Radiation represents an important contribution to the heat transfer through a packed bed. The packing structure is important in the determination of the radiation heat transfer, but this aspect is not explicit in most of the mathematical models proposed in the literature. Here, a new numerical approach is proposed to calculate the radiation heat transfer in such a bed from its structure using the Voronoi network model. On this basis, the effective thermal conductivity (ETC) is evaluated by taking into account the effects of the radiation heat transfer and conduction through neighbor particles and stagnant fluid. The validity of this approach is verified by comparing the calculated and measured ETCs under different conditions. This approach is used to investigate the effects of variables such as particle thermal conductivity, emissivity and size, and bed temperature as well. The relative contributions of different heat transfer mechanisms are analyzed. It is shown that there is a similarity in the probability density distributions of dimensionless heat flows among particles under different bed temperatures, indicating that packing structure is a dominant factor in controlling the distribution whereas the mean heat flow between particles mainly depends on bed temperature and related material properties.