# Energy storage **Energy storage** is the capture of [[Energy|energy]] produced at one moment for use at a later one. A store is always a conversion: electricity becomes chemistry in a [[Electric_battery|battery]], height in a [[Pumped-storage_hydroelectricity|pumped-hydro]] reservoir, rotation in a [[Flywheel_energy_storage|flywheel]], heat in a tank of water, or a molecule in a tank of [[Hydrogen_storage|hydrogen]]. Because every conversion is imperfect, storage by conversion is always lossy, and the energy that comes back out is always less than the energy that went in.[^kerlin-lossy] What storage buys is not energy but timing, and the price is paid in round-trip efficiency, in mass and volume, and in capital. In the microsim below the reader works on the Ragone plane: specific energy in watt-hours per kilogram against specific power in watts per kilogram, both on logarithmic axes, with one point per technology and a family of diagonals at constant discharge time, since t = E/P. The reader sets the duration the job needs — from one second to one week — and the power it needs, and the plane answers which stores fall on the right side of the diagonal, how much each would weigh, and roughly what fraction of the input would come back out. The diagonals are the whole idea: **duration picks the store**, and a technology that is excellent for a second is often useless for a day. On the [[Energy]] flagship this article is the root of Part VII — Energy transfer, section *Energy storage* (row E62), the C42 embed shared with the flywheel and hydrogen rows that follow it. Its two closest siblings are [[Flywheel_energy_storage]], which takes the high-power corner of this same plane, and [[Hydrogen_storage]], which takes the high-energy corner. ## History The oldest stores are the ones a landscape provides for free. A mill pond holds a stream's output overnight and releases it at the wheel next morning; a woodpile holds a summer's growth until winter. The arithmetic of the mill pond is still that of the largest store ever built: gravitational potential energy is m·g·h, so a kilogram of water lifted 500 m holds 4,905 J, or 1.36 watt-hours (derived).[^murphy-mgh] Raccoon Mountain, with a 230-foot dam, a 528-acre reservoir and a 990-foot drop, delivers more than 1,500 MW on that principle, its water carrying 0.82 Wh/kg (derived).[^kerlin-pumped] Chemical storage arrived with the [[Electrochemistry|electrochemical]] cell and became quantitative when the [[Voltage|volt]] and the ampere-hour were tied together: energy is ampere-hours times volts, so a 9-volt cell rated at 0.5 Ah holds 4.5 Wh, or 16.2 kJ.[^murphy-units] The twentieth century added the [[Thermal_energy_storage|thermal]] tank, the [[Supercapacitor|supercapacitor]] and the high-speed flywheel; the twenty-first added grid-scale [[Lithium-ion_battery|lithium-ion]] and the attempt to run the whole cycle through hydrogen. ## Methods ### Outline Storage mechanisms sort by the form the energy takes while it waits, and that form fixes almost everything else — specific energy, how fast the store empties, how long it holds, and what happens when it fails. Five families cover nearly all of it: mechanical (height, rotation, compressed gas), thermal (sensible and latent heat), electrochemical (batteries and flow batteries), chemical (fuels, including hydrogen and [[Ammonia|ammonia]]), and electrical (capacitors and [[Superconducting_magnetic_energy_storage|superconducting magnets]]).[^mitofsky][^theis] A sixth question cuts across all five: whether the store is *self-discharging*. A flywheel in imperfect vacuum, a hot tank and a charged capacitor all leak over minutes to days; a reservoir, a fuel tank and a lithium cell leak over months. That timescale, not the specific energy, usually disqualifies a technology for seasonal storage. ### Mechanical Mechanical stores hold energy as position or motion. Pumped hydro holds it as height, and its power is set by head and flow together: P[W] = 84.6 × Q[ft³/s] × h[ft], with large turbine–generators reaching about 90 % or better in each direction.