Funding organisation: German Research Foundation/ Deutsche Forschungsgemeinschaft (DFG)
Contact Person: Christoph Thon
Summary of the overall project:
Stirred media mills occupy a central role in various processes of ultrafine comminution. The combination of selected operating parameters and suspension properties determine the kinetic energy of grinding beads and influence both particle transport as well as the uneven axial distribution of beads in the mill during continuous operation. These factors are decisive for the comminution behaviour and energy consumption. Especially when considering the comminution of product particles down to the nano-range, particle size distributions in the mill change dynamically. Increases in suspension viscosity directly influence the velocity of grinding beads and therefore also the energy transferred to the product particles. In addition, the motion characteristics of the grinding beads are influenced by the viscosity of the suspension. As the particle size changes, the required stressing energy for ideal particle comminution therefore shifts. Consequently, the optimum operating point in mills changes dynamically, which is currently rarely taken into account in the design and operation of stirred media mills. The control of fine grinding processes, considering the interactions between grinding bead damping, particle strength and optimum operating point, thus requires real-time capable models that incorporate these dynamic properties. In this project, population balance models are to be used for this purpose, and terms for viscosity and spatial grinding bead dynamics are to be derived and supplemented. The data for modelling will be obtained experimentally and numerically via CFD-DEM simulations. Predictive models using neural networks and AI transfer functions between simulations and experiments are applied for data augmentation. Based on the extended population balance models, our project partner, the Institute for Mathematical Optimization, will develop real-time control for the stirred media mill.
Goals and tasks of iPAT
Establishment of real-time measurement of particle size and viscosity, directly or via an AI-driven soft-sensor approach
Systematic experimental investigation of the dynamic particle size and viscosity properties
Systematic numerical investigation of the spatial grinding bead distribution in the mill
Predictive models based on experimental data and numerical simulations, data augmentation and derivation of transfer functions via AI (especially neural networks and genetic programming)
Derivation of additional terms in population balances and transfer / integration into the control system of our project partner
Project partner: Institut für Mathematische Optimierung (TU Braunschweig)