Energy Storage: A Practical Reference

Pumped hydro is the oldest and largest grid storage technology: pump water uphill when power is cheap, run it back through turbines when it isn't. It accounts for roughly 90% of all grid storage ever built (~180 GW worldwide), plants last close to a century, and durations of 6–20+ hours are set by reservoir size. Historically these were slow bulk-shifting assets, but modern variable-speed machines add fast, precise frequency response.

Pumped hydro is proven at gigawatt scale, gets 70–85% round trip efficiency, has essentially unlimited cycle life, and offers the cheapest energy capacity of any mature electricity-in/electricity-out technology. The weaknesses: it needs two reservoirs with an elevation difference between them, so siting decides everything. Permitting and construction typically run 8–15 years, capex is huge and upfront, drought and environmental opposition are real risks, and it is nearly impossible to finance merchant.

Choose PSH when you need bulk 6–20+ hour shifting at hundreds of megawatts to gigawatts, have two-reservoir topography and water rights, and can carry a decade of development on rate-based, state, or otherwise patient capital. Once built, the sub-$100/kWh energy capacity and century asset life are very hard to beat. It works well as the backbone for hydro-friendly geographies planning decades ahead, and closed-loop designs unlock sites that open-loop permitting would kill. Avoid it for merchant financing, timelines under 8 years, flat or water-scarce geography, or anything below ~6 hours. In those cases, Li-ion deploys in months, and advanced CAES or geomechanical storage offers cavern-class economics without needing a mountain.

Power 100 MW to 3.6 GW per plant · duration 6–20+ hours · round trip 70–85% · response in seconds on variable-speed machines · asset life close to 100 years with essentially unlimited cycles · capex $1,500–4,000/kW, or $20–80/kWh at 10+ hour durations · permitting and construction 8–15 years.

Bath County (3 GW, Virginia — long the world's largest), Fengning (3.6 GW, China), Dinorwig (UK fast-response), dozens of GW under construction in China; Australia's Snowy 2.0 as the well-known cost-overrun case.

Capex runs $1,500–4,000/kW but often only $20–80/kWh at 10+ hour durations, which is the long-duration benchmark every other technology gets measured against. A multi-decade asset life amortizes almost anything, but only where geology, water, and patient capital coexist. China builds it state-directed, while Western merchant markets struggle to finance 10-year construction.

5 minHow Pumped Storage Power Plants Work (Hydropower)saVRee · 100k+ views

9 minWorld's Largest Batteries - (Pumped Storage)Practical Engineering · 1m+ views

2 minWhat is Pumped-Storage HydropowerU.S. Department of Energy · 50k+ views

Pumped Storage Hydropower (DOE Water Power Technologies Office) · ESGC Cost & Performance: Pumped Storage Hydropower (PNNL)

Solid gravity storage is pumped hydro without the water: lift mass (concrete blocks, compacted-earth bricks, or weights in abandoned mine shafts) with electric winches, and recover the energy by lowering it. The physics works well: 85%+ round trip, no degradation, instant response, and flexible siting. The challenge is that the energy stored per tonne-meter is tiny, so meaningful capacity requires moving enormous mass through large heights with cranes and structures that must survive millions of cycles.

Solid gravity offers high efficiency, no cycle fade, no exotic materials, and works anywhere with height or a shaft. The weaknesses: energy density is very poor, so structure cost per kWh is hard to get below batteries. Mechanical handling of thousands of blocks is a reliability and O&M question that no one has answered at fleet scale. Mine-shaft versions are capacity-limited by shaft volume. And the sector carries credibility baggage from hyped early ventures.

Consider solid gravity only where an existing deep shaft or structure supplies the height for free (decommissioned mines with 500+ m drops) and the buyer specifically needs chemistry-free storage with zero fade and cold-weather indifference. Greenfield block towers are almost never the answer: realized FOAK costs sit above Li-ion, fleet-scale mechanical reliability is unproven, and an LFP container wins every intraday application. If you need cheap bulk capacity at real scale, go to pumped hydro or CAES. Treat gravity as a site-specific reuse play, not a category bet.

Round trip 85%+ · duration a few hours · response effectively instant · no cycle fade over the life of the machinery · vendor capex claims $150–300/kWh, with realized FOAK costs above Li-ion · mine-shaft designs want drops of 500 m or more.

Energy Vault's EVx towers (Rudong, China — the first commercial-scale build; Energy Vault has since pivoted toward batteries), Gravitricity mine-shaft pilots (UK/Czech), various Chinese EPC-led block towers.

Vendors claim $150–300/kWh with long life and zero augmentation; realized FOAK numbers land well above Li-ion. The bet only works if structure and machinery costs fall faster than lithium's learning curve, and the category is currently losing that race. Mine-shaft reuse in specific geographies is the most defensible niche.

12 minHow to use gravity to store energyDW Planet A · 500k+ views

2 minEnergy Vault 3D SimulationEnergy Vault · 500k+ views

11 minGravity Energy Storage. Who's right and who's wrong?Just Have a Think · 100k+ views

ESGC Cost & Performance: Gravitational Storage (PNNL) · G-VAULT Gravity Energy Storage (Energy Vault)

Geomechanical storage is pumped hydro turned sideways and underground: pump water at high pressure into an engineered fracture in deep rock, and the rock's elastic rebound drives the water back through a turbine on demand. The earth itself acts as the upper reservoir. Quidnet Energy pioneered the approach using oilfield drilling and pressure-pumping techniques, and it aims to deliver pumped-hydro economics without mountains, using flat land and standard rigs.

Geomechanical storage reuses mature oil-and-gas drilling, siting is far more flexible than PSH, energy capacity scales with fracture volume at low marginal cost, and durations of 10+ hours come naturally. The weaknesses: round-trip efficiency runs below classic PSH because of pressure losses in the rock. Each site's geology must hold pressure without leak-off, a per-site risk similar to geothermal's. Induced-seismicity and groundwater questions follow anything that pressurizes the subsurface. And the concept has megawatts, not gigawatts, of field history.

Deploy geomechanical storage where flat land sits over competent, pressure-tight sedimentary rock with oilfield services nearby, for 10+ hour daily-to-multi-day shifting targeting cavern-class $20–50/kWh without mountains or salt. Budget for per-site geological derisking (test wells, pressure holds) before committing capacity. It suits utilities in Texas-like geographies that want PSH economics on commodity drilling timelines. Avoid it where induced-seismicity or groundwater politics are hot, where the geology hasn't been proven to hold pressure, or for sub-8-hour duty. Li-ion owns that band, and where salt caverns exist, advanced CAES carries far more operating history.

Duration 10+ hours · round trip below pumped hydro's 70–85% · target energy capex $20–50/kWh · flagship offtake 300 MWh with CPS Energy · field history measured in megawatts, not gigawatts.

Quidnet's Texas pilots and its 300 MWh CPS Energy (San Antonio) offtake — the sector's flagship deal; Hunt Energy backing.

The target is $20–50/kWh energy capacity using commodity oilfield services, which is genuinely pumped-hydro-class if the geology cooperates. The oilfield supply chain means fast scaling with no novel manufacturing. The open questions are per-site derisking cost and whether efficiency and cycling hold over decades. It is the most credible "PSH anywhere" thesis and worth watching.

4 minRevolutionary Grid-Scale Energy Storage TechnologyClearPath Action · 1k+ views

Geomechanical Energy Storage Technology (Quidnet Energy)

Subsea pressure storage uses the ocean as the upper reservoir: anchor hollow concrete spheres on the deep seabed, let seawater rush in through a turbine to discharge, and pump it back out against hundreds of meters of hydrostatic head to charge. Depth substitutes for elevation: at 700 m, each cubic meter stores what pumped hydro gets from a 700 m mountain. It co-locates naturally with offshore wind, storing energy where it's generated.

The theoretical siting resource along continental slopes is enormous, efficiency and life are pumped-hydro-like, there's no land use, and modular concrete spheres suit series manufacture. The weaknesses: everything subsea costs multiples of its onshore equivalent (installation vessels, wet-mate connectors, marine maintenance). Concrete spheres must survive decades of pressure cycling and biofouling. And the concept remains at single-sphere pilot scale with no commercial project.

Treat this as a research position, not a procurement option. It is relevant only when co-located with deep offshore wind on continental slopes at 600–800 m, where the hydrostatic head is free and the export cable is already paid for. Fund it through grants or strategic offshore-wind capital with offshore-operations diligence, not storage-market underwriting. If a near-term project needs actual capacity, use onshore alternatives: Li-ion for hours, CAES or PSH for bulk. Revisit subsea spheres only if multi-sphere pilots demonstrate marine O&M costs low enough to preserve the advantage of the free storage medium.

Deployment depth 600–800 m · pilot sphere roughly 9 m across · duration a few hours · round trip in pumped hydro's 70–85% band on paper · still at single-sphere pilot scale, with no commercial project built.

Fraunhofer IEE's StEnSea program: Lake Constance model tests, and the ~9 m sphere pilot off California (with Sperra and Pleuger) targeting mid-2020s deployment; Ocean Grazer's seabed bladder variant.

