What are the main types of energy storage and how long does each last?
Storage technologies sort into three duration tiers: seconds to minutes (flywheels and supercapacitors), hours to about half a day (lithium-ion, sodium-sulfur and flow batteries), and days to seasons (pumped hydro, compressed air and hydrogen). Pumped hydro still accounts for roughly 90% of global grid storage capacity — over 170 GW — with 70%–85% round-trip efficiency and a 50–100 year operating life.
Selection starts with separating power from energy. Picture a tank and a pipe: power (kW) is the pipe diameter, setting how fast you can charge or discharge; energy (kWh) is the tank volume, setting how long you can keep going. The formula is energy (kWh) = power (kW) × time (hours). A hospital backup system needing 5 MW for two hours needs 10 MWh of capacity. Frequency regulation wants a fat pipe and a small tank; seasonal storage wants a thin pipe and an enormous tank. No single technology serves both.
The seconds-to-minutes tier belongs to flywheels. Modern systems spin in a vacuum on magnetic bearings at up to 60,000 rpm, respond in milliseconds and survive more than a million cycles with almost no capacity fade. Their weaknesses are low energy density and high self-discharge, which confines them to fast, short duties like frequency regulation. Battery energy storage systems (BESS) matter in this tier too: conventional black start relies on diesel generators or hydro plants and takes hours to tens of hours, whereas a BESS can energise the first gas turbine in milliseconds, compressing black start from hours to minutes.
The multi-hour tier is where most new capacity is being built. Lithium-ion — especially lithium iron phosphate (LFP) — dominates on cost, with LFP cycle life of 8,000–12,000 cycles; CATL's 2023 Cell-to-Pack design packs 5 MWh into a single 20-foot container with liquid cooling. Sodium-sulfur batteries run at 300–350°C with 75%–90% round-trip efficiency and suit 6–8 hour duty; Japan's NGK has built over 200 projects totalling more than 600 MWh. Vanadium redox flow batteries fully decouple energy capacity (tank size) from power capacity (stack size) so each scales independently, the vanadium electrolyte cycles indefinitely without degrading, theoretical life exceeds 25 years, and there is no thermal runaway risk. Liquid air energy storage reached commercial demonstration in 2022 when Highview Power completed a 50 MW / 250 MWh plant near Manchester.
Beyond ten hours lies long duration energy storage (LDES). Even a 12-hour BESS cannot cover a week of overcast weather or a windless winter. Iron-air batteries store energy in the rusting and reduction of iron, an extremely cheap active material, and are markedly more cost-competitive than lithium in the 100–150 hour discharge window, with the first commercial system entering testing in 2022. Nickel-hydrogen cells, once standard on NASA satellites, are being scaled by EnerVenue toward a 30-year, 30,000-cycle target. Electrolysing surplus power into hydrogen stored in salt caverns returns only 25%–40% round-trip, yet may be the only viable seasonal option where there is no suitable terrain. Gravity storage (Energy Vault's concrete-block towers) escapes geographic constraints but has very low energy density, while Rondo Energy's hot-brick storage converts renewable electricity into heat held in 1,500°C solid oxide bricks to displace industrial gas boilers — industrial heat is roughly 30% of global energy consumption.
| Technology | Typical duration | Round-trip efficiency | Lifetime and deployment status |
|---|---|---|---|
| Pumped hydro storage (PHS) | Hours to seasonal (the book's main seasonal storage workhorse) | 70%–85% | 50–100 year life, 10–15 year build time; ~90% of global grid storage capacity, over 170 GW |
| Lithium-ion (LFP / NMC) | Hours (grid BESS typically configured under 12 hours) | Higher than pumped hydro's 70%–85% | LFP 8,000–12,000 cycles; NMC 2,000–3,000 cycles |
| Vanadium redox flow battery (VRFB) | Long duration (energy and power decouple; scale by tank size) | — | Theoretical life over 25 years; vanadium electrolyte cycles indefinitely; no thermal runaway risk |
| Compressed air (CAES) | Hours to seasonal | ~42% conventional; ~70% adiabatic (AA-CAES) | Two commercial plants: Huntorf, Germany (290 MW, 1978) and McIntosh, USA (110 MW, 1991) |
| Sodium-sulfur (NaS) | 6–8 hours | 75%–90% | Operates at 300–350°C; NGK has over 200 projects and more than 600 MWh installed |
| Flywheel | Seconds to minutes | — | Over 1 million cycles with almost no fade; up to 60,000 rpm; high self-discharge |
| Thermal storage (district cooling / hot bricks) | Within-day (charge at night, discharge by day) | — | Marina Bay district cooling stores ~70,000 RT·h (about 250 MWh thermal); Rondo hot bricks at 1,500°C |
| Hydrogen in salt caverns (seasonal) | Days, weeks to seasons | 25%–40% | Still uneconomic against pumped hydro and CAES, but may be the only seasonal option without suitable terrain |
The twentieth-century grid was all pipes and no tanks — large generators supplied abundant power but stored no energy at all, so every unit of electricity had to be consumed in the same second it was produced.
Sources
- Form Energy, Multi-day Energy Storage: The Iron-Air Breakthrough (100–150 hour discharge window)
- CATL, EnerD: The World's First Liquid Cooled BESS with 5MWh+ Density
- BNEF, Battery Price Survey 2023–2024
- SP Group, District Cooling: Marina Bay (stored cooling capacity)
- The Full Spectrum: Every Energy Source Explained — A Singapore Perspective, Chapter 13.2, 13.4 and 13.8
This question is covered in depth in The Full Spectrum Every Energy Source Explained — A Singapore Perspective,第十三章 13.2 / 13.4 / 13.8