A Data-Driven Optimization of Materials Recovery in Battery Recycling Processes

The electrification transition has increased the demand for raw materials for batteries, and superior recycling practices could allow us to retrieve more raw materials from battery waste. This study introduces a data-driven approach to optimizing battery recycling processes, explicitly targeting the recovery and purity of nickel-manganese-cobalt (NMC) and graphite during the mechanical separation stage. The main objective is to identify the sets of operational parameters of the recycling process that maximize the grade and recovery of the materials. We utilized HSCSim simulation software to generate an extensive dataset via the simulation of thousands of randomly generated operational scenarios. This dataset enabled advanced analytics to identify the optimal process conditions and parameters.<br/>Our results show that precise adjustments in operational parameters, such as magnetic field strength, flotation cell residence time, and optimization of the flotation cell conditioner, can significantly boost the separation efficiency of NMC and graphite. We also showed that data analytics can help us iteratively optimize and improve the process design. The optimal process achieved simulated recovery rates of up to 79% for NMC and 91% for graphite, with purity levels of graphite surpassing 99%. The next step is to incorporate machine learning models in the workflow to optimize the process on the fly without human intervention. These advancements are crucial for strengthening the circular economy by increasing materials recovery and tackling the urgent requirement for sustainable and efficient battery recycling processes.