Numerical modelling of a packed bed thermal energy storage for applications in a Carnot battery

Abstract

This thesis explores the heat transfer characteristics of a packed bed thermal energy storage device through the development of a 3D model in COMSOL Multiphysics. The model is based on a laboratory-scale device consisting of an aluminum box filled with sand, which acts as the thermal storage medium. The medium is heated using two resistance heaters that are embedded into the sand. The heaters are raised to approximately 500°C during the experiment. Three thermocouples placed in the sand record the temperature during charging, thus generating a time-dependent temperature curve at three points. The model simulates this experimental system, and the resulting temperature distribution is compared to the experimental data to assess the model validity. To determine the temperature throughout the storage system at all times, the model solves the heat equation. The density and heat capacity of the sand are measured and added to the model. The model uses the effective medium technique to estimate the thermal conductivity of the sand. A secondary model is developed in Excel to calculate the effective thermal conductivity using twelve different correlations, and the results are compared. The Zehner-Bauer correlation with an added Damköhler radiation term predicted an effective thermal conductivity of 0.218 W/(m.K) at 25°C and 0.451 W/(m.K) at 500°C, with slightly exponential growth due to the influence of thermal radiation. When these values for conductivity are applied to the COMSOL model, there is strong agreement (average percent error < 5%) with the experimental temperature distributions. A parametric study is performed, and shows that increasing grain size and emissivity could increase the thermal conductivity by making use of the exponential growth of thermal radiation. The model serves as a design and learning tool, and a starting point for improving the performance of the thermal energy storage system.