Project Details
Projekt
AI-PRE: AI for Explainable Roundtrip Multiscale Emulsion Polymerization Processes
Subject Area
Technical Chemistry
Image and Language Processing, Computer Graphics and Visualisation, Human Computer Interaction, Ubiquitous and Wearable Computing
Image and Language Processing, Computer Graphics and Visualisation, Human Computer Interaction, Ubiquitous and Wearable Computing
Term
since 2021
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 466601458
Emulsion polymerizations are an important field of application in polymer reaction engineering (PRE) and constitute the most important process for the production of polymer colloids with applications ranging from cosmetics to construction. Advantages of the highly versatile process are access to complex polymer structures and polymers with targeted properties, high scalability, excellent heat control, limited viscosity, and water as process medium reducing the use of low volatile organic compounds. Polymers from emulsion polymerizations are frequently referred to as “products by process” due to the strong correlation of the process conditions and polymers obtained. Thus, modeling of the reactions and prediction of reaction conditions to yield polymers with targeted properties is of high importance. The main goal of the project is to establish a novel AI-integrated polymer reaction engineering (PRE) approach for the multiscale emulsion polymerization with special emphasis on bio-based monomers, which enables AI-based reverse engineering to find optimal recipes and process conditions to obtain a polymer with targeted microstructure. This requires: (1) Description of multiscale emulsion polymerization processes by a kinetic Monte Carlo (kMC) simulation approach that considers reaction kinetics and mass transfer processes at all scales to represent the detailed polymer microstructure as a function of the process parameters. (2) Proposition of a novel coherent suite of artificial intelligence (AI)-based models that enables integrated multiscale emulsion polymerization process modeling (PPM) and reverse engineering, the latter is referred to as roundtrip PPM. An in-house developed open-source kMC simulator for reaction kinetics in homogeneous phase will be extended to emulsion polymerization in order to generate a large body of data for training and testing the AI models. To reach this overall goal the following scientific objectives will be addressed: (1) Create a coherent suite of scalable AI-based models for multiscale emulsion polymerization modeling, which will facilitate efficient kMC simulation-based learning of AI models. (2) Propose AI-based approaches for reverse engineering of emulsion polymerization processes with modeling and optimization capabilities at different scales (macroscale, microscale) based on experimental and simulation data. (3) Establish a general methodology to explain the AI models. (4) Establish the kinetics and mechanisms for dibutyl itaconate polymerization systems and extend the findings to itaconates with widely varying ester structures. (5) Establish a kMC simulation environment for multiscale homo- and copolymerizations of itaconate monomers. (6) Generation of experimental data for validation of the kMC and AI models. The focus of the project is on bio-based itaconate monomers, and the novel polymers obtained will be characterized with respect to their properties.
DFG Programme
Priority Programmes
Subproject of
SPP 2331:
Machine Learning in Chemical Engineering. Knowledge Meets Data: Interpretability, Extrapolation, Reliability, Trust
Co-Investigator
Dr. Marco Drache
