Mesoscopic simulation of heat conduction in randomly packed beds through lattice-based computation and realistic particle reconstruction

Packed beds are heterogeneous media inspiring the design of novel thermal devices. The heterogeneity leads to the anisotropic heat transfer behaviour, which raises the complexity of the analysis and design process. This paper presents an originally developed framework for the computation of heat conduction behaviour in packed beds. The formation of the packing structures was realised with practical rock data. The core computation adopted a three-dimensional lattice-based algorithm. Several digital tests were introduced, including temperature tracking, energy-tracking, round-trip efficiency and heat flux distribution. It was found that the random close packing of natural rocks can maintain a relatively stable heat release and storage performance in the representative volume. An increment of the solid phase fraction does not always enhance the heat release and storage performance, and there exists a threshold between 34.93 % and 42.33 % solid phase fraction to trigger the positive correlation. A decrease in the solid phase fraction improves the heat intake efficiency but reduces the temperature maintenance performance. The critical time threshold for balanced round-trip efficiency is located at around 13.89 lattice hours without additional thermal interference. Energy round-trip efficiency is an intrinsic property of the packed structure. It was statistically found that the power function is highly likely to be the energy evolution function from upscaling characterisation. The generalised logistic function can provide a fitting recovery of the heat flux distribution from downscaling characterisation. The developed computational framework demonstrates a feasible bottom-up routine for thermal device design and heat transfer analysis.