Generative AI in chemical process engineering

Generative AI can support chemical engineers in modeling, design, and operation of processes. We develop generative AI methods for engineering diagrams and process models, from autocompletion and autocorrection of flowsheets to P&ID generation and AI-assisted HAZOP studies.

Challenge

Generative AI models learn a parametric approximation of an underlying data distribution, which allows them to generate new samples such as text, code, images, or molecules. Tools like ChatGPT, GitHub Copilot, or IBM RXN show what this technology can do. In chemical process engineering, however, generative AI is still in its infancy. Engineering knowledge is stored in diagrams, simulation models, and documents that general-purpose models cannot read reliably, data are scarce and often confidential, and results must be correct and safe.

Our approach

We expect generative AI to affect all aspects of chemical process engineering, including modeling, optimization, design, control, and operation. Existing chemical engineering knowledge and data can become accessible through domain-specific or even company-specific foundation models, either by training or by letting generative models write and execute queries to databases. Generative models can also interact with simulation and optimization environments: they generate models, run simulations, and use the results to solve problems. This ability to incorporate mechanistic knowledge could lead to a new generation of hybrid models.

Our work focuses on four applications:

• Autocompletion of flowsheets: suggesting unit operations, topology, and design variables while an engineer draws a flowsheet.
• Autocorrection of flowsheets and P&IDs: identifying errors in engineering documents and explaining the suggested corrections.
• P&ID generation: generating P&IDs, including control structures, from process flow diagrams (PFDs).
• AI-assisted HAZOP: pre-filling HAZOP worksheets from engineering documents, previous HAZOP reports, and simulations.

All of these require machine-readable engineering diagrams. We therefore also work on the digitization of engineering diagrams, on Smart P&IDs based on the DEXPI standard (pyDEXPI), and on LLM interaction with P&IDs (ChatP&ID).

Partners & funding

• Collaboration with Linde and Siemens on the P&ID Co-Pilot, a generative AI tool for creating P&IDs (announced in March 2024).
• Research results are transferred to practice through the Digitization Companion (DigiCo).

Key publications

• Schweidtmann, A. M. (2024). Generative artificial intelligence in chemical engineering. Nature Chemical Engineering, 1, 193. doi:10.1038/s44286-024-00041-5 · Details
• Schulze Balhorn, L., Weber, J. M., Buijsman, S., Hildebrandt, J. R., Ziefle, M., & Schweidtmann, A. M. (2024). Empirical assessment of ChatGPT's answering capabilities in natural science and engineering. Scientific Reports, 14, 4998. doi:10.1038/s41598-024-54936-7 · Details
• Alimin, A. A., & Schweidtmann, A. M. (2026). GraphRAG for engineering diagrams: ChatP&ID enables LLM interaction with P&IDs. AIChE Journal, 72, e70540. doi:10.1002/aic.70540
• Rupprecht, S., Gao, Q., Karia, T., & Schweidtmann, A. M. (2026). Multi-agent systems for chemical engineering: A review and perspective. Current Opinion in Chemical Engineering, 51, 101209. doi:10.1016/j.coche.2025.101209