energy · infrastructure · family: the solution exists but nobody can afford it
four hours of storage fora hundred hours of need
Grid-scale energy storage beyond 4 hours remains economically unviable
Problem statement
As variable renewable energy (wind and solar) grows beyond roughly 50% of annual grid generation — the level where ARPA-E's DAYS program overview notes that most high-penetration studies stop, and beyond which multi-day and even seasonal storage becomes necessary — the electric grid needs storage that can discharge for 10 to approximately 100 hours to cover multi-day periods of low wind or cloudy weather. New commercial storage installations today are predominantly lithium-ion batteries with durations of roughly 1–6 hours, and pumped-storage hydro (by far the largest installed storage resource, about 22 GW in the U.S., typically discharging for 6 to 10 hours) has seen few new plants in 25 years because of siting, permitting, and financing barriers. ARPA-E's target for this duration class is a levelized cost of storage (LCOS) of 5 cents/kWh-cycle held fixed across the full 10–100 hour range; lithium-ion's cost structure fundamentally cannot reach that goal at long durations, and while pumped hydro's cost structure can, siting limitations and the cost of financing very large projects keep it from filling the gap.
Why this matters
Without affordable long-duration storage, grids with high renewable penetration must maintain fossil-fuel backup capacity for reliability during extended low-generation periods ("Dunkelflaute" events). This undermines the emissions reduction potential of renewable buildouts and creates a structural barrier to decarbonization targets. DOE's Pathways to Commercial Liftoff analysis estimates the U.S. grid may need 225–460 GW of long-duration storage capacity for a net-zero economy by 2060, representing roughly $330 billion in cumulative capital — and finds that multi-day technologies must fall from today's $1,900–2,500 per kW (at ~45% round-trip efficiency) to about $1,100 per kW by 2030 to compete. Net-zero pathways that deploy long-duration storage save an estimated $10–20 billion per year by 2050 in operating costs and avoided capital expenditures compared to pathways that do not; without the cost decline, grids default to fossil backup.
What’s been tried and why it hasn’t worked
Lithium-ion batteries excel at short-duration applications but their energy-storing component is expensive (about $150/kWh even in ARPA-E's optimistic future case, against about $2/kWh for pumped hydro), so costs climb with duration and their cost advantage over pumped hydro ends at roughly 4 to 6 hours. Flow batteries (aqueous-electrolyte and hybrid designs, using reactants such as vanadium and iron salts) decouple power and energy, but the dilute solutions that flow through their stacks have low energy density, which raises tank, balance-of-plant, and installation costs. Compressed air energy storage (CAES) loses nearly half the energy put into it — DOE/PNNL's 2020 grid-storage assessment uses a round-trip efficiency of approximately 52% (literature range 50–70%) — and depends on suitable underground geological formations; liquid air energy storage (LAES) stores above ground instead, at higher upfront capital cost. Hydrogen-based storage (electrolysis → storage → fuel cell) has low round-trip efficiency (35% in the same DOE/PNNL assessment) and high capital costs for both the electrolyzer and fuel cell. Thermal storage concepts (molten salt, sand, concrete) are inexpensive per kWh of stored heat but converting back to electricity introduces thermodynamic losses. Each approach fails on a different dimension — cost, efficiency, siting flexibility, or durability — and no single technology has cracked the combination.
What would unlock progress
ARPA-E's DAYS program overview expects the majority of approaches able to hit its cost target to have fully decoupled power and energy components, built around very cheap bulk storage media — its candidate-media analysis spans nearly-free rock and water through hydrogen, ammonia, liquid air, molten salts, and iron-based flow reactants. ARPA-E expects chemical, electrochemical, thermal, and mechanical approaches all to be candidates, and lists among its approaches of interest systems whose storage capital cost falls as duration grows — for example, by transforming active materials during operation into a denser stored form than the power stack itself needs. Equally important is round-trip efficiency: at the program's assumed 2.5 cents/kWh input electricity, 33% is a hard floor below which nothing is left to pay for capital, and ARPA-E expects funded technologies to exceed 50%.
Entry points for student teams
A team could model the techno-economics of a specific long-duration storage concept (e.g., metal-anode batteries, gravity-based storage, sensible-heat thermal storage in molten salt, rock, or concrete) for a realistic grid scenario, identifying the cost and efficiency thresholds required for breakeven against natural gas peakers. Chemical engineering, electrical engineering, and systems modeling skills would be valuable. A second team, needing nothing beyond the public ARPA-E DAYS overview, could take the program's LCOS equation and assumptions (5 cents/kWh-cycle, 2.5 cents/kWh input electricity, 10% discount rate, 20-year life) and work out, for one storage medium in its Figure 5 list, whether the medium's capital cost falls in the roughly 5 to 20 $/kWh range the overview treats as of interest and what energy density would keep its containment cost under the roughly 20 $/kWh the overview gives for containment.