Paper economics reach pumped-hydro territory only at large sphere counts and depths; near-term reality is R&D funded by grants and offshore-wind strategic interest. The relevant diligence lens is offshore-industry cost discipline, not storage chemistry: the medium is free, but the marine operations are not.

12 minUnderwater Battery Storing Power at 600m depth!German Science Guy · 10k+ views

10 minStensea - Energy Storage of the Future!?The Futurist · 10k+ views

14 minHow the Ocean Could be the Future of Energy StorageUndecided with Matt Ferrell · 500k+ views

StEnSea — Stored Energy in the Sea (Fraunhofer IEE) · California Sphere Pilot Announcement (Fraunhofer IEE)

Diabatic CAES is the classic compressed-air plant: compress air into a salt cavern off-peak, then feed it to a natural-gas combustion turbine at discharge. The stored air replaces the compressor work that normally consumes two-thirds of a gas turbine's output. Only two Western plants were ever built (Huntorf 1978, McIntosh 1991), both still running, which makes CAES proven and, in its original form, a commercial dead end at the same time.

Cavern energy capacity is nearly free ($1–10/kWh), plant life is measured in decades, and the technology is boringly bankable. The weaknesses: it burns gas, and the emissions and fuel exposure make classic diabatic CAES effectively unfinanceable in decarbonizing markets. Round-trip framing is awkward because the plant is half storage, half gas peaker. Salt-cavern geology is required. And 45 years produced exactly two Western plants, which tells you something about demand.

Build new diabatic CAES only where salt caverns are on hand, gas is cheap and unpriced for carbon, and the buyer wants dispatchable capacity more than clean storage. In practice that is a shrinking set of markets, mostly outside the West. If you already own Huntorf- or McIntosh-class assets, keep running them; the caverns and machinery are paid for. In any decarbonizing market, skip straight to advanced CAES or hydrogen-fired turbines on the same salt geology. The gas burner is what makes this design unfinanceable, not the cavern.

Plant power 110 MW at McIntosh and 321 MW at Huntorf · duration roughly 2 hours at Huntorf and 26 hours at McIntosh · cavern energy capacity $1–10/kWh · plant life 35–48 years so far on both Western units · two Western plants built in 45 years · China's revival plants at 100–300 MW.

Huntorf (Germany, 321 MW), McIntosh (Alabama, 110 MW); China's Zhangjiakou and Yingcheng plants revive the concept at 100–300 MW scale with reduced or zero gas input.

Where salt geology exists, the cavern is the cheapest electricity reservoir ever built, and the entire modern LDES cavern thesis (advanced CAES, hydrogen) descends from it. Diabatic CAES itself is a transitional design: its economics were built on cheap off-peak baseload and peak gas prices, a spread modern markets no longer offer in the same shape.

5 minTurning Compressed Air Into Electricity - CAES 2015PowerSouth Energy Cooperative · 100k+ views

2 minCAES - Compressed Air Energy Storage - IMAGES Project - animationIMAGES Project · 50k+ views

16 minThe Genius Energy Storage Idea Everyone ForgotEngineering with Rosie · 10k+ views

ESGC Cost & Performance: Compressed Air Energy Storage (PNNL) · 2022 Grid Energy Storage Technology Cost and Performance Assessment (DOE)

Advanced CAES is CAES without the gas burner: capture the heat of compression (in thermal stores or, in Hydrostor's advanced design, manage pressure with a hydrostatic water column) and return it to the air on expansion, eliminating fuel entirely. Modern designs use purpose-built caverns in hard rock rather than salt, trading some cavern cost for siting freedom. Vendors claim round-trip efficiencies of 60–70%; early plants demonstrate low-to-mid 50s.

Advanced CAES is fuel-free and emission-free, lasts decades, has cheap marginal energy capacity, and draws on industrial turbomachinery supply chains. The weaknesses: thermal stores and custom caverns push FOAK capex high. Efficiency claims usually shrink once plants actually run. Purpose-mined caverns cost far more than solution-mined salt. And project timelines run as long as pumped hydro's without pumped hydro's track record.

Choose advanced CAES for 8–24 hour shifting where salt or competent hard rock exists, the offtaker can wait out a 5–7 year development cycle, and a 30–60 year fuel-free asset at tens of $/kWh justifies FOAK project risk. Underwrite at the demonstrated low-to-mid-50s round trip, not the 60–70% brochure number. It fits utilities and state-backed programs replacing retiring thermal plants with long-lived firm capacity. Avoid it below 8 hours or on merchant timelines; Li-ion wins both. And if no cavern-grade geology exists, reach for LAES or a CO2 battery instead of paying for a purpose-mined hole.

Duration 8–24 hours · round trip in the low-to-mid 50s in early plants, against 60–70% vendor claims · energy capex in the tens of $/kWh at 8+ hours · asset life 30–60 years · development cycle 5–7 years · flagship project 500 MW / 4,000 MWh at Willow Rock.

Hydrostor's Goderich pilot (Ontario) and its flagship Willow Rock project (California, 500 MW/4,000 MWh, contracted); China's 300 MW-class Zhangjiakou (thermal-store) and Shandong salt-cavern plants — currently the world's largest operating new CAES fleet.

The pitch is pumped-hydro economics with easier siting: $/kWh in the tens at 8+ hours. The risk is mostly project-development risk (mining, permitting, FOAK turbomachinery) rather than science risk. China's state-backed buildout will generate the fleet data Western merchant financiers are waiting for, a rare case where someone else is doing the derisking for free.

2 minHow A-CAES Long Duration Energy Storage WorksHydrostor · 5k+ views

14 minHow Compressed Air Batteries are FINALLY HereUndecided with Matt Ferrell · 100k+ views

Hydrostor A-CAES Technology (Hydrostor) · ESGC Cost & Performance: Compressed Air Energy Storage (PNNL)

Liquefy air at -196 °C using off-peak power, store it in insulated tanks at atmospheric pressure, then pump, reheat, and expand it through turbines to discharge. The reservoir is a standard cryogenic tank rather than a cavern or mountain, so LAES can be sited anywhere, and the entire plant is assembled from air-separation and LNG components that the industrial-gas industry has been building for a century. Recycling the waste heat and cold is what lifts round-trip efficiency from ~25% to 50–60%.

LAES has no geology or geography constraints, uses a mature industrial supply chain, adds duration cheaply (just more tanks), and lasts 30–40 years; sharing cold and heat with industrial neighbors can push effective efficiency higher. Weaknesses: round-trip efficiency is the worst of the mechanical class (50–60% best case); liquefaction plants prefer running steadily, not flexibly; and capex per kW is high, so LAES needs long durations and cheap charging power to pencil out.

Pick LAES when you need siting freedom (no salt, no mountains, brownfield industrial land works fine) for 8–12+ hour duty on a grid with abundant cheap surplus to charge from, ideally with an industrial neighbor to buy the waste cold or supply waste heat. Wind-heavy systems like Britain's are the archetype. The mature industrial-gas supply chain makes it the most bankable of the geology-free mechanical options. Avoid it where charging power isn't reliably cheap or duty is under 8 hours: you pay the 50–60% round trip on every cycle, while Li-ion does better intraday and advanced CAES does better wherever caverns exist.

Liquefaction at -196 °C · round trip 50–60% at best, about 25% without heat and cold recycling · duration 8–12+ hours · asset life 30–40 years · tank energy capacity in the tens of $/kWh, with the liquefier and power island driving $/kW · flagship plant 50 MW / 300 MWh at Carrington.

Highview Power: Pilsworth pilot, the 50 MW/300 MWh Carrington plant (UK, under construction with UK IB backing), and gigawatt-hour follow-ons; Sumitomo (industrial gases) as an investor.

Tank capacity costs tens of $/kWh; the expensive parts are the liquefier and power island ($/kW). LAES therefore suits 8–12+ hour duty with abundant cheap surplus power, which is exactly the profile of wind-heavy grids like Britain's. You accept the efficiency penalty in exchange for siting freedom and a mature supply chain.

5 minLiquid Air Energy Storage AnimationHighview · 100k+ views

12 minLiquid Air Battery Explained - Rival to Lithium Ion Batteries?Undecided with Matt Ferrell · 500k+ views

11 minCryogenic Air Energy Storage - the new long duration energy solution?Just Have a Think · 100k+ views

Liquid Air Energy Storage Technology (Highview Power) · ESGC Cost & Performance: Thermal Energy Storage incl. LAES (PNNL)

A closed thermodynamic cycle using CO2 as the working fluid: compress gaseous CO2 until it liquefies at ambient temperature (storing the heat of compression), hold it as a compact liquid in pressure tanks, then evaporate and expand it back through a turbine, re-inflating a giant atmospheric gas-holder dome. CO2 liquefies at manageable pressure at room temperature and air does not, so the cycle gets cryogenic-class density without any cryogenic equipment.

