Material Insights · Understanding the Material World

Storing energy: from fuel stocks to batteries: Mechanical, thermal and chemical stores

Understanding Energy / Module 10: Storing energy: from fuel stocks to batteries Part 3 of 4

Mechanical, thermal and chemical stores

Batteries are not the only way to keep energy for later, and for long durations they may not be the cheapest. Water raised uphill, air squeezed into caverns, heat held in water, rock or salt, and hydrogen made by electrolysis all store energy. Their round-trip efficiencies range from over 90% to under 40%, and their best uses differ accordingly.

3.1 Pumped hydro is the old workhorse

Pumped hydro is simple in principle. Water is pumped from a lower reservoir to an upper one when electricity is plentiful and released through turbines when it is needed. The energy stored depends on the mass of water and the height it falls. A cubic metre falling 500 m releases about 1.4 kWh before losses (calculated), so large plants need big reservoirs and steep terrain. Modern reversible machines serve as both pump and turbine.

Smil's account stresses scale and maturity. Pumped hydro plants last 50 to 100 years, run for six hours to several days depending on reservoir size, and can switch from pumping to generating in minutes. Their round-trip efficiency is about 75 to 80%: roughly a quarter of the energy is lost in pumps, turbines and pipes. That is lower than a battery, but the energy capacity of a big reservoir is cheap, so pumped hydro suits longer durations than most batteries.

The drawbacks are geography, time and money. Sites need height, water and land; dams and reservoirs flood valleys and face opposition. Projects take five to ten years from planning to operation and need large investment up front. To ease these limits, developers are turning to closed-loop schemes that use two artificial reservoirs away from rivers, and to old mines and quarries. Seawater and underground variants have also been built or studied.

China leads construction by a wide margin, with 218 GW of the roughly 243 GW under construction worldwide at the end of 2025. India has made pumped storage a priority, with large closed-loop projects being developed by private companies and state utilities, often to back solar and wind contracts that require power to be available in the evening. Because pumped hydro sells flexibility rather than energy, its revenue depends on market rules that reward shifting power in time and providing reserves, and in many countries those rules are still being written.

Process flowchart Pumped hydro round trip
Inputs Surplus electricity100 units Process steps 1 Motor and pump

lift water to the upper reservoir; about 90% efficient

2 Upper reservoir

energy held as height; small losses to evaporation and leakage

3 Turbine and generator

water falls back through the same machines; about 90% efficient

Outputs Electricity returnedabout 75 to 80 units Source: Smil, Numbers Don't Lie: 71 Stories to Help Us Understand the Modern World (2020), via Smil Mastery Course, M7 'Grids, storage and the real cost of electricity'

3.2 Compressed air, liquid air, gravity and flywheels

Compressed-air energy storage pumps air into an underground salt cavern at high pressure and later releases it through a turbine. The difficulty is heat: compressing air makes it hot, and expanding it makes it cold. The two oldest plants, Huntorf in Germany (1978, 290 MW) and McIntosh in Alabama (1991, 110 MW for up to 26 hours), throw away the heat of compression and burn natural gas to reheat the air during discharge. Their efficiencies are about 42% and 54% respectively. Newer adiabatic designs store the compression heat in tanks of water, oil or rock and reuse it, aiming for 60 to 70%. China has built several large plants in salt caverns in recent years, and the method suits regions with suitable geology. The tile "Compressed-air and CO2 storage" covers both.

Liquid air storage cools air until it liquefies at about minus 196 C and keeps it in insulated tanks at low pressure. To discharge, the liquid is pumped, warmed and expanded through a turbine. It needs no special geology and uses standard industrial gas equipment, but its efficiency is modest, about 50 to 60% for a stand-alone plant that recycles its own heat and cold, rising if waste heat or cold from nearby industry or an LNG terminal can be used.

A CO2 variant, demonstrated in Sardinia, stores carbon dioxide as a liquid in tanks and as a gas in a large flexible dome. Because CO2 liquefies at ordinary temperatures under pressure, its developer claims a round-trip efficiency of about 75%.

Gravity storage lifts heavy blocks in towers or weights in mine shafts. The physics matches pumped hydro, but solid masses are far more expensive to lift and hold than water is to pump, so these systems store little energy for their cost. Flywheels, heavy rotors spinning in vacuum enclosures, respond in milliseconds and survive millions of cycles but hold energy only for minutes. Like supercapacitors, they help with fast frequency control rather than with shifting energy over hours.

3.3 Heat is the cheapest energy to store

Much of the energy people use ends up as heat, and heat is far cheaper to store than electricity. A tank of hot water, a bed of crushed rock or a silo of sand costs little per kWh held. Water stores about 1.16 kWh per cubic metre for every degree it is heated, so a large insulated tank warmed by 40 C holds about 46 kWh per cubic metre (calculated). Danish district heating networks use huge water tanks and lined pits to store heat for days, and in a few cases from summer to winter.

