Thermal energy storage

IRENA, quoting the European Association for Storage of Energy, defines thermal energy storage as "the temporary storage of energy by heating or cooling a storage medium so that the stored energy can be used at a later time" for power generation, heating or cooling. The medium can be water in a tank, rock or ceramics, a phase change material, or molten salt.

IRENA's 2020 outlook groups 13 TES technologies into four families. Sensible storage raises or lowers the temperature of a material. Latent storage uses the heat absorbed or released when a material melts or freezes. Thermochemical storage keeps energy in a reversible chemical reaction, and mechanical-thermal systems combine heat storage with compressed or liquefied air. The same report counts about 234 GWh of TES installed worldwide in 2019 and expects the market to more than triple, to over 800 GWh, by 2030.

Molten salt is the most widely used TES technology in the power sector, mainly because concentrated solar power (CSP) plants use it. IRENA puts installed molten-salt capacity at over 21 GWh, with Spain, the United States and South Africa leading at the end of 2019. The thermal stability of the salts limits the maximum operating temperature to 565°C.

Industrial heat is a newer use. A unit charges from the grid through electric heaters when power is cheap or in surplus, keeps the heat in insulated tanks, and discharges through a steam generator when a process or a district heating network needs it. Kyoto Group's Heatcube works this way: its standard 52 MWh configuration delivers steam at 150-300°C and the manufacturer quotes round-trip efficiency above 93% at one cycle per day. The two efficiency figures on this page describe different things: IRENA's over 98% in the table below is for molten-salt storage as a technology class, while 93% is Kyoto's figure for a complete Heatcube unit, which charges through electric heaters and discharges through a steam generator.

A thermal store is a set of tanks, pumps, valves, heaters and a steam generator, and somebody has to decide every few minutes what each of them does. The software around it usually has four layers, and you should know which ones you are buying.

The first is control and safety: the plant's control system, which Kyoto calls the BMS of its thermal battery, and the SCADA screens operators use. It reads tank temperatures and levels, pump and valve states and heater power, and it keeps the salt inside its operating range. A molten-salt store cannot be allowed to cool below its lower temperature limit, so heater control, alarms and safe shutdown sequences belong in this layer, and they need testing on real hardware as well as in simulation.

The second is the interface to the site. Kyoto's Heatcube connects its BMS to the customer's energy management system (EMS) through a standard industry interface. Your tender should name that interface and the signals that cross it: available heat in MWh, maximum charge and discharge power, steam demand, alarms. If the store will also offer grid services, response time matters too. Heatcube's charge response is under 3 seconds.

The third is optimization. The value of TES comes from moving electricity consumption from expensive hours to cheap ones, so the scheduler needs day-ahead and intraday prices, a weather forecast and a forecast of heat demand. Market data feeds fail and change format, so ingestion needs validation, retries and a fallback schedule.

The fourth is data and simulation: historian, dashboards and often a digital twin used to test charge strategies and to size the next plant before it is built.

Molten salt also appears in electrochemical cells, a different technology from the heat stores described here. The article on molten salt batteries covers both.

We worked with Kyoto Group for three and a half years on industrial IoT and digital twin software for its molten-salt thermal storage, taking it from a UX/UI MVP to a SCADA-based system that monitors and controls the flow of energy in real time.

In the 2024 phase the scope included what Kyoto calls the BMS of its thermal battery: control of the valves, pumps and the flow of hot fluid in the salt store. Around it the team worked on secure cloud infrastructure for the plant's devices and sensors, surveillance with infrared cameras and image recognition, and management dashboards. The system also collected weather data and analyzed energy prices. The stack was React.js, Python and AWS, with Terragrunt and Jenkins for infrastructure. Two lessons from that work carry over to any TES project. Market data APIs, ENTSO-E among them, returned inconsistent file formats and were not always available, so the price feed needed validation and a fallback. Data quality from the plant also has to be settled at the start, because the simulation and the scheduler depend on it.