The claimed round-trip efficiency is ~75% (better than LAES or CAES), it needs no geology or exotic temperatures, it runs on standard turbomachinery, and the loop is fully closed with no emissions. Weaknesses: the signature gas dome is large, fragile-looking, and land-hungry; energy density still trails batteries badly; the efficiency claim depends heavily on how well the thermal store performs; and the design is young, with one commercial-scale plant's worth of operating data.

Consider the CO2 battery for 8–10 hour daily cycling on sites with cheap land but no cavern geology. Its claimed ~75% round trip and commodity turbomachinery beat LAES's efficiency and CAES's siting constraints, and a 20–30 year life without augmentation spend improves on lithium's economics at that duration. Demand operating data from Ottana-class plants before underwriting the efficiency claim. Avoid it on land-constrained sites (the dome is hectare-scale), at durations lithium already serves cheaply (≤6 hours), and for multi-day duty, where cavern systems or iron-air carry the energy capacity far cheaper.

Round trip roughly 75% claimed · duration 8–10 hours · asset life 20–30 years with no augmentation spend · capex in the mid-hundreds $/kWh FOAK, targeting $100–200/kWh at scale · first commercial plant 20 MW / 200 MWh at Ottana · gas dome footprint at hectare scale.

Energy Dome: Sardinia 2.5 MW pilot, the 20 MW/200 MWh Ottana commercial plant, an Alliant Energy (Wisconsin) project, and a strategic partnership with Google (2025) for global deployment.

The CO2 battery sits between Li-ion and cavern systems on cost: mid-hundreds $/kWh FOAK, falling toward $100–200 at scale, aimed at 8–10 hour daily cycling where lithium augmentation costs add up. The commodity-equipment supply chain is the strongest part of the story. The open question is whether a thermodynamics startup can out-execute battery container economics before lithium reaches 10 hours.

2 minCO2 Battery | Energy DomeEnergyDome · 50k+ views

14 minHow the CO2 battery could be the future of energy storage?Undecided with Matt Ferrell · 100k+ views

14 minThe GENIUS Of the Carbon Dioxide BatteryTwo Bit da Vinci · 500k+ views

CO2 Battery Technology (Energy Dome) · Cost Benchmarking for Long Duration Energy Storage Solutions (LDES Council)

A flywheel stores energy as rotation: a carbon-fiber or steel rotor spins at up to tens of thousands of RPM on magnetic bearings in vacuum, coupled to a motor-generator. Flywheels deliver and absorb full power in milliseconds, cycle millions of times without degradation, and hold their rating regardless of temperature, which makes them close to an ideal power device. What they can't do is hold energy for long: standby losses and low energy density confine them to seconds-to-minutes duty.

Flywheels offer unmatched cycle life and response speed, precise state-of-charge knowledge (it's a tachometer), no chemistry to degrade or catch fire, and decades of service life. Weaknesses: cost per kWh is extremely high (they are $/kW machines); self-discharge runs several percent per hour; catastrophic rotor failure must be contained; and Li-ion has taken most of the frequency-regulation market that was flywheels' commercial beachhead.

Buy flywheels for seconds-to-minutes power duty at very high cycling rates (hundreds of full cycles a day for decades): UPS bridging to generator start, industrial pulse loads, rail and crane regeneration, and inertia services where temperature extremes or fire codes rule batteries out. You are buying $/kW, response speed, and cycle life; size the energy content in seconds. Avoid anything beyond ~15 minutes of discharge, because the $/kWh doesn't work and self-discharge of percent-per-hour drains longer holds. For plain frequency regulation, Li-ion is now the cheaper default; flywheels win only where cycle count or environment would destroy a battery.

Rotor speed up to tens of thousands of RPM · discharge duration seconds to about 15 minutes · full power in milliseconds · millions of cycles with no degradation, over decades of service life · self-discharge several percent per hour · grid plants around 20 MW.

Beacon Power's 20 MW regulation plants (Stephentown NY, Hazle PA); Amber Kinetics' multi-hour steel flywheels (unusual for the category); UPS bridging (Piller, Active Power); fusion-experiment and rail pulse power.

Flywheels are a niche, profitable power-quality business rather than an energy-storage market: data centers, microgrids, and industrial pulse loads pay for cycle life and reliability that batteries can't match. Every attempt to push flywheels into hours-long duty has failed on $/kWh; the survivors sell $/kW and durability.

2 minHow an Amber Kinetics flywheel worksAmberKinetics · 50k+ views

14 minHow 40-Ton Spinning Wheels Are Saving the Power GridUndecided with Matt Ferrell · 500k+ views

15 minHow NASA Reinvented the Mechanical BatteryZiroth · 100k+ views

Flywheel Energy Storage Systems (Beacon Power) · U.S. DOE Energy Storage Handbook (Sandia National Laboratories)

Molten salt is the workhorse of concentrated solar power: nitrate salts are pumped between a "cold" tank (~290 °C) and a hot tank (~565 °C), storing heat that later raises steam for a turbine. Gigawatt-hours are in daily commercial operation, which makes molten salt by far the most deployed non-hydro storage on earth. The same tanks are now being repurposed for standalone electric storage (charge with resistance heaters) and industrial steam supply.

Energy capacity is cheap and proven ($20–50/kWh-thermal), duration extends by adding salt, tanks last decades, and the output is direct high-grade heat. Charged with resistance heaters and discharged as steam on a daily cycle, the tanks return roughly 90% of the input energy as heat, because nothing goes through a heat engine. Weaknesses: as electricity storage, the steam cycle caps round trip near 40%, which is a very heavy penalty; salts freeze at ~220 °C (so trace heating runs forever) and decompose above ~600 °C, capping temperature; nitrate salt prices spike with fertilizer markets; and CSP, its host application, keeps losing to PV-plus-batteries.

Use two-tank salt when the product is heat: CSP plants adding dispatch hours, or standalone tanks charged with cheap surplus power delivering 300–550 °C industrial steam where gas is expensive or carbon is priced. You get proven tanks, $20–50/kWh-thermal, and duration added by buying more salt. Stay inside the 290–565 °C window; below it, cheaper media exist, and above ~600 °C nitrates decompose. Avoid it entirely for power-to-power arbitrage, because the ~40% steam-cycle round trip loses to any battery. And if your heat duty tolerates variable discharge temperature, solid brick and rock stores undercut salt on capex with no freeze-protection burden.

Hot tank roughly 565 °C, cold tank roughly 290 °C · nitrate salts freeze near 220 °C and decompose above 600 °C · energy capacity $20–50/kWh-thermal · power-to-heat delivery roughly 90% on a daily cycle, against a power-to-power round trip near 40% through the steam cycle · tank life measured in decades · gigawatt-hours already in daily commercial operation.

Every salt-storage CSP tower and trough: Noor III (Morocco), Cerro Dominador (Chile), Crescent Dunes (US, troubled), dozens of Chinese CSP hybrids; Malta and others reuse the tank tech; Denmark's Hyme pushes hydroxide salts hotter.

For electricity arbitrage, the 40% steam-cycle round trip usually loses to batteries. The living market is heat: charged with cheap surplus power, salt tanks deliver industrial steam at costs gas struggles to match when carbon is priced. Most salt ventures have now pivoted to exactly this.

3 minSolarReserve - Concentrated Solar Power Technology AnimationSolar Reserve · 50k+ views

4 minAndasol Power Plant Salt Storage Solar Millenium - solar thermal parabolic trough plantMatthew Wright · 10k+ views

12 minHow we can make solar power at nightDW Planet A · 500k+ views

Thermal Storage System CSP Basics (DOE Solar Energy Technologies Office) · ESGC Cost & Performance: Thermal Energy Storage (PNNL)

This is the cheapest stored joule on the market: resistively heat a pile of refractory brick, crushed rock, sand, or graphite to 400–1,800 °C with surplus electricity, insulate it well, and discharge as hot air, steam, or radiant heat for industrial processes. Electric resistance charging is ~100% efficient, and delivered heat retains 90%+ of input energy because nothing is converted back to electricity, so the big heat-engine loss never occurs.

Storage media cost dollars per tonne (sub-$10/kWh-thermal), don't degrade, and have no supply-chain constraints, and industrial heat is half of global final energy, an enormous market lithium cannot serve. Weaknesses: it is heat-out only in most designs (adding a turbine collapses efficiency and economics); discharge temperature declines as the store depletes, which some processes can't tolerate; and the business sells to conservative industrial buyers with gas at marginal cost, so adoption comes down to spreadsheet comparisons rather than technology.

Deploy heat bricks wherever an industrial process burns fuel for 200–1,500 °C heat and cheap surplus electricity is available daily: high-curtailment renewable grids, or anywhere the spark spread against gas-plus-carbon clears. Charged nightly, the sub-$10/kWh-thermal medium delivers the cheapest stored energy on this sheet. Size the store for daily cycling and verify the process tolerates the declining discharge temperature as the pile depletes. Avoid it when you need electricity back out (the turbine collapses the economics; that's pumped thermal's unsolved problem), when the process demands constant-temperature delivery (use PCM), or where gas stays cheap and carbon unpriced, because then the spreadsheet comparison never comes out in your favor.

