Energy Storage Systems — Family Index

1. At a glance

Grid-scale and behind-the-meter energy storage is the missing piece for high-renewable grids: it shifts solar from noon to evening, firms wind through low-pressure days, and stabilizes frequency on milliseconds. By 2026 there is no single dominant technology — the storage stack is segmented by duration and service, with different chemistries and physics serving wildly different roles.

Three functional roles:

• Power-class (milliseconds to minutes discharge): frequency regulation, grid inertia, ride-through, UPS. Sized in MW, not MWh. Flywheels, supercapacitors, and Li-ion fall here.
• Energy-class (hours to days): daily arbitrage, peak shift, peaking capacity replacement, renewable smoothing. Sized in MWh. Li-ion BESS dominates 1–4 hr; flow batteries, CAES, and LAES are growing for 4–24 hr.
• Long-duration / seasonal (weeks to months): firming through low-renewable seasons, capacity for multi-day weather events. Pumped hydro, hydrogen, iron-air, and thermal storage are the candidates.

A grid storage portfolio in 2026 is almost always a mix: Li-ion for fast response and 4-hr arbitrage, paired with a long-duration technology (LDES) for tail-end firming and a small flywheel or supercap bank for power-quality.

2. Power vs energy — the foundational split

The first design decision is whether the system delivers high power for short bursts or high energy for long durations. They are physically different machines.

Power-class systems (high P, low E):

• Flywheel — mechanical kinetic energy, ms response, 90%+ efficient, but 1–15 min duration.
• Supercapacitor (EDLC) — electrostatic, 100 000+ cycles, seconds duration.
• Li-ion BESS sized for power — 0.25 C to 1 C discharge, ms response. Sized small (MW » MWh ratio high).

Energy-class systems (high E, lower P):

• Li-ion BESS sized for energy — 0.25 C or slower, 2–4 hr discharge.
• Vanadium and other flow batteries — decoupled stack (power) and tank (energy); add tanks to extend duration.
• CAES, LAES, thermal — energy is in compressed gas, liquid air, or hot solids.
• Pumped hydro — energy is in elevated water mass.

Long-duration storage (>10 hr; LDES):

• Pumped hydro at the largest end (multi-day).
• Hydrogen + ammonia for seasonal.
• Iron-air (Form Energy) for 100-hr.
• Thermal storage (molten salt, sand, carbon block) for 10–100 hr.
• Gravity (Energy Vault, Gravitricity) for 4–24 hr.

The same chemistry (Li-ion) appears in both columns because cell selection, C-rate sizing, thermal management, and inverter (PCS) ratio can tune a Li-ion plant for either role — but you cannot economically use a flywheel as energy storage or a salt cavern as fast frequency regulation.

3. Pumped Hydro Storage (PHS)

PHS dominates global storage by capacity — roughly 95 % of all installed grid storage in 2024, ~190 GW worldwide. It is the only technology with a proven multi-decade operating record at scale.

Principle. Two reservoirs at different elevations. Off-peak: pump water from lower to upper using cheap electricity. On-peak: release through reversible Francis turbines (or pump-turbine pairs) back to lower reservoir, generating power. Round-trip efficiency 75–85 %.

Economics. Capex roughly $1500–3000/kW+$50–150/kWh — high upfront, but lifetime 50+ years and very low O&M make it cheap per MWh-delivered.

Anchor sites (2024):

• Bath County, Virginia, USA — 3 003 MW, the world’s largest in operation since 1985.
• Fengning, Hebei, China — 3 600 MW, fully commissioned 2024, world’s largest.
• Okutataragi, Japan — 1 932 MW.
• Snowy 2.0, Australia — 2 200 MW (2 GW), commercial operation date (COD) targeted 2027 after multiple delays, will link Tantangara and Talbingo reservoirs.
• Coire Glas, Scotland (UK) — 1 500 MW, SSE Renewables, in development.

