Development of an intelligent and autonomous slurry production process
3 years (01.01.2020-31.12.2023)
Funding: 503.000 € (TUBS)
Total Project Volume: 1.243.301,64 €
Institute for Particle Technology (iPAT), TU Braunschweig
Institute of Mechanical Process Engineering and Mechanics (MVM), Karlsruhe Institute of Technology (KIT)
Institute of Production Science (wbk), Karlsruhe Institute of Technology (KIT)
The use of an extruder in the industrial mass production of battery cells is a recent innovation. It allows for a continuous dispersion of battery slurries, as opposed to the previous method of dry batch-wise premixing and downstream mixing tank before conveying to the coating line. One of the benefits of using an extrusion line is the superior scalability. Additionally, the use of inline measurement technology enables the integration of digital twins and thus a process control strategy for high-level automation and process monitoring. The goal is to optimize productivity by achieving high overall equipment effectiveness even when dealing with fluctuating order situations and product variance, while also maximizing raw material utilization.
The InZePro cluster is focused on reducing costs by developing and improving automation strategies for a smart battery cell production. As the first process step the mixing of the battery slurries has a major impact on both electrode and cell performance. For this reason automating the mixing process is important to ensure well performing cells and improve the resource efficiency through product monitoring.
The goal of this project is to enhance the current mixing method by incorporating new technology that enables autonomous, continuous production. This is achieved by integrating dry mixing and dispersion techniques in a multi-chamber system, and by directly feeding the coating line. A digital twin is used to control and monitor the process, allowing for adaptability to varying order volumes and product variations. AI models and grey-box methods are employed to develop strategies for controlling the process and coping with disturbances and uncertainties. By combining process engineering experiments with artificial intelligence methods, it is possible to achieve process control based on digital twins, as well as process optimization and predictive maintenance via process control. The ultimate goal is to integrate a model-predictive control system that guarantees constant paste quality while taking uncertainties and secondary conditions into account. However, the material properties are difficult to represent directly in terms of measurement technology. Therefore, indirectly measurable quantities such as flow behavior, concentrations, and particle size distribution are measured. Machine learning methods can then be used to derive relationships that represent a correlation to the important parameters.