The effective thermal conductivity (ETC) of metal powder is a basic physical property that has an important role in industrial design for metallurgical, energy, chemical, and other types of application. To obtain this information, a numerical model based on the coupled discrete element-finite difference (DE-FD) method is proposed to precisely and efficiently predict the ETC of metal powders. A random packaged aluminum powder unit influenced by the effects of gravity is piled up concerning the particle size distribution of the real powder. Moreover, a sandwich structure model is proposed to simulate the thermal dynamic behavior during directional heat transfer. Based on the amount of heat passing through the system, the effective thermal conductivity was precisely calculated in an indirect way and the results of the simulation were consistent with corresponding experimental results. Relative error was within +/- 5%. Thus, the proposed numerical model may be useful in predicting the ETC of the different powder beds filed using various types of interstitial gas media within a wide range of temperatures. (C) 2018 Elsevier Ltd. All rights reserved.