Combustion of micron-sized aluminum dust particles was studied numerically in an environment with spatially discrete sources distributed in a random way. A thermal model was employed and developed to estimate flame propagation speed in a lean quiescent reaction medium. The model is based on conduction heat transfer mechanism. The random distribution of heat sources in a discrete environment is described by using the Gaussian random distribution function with different variance numbers. Flame propagation speed is obtained by using the one-dimensional diffusion-reaction equation which includes the source terms. Oxygen is considered as the main oxidizer and nitrogen is considered as the inert gas. Flame propagation speed, the lean flammability limit and the minimum activation energy were investigated as a function of dust concentration and particle diameter. The predicted results were compared with the experimental data and were found to be in a reasonable agreement. It can be concluded that the random model shows better predictions as compared to the uniform model. (C) 2016 Elsevier Ltd. All rights reserved.