Variants:

• Open-loop — upper reservoir is on a flowing river; environmental impact is high (fish, silt, flow regimes).
• Closed-loop — both reservoirs are isolated, often man-made; lower riverine impact and faster permitting in the US under FERC’s 2022 closed-loop pathway.
• Seawater pumped hydro — Okinawa Yanbaru (Japan, decommissioned 2016) was the first; high corrosion challenges.
• Underground / abandoned mine — proposals in flooded coal pits (e.g., Bełchatów); RheEnergise (UK) uses high-density slurry to shrink elevation needs.

OEMs. Andritz, Voith, GE Hydro Solutions (now part of GE Vernova), Toshiba ESS — all build variable-speed ternary or doubly-fed induction pump-turbines that can provide grid services while pumping, not just while generating. Variable-speed PHS is the modern standard for new builds.

4. Battery Energy Storage Systems (BESS) — Li-ion

See also battery-chemistries for the cell chemistry side.

LFP (lithium iron phosphate) has displaced NMC for grid storage by 2024–26 — lower cost ($/kWh), higher cycle life (6 000–10 000 deep cycles), better thermal stability, and the only chemistry that passes UL 9540A propagation tests without exotic enclosures. NMC retains a niche in space-constrained behind-the-meter installs.

Container formats (2025–26):

• Tesla Megapack 2 XL — 3.9 MWh per unit AC-coupled, integrated PCS + liquid cooling, single 20-ft footprint.
• CATL EnerC+ — 5 MWh container, LFP cells, liquid cooled.
• BYD MC Cube ESS — 6.4 MWh container, blade LFP cells, CTP (cell-to-pack) architecture.
• Sungrow PowerTitan 2.0 — 5 MWh container, string PCS architecture.
• Fluence Gridstack Pro — 5 MWh, modular.

Inside a 20-ft container: LFP modules in racks, liquid cooling loops to a chiller, BMS + EMS, AC/DC PCS (often integrated in the same enclosure), fire suppression (Stat-X aerosol or Novec replacement), gas/smoke detection, deflagration vents per NFPA 68.

Cost (2025). Module-level LFP cells $80–130/kWh\ast \ast ;fullsystemAC-installed\ast \ast$200–330/kWh depending on duration (4-hr systems are cheaper per kWh than 2-hr because PCS amortizes over more energy). Levelized cost of storage (LCOS) for 4-hr Li-ion 2025: $115–180/MWh.

Anchor projects:

• Edwards & Sanborn (Kern County, California) — 875 MW / 3.3 GWh, online August 2024, world’s largest battery; LS Power developer; LG Energy + BYD cells.
• Moss Landing (California) — Vistra + LG Energy phases total 750 MW / 3 GWh by 2024; suffered fire September 2022 and again January 2025 (Vistra Phase I), driving NFPA 855 + UL 9540A revisions.
• Crimson Storage (California) — 350 MW / 1.4 GWh, Recurrent + Axium.
• Hornsdale Power Reserve (South Australia) — original 100 MW/129 MWh Tesla site from 2017, expanded to 150 MW; pioneered frequency-control ancillary services (FCAS) revenue stacking.
• Saticoy / Oxnard (California) — 100 MW / 400 MWh, Strata Clean Energy.

5. Sodium-ion grid storage

Sodium-ion (Na-ion) emerged commercially 2023–25 as a credible cheaper, more abundant alternative to LFP for stationary storage. Energy density is lower (~120–160 Wh/kg vs 160–200 for LFP), but grid storage doesn’t care about gravimetric density.

Manufacturers (2025):

• CATL — Gen-1 sodium-ion launched 2023, Gen-2 announced 2024 with hybrid Na-ion + Li-ion packs for EVs; grid storage variants in pilot.
• BYD — Na-ion BESS pilots in China 2024–25; first commercial 2025.
• HiNa Battery (China) — partnered with Sehol/JAC on Na-ion EV; grid storage products.
• Faradion (UK, owned by Reliance Industries since 2022) — supplies Na-ion to Reliance’s Gujarat gigafactory.
• Northvolt — announced Na-ion product 2023 prior to 2024 financial troubles.
• Natron Energy (US) — Prussian-blue-electrode Na-ion, focus on data-center UPS and short-duration grid; opened Holland, Michigan plant 2024.