Storage temperature 400–1,800 °C · delivered process heat 200–1,500 °C · resistive charging close to 100% efficient · 90%+ of input energy returned as heat · media cost under $10/kWh-thermal, a few dollars per tonne of brick or rock · daily charge and discharge with no media degradation.

Rondo (brick, cement and biofuel plants), Antora (graphite blocks, also exploring thermophotovoltaic re-electrification), Polar Night Energy (sand, Finland district heat), Kraftblock, Brenmiller; Siemens Gamesa's rock-store ETES pilot was an early demonstration.

Where surplus renewable power is cheap and gas or carbon is expensive, stored electric heat beats fuel today, without any subsidy math. The unit economics are dominated by insulation, containment, and heat-delivery engineering rather than the medium. The strategic risk is commoditization: the storage media are cheap materials anyone can buy, so winners will win on execution, financing, and industrial trust.

10 minCan a Simple Brick Be the Next Great Battery? | John O'Donnell | TEDTED · 100k+ views

14 minHow A Brick & Rock Battery Is Changing Energy StorageUndecided with Matt Ferrell · 500k+ views

8 minHow 3000 Degree Bricks Will End Battery StorageZiroth · 500k+ views

Rondo Heat Battery: How It Works (Rondo Energy) · Thermal Batteries: Decarbonizing U.S. Industry (Energy Innovation)

Pumped thermal is a heat pump run forward, then backward: charge by pumping heat from a cold store into a hot store (salt, rock, or gravel), discharge by running the temperature difference through a heat engine. A heat pump moves 2–3 units of heat per unit of electricity, so the big heat-engine loss on discharge is partially bought back on charge. That yields theoretical round trips of 60–70% from equipment that is all standard turbomachinery and tanks.

It sites anywhere, uses cheap thermal media, lasts decades, and runs on a fully mechanical supply chain. Weaknesses: real round-trip efficiency depends on aggressive turbomachinery performance at temperatures where compressors and expanders are custom; the cycle's complexity (four heat exchangers, two stores, reversible machines) multiplies cost and failure modes; and no plant beyond pilot scale has demonstrated the claimed numbers. The list of failed and struggling attempts (including 1414 Degrees and Stiesdal) is worth studying.

Treat pumped thermal as a pilot-stage bet, not a procurement option. Back it only if you're funding geology-free 10–24 hour storage development and can condition capital on measured (not modeled) round trip from an integrated plant; 55%+ is the threshold that makes the arbitrage math work. Its promise is CAES-class $/kWh anywhere, which is worth option money. For capacity you need this decade, buy Li-ion below 8 hours and advanced CAES or LAES beyond it. And if the site's real product could be heat rather than electricity, skip the heat engine entirely and buy solid heat bricks.

Charging heat pump moves 2–3 units of heat per unit of electricity · modeled round trip 60–70%, with 55% the threshold that makes the arbitrage work · duration 10–24 hours · target energy capex in the CAES range, tens of $/kWh · asset life measured in decades · nothing demonstrated beyond pilot scale so far.

Malta Inc (X/Alphabet spinout, salt-based), Echogen (supercritical CO2 cycle), Stiesdal GridScale (rock), MAN ETES; academic origins in Isentropic Ltd's gravel design (an early venture that failed).

On paper it's the best of both worlds: cavern-free siting with CAES-class $/kWh. In practice the efficiency-versus-capex trade-off is very tight, because every compromise in turbomachinery drops RTE toward values where cheap thermal media can't save the arbitrage math. In this category, the diligence question is always the measured round trip, not the modeled one.

4 minMalta - Pumped Heat Energy Storage (PHES) SystemMalta Inc. · 1k+ views

4 minNational Facility for Pumped Heat Energy StorageNational Facility for Pumped Heat Energy Storage · 5k+ views

16 minPumped Thermal Energy StorageSouthwest Research Institute · 1k+ views

Task 36: Carnot Batteries (IEA Energy Storage TCP) · Malta Thermal Energy Storage & Heat Pump Technology (Malta Inc)

The idea is simple: run chillers at night, when power is cheap and heat is easy to reject, freeze water or chill a large tank, then melt or circulate it to carry the building's cooling load through the afternoon peak. It stores cooling demand rather than electricity, but the grid effect is the same: megawatts shifted out of the peak. It has operated at scale in commercial buildings and district cooling for decades.

The technology is mature and cheap: it uses standard HVAC equipment, tanks last 30 years, and it often improves chiller efficiency because night air is cooler. Effective round trip is high because the "discharge" is the cooling service itself. The weaknesses: it only shifts cooling, not general electricity; the economics depend on time-of-use rate spreads that utilities keep flattening; and in dense buildings, tanks compete for floor space that is often worth more than the savings.

Install ice or chilled-water storage in any building or district system with a large afternoon cooling peak, meaningful time-of-use or demand-charge spreads, and room for tanks. New construction and district cooling usually pencil best, because the tank can displace chiller capacity rather than add to it. Wherever cooling drives the peak, this is the default first move before batteries. Skip it if rate structures are flat, cooling load is small, or floor space is worth more than the demand savings. And remember that it only shifts cooling; if you need general load shifting or backup, use a behind-the-meter battery instead.

Duration typically 4–8 hours, sized to the afternoon cooling peak · ice holds 334 kJ/kg (93 Wh/kg) of latent heat · tank life about 30 years · effective round trip is high, because the discharge is the cooling service itself · cost in the tens of $/kWh-equivalent.

Thousands of Calmac/Trane ice-tank buildings, Chicago and Gulf-region district cooling plants, Ice Energy's residential "Ice Bear" (bankrupt), Singapore's Marina Bay district system.

When a building needs cooling anyway, this is among the cheapest peak-shifting storage available: tens of $/kWh-equivalent, with utility-grade reliability. The market is limited by cooling demand and rate design, not by the technology. As air conditioning drives grid peaks higher, this old idea keeps getting rediscovered.

9 minThermal Ice StorageEvapcoInc · 10k+ views

4 minEnwave Chicago District Cooling System features large-scale Thermal Energy Storage (TES)International District Energy Association · 10k+ views

6 minThermal Energy StorageMEP Academy · 10k+ views

How Ice Energy Storage Works (CALMAC/Trane) · Thermal Energy Storage for Buildings (DOE Building Technologies Office)

These are the high-density options in heat storage. Phase-change materials (PCMs) like salt hydrates, paraffins, and metal alloys absorb a lot of latent heat at a constant temperature, so they can discharge at a steady temperature in a way that sensible-heat bricks can't. Thermochemical storage goes further: it drives a reversible reaction (salt hydration, metal-oxide redox, ammonia dissociation) and stores the energy in chemical bonds, with effectively zero standby loss and several times the density of hot rock.

PCMs deliver heat at a constant temperature, which suits processes that need exact conditions. Thermochemical stores hold energy for months without insulation and can be transported. The weaknesses: PCMs suffer from cycling degradation, phase separation, and poor thermal conductivity, which forces expensive encapsulation. Thermochemical systems add reactors and mass handling that swamp the density advantage. And both remain far from the cost floor that plain bricks set. Density is rarely what industrial heat customers are actually short of.

Pick PCM when the application needs constant-temperature discharge, or when compactness is worth paying a multiple of brick cost: domestic hot water in space-constrained homes, cold-chain shipping, or processes that can't tolerate a sliding temperature. Choose thermochemical only if a months-long hold or transportable heat is the actual requirement, and price in the reactors and mass handling honestly. For bulk industrial heat storage, avoid both; sensible bricks, rock, or salt win on $/kWh-thermal nearly everywhere. Back specific applications where someone will pay the density or precision premium, not the category as a whole.

PCM discharge temperature is fixed by the material's melting point · latent heat of typical salt-hydrate and paraffin PCMs runs roughly 150–250 kJ/kg · thermochemical stores hold energy for months with effectively zero standby loss · energy density several times that of hot rock · cost a multiple of sensible brick storage, which sits under $10/kWh-thermal.

Sunamp's heat batteries (UK homes, salt-hydrate PCM), PCM cold-chain shipping packs, aluminum-alloy PCM pilots (Azelio, wound down), university metal-oxide and salt-hydrate programs, solar-ammonia loops (historic ANU work).

These technologies win niches where temperature precision, compactness, or a months-long hold justifies paying a multiple of brick costs: domestic heat, cold chain, and mobile heat delivery. As bulk storage they lose to sensible heat on simple $/kWh-thermal nearly everywhere. If you're investing, back specific applications, not the category.

