Development of a reduced-order dynamic model for large-scale seasonal thermal energy storage applications

Abstract

This study introduces an efficient simulation model for large-scale pit seasonal thermal energy storage (PTES) applications, designed to retain accuracy while significantly reducing computational demands. Being implemented in Modelica/Dymola, the reduced-order model is compared against an experimentally validated COMSOL Multiphysics simulation model based on key performance indicators including energy balance, thermal losses, temperature stratification and computational time. Energy balances of both models show good agreement, with deviations of less than 6 % in terms of charged energy and under 5 % in discharged energy. Total thermal losses align closely, with discrepancy below 2 %, underscoring the model's reliability. Temperature stratification analysis reveals strong alignment of both models under idle conditions, especially in the upper layers of the storage. During dynamic charging and discharging phases, minor discrepancies are observed, with root mean square error values ranging from 1.2 K in the upper layers to 2.4 K at the bottom. Additionally, the reduced-order model demonstrates a substantial reduction in computational time, making it up to 98 % faster than the COMSOL model. The model is therefore established as a highly efficient yet accurate tool for large-scale sTES simulations, particularly suited for iterative system design, optimization processes, and real-time control.