Cost target: <$60/kWh at cell level. Cycle life: 3 000–5 000 in early product, targeting LFP parity. Cold-weather performance is markedly better than Li-ion.

6. Vanadium Redox Flow Battery (VRFB)

The most mature flow battery. Energy stored in two tanks of vanadium sulfate electrolyte at different oxidation states (V²⁺/V³⁺ and V⁴⁺/V⁵⁺); cell stack pumps electrolyte across an ion-exchange membrane to charge/discharge.

Key property: power (stack size) and energy (tank size) are decoupled. Want more hours of duration? Buy more electrolyte; the same stack discharges longer. This makes VRFB economic at 8+ hr durations where Li-ion’s per-kWh cost dominates.

Performance: 20 000+ cycles with no capacity fade (electrolyte does not degrade — the vanadium ions are not consumed). Round-trip efficiency 65–80 % (parasitic pumping losses). Calendar life > 20 years on the electrolyte; stack membrane is the wear item, ~10 yr.

Capex: $400–700/kWh at scale (2025), with electrolyte responsible for a large fraction — vanadium pentoxide commodity price drives it.

Manufacturers + anchor projects:

• Rongke Power (Dalian, China) — 200 MW / 800 MWh Dalian Flow Battery Energy Storage Peak Shaving Station, world’s largest VRFB, fully commissioned 2024.
• Sumitomo Electric (Japan) — Hokkaido and Yokohama projects; supplied 60 MWh to Hokkaido Electric.
• Invinity Energy Systems (UK) — VS3 modular product; deployed at Energy Superhub Oxford and elsewhere.
• Pu Neng (China, formerly Prudent Energy) — large pipeline.
• CellCube (Austria) — modular flow systems for commercial-industrial.
• VFlowTech (Singapore) — distributed flow systems.

Vanadium electrolyte can be leased rather than purchased (Bushveld Energy / Largo Inc. models), shifting capex to opex and aligning with project finance.

7. Iron flow and iron-air

Iron flow (ESS Inc., Oregon) — uses iron chloride / iron metal redox couple. Earth-abundant, non-flammable, non-toxic. Duration sweet spot 5–12 hr. Round-trip efficiency 65–75 %. Cycle life 25 000+. ESS Inc.’s Energy Warehouse (75 kW / 500 kWh) and Energy Center (MW-scale) products are deployed at Stoke-on-Trent (UK), Sacramento Municipal Utility District, and Portland General Electric.

Iron-air (Form Energy, Massachusetts) — uses reversible iron rust (Fe ↔ Fe₂O₃·H₂O) in an aqueous electrolyte. Discharges over 100 hr (“multi-day storage”) at low round-trip efficiency (~40–50 %) but with a target capex of ~$20/kWh — an order of magnitude below Li-ion. Form’s Weirton, West Virginia factory began production 2024.

Form Energy anchor projects (2024–25):

• Big Rivers Electric (Kentucky) — 10 MW / 1 GWh, contract 2023, construction 2024–25.
• Xcel Energy (Minnesota + Colorado) — multiple 10 MW pilots.
• Georgia Power — 15 MW iron-air at the Robins Air Force Base resilience project, 2024 award.
• Great River Energy (Minnesota Cambridge site) — original 1.5 MW pilot, retired in favor of the larger commercial deployment 2024.
• National Grid (Massachusetts) — pilot tranche.

8. Zinc-based flow and hybrid

Zinc-bromine (Zn-Br) flow — Redflow (Australia, in administration 2024 then reorganized), Primus Power (US). Hybrid flow: zinc plates onto the negative electrode during charge (capacity-limited by plate thickness, unlike VRFB). 8–12 hr duration; flammability low.