Genes
Structural cousins: stuck the same way, in other fields
Sources
ARPA-E DAYS (Duration Addition to electricitY Storage) program page, U.S. Department of Energy, ARPA-E, "Duration Addition to electricitY Storage (DAYS) Overview" (program overview document), U.S. Department of Energy (Sept 18, 2018), "Department of Energy Announces New Projects to Extend Grid Energy Storage," U.S. Department of Energy (March 2023), "Pathways to Commercial Liftoff: Long Duration Energy Storage," Kendall Mongird, Vilayanur Viswanathan, Jan Alam, Charlie Vartanian, Vincent Sprenkle (PNNL) & Richard Baxter (Mustang Prairie Energy), "2020 Grid Energy Storage Technology Cost and Performance Assessment," U.S. DOE Publication No. DOE/PA-0204, December 2020 (CAES and Hydrogen chapters), and Accessed 2026-08-21. go to source 1 ↗ go to source 2 ↗ go to source 3 ↗ go to source 4 ↗ go to source 5 ↗ go to source 6 ↗
Working notes
Notes made while writing and checking this brief, shown as written. They use working shorthand.
Related to energy-grid-battery-scale-failure-prediction (battery failure at grid scale) and energy-grid-inertia-loss-frequency-instability (grid stability challenges from renewables). The DAYS program awarded $28 million to 10 projects (DOE announcement, Sept 18, 2018) spanning electrochemical, thermal, and mechanical approaches.
Reconciliation 2026-08-21: This brief carried quantitative claims well beyond its sole cited source (the ARPA-E DAYS program page), and several drifted. The core ARPA-E framing verified clean against the DAYS program page (meta description confirms the program name and the 10-to-approximately-100-hour duration range) and the full "Duration Addition to electricitY Storage (DAYS) Overview" PDF — but the cost target was misstated: it is an LCOS of 5 cents/kWh-cycle held fixed across all durations (with an assumed 2.5 cents/kWh input electricity price), not a generic "below 5¢/kWh" competitiveness threshold, and the claim "no technology exists that can provide 10–100 hour storage at that cost" contradicted the overview itself, which shows pumped-storage hydro's cost structure "meets and even exceeds the goal across the full range of durations" while siting/financing has all but halted new PSH (~22 GW installed, few new plants in 25 years). "Nearly all deployed grid storage today is lithium-ion designed for 2–4 hours" was also off — the overview says PSH is by far the largest installed storage and that new commercial installations are predominantly Li-ion at 1–6 hour durations; rewritten to match. The renewable-penetration trigger ("50–60%") was tightened to the overview's actual statement (most high-penetration studies consider up to ~50%; higher requires multi-day storage). The unsourced "100–400 GW by 2050 / trillions of dollars" projection was replaced with DOE's Pathways to Commercial Liftoff: Long Duration Energy Storage (March 2023) figures, confirmed in the report's executive summary: 225–460 GW of LDES for a net-zero economy by 2060, ~$330B cumulative capital, $10–20B/yr savings by 2050, and multi-day cost/RTE milestones ($1,900–2,500/kW and 45% RTE today → ~$1,100/kW and 55–60% RTE by 2030). CAES/hydrogen efficiency figures were re-anchored to the PNNL/DOE 2020 Grid Energy Storage Technology Cost and Performance Assessment (Mongird et al., DOE/PA-0204): CAES RTE ~52% assumed (literature range 50–70%), hydrogen bidirectional RTE 35% — the old "CAES/LAES losses of 40–50%" and "H₂ ~30–35%" were close but uncited. The DAYS media list ("water, sulfur, iron, sand, cement") did not match the overview's Figure 5A candidates (water, rock, hydrogen, ammonia, liquid air, molten salts, silicon, refractory brick, FeCl₂); corrected. One Source Notes correction of record: DAYS awarded $28M to 10 projects (DOE announcement, Sept 18, 2018) — the $30M figure above was the FOA's announced up-to amount, not the awarded total. All Source-line URLs fetched and read 2026-08-21.
