A novel and refined particle-scale thermal resistance network model for high-resolution heat transfer analysis in pebble beds

A novel thermal resistance network model based on the analytical solution of thermal resistance within particles has been proposed, aimed at achieving a more refined temperature calculation at the particle scale for pebble beds. A complete solution process has been established by coupling a random forest model, and a high level of consistency has been demonstrated when compared to the precise results obtained from the finite volume method (FVM). The model has been applied to the temperature calculations of the High-Temperature Test Unit (HTTU) pebble bed, yielding results that align well with experimental values. The R 2 value for the thermal resistance within individual particles, trained using the random forest model, reached 0.99494 in the test set, with both the mean absolute error and the root mean square error being very small, thus supporting calculations for large-scale pebble beds. For simple cubic (SC), body-centred cubic (BCC), and face-centred cubic (FCC) structures, both fixed and variable material thermal conductivity scenarios have been considered. It was found that the relative deviation of calculated excess temperatures at most nodes and particle centres did not exceed 0.5 % when compared to the precise FVM results. The modelling of the HTTU pebble bed provided histograms of frequency distributions for contact radius, angles between contact surfaces, and coordination numbers, along with the distribution of thermal resistance within particles. Under a 20 kW operating condition, the proposed thermal resistance network model was shown to yield better alignment with experimental values, while also providing a more precise and detailed temperature distribution compared to the thermal discrete element method, all within an acceptable computational load. The temperature differences between surface nodes of the HTTU pebbles were analyzed, revealing an average maximum temperature deviation of 24.97 °C, thereby highlighting the necessity of calculating node temperatures rather than solely focusing on the temperature at the particle centre.