The concept of storing energy with a heat pump and the engine has been developed for several generations since Marguerre first introduced a system consisting of two Thermal Energy Storage Systems (TESs) filled with wet steam in 1924 [14]. Two patents were filed in the 1930s that described Marguerre’s system in English [15].

In 1972, Babcock [16] first patented a process that uses nuclear power plants’ superheated steam. At 450 °C and 200 bars, a ceramic refractory accumulator was employed to compress and store the steam. A second early heat pump system was described in 1977 by Smith [17]. In the process of charging, the air is compressed, cooled, and then liquefied. Regenerators or packed beds are used to store the energy extracted during cooling, while tanks are used to store the liquefied air produced by the process. An open supercritical system is discharged by the Rankine cycle; compressed liquid air is reheated in a regenerator and then expanded to produce work through a turbine. During the charging process, the cool air obtained from the process is stored in the cool store and used for the liquefaction process. Similar to Babcock’s approach, a system was developed by Cahn [16] in 1978. Cahn proposed an energy system that could use thermal waste and would be independent of a power plant. In contrast to solid stores such as Babcock’s, the Cahn design does not have thermal fronts (a naturally unsteady phenomenon that reduces the maximum capacity utilisation of the tanks, and enhances self-discharge losses during storage, making the operation of the system relatively complex) and does not experience conductive losses.

The Highview Power system was developed between 2011 and 2014 and was a precursor to the LAES [18], as described by Morgan et al. [19]. However, there are some differences between these systems including: (i) compression–expansion processes (adiabatic or isothermal); (ii) liquefication processes (Claude cycle or Linde cycle); (iii) the number of compression and expansion phases; as well as the manner in which energy is recuperated. According to [20], Highview Power’s prototype was only 8% efficient; this value is low due to the small size of the plant and the ineffective use of the regenerator.

Park et al. [21] stated that the Highview design was constrained by pinch points that restricted the RTE (round-trip efficiency) to 36%. It is expected that further enhancements of  the system will result in efficiency levels that exceed 50%. Accordingly, LAES may be able to achieve similar efficiencies to other PTES systems and also have the advantages of using proven technologies, geographical flexibility, and abundant storage media [21].

Joule–Brayton cycles and Rankine cycles have been used in recent studies to develop PTES systems. As part of the ISENTROPIC project, construction of a prototype based on a Joule–Brayton cycle began in 2015, while ABB is interested in building a prototype based on a Rankine cycle [21]. In the commercial sector, a number of PTES solutions exist or are being developed. These include the Brayton PTES developed by Isentropic [15], the transcritical PTES developed by MAN Energy Solutions [22] and the Rankine PTES designed by Malta. While development and research are currently at an early stage, some controversial results regarding key performance indicators, such as round-trip efficiency, have been reported. There has been no systematic investigation of how thermal energy storage impacts system performance; studies are urgently needed to maximise the potential of PTES [23]. The PTES development timeline is illustrated in Figure 2.

The PTES systems can be broadly categorized into three groups based on the type of cycle they incorporate. Within each category, several variations have been proposed, including Joule–Brayton cycles, Rankine cycles, and transcritical cycles. Figure 3 demonstrates the difference between the three types of PTES thermodynamic cycles.