The application of residence time density distributions gained with discrete element method (DEM) simulations for predicting the mixing behavior of gravity blenders in different operation modes is presented. Knowing these distributions, the performance of a blender can be predicted during different blending tasks in continuous and discontinuous operation mode. Up to now, the residence time distributions needed for these predictions could only be measured. It will be explained how a gravity blender containing more than 1 million of particles is mapped to simulation considering the computational effort as well as the quantitative validity of the results. Based on these data, the discontinuous blending behavior is predicted. The results indicate good agreement between simulations and experimental data and can serve as a basis for designing gravity blenders.