Zinc-air rechargeable — NantEnergy (formerly Fluidic Energy) and Eos Energy Enterprises (Zynth product line, znyc cathode in aqueous electrolyte). Eos Z3 battery: 3–12 hr, $40–50/kWh module target. Stryten Energy acquired some NantEnergy assets in 2024.

Zinc-ion (non-flow) — Salient Energy and others; early commercial 2025.

9. Supercapacitors / EDLC

Electric double-layer capacitors store charge electrostatically at the electrode-electrolyte interface (no faradaic reaction). Energy density 5–10 Wh/kg, power density 5–10 kW/kg, >500 000 cycles, –40 to +65 °C operating range.

Manufacturers: Skeleton Technologies (Germany/Estonia, graphene curved electrodes), Maxwell Technologies (acquired by Tesla 2019), Eaton, Nesscap (acquired by Maxwell), Yunasko, CAP-XX.

Uses: UPS bridging (15 s–2 min before genset start), wind-turbine pitch control, regen braking on trams + cranes, voltage ride-through, power-quality (sag/swell correction), and hybrid with Li-ion to handle peak power on the supercap and energy on the battery.

Lithium-ion capacitors (LIC) — hybrid: pre-lithiated graphite negative electrode + EDLC positive; 10–20 Wh/kg; bridges the EDLC–Li-ion gap.

10. Flywheel energy storage (FESS)

A high-strength rotor spinning in a vacuum chamber on magnetic bearings stores kinetic energy E = ½ I ω². Carbon-fiber composite (CFRP) rotors push tip speeds to 1000 m/s for energy density 100+ Wh/kg.

Anchor deployments:

• Beacon Power Stephentown, New York — 20 MW frequency regulation, online 2011, oldest large-scale FESS still operating.
• Beacon Power Hazle Township, Pennsylvania — 20 MW.
• Amber Kinetics — M32 (32 kWh/8 kW, 4 hr) steel rotor flywheels deployed in Hawaii and the Philippines for 4-hr energy storage, an unusual long-duration FESS approach.
• Active Power (now part of Piller Group) — CleanSource flywheel UPS, 15 s ride-through for data centers.
• Calnetix VYCON (now part of Calnetix) — VDC flywheel UPS.
• Temporal Power (Ontario, closed 2018) — early grid-scale provider.

90 %+ round-trip efficiency, ms response, but self-discharge ~1 %/hr through bearing and windage losses confines them to sub-15-min duration. Best matched with frequency regulation, microgrid stabilization, and rail regen braking.

11. Compressed Air Energy Storage (CAES)

Charge: compress air into a salt cavern, hard-rock cavern, or pressure vessel. Discharge: release through a turbine, mixed with fuel (diabatic) or with stored heat (adiabatic).

Diabatic CAES — compression heat is dumped to atmosphere; expansion requires combustion of natural gas to prevent freezing. Round-trip ~50 % electrical efficiency.

• Huntorf, Germany — 290 MW, online 1978, 2 salt caverns at 600 m depth, the world’s first commercial CAES.
• McIntosh, Alabama, USA — 110 MW, 2 600 MWh, online 1991, Alabama Electric Cooperative, salt cavern.

Adiabatic CAES (A-CAES) — compression heat captured and stored (in thermal oil, molten salt, or packed bed) and returned during expansion, eliminating gas combustion. Round-trip ~70 %.

• Hydrostor (Canada) — Goderich, Ontario pilot (1.75 MW, online 2019, decommissioned 2024 after data collection); Willow Rock (Kern County, California) 500 MW / 4 000 MWh A-CAES under permitting; Silver City (Broken Hill, NSW Australia) 200 MW / 1 600 MWh under construction 2025.
• Storelectric (UK) — concept developer, no commercial site yet.
• CHEYESS (China) — Zhangjiakou 100 MW A-CAES online 2022; Shandong Feicheng 300 MW under construction 2025.

Capex $1000–3000/kW; siting limited by salt or hard-rock cavern geology.

12. Liquid Air Energy Storage (LAES)

Charge: compress and cool ambient air through a Linde / Claude liquefaction cycle to −196 °C, store as liquid in insulated tanks at near-atmospheric pressure. Discharge: pump liquid, re-gasify through ambient + waste heat, expand through turbine.

