The grinding process, used since antiquity to crush seeds and nuts, is vital across various industries, such as mining, civil engineering, and pharmacy. Current research aims to optimize this procedure by enhancing the properties of powders while reducing energy costs. Experimental methods for studying grinding face complexities due to dynamic forces and the continual changes in materials. Fortunately, recent advancements in simulation, using the Discrete Element Method (DEM), offer a perspective to investigate these mechanisms, especially in the context of co-grinding for nuclear fuel manufacturing.

This thesis topic specifically aims to accelerate the simulation of these mechanisms for industrial use. The objective is to develop a fragmentation metamodel based on artificial intelligence. To achieve this, it will be necessary to create a database simulating particle fragmentation and to define the essential features of the process. The approach will encompass several phases, including predicting a particle's fragmentation and learning the fragmentation mode using advanced techniques, such as neural networks.

The research will build upon previous works, notably those of D.-C. Vu (CEA thesis 2020-2023), and will be validated using experimental data associated with other academic endeavors. The doctoral candidate will have access to significant simulation resources, with access to the computing resources of the IRESNE Institute (CEA-Cadarache) and other platforms. In essence, this thesis project aims to merge expertise in grinding with artificial intelligence techniques to innovate in the field of particle fragmentation.