[^kerlin-hydro] The reason pumped storage hunts for mountains is visible in that formula — at Raccoon Mountain's 990 ft the same 1,500 MW needs about a fifth of the water a 200-foot river dam would pass. Pumped storage also pairs unevenly with generation: it suits plants that cannot cheaply throttle down at night, such as [[Wind_power|wind]], [[Geothermal_energy|geothermal]] and [[Nuclear_power|nuclear]], and suits [[Fossil_fuel|fossil]] and [[Solar_power|solar]] plants poorly.[^kerlin-pumped] Rotation is the other mechanical route. A [[Flywheel|flywheel]] holds E = ½·I·ω², and because both the energy and the stress in the rim go as the square of angular speed, the useful limit is set by material strength rather than by size — the subject of the sibling article. [[Compressed-air_energy_storage|Compressed air]] holds energy as pressure in a cavern and pays a thermal penalty: compressing a gas heats it, and unless that heat is stored too it is lost before the air is used. ### Thermal Thermal storage is the cheapest storage per joule and the least transportable. Sensible heat follows Q = m·c_p·ΔT, so a tank of water swinging 40 °F holds 320 Btu per gallon; supplying 125,000 Btu a day from such a tank needs roughly 390 gallons, about 50 cubic feet.[^kerlin-solarstore] Latent stores — ice, paraffin, a [[Salt_(chemistry)|salt]] that melts near the working temperature — pack more into the same volume because the [[Latent_heat|latent heat]] of a [[Phase_transition|phase change]] is large compared with the sensible heat over a workable temperature swing. The catch is that heat is a low-grade form. Getting work back out of a thermal store means running a [[Heat_engine|heat engine]] between it and the surroundings, and the [[Second_law_of_thermodynamics|second law]] caps that at 1 − T_L/T_H.[^yan-ch6] A solar thermal chain illustrates the cost: only the direct beam can be focused, so a net efficiency of about 0.51 × 0.30 ≈ 0.15 is realistic, and a thermal power plant typically rejects two and a half to three times its electrical output as waste heat.[^kerlin-solarchain] ### Electrochemical A battery stores energy in the free-energy difference between two [[Chemical_reaction|chemical]] states held apart by an [[Electrolyte|electrolyte]] that conducts ions but not electrons. Its cell voltage comes from ΔG = −n·F·V_eq, with the Faraday constant F ≈ 96,485 C per mole of electrons, so energy per unit mass is fixed by the reaction and by how much inert material — casing, collectors, separator — rides along with it.[^haverkort-electro][^haverkort-batteries] Electrochemical stores dominate the middle of the Ragone plane because they are the only family that is good at both energy and power without being good at either extreme. Their characteristic failure is ageing, not leakage: the reaction that stores the charge slowly changes the electrodes, so capacity and power both fade with cycling. Flow batteries answer that by moving the active material into external tanks, decoupling the energy rating from the power rating — tanks set one, stack the other.[^haverkort-batteries] ### Chemical Chemical storage means making a fuel. [[Electrolysis_of_water|Water electrolysis]] is the cleanest example: ΔG° = 237 kJ/mol gives an equilibrium voltage of 1.23 V, while ΔH° = 286 kJ/mol corresponds to a thermoneutral voltage of 1.48 V, so a cell driven at 1.8 V is rejecting n·F·(1.8 − 1.48) = 61.7 kJ of heat per mole of [[Hydrogen|hydrogen]] made (derived).[^haverkort-electro] Running the chain backwards through a [[Fuel_cell|fuel cell]] can recover at most ΔG/ΔH ≈ 85 % of the fuel's enthalpy as work, so an optimistic enthalpy-basis round trip is (1.48/1.8) × 0.85 ≈ 0.70, and real chains land well below that (derived; ILLUSTRATIVE). What the chemical route buys is specific energy. Hydrogen carries about 33 kWh per kilogram, some 24,000 times the 1.36 Wh/kg of water lifted half a kilometre (derived).