Highview Power (UK) — CRYOBattery product. Pilot (5 MW / 15 MWh) at Pilsworth, Manchester since 2018. Carrington, Trafford (Manchester) 50 MW / 300 MWh plant under construction 2024–26, financed in 2022, COD targeted 2026. Highview’s Hunterston, Scotland site (50 MW / 400 MWh) is also in development.

Round-trip efficiency 60–70 % (without external heat input — higher when paired with industrial waste heat or LNG regasification, which contributes cold). Footprint is small per MWh, the technology is siting-flexible (no salt caverns or topography required), and storage media (air, insulated tank) is cheap and abundant.

13. Thermal energy storage

Store energy as heat (or cold) and discharge as heat — directly into a thermal load — or back to electricity through a Rankine cycle.

Molten salt

The mature CSP (concentrated solar power) storage medium. Solar Salt is 60 % NaNO₃ + 40 % KNO₃, liquid 220–600 °C. Two-tank arrangement: cold tank (290 °C) → solar receiver → hot tank (565 °C) → steam generator → cold tank.

• Crescent Dunes (Nevada) — 110 MW + 10 hr storage, online 2015, ~6 000 MWh thermal.
• Solana Generating Station (Arizona) — 280 MW + 6 hr storage.
• Noor Ouarzazate (Morocco) — 580 MW total CSP across phases, Noor III with 7.5 hr molten salt.
• Cerro Dominador (Chile) — 110 MW + 17.5 hr storage, online 2021.

Capex $30–50/kWh_th (orders of magnitude below Li-ion per kWh stored), but you’re storing thermal energy with a Rankine round-trip ~35 %.

Sensible thermal

• Water tank (chilled or hot) — ice storage and chilled water are standard HVAC peak-shift in commercial buildings.
• Packed bed of rocks or ceramic pellets — high-temp air or gas flows through, charges/discharges by temperature gradient. SiestaSys, EnergyNest, Brenmiller bGen, Echogen.
• Concrete — Siemens Gamesa ETES (Hamburg pilot, 130 MWh).

Latent (PCM — phase change materials)

Store at the phase transition (solid ↔ liquid) — paraffin (50–80 °C), salt hydrates (Na₂SO₄·10H₂O at 32 °C), salt eutectics (KNO₃/NaNO₃ at 220 °C). Energy density 100–250 kJ/kg at constant temperature.

Sand battery

Polar Night Energy (Finland) — first commercial unit at Kankaanpää (8 MWh, online 2022), now scaled to Pornainen 100 MWh sand silo at 500–600 °C, online March 2025, supplying district heating to displace oil and woodchip. The hot sand is heated resistively by surplus wind power; output is heat to a heat exchanger.

Thermo-photovoltaic discharge

Antora Energy (California) — solid carbon block heated resistively to >2 400 °C, discharges via thermo-photovoltaic (TPV) cells that convert blackbody radiation back to electricity at ~40 % efficiency. First commercial unit announced 2023, demo at the Antora San Jose facility 2024, customer pilot 2025.

Rondo Energy (California) — brick-based thermal battery (Rondo Heat Battery) for industrial process heat (300–1500 °C), online 2023, deployed at Calgren ethanol plant and Diageo.

14. Gravity / mechanical storage (excluding PHS)

• Energy Vault (Switzerland/US) — gravity tower of composite blocks lifted by a multi-cable crane. The first commercial 25 MW / 100 MWh unit at Rudong, China online 2023; pivoting more toward hybrid (battery + gravity) and software (energy management) by 2025.
• Gravitricity (UK) — heavy weight (500–5 000 t) suspended in a deep mineshaft; fast response (~1 s). Pilot at Edinburgh 2021; first commercial mine site under permitting (Czechia 2024 pilot).
• ARES (Advanced Rail Energy Storage) — now part of Quidnet (Texas) — original concept: rail cars on incline; Quidnet has pivoted to “geomechanical pumped storage” (water pumped into a sealed underground rock fracture acts like a spring).
• Subsea pumped hydro — Subhydro / Ocean Battery (StEnSea project Germany) — concrete spheres on deep seabed; pump water out (charge) against ocean pressure, let water in through turbine (discharge). 0.5 MW pilot off California 2023, scaling for North Sea by 2026.

