A leap forward in chemical process design: Introducing an automated framework for integrated AI and CFD simulations

Abstract

Despite the numerous possibilities of integrating AI and CFD simulations for chemical process design, researchers often rely on manual techniques, resulting in suboptimal models and time-consuming processes. To address these challenges, we propose an automated framework that combines high-fidelity AI modeling with hyperparameter optimization, automated CFD simulations using OpenFOAM, and effortless post-processing for data extraction. This framework was tested on a reactor scale-up process involving 51 different configurations for butadiene synthesis, achieving 98.8 % accuracy in CFD validation and over 99 % accuracy in AI models. The automation pipeline streamlines geometry generation, meshing, simulation, data extraction, and AI-driven optimization, significantly reducing manual effort. Our framework is versatile, customizable for various types of process equipment design, and employs open-source software for ease of adoption and reproducibility. This approach not only enhances accuracy and efficiency but also opens up AI and CFD integration to a broader range of researchers.