Energy Storage: Moving Energy Through Time
The thing to understand about storage is that there is no such thing as a best technology, only a best technology for a given duration. A device that can deliver enormous power for two seconds and one that can deliver modest power for two months are both called storage, and they have almost nothing in common.
Why Duration Is the Organising Principle
A storage device has two independent ratings. Power, in watts, is how fast it can deliver energy. Capacity, in watt-hours, is how much it holds. Their ratio is duration, and it decides everything about what the device is for.
At the shortest end, milliseconds to seconds, the task is stability. Grids run at a fixed frequency, and any mismatch between generation and demand pushes that frequency off target within a fraction of a second. Supercapacitors and flywheels respond almost instantly and are used to hold the frequency steady. They store very little energy; that is not their purpose.
In the hours range the task is arbitrage and peak shaving: absorb solar output at midday, release it in the evening. This is where lithium-ion has won decisively, and where most of the world's new storage investment goes.
From days to months the task is seasonal balancing - covering a fortnight of still, overcast winter weather in a grid that depends on wind and sun. Here every electrochemical option becomes prohibitive, because the cost scales with the energy stored and the device sits idle most of the year. This range remains the open problem.
The Technologies and What Each Is Good At
Lithium-ion batteries store energy by moving lithium ions between two electrodes. They achieve round-trip efficiencies of 85 to 95 percent, respond in milliseconds, and have fallen in cost by roughly 90 percent since 2010 - driven by electric vehicles rather than by the grid, which then inherited the manufacturing scale. Their limitation is duration: beyond about eight hours the economics deteriorate, because every additional hour requires proportionally more cells.
Pumped hydro stores energy as water at height, and its dominance is easy to miss because it is old and unglamorous. It accounts for over 90 percent of installed grid storage energy worldwide, offers 70 to 85 percent round-trip efficiency, and lasts 50 to 100 years. Its constraint is geography, as described under hydropower - two reservoirs, a height difference, and permission to build.
Flow batteries separate power from capacity by storing liquid electrolyte in external tanks. Want more energy? Use bigger tanks. This makes them attractive for durations of eight to twelve hours where lithium-ion falters, though energy density is poor and the technology is far less mature.
Thermal storage is often overlooked because it does not return electricity. Molten salt in concentrated solar plants, hot water in district heating, and increasingly heated rock or sand for industrial process heat all store energy at very low cost per kilowatt-hour. Since roughly half of final energy demand is heat rather than electricity, storing heat as heat avoids two conversions and their losses.
Compressed air, gravity systems lifting solid masses, and hydrogen as a chemical store each occupy specific corners. Hydrogen's 30 to 40 percent round-trip efficiency is poor, but it is one of the few options that can hold energy for months without loss.
Round-Trip Efficiency and What It Does Not Tell You
Round-trip efficiency is the fraction of energy returned after a complete store-and-release cycle. Lithium-ion achieves around 90 percent, pumped hydro 70 to 85, compressed air 40 to 70, and hydrogen 30 to 40. Read alone, that ordering suggests hydrogen is simply bad.
It is the wrong way to read it. Efficiency matters in proportion to how often the device cycles and how expensive the input energy is. A battery cycling daily loses 10 percent of a lot of energy, 365 times a year, so efficiency dominates its economics. A seasonal store cycling once a year loses 65 percent of a comparatively small amount, once - and the energy it stored was surplus that would otherwise have been curtailed and wasted entirely.
The relevant comparison for long-duration storage is not against a battery but against the alternative of building more generation, or of running a gas plant for those two weeks. Against that benchmark an inefficient store can still be the cheaper answer.
Self-discharge matters for the same reason. A lithium-ion cell loses a few percent per month, which is irrelevant over a day and fatal over a season. Pumped hydro loses to evaporation. Hydrogen in a salt cavern loses almost nothing, which is precisely why it is discussed for seasonal use despite its efficiency.
The Unsolved Range
A grid running largely on wind and solar faces a specific worst case: a prolonged period of low wind and low sun, in winter, across a wide region. Northern Europe experiences these regularly, and they can last a week or more. Covering them requires storage measured in terawatt-hours, at a scale no battery fleet approaches.
The candidates all involve accepting poor efficiency in exchange for cheap capacity. Hydrogen in salt caverns is the most developed: the storage medium costs almost nothing once the cavern exists, and the energy can sit for months. Thermal stores in rock or molten salt are being built at industrial scale. Synthetic methane made from hydrogen and captured carbon dioxide could use the existing gas network unchanged, at the cost of another conversion step.
There is a second answer that avoids the problem rather than solving it, and it is why storage and generation cannot be discussed separately. A source that produces continuously regardless of weather removes much of the demand for storage in the first place - which is the structural role that nuclear fission, geothermal and research into continuously available ambient energy all occupy from different directions.
In practice the answer will be a portfolio. Flywheels for the milliseconds, lithium-ion for the hours, pumped hydro and flow batteries for the days, something chemical or thermal for the weeks, and enough always-on generation that the weeks are rare. No single technology covers the range, and the useful question about any storage proposal is which part of the range it is for.
Frequently asked questions
Why can't one storage technology do everything?
Because power and capacity are separate ratings and the cost structures differ. A flywheel delivers enormous power for seconds; a hydrogen cavern holds energy for months but returns only a third of it. Optimising for one duration makes a device poor at the others, which is why grids use several technologies together.
What is round-trip efficiency?
The fraction of energy you get back after storing and retrieving it. Lithium-ion returns about 90 percent, pumped hydro 70 to 85, hydrogen 30 to 40. Efficiency matters most for devices that cycle frequently; for a store used once a year on surplus energy that would otherwise be wasted, it matters much less.
Is pumped hydro really bigger than all batteries?
By stored energy, yes, and by a wide margin. Pumped hydro accounts for more than 90 percent of the world's grid-scale storage energy, with capacity measured in hundreds of gigawatt-hours. It is old, geographically constrained and easy to overlook, but it remains the dominant store on every continent that has suitable sites.
Why is seasonal storage so difficult?
Because the cost of an electrochemical store scales with the energy it holds, and a seasonal store holds a great deal while cycling perhaps once a year. Spreading the capital cost over one cycle instead of 365 makes the economics collapse. The viable candidates - hydrogen in caverns, thermal stores - are cheap per unit of energy precisely because they are inefficient in other ways.
Does more renewable generation always mean more storage?
Up to a point. Modest shares of wind and solar can be absorbed by existing grid flexibility. Beyond roughly 50 to 60 percent, storage and transmission requirements rise steeply, because the mismatch periods grow longer. Generation that runs independently of weather reduces that requirement rather than adding to it.
