Storage technologies and selection

1. What is being stored, and what must be delivered?

Energy storage shifts an energy service in time. Electricity may become electrochemical potential, elevated water, compressed gas, rotational motion, an electric field, heat or hydrogen. Classify the physical storage medium separately from the delivered service. A hot-water tank stores energy but does not normally deliver electricity; a hydrogen tank is not an electricity store unless the charging and reconversion equipment are included. Define input, storage and output boundaries before comparing capacity.

MW describes a rate; MWh describes a quantity. Duration E/P is meaningful only when energy and power share a usable output boundary. Response delay, ramp rate, minimum stable output, standby losses, recharge time, cycling duty and availability are additional constraints. A high energy quantity cannot replace converter power, black-start capability or a transfer-time guarantee. Compare energy delivered after losses at the same service point, including the reserve that cannot be sold twice.

Project check: Write the required service as net MW, duration, interruption limit, recharge window and evidence owner.

2. Batteries: lithium-ion is one family, not the whole category

A rechargeable battery reverses electrochemical reactions between electrodes. Lithium-ion cells move lithium ions through an electrolyte while electrons travel through the external circuit. LFP and nickel-rich lithium-ion chemistries have different material properties, but a chemistry name alone establishes neither system safety nor project life. Compare validated temperature range, power at the required SOC, capacity retention, thermal management, protection, fire-test configuration and supplier warranty conditions. Lead-acid, sodium-based and zinc-based systems have their own reaction systems and operating requirements; their performance cannot be copied from a lithium-ion datasheet.

In a flow battery, electrolyte is circulated between tanks and an electrochemical stack. Tank inventory helps determine energy capacity; stack and balance-of-plant design constrain power. This offers some separation of energy and power sizing, but pumps, membranes, electrolyte condition, crossover, shunt currents, footprint and hydraulic limits remain coupled. Distinguish all-liquid redox systems from hybrid systems involving deposited material: scaling rules are not identical. Net AC efficiency must include pumps and other auxiliaries at the stated duty.

Project check: Request a whole-system service guarantee; ask which capacity, current, SOC and temperature limits apply simultaneously.

3. Pumped hydro: store elevation, not merely water

During charging, a motor-driven pump raises water from a lower reservoir to an upper reservoir. During discharge, water flows through a turbine and generator. Reversible pump-turbine units and separate pump/turbine arrangements are possible. Conventional reservoir hydro primarily uses natural inflow; pumped storage deliberately consumes energy to restore elevated water. A plant may combine both functions, so account separately for natural-inflow generation and energy returned from pumping. Open-loop and closed-loop arrangements differ in their interaction with natural water bodies; closed-loop does not eliminate water, land or environmental obligations.

The first approximation is E = ρgHV and P = ρgHQ, where H is effective head, V is usable transferred volume and Q is flow. Convert joules to MWh by dividing by 3.6 billion. Generation efficiency reduces deliverable energy; pumping efficiency increases charging electricity. Real head changes with reservoir levels and hydraulic losses, so a single mean-head calculation is an educational screen. Geological investigations, slope stability, waterways, sediment, evaporation, ecological flow, machine operating zones and permitting can decide feasibility before a low equipment price matters.

Project check: For an island microgrid, verify surveyable head, usable volume, land and water rights, construction access and machine behaviour before treating a small pumped-hydro concept as feasible.

4. Compressed and liquid air: the thermal balance matters

CAES uses electricity to compress air into a suitable store and later expands it through machinery to produce electricity. Compression creates heat, and expansion cools the gas. A diabatic arrangement may reject compression heat and burn fuel before expansion; an adiabatic concept seeks to recover and reuse heat. An isothermal concept seeks heat exchange during compression/expansion. These are different process designs. If fuel is consumed, record it explicitly rather than attributing all generation to charging electricity or claiming a purely electrical round-trip efficiency.

Storage pressure is not an energy guarantee: pressure range, temperature, air inventory, cavern integrity and expander performance determine usable output. Liquid-air storage instead uses refrigeration to liquefy air, stores it cryogenically, and later pressurises and warms it before expansion. Cold recovery and external waste heat can change performance, but those resources require their own availability and cost evidence. Pressure systems, cryogenic materials, leakage, geology, thermal stores and process controls create project requirements quite different from a battery container.

Project check: Request a complete mass/heat/electricity balance and distinguish electricity-only operation from fuel-assisted or externally heated operation.