15. Hydrogen + ammonia (power-to-X-to-power)

See hydrogen-fuel-cells for fuel-cell side.

Round trip: electrolysis (60–75 % efficient) → compression/liquefaction/conversion to NH₃ (additional 10–20 % parasitic) → storage (salt cavern, pressure vessel, cryogenic) → reconversion via fuel cell or H₂ turbine (50–60 %). Total ~30–40 % round-trip electrical.

Economically a poor short-duration store but the only credible seasonal storage at TWh scale; the molecule has secondary value (industrial feedstock, fuel) that subsidizes the round trip.

ACES Delta (Utah, US) — Mitsubishi Power + Magnum Development — 220 MW electrolyzer + 2 salt caverns at the Intermountain Power Project replacement (a hydrogen-capable combined-cycle), COD 2025.

HyStock (Netherlands) — Gasunie salt cavern hydrogen storage, pilot 2024.

16. CO₂ battery

Energy Dome (Italy) — CO₂ stored in a low-pressure flexible reservoir (“dome”) at ambient. Charge: compress CO₂ to a liquid (using surplus electricity to drive compressors), recover compression heat. Discharge: evaporate liquid CO₂ using stored heat, expand through turbine. Round-trip 75 % claimed.

• Ottana, Sardinia, Italy — 2.5 MW / 4 MWh demonstration online 2022.
• Sardinia commercial unit — 20 MW / 200 MWh under construction with A2A as offtaker, COD 2024.
• Columbia Energy Storage (US Midwest) — 25 MW with Alliant Energy.
• NTPC (India) — first international license 2024.

17. Round-trip efficiency comparison (2026)

Technology	RTE (electric in → electric out)
Flywheel	85–95 %
Li-ion (LFP/NMC)	85–92 %
Supercapacitor	85–95 %
Lead-acid	75–85 %
Pumped hydro (PHS)	75–85 %
Vanadium flow (VRFB)	65–80 %
LAES	60–70 %
A-CAES (adiabatic)	~70 %
Diabatic CAES	~50 % (electrical, ignoring gas heat)
Iron-air	40–50 %
Hydrogen + fuel cell	30–40 %
Thermal (resistive in → Rankine out)	30–45 %
Thermal (resistive in → heat out)	70–90 % (no Rankine penalty)

18. Duration economics (2026)

Duration	Dominant / emerging technologies
<1 min	Supercap, flywheel
1–15 min	Flywheel, Li-ion (high-C)
15 min – 4 hr	Li-ion LFP BESS (dominant)
4–10 hr	Li-ion (cheaper LFP), early flow batteries, A-CAES, LAES, thermal
10–100 hr	Iron-air (Form), VRFB, thermal, A-CAES
100 hr – days	Iron-air, thermal (large), PHS
Weeks – seasonal	Hydrogen, ammonia, PHS reservoirs (limited sites)

19. Grid services storage provides

• Frequency regulation (response ms–sec, duration min) — flywheel + Li-ion + supercap.
• Spinning reserve / contingency reserve (10 min, called rarely) — Li-ion.
• Energy arbitrage (charge cheap, discharge expensive) — Li-ion, flow, PHS.
• Peaking capacity (replace gas peakers, 4–6 hr) — Li-ion BESS, PHS.
• Renewable smoothing / ramp control — Li-ion + flywheel hybrid.
• Black-start (restart a dead grid) — PHS, large Li-ion BESS with grid-forming inverters.
• Capacity firming / dispatchable renewables — Li-ion 4-hr paired with solar/wind PPA.
• Transmission and distribution deferral — site storage at congested nodes to delay line upgrades.
• Demand-charge management (commercial / industrial) — behind-the-meter Li-ion.
• Virtual power plant (VPP) — aggregate distributed batteries (Tesla Powerwall, Sunrun, sonnen) for wholesale market participation.
• Synthetic inertia + grid-forming — BESS with grid-forming inverters (Tesla VPP Hawaii, Hornsdale upgrade) provide synthetic inertia where rotating generators were retired.

