This work develops a method for prediction of dynamic changes in particle size distributions (PSD) of high-shear granulation systems using a discrete element simulation technique. This method allows for direct evaluation of particle interactions based on multi -dimensional descriptions of particle parameters. Pouw et al. [38] [G.A. Pouw, D. Verkoeijen, G.M.H. Meesters, B. Scarlett, Population balances for particulate processes-a volume approach. Chemical Engineering Science 57 (2002) 2287-2303] proposed the use of a volume-based Population Balance Equation (PBE) model with the volume of solid, liquid, and gas in each particle as internal parameters to predict result of particle interactions. This paper extends on the work of Pouw et al. by using a discrete element simulation approach rather than direct application of population balance equations to determine the evolution of particle size distributions. This is accomplished by simulating the effects of particle interactions based on physically significant coalescence criteria. Three granule modification mechanisms are used in the proposed method: coalescence, consolidation, and breakage. Two types of coalescence are modeled in this simulation. In Type I coalescence, granules are stopped solely by viscous dissipation of the binder layer before the granule solid surfaces touch, whereas Type II coalescence occurs when deformable granules come into contact with their solid surfaces and the granule surfaces then bind together. Consolidation, the escape of air from granules due to compaction following collisions, is described by an exponential relationship related to the porosity of each individual simulated particle. One may assume that breakage will occur when there is sufficient externally applied kinetic energy to deform and shear a granule. Breakage can be determined based on whether the Stokes deformation number for a particle exceeds a critical value. One objective of this work is to move toward modeling and simulation methods that allow for dynamic changes in operating conditions at any time in a batch run. Current empirically based coalescence kernels used in population balance based models are generally developed using static operating conditions, limiting the model validity for online control applications. The main contribution of this work lies in the observation that when using this type of simulation model, the physics of the granulation system can be altered more easily than modeling the granulation process using traditional population balance. Systems exhibiting a wide range of yield strength can be modeled to determine boundaries for Type I coalescence, Type II coalescence, or rebound events based on physical arguments rather than extrapolation of empirically based coalescence kernels. Using similar initial and operating conditions, the discrete element simulation is shown to produce results similar to population balance results. To examine the extended flexibility of the new modeling method, several open-loop simulations using this method are presented in this paper to display how a process would dynamically react to changes in operating conditions. (c) 2005 Elsevier B.V. All rights reserved.