In packed beds, bed structure significantly influences heat transfer between particles and fluids. A pore-network model (PNM) incorporating conduction, convection, and radiation is developed to investigate heat transfer in packed beds at the particle-pore scale. The model reveals how structural variations, such as bed porosity and pore geometry, influence heat transfer mechanisms. Validation against experimental data from demonstrates strong agreement in temperature evolution and heat transfer coefficients, confirming the model's accuracy. Bed-scale simulations reveal that bed porosity, gas velocity, and temperature collectively determine the dominant heat transfer mode. Convective heat transfer prevails at higher gas velocities, accounting for over 80 %, particularly in loosely packed beds. Conduction is more significant in denser beds and at lower velocities, contributing up to 40 %. Radiative heat transfer becomes substantial, accounting for up to 30 % only at elevated temperatures (e.g., 1000 °C), surpassing conduction in loosely packed configurations. At the pore scale, denser beds exhibit more uniform pore geometries that enhance local convective transfer through pores with near-regular shapes, such as smaller pores approximately half the particle size, typically observed in cells with porosities below 0.4. Conversely, increased heterogeneity in high-porosity beds promotes advective transport through larger pore throats. This model offers convenience in exploring and quantifying how bed porosity and local structural heterogeneity governs heat transport in granular systems and offers a flexible modelling framework for exploring structure–property relationships under diverse thermal and flow conditions.