20. Standards + safety

• UL 9540 — Energy Storage Systems and Equipment (the listing standard for the integrated system).
• UL 9540A — Test Method for Evaluating Thermal Runaway Fire Propagation in Battery Energy Storage Systems — required by NFPA 855 for cell, module, unit, and installation-level testing; tightened after 2022 + 2025 Moss Landing fires.
• NFPA 855 — Standard for the Installation of Stationary Energy Storage Systems (2023 edition adopted in most US AHJs) — covers spacing, ventilation, deflagration, fire detection/suppression.
• NFPA 68 / 69 — explosion protection (deflagration vents, inerting).
• IEC 62933 series — Electrical Energy Storage (EES) Systems (general, terminology, testing, planning, safety).
• IEEE 1547 — Interconnection of Distributed Energy Resources with the grid.
• IEEE 2030.2 — Smart grid interoperability for storage.
• NETA ATS — acceptance testing.
• OSHA, NEC Article 706 — electrical installation.

21. Cost trajectories (2026)

• Li-ion BESS: module ~$80–130/kWh;fullsystemAC$200–330/kWh installed for 4-hr; LCOS $115–180/MWh. Down ~80 % over the past decade.
• VRFB: $400–700/kWh, falling slowly; vanadium price-elastic.
• Iron-air (Form): target ~$20–30/kWh at scale for 100-hr; commercial deployments at higher initial cost.
• PHS: $1500–3000/kW+$50–150/kWh; stable, sunk-cost long-life amortization makes it cheap per MWh over 50 yr.
• CAES (A-CAES): $1000–3000/kW; site-dependent.
• LAES: $600–1000/kWh of energy capacity, falling.
• Thermal (sensible / molten salt): $20–50/kWh_th — but Rankine round-trip cuts effective electrical capex by ~3×.
• Hydrogen LCOS: $200–500/MWh for round-trip storage 2026; competitive only when the H₂ molecule has a secondary use (industrial feedstock, transport fuel, ammonia export).

22. Selection heuristics

• <15 min frequency regulation, ride-through → flywheel + supercap (or Li-ion sized for power).
• <4 hr daily arbitrage / peak shift → Li-ion LFP BESS (default).
• 4–12 hr arbitrage → Li-ion LFP + emerging flow (VRFB, iron flow); evaluate LAES/CAES if siting permits.
• >12 hr long-duration → iron-air (Form), VRFB, A-CAES, or thermal (sand, molten salt, brick).
• Frequency regulation + arbitrage hybrid → flywheel + Li-ion in same plant (Beacon-style).
• Geographic favorability (mountain + reservoir + grid tie) → PHS still cheapest at 100+ MW, 8+ hr.
• Existing thermal infrastructure (decommissioned coal, CSP) → repurpose with molten-salt or sand storage; reuse steam turbine.
• Multi-day seasonal firming → hydrogen (with secondary use), iron-air (if Form costs hit target), or oversized PHS.
• Industrial process heat displacement → Rondo brick, Antora carbon block, Polar Night sand — avoid the Rankine round-trip entirely.
• Behind-the-meter commercial/industrial → Li-ion (peak-shave) + supercap (power quality) + optional thermal (ice-storage HVAC).