Storing heat by raising the temperature of a material is called sensible heat storage. Concentrated solar power plants use it at grid scale: they heat molten salts to about 565 C during the day and use them to raise steam after sunset, so a solar plant can keep generating into the night. In Finland, sand batteries heat sand to several hundred degrees with cheap electricity and feed the heat into district heating networks.

Latent heat storage uses the energy absorbed when a material melts or freezes. Ice storage cools office buildings: chillers make ice at night when power is cheap, and the ice melts to cool the building during the afternoon peak. Phase-change materials do the same in smaller units for buildings and cold chains.

The catch comes if the heat must be turned back into electricity. A turbine converts only about 35 to 45% of high-temperature heat into power, so Carnot batteries, which store electricity as heat and later regenerate electricity, have low round-trip efficiency unless their heat is also used directly. The best use of heat storage is therefore to supply heat: for buildings, for industrial processes, for drying and cooking. Used that way, heat stores can return over 90% of what they take in over short periods, and they can soak up surplus solar and wind power cheaply. The "Sensible heat storage", "Latent heat and cold storage" and "Carnot batteries" tiles compare them.

3.4 Hydrogen and other fuels for long gaps

For gaps of weeks or seasons, batteries become too costly, because they would be charged and discharged only a few times a year and each cycle would have to repay a large share of their capital. Here chemical fuels have an advantage: once the conversion equipment exists, extra energy capacity is cheap. Hydrogen made by electrolysis can be stored in salt caverns, as natural gas is today, for months with very small losses. Caverns in Texas and at Teesside in England have stored hydrogen for refineries and chemical plants for decades.

The weakness is efficiency. An electrolyser turns roughly 60 to 70% of the electricity into hydrogen energy; compression and storage lose a few per cent; a gas turbine or fuel cell then turns about 40 to 60% of the hydrogen back into electricity. The round trip returns roughly 25 to 40%. For every three units of surplus wind stored in spring, about one comes back in winter. Converting hydrogen into ammonia or methanol makes it easier to keep in ordinary tanks, but adds further losses.

For daily cycling that would be wasteful. But for rare, long gaps the comparison is with building enough batteries to cover a week of calm, cloudy winter weather, or keeping gas plants on standby. Studies of systems with very high shares of wind and solar usually find a role for some long-duration chemical storage, though often a modest one. Smil adds a caution from history: forecasts of a hydrogen economy, like many forecasts of fast energy transitions, have repeatedly run far ahead of what was built.

The cheapest long-term stores remain the ones that already exist. Hydro reservoirs hold months of water. Gas fields, oil tanks and coal piles hold months of fuel. Biomass can be stockpiled. As fossil fuels decline, the system will need to rebuild some of this cheap, bulky, seasonal storage in new forms, and that is one of the least solved parts of the transition. The "Hydrogen storage" and "Ammonia, methanol and e-fuels" tiles follow the options.

Process flowchart Hydrogen power to power
Inputs Surplus electricity100 units Process steps 1 Electrolyser

splits water; about 60 to 70% of the energy ends up in hydrogen

2 Compression and cavern storage

a few per cent used; months of storage with little loss

3 Gas turbine or fuel cell

about 40 to 60% of the hydrogen energy becomes electricity

Outputs Electricity returnedabout 25 to 40 units (calculated)

Heat from the electrolyser and fuel cell can be used, which raises total efficiency if there is a use for it.

Source: Pacific Northwest National Laboratory, Hydrogen (bi-directional) methodology, Energy Storage Grand Challenge Cost and Performance Assessment, https://www.pnnl.gov/sites/default/files/media/file/Hydrogen_Methodology.pdf (round-trip efficiency about 35%)

See it in the table

H₂O WaterGas Natural gasSt SteelNaCl SaltH2 HydrogenCO2 Carbon dioxideNH3 AmmoniaMeOH MethanolCSP Concentrated solar powerGrv Gravity storageCAES Compressed-air and CO2 storageLAES Liquid air storageSHS Sensible heat storagePCM Latent heat and cold storage

Check your understanding

Roughly what is the round-trip efficiency of pumped hydro?

About a quarter of the energy is lost in pumps, turbines and pipes, as Smil notes.

Why do the older compressed-air plants such as Huntorf burn natural gas?

Without stored heat the expanding air would chill the turbine, so gas is burned to reheat it, which lowers efficiency.

When is heat storage most efficient?

Heat-to-heat storage can return over 90%; turning heat back into power recovers only about 35 to 45%.
← Part 2: Batteries: chemistry, cost and wearMark read · Part 4: Choosing a store: efficiency, duration, cost and minerals →