Chemical Process Flowsheets

Chemical process flowsheets, diagrams representing the sequence and interconnection of unit operations in a chemical process, are being revolutionized by artificial intelligence. Current research focuses on using machine learning, particularly deep reinforcement learning and transformer-based language models, to automate flowsheet design, error correction, and optimization, often leveraging text-based representations like SFILES 2.0 for efficient data handling and model training. These advancements aim to accelerate process design, improve efficiency, and enhance safety by automating tasks previously reliant on manual, time-consuming methods. The development of large, standardized flowsheet datasets is crucial for further progress in this field.

Papers

December 5, 2023

Toward autocorrection of chemical process flowsheets using large language models
Lukas Schulze Balhorn, Marc Caballero, Artur M. Schweidtmann
Large Language Model Error Correction Flow Diagram Chemical Process Flowsheets

October 10, 2023

Deep reinforcement learning uncovers processes for separating azeotropic mixtures without prior knowledge
Quirin Göttl, Jonathan Pirnay, Jakob Burger, Dominik G. Grimm
Reinforcement Learning Mixture Component Prior Knowledge Deep Reinforcement Learning Approach Analysis by Synthesis Chemical Process Flowsheets

February 7, 2023

Data augmentation for machine learning of chemical process flowsheets
Lukas Schulze Balhorn, Edwin Hirtreiter, Lynn Luderer, Artur M. Schweidtmann
Machine Learning Data Augmentation Process Design Chemical Process Flowsheets

August 1, 2022

Learning from flowsheets: A generative transformer model for autocompletion of flowsheets
Gabriel Vogel, Lukas Schulze Balhorn, Artur M. Schweidtmann
LeArning Abstract Auto Completion Transformer Based Generative Analysis by Synthesis Chemical Process Flowsheets

July 25, 2022

Chemical Process Flowsheets

Papers

Toward autocorrection of chemical process flowsheets using large language models

Deep reinforcement learning uncovers processes for separating azeotropic mixtures without prior knowledge

Data augmentation for machine learning of chemical process flowsheets

Learning from flowsheets: A generative transformer model for autocompletion of flowsheets

SFILES 2.0: An extended text-based flowsheet representation

Flowsheet synthesis through hierarchical reinforcement learning and graph neural networks