect the optimal deployment, operation, and value of LDES in a zero-emissions system. The study aims to inform planners and policymakers on geographically appropriate and policy-relevant strategies for LDES under diverse future conditions in the Western Interconnect.

Existing LDES studies largely evaluate cost and performance sensitivities or technology portfolios but often simplify system representation (e.g., limited durations, single-region models, averaged resource data, omission of transmission, or separate capacity expansion and dispatch). For example, de Sisternes et al. analyzed storage for ERCOT emission reductions but did not model >10 h storage or transmission; Dowling et al. examined 100% renewables with LDES but used averaged national capacity factors and no transmission; Guerra et al. included transmission but treated LDES exogenously or with separate revenue models; Sepulveda et al. systematically explored LDES cost/efficiency design space with firm generation in stylized regions. In sum, prior work clarifies how LDES helps decarbonization given assumed grids, but less is known about how different grid compositions and constraints shape LDES value and deployment. This paper addresses that gap by co-optimizing investment and dispatch with multi-nodal transmission, broad storage durations, high geographic resolution of renewables, and policy-like constraints across 39 scenarios.

The study models the Western Electricity Coordinating Council (WECC) in 2050 under a zero-emissions constraint using the Switch capacity expansion model. The system includes 50 load zones connected by 126 aggregated transmission lines and represents 3580 existing plants (1010 expected to remain in 2050, primarily hydro and nuclear) and 4908 candidate plants (onshore/offshore wind, multiple solar technologies, biomass, geothermal). Renewable siting leverages high-resolution datasets and screening; candidate utility-scale PV and onshore wind are aggregated to one candidate per zone to manage complexity. Dispatch, investment, and transmission expansion are co-optimized to minimize total system cost with energy balance enforced at 4-hour intervals for all days in 2050 (supplementary sensitivity compares to a 1-hour baseline). Storage modeling: Each load zone has a technology-agnostic candidate storage asset with unconstrained power and energy capacities (optimized by the model). Baseline storage parameters: power capex 19.58 $/kW, energy capex 22.43 $/kWh, fixed O&M 6.096 $/kW-yr, variable O&M none, round-trip efficiency 85% (92% charge/92% discharge), equal charge/discharge rate, 15-year lifetime, no idle losses, scheduled outage rate 0.0055, forced outage rate 0.02. Costs reflect achieving DOE’s Energy Storage Grand Challenge (90% cost reduction by 2030 from NREL 2020 ATB lithium-ion 10-h baseline, with additional reductions to 2050 per ATB moderate). Demand uses the WECC-wide 2050 “Compliant” scenario (high efficiency, electrification, ZEV adoption) from Wei et al. No demand response, hydrogen production, EV charging management, or cross-sector coupling is modeled. Scenarios (39 total): Baseline plus sets A–E: (A) wind-vs-solar capacity share constraints (varying from 91% solar/9% wind to 40% solar/60% wind); (B) reduced hydropower availability by uniform monthly derates (15% to 100% reductions); (C) transmission expansion costs varied (10× increase; zero cost implying unconstrained expansion but still with losses); (D) storage energy capacity cost varied from 102 to 0.5 $/kWh (power cost unchanged); (E) LDES mandates imposing WECC-wide storage energy capacity totals from 2 to 64 TWh (baseline 1.94 TWh). Temporal resolution: All main runs use 4-hour intervals; a 1-hour sensitivity indicates the 4-hour model modestly underestimates capacities (<10% most technologies), but preserves key trends and storage durations. Notable modeling limitations include pumped hydro modeled as non-pumped hydro (unable to charge from grid), and residential/commercial PV treated as candidate capacity rather than exogenous adoption.