Reconciliation 2026-09-30: Fact repair after reading the whole write-up against every source it names. Fetched 2026-09-30: the ARPA-E DAYS program page (https://arpa-e.energy.gov/programs-and-initiatives/view-all-programs/days; only its title and description text load without a browser, so it was read in part); the DAYS Overview PDF (https://arpa-e.energy.gov/sites/default/files/migrated/documents/files/DAYS_ProgramOverview_FINAL.pdf, full text plus Figures 3 and 5 read as images); the DOE announcement of Sept 18, 2018 (https://www.energy.gov/articles/department-energy-announces-new-projects-extend-grid-energy-storage, full text); the DOE Pathways to Commercial Liftoff report (the formenergy.com address now refuses automated requests, so the Internet Archive's copy of that same file was read in full, 80 pages, at web.archive.org/web/20240716061522id_/formenergy.com/wp-content/uploads/2023/06/doe-pathways-to-commercial-liftoff-report-long-duration-storage95.pdf); and the PNNL CAES and Hydrogen chapters (https://www.pnnl.gov/sites/default/files/media/file/CAES_Methodology.pdf and https://www.pnnl.gov/sites/default/files/media/file/Hydrogen_Methodology.pdf, full text). (1) The text said "nearly all" new commercial installations are lithium-ion; the DAYS overview says new commercial installations are "predominantly Li-ion"; the sentence now says predominantly. (2) Pumped hydro was called "the largest proven long-duration resource"; the overview calls it "by far the largest source of electricity storage on the grid today" with a typical discharge of 6 to 10 hours, which is below the program's 10-hour floor; the parenthesis now says largest installed storage resource, 6 to 10 hours. (3) The text said pumped hydro's "geographic constraints" keep it from filling the gap; the overview gives two reasons, siting limitations and the uncertainty and cost of financing very large projects; the sentence now gives both. (4) The text said lithium-ion costs "scale linearly with duration" and become uneconomical "beyond ~4 hours". The overview does not say linearly; it says Li-ion's installed energy cost is high (about $150/kWh assumed for future Li-ion against about $2/kWh for pumped hydro) and that Li-ion costs less than pumped hydro up to about 4 to 6 hours. The sentence now says that. (5) The flow-battery sentence named vanadium redox and zinc-bromine and blamed "electrolyte degradation" and "high balance-of-system costs". Zinc-bromine and electrolyte degradation are in none of the sources. The overview says the dilute solutions flowing through the stacks have low energy density and that raising it would cut tank, balance-of-plant, and installation costs; the Liftoff report lists aqueous-electrolyte and hybrid flow batteries and names vanadium as a raw material. The sentence now says that. (6) The text said CAES, "like liquid air energy storage (LAES), requires large physical plant". The Liftoff report says CAES needs underground geological formations and has a footprint significantly smaller than other mechanical storage, and that LAES stores above ground, with a smaller footprint still but higher upfront capital cost. The sentence now follows the report. (7) Hydrogen's round-trip efficiency was called "extremely low"; the PNNL chapter says "low RTE"; now low. (8) The "What Would Unlock Progress" sentence on "reversible thermochemical reactions, low-cost electromechanical systems, or novel electrochemistry using non-critical minerals" and the claim that power conversion "is where most systems lose economic viability" are in none of the sources. They are replaced with what the overview says: chemical, electrochemical, thermal, and mechanical approaches are all in scope; systems whose storage capital cost falls with duration are listed among the approaches of interest; and at 2.5 cents/kWh input electricity, 33% round-trip efficiency is a hard floor and the program expects more than 50%. (9) The entry-point examples "iron-air batteries" and "sulfur-based thermal storage" are in none of the sources; they are replaced by the Liftoff report's own categories (metal-anode batteries; sensible heat in molten salt, rock, or concrete). Gravity-based storage and natural gas peakers are confirmed there. A second starting point built only from the DAYS overview's LCOS assumptions and Figure 5 (media capital cost of roughly 5 to 20 $/kWh and containment cost up to roughly 20 $/kWh as the ranges of interest) was added so that the section offers two doors. (10) Source Notes said DAYS "awarded $30M"; the DOE announcement says $28 million to 10 projects, and no source gives $30 million; corrected. (11) The sentence on Form Energy's iron-air battery was removed: no source names Form Energy or iron-air (the Liftoff report was hosted on Form Energy's site, which is not a statement about the company). Confirmed against the sources and left as written: the 10 to approximately 100 hour range; the 5 cents/kWh-cycle LCOS held fixed across durations; "most studies of high wind and solar penetrations consider levels up to about 50%" and that higher penetration requires multi-day and in extreme cases seasonal storage; about 22 GW of pumped hydro and few new installations in 25 years; 1 to 6 hour durations for current commercial installations (Figure 3); 225–460 GW by 2060, $330B cumulative capital, $10–20B annualized savings by 2050, and the $1,900–2,500/kW and 45% RTE today to $1,100/kW by 2030 milestones (Liftoff report, executive summary); CAES round-trip efficiency about 52% with a literature range of 50% to 70%, and hydrogen 35%, in the PNNL chapters; the decoupled power-and-energy expectation and the Figure 5 media list (water and rock through hydrogen, ammonia, liquid air, molten salts, and FeCl2); thermal media of molten salt, sand, and concrete (Liftoff report; DOE announcement); and every author, title, report number, and date on the Source line. Not in any source and left as a general term: the word "Dunkelflaute". Titles unchanged. Genes untouched. A second reader went over these repairs the same day against the same six sources, fetched independently; all eleven stand, with two rewordings: the "What Would Unlock Progress" sentence now says the overview lists declining-cost-with-duration systems among its approaches of interest rather than that it "singles out" them, and the second entry point now keeps the overview's media budget (roughly 5 to 20 $/kWh) and containment budget (up to roughly 20 $/kWh) separate instead of folding the whole energy side into the media figure.