[^murphy-app] Even the carbon in an ordinary fuel does well: complete combustion releases 394 kJ per 12 g of carbon, which is 32.8 MJ/kg, or 9.1 kWh/kg (derived).[^murphy-units] That gap — four orders of magnitude between a gravitational store and a molecular one — is why fuels won transport and reservoirs won the grid. ### Electrical methods A [[Capacitor|capacitor]] stores energy directly in an electric field, a superconducting magnet directly in a magnetic one. Neither converts anything, which is why both are fast: a [[Supercapacitor|supercapacitor]] empties in seconds and cycles a million times without the chemical ageing a battery suffers. The price is specific energy an order of magnitude or more below a battery's and, for the magnet, a [[Cryogenics|cryogenic]] plant drawing power whether or not the store is used. ## Applications The list below runs roughly by the duration each application needs — the axis the microsim puts under the reader's hand. ### Mills The water mill is the original store-and-release machine, and the [[Flywheel|flywheel]] on its shaft the original short-duration one, smoothing the stroke-to-stroke variation of a saw or a hammer. The pond covers hours, the flywheel a fraction of a second; both survive, scaled up, in the modern grid. ### Homes Domestic storage is mostly thermal and mostly invisible: a hot-water tank is a store, and so is a well-insulated house. The solar-thermal sizing case makes the point — about 390 gallons carries a day's 125,000 Btu at a 40 °F swing, a cheap store by any electrical standard.[^kerlin-solarstore] Home batteries answer a different question, covering outages and shifting rooftop [[Solar_cell|solar]] into the evening. ### Grid electricity and power stations Grid storage is where duration matters most, and pumped hydro still holds the overwhelming majority of installed capacity. The United States has roughly 2,400 large dams totalling about 35,400 MW, some 0.9 quad a year or 6.5 % of national electricity, and pumped-storage plants use the same hydraulics.[^kerlin-hydro] Gas stored underground in porous formations obeys the reservoir engineering that governs petroleum production, which is why depleted fields are the preferred sites.[^zeidouni] [[Grid_energy_storage|Grid storage]] has its own dense child article on this flagship. ### Air conditioning Cooling is where storing the *cold* rather than the electricity wins. Making ice at night and melting it through the afternoon shifts load off the peak without touching the building's wiring, and it exploits latent heat: melting [[Ice|ice]] absorbs far more per kilogram than warming water through any tolerable range. The same logic runs in reverse for a [[Heat_pump|heat pump]] with a buffer tank. ### Transport Transport is the application fuels won, for the reason the Chemical section gives: a vehicle carries its store, so specific energy dominates. Electrification changes the accounting, not the physics — a [[Electric_vehicle|battery vehicle]] trades specific energy for round-trip efficiency and grid refilling, while a [[Hydrogen_vehicle|hydrogen vehicle]] keeps a fuel's specific energy and pays in tank mass and chain losses. ### Electronics At the small end the constraint inverts: what matters is shelf life and predictable delivery at low current, not specific energy. A 9 V cell holding 4.5 Wh runs a one-watt load for four and a half hours — and will still do it after years in a drawer.