S. Hyun, J. Park, G-A Ryu, Y. Han, S. Baik
Korea Institute of Ceramic Engineering and Technology,
Korea
Keywords: sintering, multiphysics simulation, AI, realtime digital twin model
Summary:
Virtual characterization such as computer simulations was successfully applied to characterize various manufacturing processes especially performed in extreme conditions, e.g. under high temperatures and pressure. One of the crucial processes in manufacturing ceramic devices is the sintering process. The sintering process involves the heating of powdered materials to form a solid mass. Accompanying the thermomechanical phenomena during the process, it is highly demanded to control precisely the temperature profile in the furnace, otherwise, the products out of the process would fail due to various faults such as undesired deformations, and cracks. Designing an efficient sintering furnace involves considerations of heat transfer, temperature profiles, and optimizing the overall process. Combining computational fluid dynamics (CFD), heat transfer, electromagnetic simulations, and artificial intelligence (AI) for real-time temperature profile prediction can significantly enhance the design and control of the sintering furnace. A.I. algorithms, particularly machine learning, can be trained using virtual data as well as experimental data from the sintering process. These models can predict temperature profiles based on various input parameters, allowing for real-time adjustments and control. This digital twin model can be used to optimize the sintering process by adjusting parameters such as temperature, airflow, and material composition in real-time to achieve desired outcomes. The real-time predictions and adjustments allow for precise control of the sintering process, reducing waste and improving overall efficiency. This integrated approach contributes to energy efficiency, process optimization, and improved product quality in industries that rely on the sintering process. In this work, we introduce an automatic generation platform of virtual data for the sintering process under various manufacturing conditions to achieve optimal temperature profiles. We developed a digitized and automatic workflow accompanying the multi-scale and multi-physics simulations and AI algorithms to characterize and optimize the temperature profile in the process. The accelerated digital twin model can predict the temperature profile significantly faster than the actual computer simulations. In the wide range of manufacturing technology based on ceramic materials, this multiscale and data-driven approach can be used as a real-time characterization and monitoring tool for optimizing processes for complex geometry and operating conditions. The ceramics manufacturing DX platform (VECTOR) is developed to provide optimization digital tools customized for fundamental study as well as industrial needs.