Hybrid Artificial Intelligence-based Process Flowsheet Synthesis and Design using Extended SFILES Representation

• Published in: Computer Aided Chemical Engineering Vol. 53; pp. 1279 - 1284
• Main Authors: Mann, Vipul, Sales-Cruz, Mauricio, Gani, Rafiqul, Venkatasubramanian, Venkat
• Format: Book Chapter
• Language: English
• Published: 2024
• Subjects: artificial intelligence; computer-aided flowsheet synthesis; flowsheet modeling; process design; flowsheet modeling; artificial intelligence; process design; computer-aided flowsheet synthesis
• DOI: 10.1016/B978-0-443-28824-1.50214-3
• ISBN: 9780443288241; 0443288240
• ISSN: 1570-7946

Process flowsheet synthesis and design involves simultaneously solving several problems, including determining the unit operations and their sequence, underlying reactions and reaction stoichiometry, downstream separation design and operation parameters, sustainability factors, and many more. Naturally, this results in a large amount of data being associated with a given process flowsheet that captures the relevant process context and should be readily accessible. This data is useful for solving related problems both using data-driven and process knowledge-based methods. A hierarchical framework, called the extended SFILES (or eSFILES), proposed recently stores this information using a combination of text-based, graph-based, and ontology-based representations. Here, we provide details on a prototype software for automated flowsheet representation and generation across various levels in the eSFILES framework. The underlying methods include a novel flowsheet grammar, a set of inferencing algorithms, and interfacing with a commercial process simulator facilitating rigorous flowsheet simulation.