7 minHow phase change materials (PCMs) work in Sunamp thermal batteriesSunamp · 5k+ views

13 minAre Phase Change Materials the Future of Water Heaters?Undecided with Matt Ferrell · 100k+ views

4 minThe Heat Battery using Thermochemical Storage | TNOTNO · 10k+ views

Task 40 Final Report: Compact Thermal Energy Storage (IEA Energy Storage TCP) · How Heat Batteries Work (Sunamp)

Split water with renewable electricity and store the energy in the hydrogen bond. As the front end of a storage chain, electrolysis converts unlimited surplus power into a molecule that can sit in a cavern for months. That makes it the only practical route to true seasonal electricity storage. Three electrolyzer families compete: alkaline (cheapest, most mature), PEM (compact, dynamic, iridium-dependent), and solid-oxide (highest efficiency, hot and fragile).

Energy capacity is effectively unbounded, and the marginal cost of reservoir space is near zero. The product also doubles as industrial feedstock, fuel, and reductant, so "storage" competes with several revenue stacks at once. The weaknesses: electricity-to-hydrogen runs 60–75% efficient before compression; the full power-to-power round trip lands at 30–40%; electrolyzer capex and stack life still dominate levelized cost; and the sector's 2021–24 hype cycle left a trail of canceled projects that should discipline any forecast.

Use electrolysis as storage only for durations nothing else reaches (weeks to seasons), and only with a cheap reservoir downstream, which in practice means salt-cavern access. Under those conditions, a 30–40% power-to-power round trip is acceptable because the alternative is not storing at all. It also needs power that is both cheap and available several thousand hours a year: charging only on curtailed surplus leaves the electrolyzer idle too much, and running it flat-out erases the surplus story. For anything under ~100 hours, don't use it; Li-ion, flow, and CAES all beat it. And if the real customer is industry, sell the molecule as feedstock rather than round-tripping it.

Electricity to hydrogen 60–75% efficient before compression, roughly 50–55 kWh per kg · power-to-power round trip 30–40% · system capex $300–600/kW for Chinese alkaline and higher in the West · stack life roughly 60,000–90,000 hours · solid-oxide cells run 700–850 °C at 85%+ electrical efficiency · storage duration weeks to seasons.

NEOM's 2.2 GW plant (Air Products/thyssenkrupp), Sinopec Kuqa (China, alkaline), Ørsted and BP refinery projects; electrolyzer makers Nel, ITM, Plug, Electric Hydrogen, Longi and PERIC in China.

Everything hinges on the capacity factor of cheap power: electrolyzers running only on curtailed surplus sit idle too much to amortize, and running them flat-out kills the "surplus" story. China ships alkaline systems at a third of Western prices, which is compressing equipment margins globally. As storage, hydrogen's role is the durations lithium and CAES can't reach (weeks and seasons), where a 35% round trip is acceptable because the alternative is not storing at all.

4 minThe Hydrogen ElectrolyserPhi Suea House Project · 500k+ views

6 minUnboxing : We dismantled a PEM electrolyzer to better understand its internal structure.SENZA Hydrogen Energy And Environmental Technology · 50k+ views

15 minHow A PEM Electrolyser Works (Lesson 2/5) Hunor Kacso | Hydrogen Training SolutionsHydrogen Training Solutions · 5k+ views

Global Hydrogen Review 2025 (IEA) · Clean Hydrogen Production Cost Scenarios with PEM Electrolyzers, Record 24005 (DOE Hydrogen Program)

This is the reservoir half of the hydrogen chain. Salt caverns (solution-mined voids holding thousands of tonnes at 100–200 bar) have stored hydrogen commercially for refineries since the 1970s, and they are the only storage medium on earth whose marginal capacity costs pennies per kWh. Above ground, compressed tanks (350–700 bar), liquid hydrogen at -253 °C, and metal hydrides serve smaller, mobile, or geology-poor applications at 10–100× cavern cost.

Cavern storage is proven, vast, and extremely cheap per kWh ($0.20–2), with months of hold and no degradation. The weaknesses: salt geology is geographically concentrated (US Gulf Coast, northern Europe); hydrogen embrittles steels and leaks through everything, which adds cost to every pipe and valve; compression to storage pressure eats 5–15% of the energy; liquid hydrogen boils off daily; and above-ground tanks always cost hundreds of times more per kWh than caverns.

Use salt caverns whenever seasonal-scale hydrogen inventory is the goal and the geology exists (Gulf Coast, northern Europe). At $0.20–2/kWh they are the only reservoir cheap enough to cycle once a year, and much of the value goes to whoever secures cavern rights early. Above-ground tanks are for mobility, refueling, and small industrial buffering only; never size them for bulk grid energy, where their $/kWh runs hundreds of times cavern levels. If there's no salt, don't force it: import storability via ammonia or LOHC, test porous-rock storage if local geology permits, or concede the seasonal case to transmission.

Cavern storage $0.20–2/kWh, above-ground tanks 10–100× that · cavern pressure 100–200 bar holding thousands of tonnes · compressed tanks 350–700 bar · liquid hydrogen at -253 °C · compression to storage pressure eats 5–15% of the energy · hold time months with no degradation.

Chevron Phillips Clemens and Air Liquide Spindletop caverns (Texas, decades of operation), Teesside (UK), the ACES Delta project (Utah — 300 GWh-class caverns feeding hydrogen turbines), HyStock (Netherlands); porous-rock storage pilots in Austria and Argentina.

The cavern is why hydrogen wins seasonal storage on paper: it is the only reservoir cheap enough to cycle once a year. The value therefore concentrates in geology rights and conversion equipment, not the molecule. Where salt doesn't exist, hydrogen's seasonal case weakens sharply, because pipelines or carriers (ammonia, LOHC) have to import the storability.

3 minStorengy - Storage in salt cavernStorengy · 10k+ views

4 minFundamentals of hydrogen storageFuture Energy & Technology · 10k+ views

4 minH2 tanks type 1 - type 5, technological differences and advantagesVoith Group · 10k+ views

Subsurface Hydrogen Storage: State of Knowledge and Research Recommendations (SHASTA, NETL/PNNL/LLNL) · SHASTA Subsurface Hydrogen Storage Project (DOE/NETL)

This is the discharge half of the hydrogen chain: convert stored hydrogen back to electricity through fuel cells (PEM or solid-oxide, 45–60% efficient) or hydrogen-capable gas turbines (35–60%, mature machines burning blends today and pure H2 on vendor roadmaps). It is the component that turns a cavern into a power plant, and it decides whether "hydrogen storage" means anything for grids rather than just for industry.

Turbines bring gigawatt scale, black-start capability, and a century of fleet experience; fuel cells bring quiet, modular, high-efficiency conversion at distribution scale. The weaknesses: conversion losses multiply with electrolysis losses, which is how the round trip lands at 30–40%; hydrogen combustion adds NOx and flashback engineering; fuel-cell capex and stack life still trail turbines at scale; and utilization is inherently low for seasonal duty, so the capex per kW has to pencil at near-idle run hours.

Deploy hydrogen-to-power as firm capacity that rarely runs and gets paid to exist: seasonal backup, wind-drought insurance, capacity-market and reliability contracts. H2-ready retrofits of existing gas turbines are the capital-efficient entry point, and fuel cells fit where quiet, modular, distribution-scale units make sense. Underwrite it per kW-year against the real competition, which is gas peakers with CCS or offsets, not batteries. Never build it as an arbitrage asset; the 30–40% round trip and single-digit capacity factors kill energy-market economics. For anything cycled daily or weekly, use lithium or flow batteries instead.

Fuel cells 45–60% efficient, hydrogen-capable turbines 35–60% · power-to-power round trip 30–40% including electrolysis · turbine units 30–500 MW, fuel-cell modules usually under 10 MW · fielded turbines burn 30–50% hydrogen blends, 100% on vendor roadmaps · turbine start-up 5–30 minutes · capacity factor in single digits on seasonal duty.

Intermountain Power (Utah) — 840 MW turbines running 30% hydrogen with 100% capability, fed by ACES caverns; Mitsubishi, GE Vernova, Siemens Energy H2 turbine programs; Bloom and FuelCell Energy stationary fleets; Germany's planned "H2-ready" peaker auctions.

These are capacity assets, not energy assets: the stored energy sits in a cavern at $0.20–2/kWh, so cost per kWh of capacity is never the constraint, and the revenue comes from being available (capacity markets, reliability contracts) rather than from cycling. They compete with gas peakers with offsets or CCS, not with batteries, and the question is which decarbonized firm-power option clears cheapest per kW-year. Turbine retrofits of existing gas fleets are the capital-efficient path, and they are what keeps this category alive.

3 minCan gas turbines run on hydrogen fuel? | GE VernovaGE Vernova · 10k+ views

8 minHow Do Hydrogen Fuel Cells Work?Reactions · 500k+ views

3 minHow A Bloom Energy Server WorksBloom Energy · 100k+ views

Hydrogen Energy Storage, Energy Storage Handbook Ch. 11 (Sandia) · Hydrogen and Natural Gas Turbines: Performance and NOx Control (NETL)

Ammonia is hydrogen made shippable: react hydrogen with nitrogen (Haber-Bosch) to form ammonia, which liquefies at -33 °C or modest pressure and moves through a century-old global industry of tanks, ships, and terminals handling 180 Mt a year. As a storage medium, ammonia trades conversion losses for logistics: it carries more hydrogen per cubic meter than liquid hydrogen itself and holds indefinitely in cheap steel tanks. That makes it the leading candidate for moving renewable energy between continents.

