(80a) Synergistic Workflows of Machine Learning and Chemical Engineering for Process Model Development

The integration of machine learning (ML) into chemical engineering has impacted our approach to process modeling. Process models based on mechanistic understanding, while grounded in physical laws, can face challenges related to development and computational costs. Meanwhile, data-driven approaches offer flexible, scalable alternatives but can lack interpretability and physical consistency. This talk presents synergistic hybrid workflows, highlighting past and recent literature on process model identification and identifying open challenges. We will review and contrast different paradigms for process model development, ranging from more traditional approaches to more automated ML/AI-based approaches. Emphasis will be placed on workflow design, data requirements, model validation, and interpretability. Two case studies will be discussed: developing a hybrid model for a microbial bioprocess and building a predictive model for a reactive milling process. These examples demonstrate how combining ML and chemical engineering can accelerate model development for emerging and complex systems.