5. Flywheels and supercapacitors: power service is not long endurance

A flywheel stores rotational kinetic energy, E = ½Jω². A motor accelerates the rotor and a generator or reversible machine extracts energy as it slows. Usable energy depends on the permitted speed window, not energy at maximum speed alone. Bearings, vacuum equipment, motor losses and controls consume standby energy. Repeated short power events may suit this architecture, but rotor containment, overspeed protection, equipment maintenance and the required stand-by duration must be evaluated at system level.

A capacitor stores electric-field energy, with usable energy ½C(V_max² − V_min²) for an ideal constant capacitance. Supercapacitors can support high-power short-duration duties, but terminal voltage falls during discharge and the converter has an operating window. Leakage, balancing and standby losses matter. A hybrid system may let a fast device handle short transients while a battery handles energy, provided controls allocate duties and prevent duplicated reserve claims. Superconducting magnetic storage uses a magnetic field and requires a separate cryogenic and protection design.

Project check: Specify the event waveform, permitted voltage/frequency excursion, hold-up time, repetition interval and recharge priority; MW alone is insufficient.

6. Thermal storage: distinguish heat delivery from electricity return

Sensible storage changes material temperature; latent storage uses phase change; thermochemical storage uses reversible reactions or sorption processes. Water, solids and molten salts are examples of storage media, not interchangeable products. For sensible storage, Q ≈ mc_pΔT when properties can be treated as constant. Only the temperature interval capable of meeting the service is useful. Stratification, heat-exchanger approach temperature, flow and holding loss may reduce usable delivery before the store is empty.

Electric-to-heat-to-process service avoids a return-to-electricity stage. Electric-to-heat-to-power requires an engine or other conversion stage, whose losses, minimum load and cooling conditions change the result. Ice storage shifts a cooling service and cannot be added directly to an electrical backup ledger. Heat-pump COP measures heat delivered per electrical input at stated conditions; it is not the electrical round-trip efficiency of thermal storage. Continue to the Heat & Power centre for charge–hold–discharge examples and process temperature constraints.

Project check: State whether the output is MWh_e, MWh_th or cooling service, and name the meter and guaranteed delivery temperature.

7. Hydrogen: a conversion chain and a fuel inventory

Electricity can power an electrolyser, after which hydrogen is conditioned, compressed or otherwise stored and later used in a fuel cell, turbine or process. Electrolyser MW, stored hydrogen mass and generator MW are separate design variables. Producing a chemical feedstock is a different service from returning electricity to a microgrid. Evaluate water supply, purity, drying, compression, storage, leakage, ventilation, hazardous-area design and permissions alongside the main machines. Hydrogen is an energy carrier; its production pathway determines upstream energy and emissions.

Use one consistent lower- or higher-heating-value convention. For a fictional lower-heating-value basis of 33.3 kWh/kg and 50% net electrical conversion efficiency, 1,000 kg yields 16.65 MWh_e before additional excluded losses. At a fictional electrolyser consumption of 50 kWh/kg, producing that inventory needs 50 MWh_e before compression and other auxiliaries: the partial-chain electricity return is 33.3%, not 50%. These assumptions illustrate accounting only; they are not universal equipment performance or a project quote.

Project check: Ask whether long inventory duration compensates for the conversion chain, site infrastructure and low utilisation of reconversion equipment.

8. Select a service architecture, then a technology

Screen options in a defensible order: required service, local physical resources, safety and permissions, dynamic operation, recharge opportunity, then lifecycle economics. A commercial building without suitable elevation cannot acquire pumped-hydro capability by entering a favourable efficiency. An island may have elevation but lack usable reservoirs or construction access. A thermal store may reduce electric peak demand by moving heat production, yet still leave critical electrical backup unchanged. Each option must close its own first limiting condition.

Compare installed scope, charging energy or fuel, auxiliaries, standby losses, maintenance, replacements, civil works, decommissioning and residual value over a common service horizon. Cost per nameplate kWh alone ignores usable output and cycling. Revenue stacking needs simultaneous physical feasibility and contractual permission; it is not the sum of independent maximum revenues. Hybrid designs can combine a fast power layer with a longer energy layer, but require a controller, state estimation and reserve allocation that preserve critical service when forecasts or communications fail.

Project check: The existing procurement calculator is battery-only. Do not put water volume, thermal MWh or hydrogen inventory into its DC battery fields.

Independent exercises and answers

Three cases: A is a factory without usable elevation requiring 2 MW for 20 seconds; B is an island with possible reservoirs requiring 5 MW for six hours; C is a 150°C heat process seeking overnight charging. What should be checked first?