Quality Assurance of an Underground Cavern Thermal Energy Storage System : A Case Study in Energy, Power and Loss Calculations for Mälarenergi

A quality assurance of Mälarenergi's energy, power, and loss calculations was conducted in this study. Results from the energy quality assurance showed that the energy storage system holds around 1.26 GWh more energy than Mälarenergi's previous calculations with a temperature distribution based on historical data. Plausibly, if the bottom temperature distribution can reach 45 oC instead of 50 oC, 2.6 GWh more energy than Mälarenergi's calculations is obtained.These cases similarly give an energy capacity of 13.21 GWh and 14.66 GWh respectively, with the assumption that the top maximum temperature distribution can reach 95 oC. Saidtemperature has been reached but with a lesser energy yield.The loss calculations and their validation tested four scenarios, which in the end showed that the bedrock recovers energy to the storage during discharging, with a surprisingly high energy recovery factor of 59%, revealing additional energy content associated with the bedrock, while at the same time evidencing that the storage side power of the heat exchanger is the most accurate power in contrast to the district side. The loss analysis also showed that the gradient first - SL combined method was the best method of the five presented, as it was least sensitive to the heat exchanger power according to the sensitivity analysis and because of its steady losses assumptions. With this method, applied for Scenario 4, a resulting total cumulative energy loss of 33 GWh was found, which was slightly more than the updated predicted loss of 28 GWh. However, the power loss was slightly smaller than predicted during the last 5 months of operation, suggesting that the calculated loss will be smaller than predicted in the near future. This phenomenon is probably due to more cycling than predicted, which hinders more heat from penetrating deep into the bedrock and hence only postpones the initial losses. A power loss model was also created with an ansatz that suggests the losses depend on the energy content and time. The power loss model was applied and compared mainly to the gradient first - SL combined loss method. The result suggested that the model was very accurate over long time spans, as the magnitude of its error was around 0.4%. The loss model did not, however, capture every detail of the power loss, and a Spearman coefficient of around 0.65 was obtained. For further studies, an updated ansatz was suggested that distinguishes the different time periods of charging, discharging, and steady-state periods. Furthermore, a continuous update of the calculations over time is recommended to explore how the energy recovery and energy calculations might be affected, as well as to observe how the losses develop.