To develop a predictive dust explosion model or theory many considerations of the interaction between several complex multiscale processes are required. Due to practical considerations only some of the processes may be fully resolved, while the rest must be approximated or neglected. The current study focuses on constructing an organizational framework for dust explosion model development. The framework is organized based on progression of the explosion in time and geometric scale of the important features. Suggested methodologies are given to investigate the small-scale features and develop feed-forward approximations for predictive models at the larger system-scale. The concepts developed through constructing the organizational framework are applied to closed volume dust explosion testing and classification of several sources of experimental variance for this system is given. (C) 2013 Elsevier Ltd. All rights reserved.