23. Notable projects 2024–2026

Project	Technology	Capacity	Country / state	Year
Edwards & Sanborn	Li-ion LFP	875 MW / 3.3 GWh	USA – California	2024 (online)
Crimson Storage	Li-ion	350 MW / 1.4 GWh	USA – California	2024
Moss Landing (Vistra + LG)	Li-ion	750 MW / 3 GWh	USA – California	2020–24 (2025 fire incident)
Hornsdale Power Reserve	Li-ion (Tesla)	150 MW	Australia	2017 expanded 2020
Fengning PHS	Pumped hydro	3 600 MW	China	2024 fully online
Snowy 2.0	Pumped hydro	2 200 MW	Australia	COD 2027
Rongke Dalian	VRFB	200 MW / 800 MWh	China	2024
Form Energy / Big Rivers	Iron-air	10 MW / 1 GWh	USA – Kentucky	2024–25
Form Energy / Georgia Power Robins AFB	Iron-air	15 MW	USA – Georgia	2024 award
Energy Dome Sardinia	CO₂ battery	20 MW / 200 MWh	Italy	2024
Hydrostor Willow Rock	A-CAES	500 MW / 4 000 MWh	USA – California	Permitting 2025
Hydrostor Silver City	A-CAES	200 MW / 1 600 MWh	Australia	Construction 2025
Highview Carrington (CRYOBattery)	LAES	50 MW / 300 MWh	UK	COD 2026
Polar Night Pornainen	Sand battery	100 MWh thermal	Finland	Online 2025
Antora San Jose demo	TPV thermal block	early MW	USA – California	2024
ACES Delta	H₂ + salt cavern	220 MW electrolyzer + 300 GWh storage	USA – Utah	COD 2025
Cerro Dominador	CSP + molten salt	110 MW + 17.5 hr	Chile	2021 (operating)
Beacon Power Stephentown	Flywheel	20 MW	USA – New York	2011 (operating)

24. Cross-references

• battery-chemistries — cell-level chemistry detail (LFP, NMC, LTO, solid-state, Na-ion).
• photovoltaic-cells — the solar resource these systems shift.
• wind-turbine-types — the wind resource these systems firm.
• hydrogen-fuel-cells — power-to-hydrogen-to-power round-trip detail.
• transformers-power-systems — grid interconnection + PCS / inverter side.
• engineering-codes — NFPA 855, UL 9540 / 9540A, IEC 62933, IEEE 1547.
• heat-transfer-correlations — thermal storage charging + discharge heat transfer.
• electric-motor-taxonomy — pump-turbine motors, flywheel motor-generators.

25. Citations

• IEA — Energy Storage Insights 2024 and Renewables 2025 / Batteries and Secure Energy Transitions Report, https://www.iea.org/reports
• BloombergNEF — Long-Duration Energy Storage Outlook 2024; BNEF Battery Price Survey 2024 (cell + system $/kWh).
• NREL — Annual Technology Baseline (ATB) 2024 — Utility-Scale Battery Storage and Storage Futures Study (2022 update + 2024 supplement), https://atb.nrel.gov/
• DOE — Long Duration Storage Shot (target $0.05/kWh-cycle for >10 hr by 2030), https://www.energy.gov/eere/long-duration-storage-shot
• DOE — Energy Storage Grand Challenge roadmaps.
• NFPA 855:2023 — Installation of Stationary Energy Storage Systems.
• UL 9540A, 5th ed. (2024) — Test Method for Thermal Runaway Fire Propagation in Battery Energy Storage Systems.
• UL 9540, 3rd ed. — Energy Storage Systems and Equipment.
• IEC 62933 series — Electrical Energy Storage Systems.
• IEEE 1547-2018 and IEEE 2030.2.1-2019.
• Sandia National Laboratories — DOE Global Energy Storage Database, https://gesdb.sandia.gov/
• EPRI Energy Storage Integration Council (ESIC) reference designs and safety guidelines.
• CPUC — Self-Generation Incentive Program (SGIP) post-fire investigation reports (2022 + 2025 Moss Landing).
• Manufacturer datasheets: Tesla Megapack 2 XL, CATL EnerC+, BYD MC Cube ESS, Sungrow PowerTitan 2.0, Form Energy iron-air, ESS Inc. Energy Warehouse, Hydrostor A-CAES, Highview CRYOBattery, Energy Dome, Polar Night Energy.