Ammonia has existing worldwide infrastructure and trade, easy bulk storage, and direct uses as fertilizer feedstock and (emerging) marine fuel, so it has a market even if power-to-power never pencils. The weaknesses: synthesis adds ~10–20% energy loss and cracking back to hydrogen adds another 15–25%, which drags full round trips to 20–35%; ammonia is toxic and its combustion raises N2O/NOx issues; Haber-Bosch runs best at steady state, which clashes with variable renewables; and green ammonia currently costs 2–4× the gray product it must displace.

Choose ammonia when energy must cross oceans, or when the offtake is ammonia itself. Fertilizer decarbonization and marine bunkering need no cracking step and no new use case, which makes them the bankable entry points. As stationary storage it makes sense only where salt geology is absent and supply distances are long enough that shipping beats pipelines. Avoid power-to-power wherever local hydrogen caverns exist; cracking loses another 15–25% on top of synthesis losses, which drops round trips to 20–35%. And avoid any scheme that requires small-scale cracking or flexible Haber-Bosch operation, since both fight the process's economics.

Liquefies at -33 °C at ambient pressure · 17.6% hydrogen by mass and about 12.7 MJ/L, more hydrogen per cubic meter than liquid hydrogen · synthesis loses 10–20% and cracking back to hydrogen another 15–25%, so power-to-power lands at 20–35% · Haber-Bosch runs at 150–300 bar and 400–500 °C · green ammonia costs 2–4× gray · existing ammonia industry about 180 Mt a year.

NEOM (destination: ammonia export), Yara and CF Industries green-ammonia retrofits, Japan's co-firing program (JERA's Hekinan coal plant), first ammonia-fueled vessels (Fortescue, NYK), Australia–Japan and Gulf–Europe supply deals.

Think of ammonia as the LNG of hydrogen: value accrues to ports, ships, and offtake contracts, not the molecule. For stationary storage it only beats local caverns where geology is absent and distance is long. The bootstrap market is decarbonizing existing ammonia demand, a 180 Mt/year sink that needs no new use case at all.

11 minThe Haber-Bosch Process: Industrial Ammonia SynthesisProfessor Dave Explains · 10k+ views

1 minHow To Produce Green AmmoniaREVOLVE · 10k+ views

12 minSIP Energy Carriers -The Innovation for Ammonia Fuel in Japan-JST Channel · 10k+ views

Ammonia Technology Roadmap (IEA) · Green Ammonia Policy Briefing (Royal Society)

These technologies store hydrogen inside room-temperature liquids. Liquid organic hydrogen carriers (LOHCs) chemically bind hydrogen to oil-like molecules (benzyltoluene, methylcyclohexane) that ship in ordinary tankers and release it on demand at a dehydrogenation plant. The broader family (e-methanol, e-methane, e-kerosene) goes further: it synthesizes drop-in fuels from hydrogen and captured CO2 that work in existing engines, pipelines, and aircraft.

These fuels are fully compatible with today's liquid-fuel logistics, need no cryogenics or pressure, and store indefinitely. E-fuels monetize as premium products (aviation mandates) rather than competing with grid storage at all. The main weakness is that each chemical conversion step loses energy: LOHC dehydrogenation is endothermic and needs a lot of heat, and e-fuel chains deliver only 10–30% of the original electricity as useful work. Carrier and catalyst costs add up, and every route needs cheap CO2 or heat integration to approach viability.

Pursue LOHC and e-fuels only where a mandate or premium fuel offtake pays for the poor efficiency: SAF quotas, maritime rules, contracted e-methanol. Underwrite them as commodity-chemicals projects with secured cheap power, CO2 supply, and heat integration for the endothermic steps. LOHC specifically makes sense when hydrogen has to move through existing liquid-fuel logistics without cryogenics or pressure. Don't deploy any of this as grid storage: at 10–30% chain efficiency it loses to every alternative. If you need seasonal power, use hydrogen caverns; if you need bulk seaborne energy, ammonia's infrastructure head start makes it the better choice.

LOHC carriers hold 5–6% hydrogen by mass, roughly 50 kg per cubic meter · dehydrogenation runs 250–320 °C and consumes 25–30% of the hydrogen's energy as heat · carriers ship and store at ambient temperature and pressure · e-fuel chains return 10–30% of the input electricity as useful work · hold time months to years.

Hydrogenious LOHC (Germany, benzyltoluene), Chiyoda's SPERA methylcyclohexane demo (Brunei–Japan shipping), HIF Global's Haru Oni e-methanol plant (Chile, Porsche offtake), Infinium and Twelve e-fuel plants, EU ReFuelEU aviation mandates as the source of demand.

As electricity storage these chains don't work; the efficiency losses are too high. As regulated fuels they are viable exactly where mandates or carbon prices force them (SAF quotas, maritime rules). Investment lens: these are commodity-chemicals businesses, and the "storage" framing makes the round trip look better than it is. Underwrite the fuel offtake, not the arbitrage.

3 minThe missing link to a clean hydrogen world: LOHC Technology - Narrated VersionHydrogenious LOHC · under 1k views

3 minHaru Oni e-fuels projectEverllence · 100k+ views

15 minThe Big Problem With Synthetic FuelsEngineering Explained · 100k+ views

LOHC Technology Explained (Hydrogenious) · The Role of E-fuels in Decarbonising Transport (IEA)

This is the technology every other entry in this sheet is measured against: containerized LFP battery systems, deployed in gigawatt-hour blocks, with costs that keep falling because EV factories keep scaling. Li-ion took frequency regulation, then 1-hour, then 4-hour markets, and is now bidding 8-hour projects. Its main weakness: energy capacity and power are coupled, so every added hour of duration means buying more cells. That's the weakness every long-duration storage technology is trying to exploit. (Chemistry detail lives in the Battery Chemistries sheet.)

Li-ion offers 88–92% round-trip efficiency, millisecond response, bankable performance data, an enormous supply chain, and fast deployment (months instead of the years a mechanical system takes). Weaknesses: $/kWh stays roughly flat with duration, so long duration gets linearly expensive; calendar and cycle fade force augmentation spend; fire codes complicate urban siting; and a 20-year asset life trails the 40–80 years of mechanical rivals.

Default to Li-ion for everything up to ~8 hours (arbitrage, ancillary services, capacity, hybrid solar) whenever deployment speed, bankability, and 88–92% round-trip efficiency matter. Assume the viable duration ceiling moves right every year, and stress-test any competitor against next year's LFP price, not today's. Plan augmentation spend into the pro forma from day one. Look elsewhere if duration runs past 10 hours or multi-day (flow, iron-air, CAES, and thermal systems decouple energy from power), if fire codes block dense urban or indoor siting (zinc and sodium chemistries), or if the asset must serve 40+ years without repowering (mechanical systems).

Round-trip efficiency 88–92% · durations 1–8 hours in current projects · response in milliseconds · cycle life 6,000–10,000 on grid LFP cells, calendar life around 20 years · system prices below $150/kWh in 2024–25, with Chinese turnkey bids under $100/kWh · deployment in months rather than years.

Tesla Megapack fleets (Moss Landing 3 GWh, Hornsdale), CATL EnerOne/TENER, Fluence, Sungrow — global installs passed ~350 GWh cumulative with China deploying the majority.

System prices fell below $150/kWh in 2024–25 (China turnkey bids under $100), which resets the bar for every challenger annually. The correct planning assumption is that Li-ion owns everything up to the crossover duration and that the crossover keeps moving right. If an LDES business case doesn't model continued Li-ion price declines, it's not a credible business case.

7 minBattery Energy Storage Systems (BESS)The Power Hub · 100k+ views

20 minBattery Energy Storage System | BESS | All Components ExplainedGaurav J - TheElectricalGuy · 50k+ views

5 minInnovation: Inside a MASSIVE Battery Storage Construction Project with Tesla MegapackMortenson · 5k+ views

Batteries and Secure Energy Transitions (IEA) · Energy Storage Cost and Performance Database (PNNL / DOE Energy Storage Grand Challenge)

Flow batteries are the electrochemical way to decouple duration from power: energy lives in electrolyte tanks and power in the stack, so you add hours by adding cheap liquid rather than cells. Vanadium systems dominate deployments (China is building at gigawatt-hour scale), while zinc-bromine, all-iron, and organic chemistries chase lower electrolyte cost. Unlimited cycle life at full depth of discharge makes flow a natural choice for deep daily cycling. (Chemistry detail lives in the Battery Chemistries sheet.)

The strengths: duration scales with tank size, cycle life runs past 15,000 cycles without fade, the systems are non-flammable, and the electrolyte retains asset value (vanadium can be leased). Weaknesses: round-trip efficiency is 65–80% after pumping losses; stacks, membranes, and plumbing keep system $/kWh above lithium at short durations; power density is low, so footprint is building-scale; and Western vendors remain subscale while Chinese state-backed projects set the price benchmarks.

