When creating a digital twin of a building, all engineering system devices within the building must be digitised and integrated into the numerical model. This research focused on describing the processes occurring in a heat storage system, which was designed to accumulate heat during the summer period and utilise it by the building’s heating during the cold period. This is a long-term acting heat storage system with a soil-type filler. Several stages of the thermal energy storage system operation were distinguished: charging, discharging, and heat retention. During all these processes, heat exchange with the environment occurs. Typically, it is heat loss to the environment. Experimental research was performed in field conditions to analyse the mentioned processes. After each charge, the heat was dissipated in the soil volume. The output signals of the sensors were recorded and analysed. It was noticed that two temperature measurement sensors are necessary to estimate the type of work regime and charge or discharge intensity. More sensors are needed to determine the amount of stored energy more accurately. Creating a validated numerical model and comparing measured temperatures with simulated values at the same points enables a highly accurate assessment of the stored energy in the accumulator, as well as the description and forecast of parameter changes. By integrating the numerical model of the accumulator into the building’s digital twin and combining it with building engineering systems models, it is possible to enhance the efficiency of the building’s engineering systems and reduce energy consumption.