Baseline: Seasonal/geographic patterns emerge with high solar curtailment much of the year (weekly peaks up to 33%), and sharp marginal price spikes during high-demand periods (7-day average ~250 $/MWh in late July and ~410 $/MWh in late December). The southern WECC relies on solar plus 6–8 h storage; the northern WECC relies on hydro, wind, and transmission with more variable storage durations. Excluding Alberta—which builds about 300 GW of 18-h storage—the baseline system’s storage is 99% short-duration (<10 h). Wind vs solar share (Set A): 6–10 h storage power capacity scales roughly with solar capacity (e.g., 243 GW at 91% solar/9% wind vs 97 GW at 40% solar/60% wind). Wind-dominant grids substitute toward 10–20 h storage. Transmission expansion is lowest in a 70% solar/30% wind mix, though not least-cost overall. Hydropower reductions (Set B): A 50% hydro generation reduction raises WECC storage energy capacity by 65% and storage power by 21%, shifting average storage durations from 6.3 h to 23 h in hydro-dominant zones; hydropower availability significantly shapes LDES needs despite hydro being <15% of generation. Transmission costs (Set C): Raising expansion costs 10× modestly raises WECC storage energy (from 1.94 to 2.43 TWh), concentrated in Alberta (+474 GWh, +157%). Zero-cost expansion (no congestion) redistributes generation: WECC wind capacity drops by 54% (−53.9 GW), solar increases by 10% (+43.3 GW), and eight southwestern zones supply ~70% of WECC energy (vs 35% baseline) with little change in total WECC storage capacity. Storage energy cost (Set D): As energy cost falls from 102 to 0.5 $/kWh, optimal durations extend from ~9 h to ~825 h (34 days), with seasonal operation emerging below ~5 $/kWh. At 1 $/kWh, system storage energy reaches ~22 TWh, median duration ~127 h (5.3 days), wind capacity declines 17%, and new transmission drops 76%; at 0.5 $/kWh, storage energy ~36 TWh, median duration ~163 h, wind −30%, transmission −75% (system cost −21% vs baseline). LDES mandates (Set E): Increasing mandated WECC storage to ~20 TWh yields the largest marginal benefits: renewable curtailment falls 92% (118 GWh to 9.6 GWh), total installed power capacity declines ~10%, and transmission build falls ~74–75%. Storage operation shifts to bi-annual cycles beyond ~4 TWh and to yearly cycles beyond ~20 TWh. Price impacts: Storage mandates substantially reduce marginal price variability and peaks—e.g., with 20 TWh, 99% of marginal prices are below 130 $/MWh, and high-demand period prices fall by over 70%. Night prices remain higher than noon by 29–52% across scenarios but decline as storage increases; moving from 1.94 TWh to 64 TWh reduces average marginal prices ~22%. Regional prices remain lower in the solar-rich southern WECC.

The study shows that LDES value is highly context-dependent. Solar-dominant systems benefit primarily from 6–10 h storage to manage diurnal cycles, while wind-dominant systems benefit more from 10–20 h storage to smooth multi-day wind variability. Reductions in hydropower—a major zero-carbon flexibility source—substantially increase the need for longer-duration storage and larger capacities. Transmission and storage are partial substitutes: when expansion is constrained or costly, localized LDES becomes more valuable (notably in transmission-dependent zones); when expansion is unconstrained, generation and some storage concentrate in resource-rich regions without significantly increasing WECC-wide storage totals. Storage energy cost is pivotal: below roughly 5 $/kWh, seasonal and extra-long durations (20–400+ h) become cost-effective, displacing some wind and reducing transmission build. Policy-driven LDES mandates up to ~20 TWh deliver outsized system benefits by sharply reducing curtailment, total capacity needs, transmission build, and price volatility—especially during seasonal peaks—thereby enhancing reliability and affordability in a zero-emissions grid. These findings provide actionable guidance for planners on optimal storage durations by region, prioritization of hydropower-risk assessments, and evaluation of transmission-storage tradeoffs and LDES cost targets.

By co-optimizing investment, dispatch, and transmission in a high-resolution, zero-emissions WECC model, the paper quantifies how grid characteristics shape the deployment and value of technology-agnostic LDES. Key contributions include: identifying duration regimes suited to solar- vs wind-dominant systems (6–10 h vs 10–20 h), quantifying the strong sensitivity of LDES needs to hydropower availability, bounding storage-transmission substitution effects under constrained and unconstrained expansion, and demonstrating that energy capacity cost below ~5 $/kWh unlocks seasonal operation and very long durations. A WECC-wide LDES mandate of ~20 TWh provides the highest relative value, substantially lowering curtailment, total capacity, transmission build, and peak-period prices. Future research should: integrate hydrological and climate-informed hydro models; evaluate interconnect coordination; incorporate reserve requirements and extreme weather years; improve modeling of pumped hydro and distributed PV; explore demand-side management and cross-sectoral couplings; and assess market designs enabling seasonal arbitrage and cost recovery for LDES.

Modeling limitations include: no connections to other interconnects; uncertainty in 2050 demand and weather patterns; omission of operating reserves and extreme years; exclusion of CCS technologies (only biomass/geothermal as firm zero/low-carbon resources); 4-hour temporal resolution (bias toward underestimating capacity by <10% for most technologies vs a 1-hour sensitivity); aggregated wind/solar candidate projects; pumped hydro modeled like conventional hydro (cannot charge from grid); residential/commercial PV treated as candidate capacity rather than exogenous adoption; no demand response, hydrogen production, or EV charging management; and limited representation of market mechanisms for seasonal storage arbitrage. These constraints may affect quantitative estimates but are unlikely to overturn the qualitative trends.