Choose flow for 8–12 hour duty cycled deeply every day: solar-heavy grids with evening ramps, microgrids, and fire-sensitive or indoor sites. In those cases, 15,000+ cycles without fade or augmentation and a 20+ year electrolyte life beat lithium's repowering math. Lease the vanadium to neutralize price swings. Insist on fleet operating data (increasingly available from China's GWh-scale projects) before banking stack O&M assumptions. Avoid it below ~6 hours, where LFP wins outright, on footprint-constrained sites, and for multi-day firming, where iron-air's $20/kWh energy capacity is the better fit.

Round-trip efficiency 65–80% after pumping losses · duration set by tank size, typically 4–12 hours · cycle life past 15,000 full-depth cycles with no fade, electrolyte life 20+ years · response in seconds · vanadium price swings can move system cost 30% · crossover against lithium around 8–10 hours.

Dalian 100 MW/400 MWh VRFB and China's multi-GWh pipeline (Rongke, state grid tenders), Invinity, CellCube, ESS Inc (iron), Sumitomo Electric fleets in Japan.

Flow is the crossover argument in miniature: it loses to lithium below ~6 hours and wins on paper above ~8–10 hours, because tank costs dominate at long durations. That win depends on stack costs and O&M behaving as modeled. Vanadium price volatility can swing system cost 30%; electrolyte leasing is the standard way to manage that risk. Bankability is arriving via Chinese fleet data rather than Western projects.

9 minHow a Vanadium Redox Flow Battery Works | Sumitomo ElectricSumitomo Electric Group · 50k+ views

9 minVanadium Flow BatteriesReactions · 50k+ views

15 minWhat are redox flow batteries and why are they important?Billy Wu · 10k+ views

Redox Flow Batteries, Energy Storage Handbook Ch. 6 (Sandia) · Flow Batteries Technology Strategy Assessment (DOE Storage Innovations 2030)

Iron-air is the purest bet on multi-day storage: batteries that discharge for ~100 hours by rusting iron, then charge by un-rusting it. Iron is the cheapest energy-storage medium available in battery form, so energy capacity capex targets ~$20/kWh (no lithium roadmap comes close). The price is ~40–50% round-trip efficiency and very low power density. This is storage built to sit mostly full and carry grids through wind droughts. (Chemistry detail lives in the Battery Chemistries sheet.)

The strengths are very cheap energy capacity from a boring supply chain, non-flammable chemistry, and a natural fit for the 10–100+ hour band that lithium cannot economically reach. Weaknesses: half the energy is lost to hydrogen evolution and air-electrode overpotentials; the footprint runs to acres per megawatt; slow response limits it to energy-only services; air-electrode durability across thousands of cycles is the persistent engineering risk; and the multi-day market it serves barely exists in today's tariffs.

Deploy iron-air for 24–100+ hour firming (riding out multi-day wind droughts, deferring transmission, hardening winter reliability) where cheap land is available. Structure revenue as reliability or capacity contracts rather than merchant arbitrage, since the asset is designed to sit mostly full and cycle tens of times a year. At durations only hydrogen caverns otherwise reach, it is the option that doesn't need geology. Avoid it for daily cycling or anything RTE-sensitive: at 40–50% round trip, lithium and flow own the intraday market. Avoid it for fast services outright, since its response is too slow. And where salt caverns exist, hydrogen may still beat it at true seasonal scale.

Discharge duration up to about 100 hours · round-trip efficiency 40–50% · energy-capacity capex targeting roughly $20/kWh · footprint on the order of acres per megawatt · response in minutes rather than milliseconds · duty of tens of full cycles a year, not daily cycling.

Form Energy: Cambridge (Minnesota) first commercial install, Great River Energy and Xcel projects, Weirton (West Virginia) factory; academic alkaline-iron-electrode lineage back to NiFe.

At 100-hour duty, LCOS beats every alternative except hydrogen caverns, and it doesn't need geology. The commercial risk is timing: revenue mechanisms for multi-day firming (capacity accreditation for long-duration, reliability contracts) are only now being invented. Form's strategy of selling insurance-like reliability to utilities, rather than merchant arbitrage, matches how the asset actually behaves.

6 minA New Technology Solution for Multi-Day Energy StorageForm Energy · 5k+ views

13 minForm Energy’s Virtual Lab TourForm Energy · 10k+ views

12 minWhy Rust Batteries May Be the Future of Energy - Iron Air Battery TechnologyUndecided with Matt Ferrell · 500k+ views

Iron-Air Battery Technology (Form Energy) · Achieving the Promise of Low-Cost Long Duration Energy Storage (DOE)

These are the supporting cast of stationary electrochemistry: zinc-halide static batteries (Eos), rechargeable zinc-air, high-temperature sodium-sulfur (NGK's decades-old fleet), and the incoming wave of sodium-ion containers. Each targets a different lithium weakness: zinc goes after fire safety and non-Chinese supply chains, Na-S offers proven long-duration fleet history, and Na-ion competes on materials cost with drop-in manufacturing. (Chemistry detail lives in the Battery Chemistries sheet.)

Non-flammable aqueous chemistries allow indoor and urban siting that lithium's fire codes complicate; sodium systems avoid lithium-price exposure entirely; and Na-S carries 20 years of utility operating data. Weaknesses: every member trails lithium on efficiency, density, or cost (usually two of the three); small vendor bases create single-supplier project risk (Eos's financing saga, NGK's monopoly); and sodium-ion containers arriving from CATL-scale factories may cannibalize the niche players' safety-and-cost story before they scale.

Reach for these when siting or policy, not price, is the binding constraint: zinc's non-flammable chemistry unlocks indoor, urban, and colocation sites that lithium's fire codes rule out; Na-S suits buyers who value 20 years of utility fleet history at 6-hour durations; and domestic-content or supply-chain-diversification mandates favor all three. Before committing, make the vendor name the siting, safety, or supply-chain constraint that rules lithium out, and check that the constraint is a real requirement rather than a preference. If plain $/kWh at 2–6 hours decides, avoid the niche players. LFP today and CATL-scale sodium-ion containers tomorrow will win on price, and single-supplier vendor risk makes the bet worse.

Sodium-sulfur runs at 300–350 °C, 75–85% round trip, about 4,500 cycles over a 15-year design life · zinc-halide systems cover 3–12 hours at roughly 70–75% round trip · sodium-ion cells 100–160 Wh/kg and usable below -20 °C · typical grid durations 4–8 hours · NGK's Na-S fleet has 20 years of operating history, including 108 MW in Abu Dhabi.

Eos Z3 cubes (US utility projects, DOE loan), NGK Na-S fleets (Abu Dhabi 108 MW), BASF-NGK partnership, CATL Naxtra and BYD sodium-ion grid containers entering volume production.

The realistic outcome is segmentation: sodium-ion absorbs the cost-sensitive intraday market as a lithium sibling, while zinc and Na-S defend safety- and siting-constrained niches. For diligence, the question is always the same: what does this deliver that an LFP container at next year's price does not?

12 minHow Zinc Batteries Are Defying LimitsUndecided with Matt Ferrell · 500k+ views

3 minEnergy Storage: Battery Materials and Zinc BatteriesSandia National Labs · 10k+ views

11 minWhy CATL and BYD Are STILL Betting on Sodium-Ion Batteries.Just Have a Think · 100k+ views

Sodium-Based Battery Technologies, Energy Storage Handbook Ch. 4 (Sandia) · Rechargeable Zinc Batteries for Grid Storage, Energy Storage Handbook Ch. 5 (Sandia)

Supercapacitors store energy electrostatically, in the double layer at a carbon electrode's surface, with no chemical reaction at all. They charge and discharge in seconds, last a million cycles, and work from -40 to +65 °C, but they store an order of magnitude less energy per kilogram than batteries. Hybrid lithium-ion capacitors split the difference. A supercapacitor is a power component; it behaves like an electrical spring rather than a fuel tank.

The strengths are near-perfect efficiency, essentially unlimited cycle life, instant response, precise state-of-charge readings from voltage, and graceful aging. The weaknesses: cost runs into the thousands of $/kWh; self-discharge on the order of a percent per day scales poorly beyond minutes of duty; voltage sags linearly as the device discharges, which forces wide-range power electronics; and ever-cheaper high-power lithium cells put pressure on every application niche.

Use supercapacitors for seconds-scale power duty at extreme cycle counts or temperatures: wind-pitch backup, tram and crane regeneration, truck engine-start, ride-through bridging, grid-forming inertia. These are jobs where a million cycles from -40 to +65 °C would destroy any battery. If a load mixes sharp pulses with sustained draw, put supercapacitors in front of a battery. Size the energy for seconds to a minute or two, never more. Beyond that, avoid them: self-discharge of roughly a percent per day and costs in the thousands of $/kWh make minutes-plus duty untenable, and high-power lithium cells keep taking over the marginal niches. The recurring failure mode is picking a capacitor for a job where a battery belongs.