[^murphy-units] [[Capacitor|Capacitors]] handle the microsecond end of the same board. ## Use cases Storage is bought for one of four jobs, and each sits on a different diagonal of the Ragone plane. *Power quality* covers milliseconds to seconds — a voltage sag ridden through by a flywheel or a supercapacitor. *Bridging* covers seconds to minutes, holding a load up while a standby generator starts. *Energy management* covers hours, moving cheap night energy into an expensive afternoon, which is what pumped hydro was built for. *Seasonal storage* covers months, which no electrical or mechanical store can reach and only a fuel can. The mismatch between those durations and the technologies is what the sim makes visual. A store good for a second at high power is typically hopeless at a week, because a week of holding demands either no self-discharge or an enormous quantity of energy. The human scale calibrates it: a person eating 2,000 kcal a day is a continuous 97 W load, while the average American consumes about 10,000 W of primary power, or roughly 240 kWh a day.[^murphy-units] ## Capacity Capacity has two numbers, not one, and confusing them is the commonest error in the subject. *Energy capacity*, in kWh, says how much the store holds; *power capacity*, in kW, how fast it can be emptied. Their ratio is the discharge time t = E/P — the diagonal the reader moves along in the sim. | Store | Specific energy | Specific power | Typical duration | |---|---|---|---| | Supercapacitor | very low | very high | seconds | | Flywheel | low | very high | seconds to minutes | | Lithium-ion | moderate | high | hours | | Lead–acid | low | moderate | hours | | Compressed air | low | moderate | hours | | Pumped hydro | 1.36 Wh/kg of water at 500 m (derived) | set by turbine size | hours to days | | Hydrogen | ≈33 kWh/kg of fuel (tank excluded) | set by the fuel cell | days to seasons | | Hydrocarbon fuel | ≥9.1 kWh/kg of carbon (derived) | set by the engine | seasons | *ILLUSTRATIVE. The two derived rows are computed here from m·g·h and from the cited combustion figure; the rest are the sim's external reference point set, given as bands because every real product sits inside a wide range. The Portal Books carry the mechanisms, not a comparative table.*[^murphy-mgh][^murphy-units][^murphy-app][^haverkort-batteries] A third number, the [[Capacity_factor|capacity factor]], connects storage to the generation it serves: a 2.5 MW wind machine at a 30 % capacity factor produces 6,570,000 kWh in a year, and it is the *shape* of that output, not its total, that creates the demand for storage.[^kerlin-wind] ## Economics Storage never makes energy; it moves it in time and loses some in the move.[^kerlin-lossy] A store therefore earns its keep only where the value of energy differs enough between two moments to cover the round-trip loss and the capital. That is why the economics turn on cycles rather than capacity: a store filled and emptied daily amortises its cost over thousands of cycles, while a seasonal store gets one or two a year and must be correspondingly cheap per unit held. The plane's two axes translate into two prices — dollars per kilowatt of power capacity, set by the converter (turbine, inverter, fuel-cell stack), and dollars per kilowatt-hour of energy capacity, set by the medium (reservoir, cell stack, tank). Flow batteries and pumped hydro can be sized independently along each axis; a sealed battery cell cannot.[^haverkort-batteries][^theis] ## Research The pair groups current work by country. What follows is the physics each line turns on; the institutional detail — programmes, consortia, sites — belongs to the pair. ### Germany The German subsection is about power-to-gas: converting surplus electricity to hydrogen, and sometimes on to methane or ammonia, for storage in the existing gas system. Everything about whether it pays is in the electrolyser voltage. At the thermoneutral 1.48 V the cell neither absorbs nor rejects heat; below it the cell actually draws heat from its surroundings; above it the excess is waste, 61.7 kJ per mole of hydrogen at 1.8 V (derived).[^haverkort-electro] Pushing cells closer to 1.48 V at high current density is the research problem. ### United States The United States subsection is about grid-scale [[Electric_battery|batteries]] and the chemistry that will succeed lithium-ion. The constraint is the Electrochemical section's: specific energy is set by the reaction and diluted by every inert component carried with it, so gains come from replacing inert mass rather than from re-engineering the same cell.