The European Union strives a transition towards a circular economy, hence requiring effective and efficient treatment processes for the recycling and thermal exploitation of produced waste. Mixed commercial waste presents a significant share of the produced waste, with around 1.1 million metric tons alone in Austria in 2016. It is usually first treated mechanically, with coarse shredders performing the first processing step of comminution and liberation while also functioning as the primary dosing devices. Despite their importance for waste processing, the knowledge about the process behavior of coarse shredders and its dependence on their parametrization is limited. But understanding the influence of different configurable factors is essential for their optimized operation and enhanced processing, involving smart real-time control, in the context of the fourth industrial revolution. Physical numerical models are hardly applicable for understanding these machines’ behavior in real-scale operation. They require too detailed information about the always-changing condition of the heterogeneous input material: mixed commercial waste. Consequently, empirical modeling approaches are followed in this thesis. As a necessary precondition for successful experimentation, a procedure for material sampling was first established, based on the theory of sampling. The induced general estimation error was then determined based on a replication experiment. Subsequently, a 32 runs coarse-shredding experiment with mixed commercial waste was conducted, based on a fully randomized, D-optimal experimental design. It aimed at investigating the influences of the radial gap width, the shaft rotation speed, and the cutting tool geometry on a shredder’s throughput behavior, its energy demand, and the particle size distribution it produces. Significant models were successfully derived, applying (multivariate) multiple linear regression. For doing so, the particle size distribution was described as isometric log-ratio-transformed mass shares of three particle size classes (>80 mm, 30–80 mm, 0–30 mm). The models show significant effects of all three factors on throughput behavior and energy demand. But only the cutting tool geometry significantly influenced the shares of the particle size classes. Based on the models, conclusions on the optimized operation of coarse shredders were drawn, which contradict common operation settings. Finally, investigations on the sensor-based real-time measurement of particle size distributions, as defined by drum screening, were conducted: partial least squares regression models, based on geometric descriptors, obtained from two-dimensional RGB images of the particles, turned out to be a promising approach.