Specific energy 5–10 Wh/kg for double-layer cells and 10–30 Wh/kg for lithium-ion capacitors · power density 5–15 kW/kg · cycle life 500,000 to over a million · round-trip efficiency above 95% · self-discharge roughly 1% a day · operating range -40 to +65 °C · cost in the thousands of $/kWh.

Wind-turbine pitch backup (the classic design win), regenerative braking in trams and port cranes, engine-start modules for trucks, grid inertia pilots; vendors Skeleton (curved graphene), Maxwell (absorbed into Tesla and back out), Eaton, LS Materials.

This is a stable, profitable components business (a ~$1B-class market) that sells reliability and cycle life into industrial niches, not a storage market in the grid sense. Growth theses hinge on grid-forming inertia services and hybridization with batteries.

7 minWhat is a supercapacitor?Energy Storage CDT · 50k+ views

11 minSupercapacitors explained - the future of energy storage?Undecided with Matt Ferrell · 500k+ views

9 minNew Supercapacitors will replace Batteries? Stress Testing LICs (Lithium-Ion Capacitors)GreatScott! · 500k+ views

Supercapacitors Technology Strategy Assessment (DOE Storage Innovations 2030)

SMES stores energy as circulating current in a superconducting coil. There are no moving parts and no chemistry, just a persistent magnetic field. Discharge is instantaneous, and full power reversal takes milliseconds, faster than any other storage. The catch is that the coil must be held at cryogenic temperature forever. Refrigeration overhead, magnet cost, and structural forces have kept SMES confined to small, specialized, seconds-scale duty despite half a century of development.

SMES is the fastest, most cycle-proof storage that exists, with 95%+ instantaneous efficiency, and it is ideal for bridging sub-second power quality events. The weaknesses: energy density is dismal, and scaling up means storing dangerous amounts of magnetic energy; the cryogenics consume standing power, which wrecks net efficiency at any real duty cycle; superconductor and magnet costs are extreme; and cheaper flywheels, capacitors, and batteries have progressively absorbed every historical niche.

There is almost no commercial case for SMES today. Consider it only for sub-second power-quality duty where nothing else responds fast enough: millisecond ride-through at a fab-class critical load, or pulsed power for research. Even there, price flywheels and supercapacitors first, since they have absorbed nearly every historical SMES niche at a fraction of the cost. Treat the field as an option on cheap REBCO tape: if fusion-magnet demand collapses HTS costs, revisit SMES for grid inertia and pulse loads. For any energy-holding duty beyond seconds, the cryogenic standing losses disqualify it outright.

Stored energy 1–10 MJ in fielded units, roughly 0.3–3 kWh · instantaneous efficiency above 95% before cryogenic standing losses · full power reversal in milliseconds · low-temperature coils sit at 4.2 K, high-temperature designs at 20–77 K · cycle life effectively unlimited · practical duration seconds, not minutes.

1–10 MJ power-quality units historically deployed at semiconductor fabs and paper mills (American Superconductor lineage), lab-scale HTS-coil demos in Japan, Korea, and China; occasional fusion-adjacent pulsed-power studies.

Commercially dormant: the technology survives as a research field waiting on cheap high-temperature superconductor tape. If REBCO tape costs collapse (fusion-magnet demand could drive that), niche SMES could re-emerge for grid inertia and pulse loads. This is an option-value story, not a market today.

3 minThe superconducting magnetic Energy storage (VOSTFR)InstitutNEEL · 10k+ views

4 minSuperconducting Magnetic Energy Storage DemonstrationJamesJamieSMES · 10k+ views

3 minSuperconducting Magnetic Energy Storage System "SMES"ikinamo · 10k+ views

Superconducting Magnetic Energy Storage Overview (UN Climate Technology Centre) · Grid-Scale SMES with 2G-HTS Technology (OSTI / DOE)

The largest battery fleet ever built is parked in driveways: EV packs add up to terawatt-hours, and they sit idle 95% of the time. V2G discharges them back through bidirectional chargers to serve the grid; smart charging (V1G), the milder version, just times the charging. The storage capacity is effectively free, because the vehicles were bought for transport. That makes V2G the cheapest marginal kWh of flexibility in existence, if the coordination problem can be solved.

The strengths are near-zero marginal hardware cost for the energy capacity, massive scale that arrives automatically with EV adoption, and high round-trip efficiency. The weaknesses: bidirectional chargers, interconnection rules, and metering remain immature and jurisdiction-specific; consumers' cars must be plugged in, charged, and contractually available, and that kind of behavioral reliability is unproven at scale; automaker warranty and battery-wear anxieties persist (though modern data shows the impact is modest); and aggregating millions of small assets is a software-and-regulatory grind.

Start V2G with fleets whose parking is predictable and contractual (school buses, delivery depots, corporate lots), in jurisdictions where export interconnection and tariffs already pay for it. Those vehicles are parked and plugged in through the evening peak anyway, so you get near-free capacity without betting on consumer behavior. For residential programs, deploy V1G smart charging first: it captures most of the value with none of the bidirectional hardware or warranty friction. Avoid underwriting V2G as firm capacity where interconnection rules are immature or driver availability is statistical. Grid commitments that must hold on the tenth hot evening still belong to stationary batteries.

EV packs 40–100 kWh each, parked and idle roughly 95% of the time · bidirectional chargers 7–11 kW AC or 10–50 kW DC per vehicle · round-trip efficiency roughly 80–90% through the charger pair · usable duration a few hours before driving needs take priority · charger hardware a few thousand dollars per vehicle, with no marginal cost for the energy capacity.

UK's Octopus Power Pack V2G tariff, Nuvve school-bus fleets (predictable parking makes buses the easiest fleet to start with), GM Energy and Ford home-backup ecosystems, California's emergency load program precedents, Renault/Mobilize V2G in France.

The economics look great in theory (free capacity, paid twice), and the bottleneck is entirely institutional: interconnection, market access rules, and consumer trust. School buses, depots, and fleet vehicles with predictable schedules pencil first. For grid planning, even V1G smart charging captures most of the value at a fraction of the friction.

5 minHow does Vehicle to Grid (V2G) work?Best Electric Vehicle · 50k+ views

9 minBidirectional Charging (V2G, V2H & V2L) Explained: All You Need To Know!go-e International - Smart EV Charging · 5k+ views

3 minUnlocking new energy benefits of EVs with vehicle-to-grid technology | NissanNissan · 500k+ views

Vehicle-to-Grid Technology Analysis (IEA) · Vehicles-to-Grid Integration Assessment Report (DOE)

Demand flexibility is storage's accounting twin: from the grid's perspective, shifting when energy is consumed is indistinguishable from storing it. Virtual power plants aggregate thermostats, water heaters, home batteries, EV chargers, and industrial loads into dispatchable blocks. A pre-heated building or a delayed electrolyzer run works like a discharge, without needing any storage hardware. Every serious storage analysis has to price this competitor, because it sets the ceiling on what physical storage can charge.

The strengths: it has the lowest capital cost of any flexibility resource (the assets already exist), it deploys in months, and the thermal inertia in buildings and industrial processes is a vast, unmonetized battery. The weaknesses: capacity is statistical rather than guaranteed, and performance during the tenth consecutive hot evening is the perennial doubt; customer fatigue erodes participation; measurement and baselining invite gaming; and depth is bounded, since demand can shift by hours but rarely by days.

Procure demand flexibility before any physical storage. At $10–50/kW-year it is the cheapest first tranche of peak capacity on any system, it deploys in months, and every storage business case should be built net of it. Use it for hour-scale peak shaving, emergency capacity, and shaping load around renewable output, with a derated capacity credit that reflects statistical rather than guaranteed delivery. Do not lean on it for firm multi-hour commitments under sustained stress; the tenth consecutive hot evening is when enrolled megawatts evaporate. And do not use it for anything beyond hours of depth. Multi-day gaps belong to batteries, iron-air, and other physical LDES.

Delivered cost $10–50/kW-year · event duration 1–4 hours, typically a few dozen events a year · per-home contribution 0.5–2 kW from thermostats, water heaters and EV chargers · response from seconds on frequency services to 10–30 minutes on economic dispatch · deployment in months · depth bounded to hours, not days.

Tesla and Sunrun VPPs in California (ERCOT and CAISO emergency events), Octopus's Kraken platform, EnelX/CPower commercial-industrial DR, Texas's residential VPP pilots, Google Nest Renew, since discontinued.

Delivered flexibility at $10–50/kW-year undercuts any battery for the first tranche of need, which is why storage business cases should always be built net of demand response. The market failure mode is also classic: paying for promised megawatts that evaporate under stress. For investors, the moat is the customer relationship and dispatch software, not the concept.

9 minPowering a real city with a virtual power plantVerge Science · 100k+ views

5 minVirtual Power PlantEnergy Research Institute at NTU · 10k+ views

3 minThe Basics of Demand ResponseEnel X · 10k+ views

Virtual Power Plants, Real Benefits (RMI) · Unlocking the Potential of Distributed Energy Resources (IEA)