[^kerlin-batteries][^haverkort-batteries] Pumped hydro remains the incumbent to displace.[^kerlin-hydro] ### United Kingdom The United Kingdom subsection collects demonstration projects whose common thread is duration rather than power: an island grid with a large variable-renewable share needs stores that hold for days. That points to compressed and liquefied gases, to the remaining pumped-hydro sites, and to hydrogen — each a bet that a low cost per kilowatt-hour of capacity beats a high round-trip efficiency.[^theis] ## See also - [[Supercapacitor]] - [[Compressed-air_energy_storage]] - [[Thermal_energy_storage]] - [[Grid_energy_storage]] - [[Flywheel_energy_storage]] - [[Hydrogen_storage]] - [[Pumped-storage_hydroelectricity]] - [[Energy_density]] ## References [^kerlin-lossy]: Kerlin, Thomas (2013). *Future Energy: Opportunities & Challenges*. Chapter 3, p. 188 ("storage by conversion is always lossy"). Portal Book 048, https://open.umn.edu/opentextbooks/textbooks/future-energy-opportunities-challenges [^kerlin-pumped]: Kerlin (2013), *Future Energy*, Chapter 6, pp. 240–241 (Raccoon Mountain: a 230 ft dam, a 528-acre reservoir, a 990 ft drop and more than 1,500 MWe; p. 240 on the generation types pumped storage does and does not pair well with). Portal Book 048. [^kerlin-hydro]: Kerlin (2013), *Future Energy*, Chapter 6, pp. 236–239 (Eq. 9-1 `P = g·ṁ·h` and the unit forms `P[W] = 84.6·Q[ft³/s]·h[ft]` and `P[W] = 9.81·ṁ[kg/s]·h[m]`; large turbine–generators at about 90 % or better; ≈2,400 large US dams totalling ≈35,400 MWe, ≈0.9 quad/yr, ≈6.5 % of US electricity). Portal Book 048. [^kerlin-solarstore]: Kerlin (2013), *Future Energy*, Chapter 4, p. 181 (sensible storage `Q = m·c_p·ΔT` with c_p = 1 Btu/(lb·°F) and 8 lb/gal: a 40 °F swing gives 320 Btu/gal, so 125,000 Btu/day needs ≈390 gal ≈ 50 ft³). Portal Book 048. [^kerlin-solarchain]: Kerlin (2013), *Future Energy*, Chapter 4, p. 185 (only the unscattered direct beam can be focused, so a solar heat-engine chain runs at about 0.51 × 0.30 ≈ 0.15; a thermal plant rejects 2.5–3 times its electrical output, which corresponds to η ≈ 25–29 %, computed). Portal Book 048. [^kerlin-wind]: Kerlin (2013), *Future Energy*, Chapter 5, p. 229 (one 2.5 MWe machine at a 30 % capacity factor over 8,760 h gives 6,570,000 kWh/yr). Portal Book 048. [^kerlin-batteries]: Kerlin (2013), *Future Energy*, Chapter 9, pp. 354–466 (batteries and storage for transport; page to pin). Portal Book 048. [^haverkort-electro]: Haverkort, Willem (2024). *Electrolysers, Fuel Cells and Batteries: Analytical Modelling*. Chapter 1, "Electrochemistry", pp. 25–28 (ΔG = −n·F·V_eq with F ≈ 96,485 C/mol e⁻; 237 kJ/mol → 1.23 V; 286 kJ/mol → 1.48 V; the cell absorbs heat below V_tn and rejects it above; voltage efficiency φ_e = V_eq/V_cell; p. 27 gives ΔG/ΔH ≈ 85 % as a fuel cell's ceiling and ΔH/ΔG ≈ 118 % for electrolysis). The 61.7 kJ per mole at 1.8 V is n·F·(1.8 − 1.48), computed. Portal Book 053, https://open.umn.edu/opentextbooks/textbooks/electrolysers-fuel-cells-and-batteries-analytical-modelling [^haverkort-batteries]: Haverkort (2024), Chapter 7, "Batteries", pp. 110–129 (cell construction, energy density and the split between energy-rated and power-rated components; page to pin). Portal Book 053. [^murphy-units]: Murphy, Thomas (2021). *Energy and Human Ambitions on a Finite Planet*. Chapter 5, pp. 92–98 (1 kWh = 3.6 MJ; ampere-hours × volts = watt-hours; a 9 V, 0.5 Ah cell holds 4.5 Wh = 16.2 kJ and runs a 1 W load for 4.5 h; 2,000 kcal/day = 96.85 W, "very close to 100 W"; the average US citizen ≈10,000 W; combustion releases 394 kJ per 12 g of carbon, ≈4 eV per atom). The 32.8 MJ/kg and 9.1 kWh/kg figures are computed from the 394 kJ per 12 g. Portal Book 097, https://open.umn.edu/opentextbooks/textbooks/energy-and-human-ambitions-on-a-finite-planet [^murphy-mgh]: Murphy (2021), Chapter 5, pp. 89–91 (Table 5.2's list of energy forms: `m·g·h` for hydro and tidal, `½·m·v²` for wind and current, `h·ν` for solar, `H − T·S` for chemical, `c_p·m·ΔT` for thermal, `q·V` for electric, `m·c²` for nuclear; the falling-apple potential/kinetic ledger). The 1.36 Wh/kg at 500 m and 0.82 Wh/kg at 990 ft are computed from `m·g·h`. Portal Book 097. [^murphy-app]: Murphy (2021), Appendices, pp. 378–431 (battery, gravitational and kinetic storage; the ≈33 kWh/kg energy content of hydrogen; page to pin). Portal Book 097. [^theis]: Theis, Tom; Tomkin, Jonathan, eds. (2015). *Sustainability: A Comprehensive Foundation*. Chapter 10, "Sustainable Energy Systems" (storage in the context of variable renewable supply; page to pin). Portal Book 098, https://open.umn.edu/opentextbooks/textbooks/sustainability-a-comprehensive-foundation [^mitofsky]: Mitofsky, Andrea (2018). *Direct Energy*. Part I, "Survey of Energy Conversion Devices", pp. 33–254 (the device-by-device survey that sorts conversion and storage by the form the energy takes; page to pin). Portal Book 055, https://open.umn.edu/opentextbooks/textbooks/direct-energy [^yan-ch6]: Yan, Claire Yu (2022). *Introduction to Engineering Thermodynamics*. Chapter 6, "Entropy and the Second Law of Thermodynamics", pp. 239–348, especially p. 272 (η = 1 − T_L/T_H and the impossibility of exceeding it). Portal Book 115, https://open.umn.edu/opentextbooks/textbooks/introduction-to-engineering-thermodynamics [^zeidouni]: Zeidouni, Mehdi (2025). *Petroleum Reservoir Dynamics*. Chapter 3, pp. 2–167 (flow and storage in porous formations; page to pin). Portal Book 084, https://open.umn.edu/opentextbooks/textbooks/petroleum-reservoir-dynamics ## Further reading - Kerlin, *Future Energy: Opportunities & Challenges* (2013), Chapter 6 (hydro and pumped storage) and Chapter 9 (batteries) — Portal Book 048. - Murphy, *Energy and Human Ambitions on a Finite Planet* (2021), Chapter 5 (units and scale) and the Appendices (storage mechanisms) — Portal Book 097. - Haverkort, *Electrolysers, Fuel Cells and Batteries: Analytical Modelling* (2024), Chapter 7 (batteries) and Chapter 11 (electrolysers) — Portal Book 053. - Mitofsky, *Direct Energy* (2018), Part I, the survey of conversion devices — Portal Book 055. - Theis and Tomkin, eds., *Sustainability: A Comprehensive Foundation* (2015), Chapter 10 — Portal Book 098. ## External links - [Future Energy: Opportunities & Challenges](https://open.umn.edu/opentextbooks/textbooks/future-energy-opportunities-challenges) (Kerlin, 2013) — Portal Book 048 - [Energy and Human Ambitions on a Finite Planet](https://open.umn.edu/opentextbooks/textbooks/energy-and-human-ambitions-on-a-finite-planet) (Murphy, 2021) — Portal Book 097 - [Electrolysers, Fuel Cells and Batteries: Analytical Modelling](https://open.umn.edu/opentextbooks/textbooks/electrolysers-fuel-cells-and-batteries-analytical-modelling) (Haverkort, 2024) — Portal Book 053 - The Wikipedia pair's external links list agency statistics and project databases <!-- MATTERSIM:BEGIN g33 — Matter & Energy Cluster microsim (framework build, specs/sims/Energy_storage.json); do not hand-edit inside --> **Microsim — three.js (Wikitube framework):** *Energy storage* <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/matter/Energy_storage.html" data-title="Energy storage"></div> *Built from `MICROSIM_GUIDE/specs/sims/Energy_storage.json`; part of the [[PORTAL_Matter|Matter portal]] spine (section sims and See-also variants).* <!-- MATTERSIM:END --> ## Wikipedia : Wikitube **Strict pair:** [Wikipedia](https://en.wikipedia.org/wiki/Energy_storage) : [Wikitube](https://en.wikitube.io/wiki/Energy_storage) · pinned revision [1371474530](https://en.wikipedia.org/w/index.php?oldid=1371474530) · 2026-09-11 ## Previous hub tags Hubs: `Life_Physics`. Portals: [[PORTAL_Energy]]. --- *Matter & Energy Cluster child articles, wave 1 · 2026-09-11 · drafted · Energy row E62 · sim pending (matter/Energy_storage).*