Machine learning for the design of chemical engineering unit operations

The project is part of the Priority Programme “Machine Learning in Chemical Engineering“ (SPP 2331) funded by the German Research Foundation (DFG). The Priority Programme brings together the chemical engineering and machine learning communities from top German universities in order to open up new methods for chemical engineers and formulate new types of problems for involved machine learning experts, with the long-term goal to jointly generate advances for methods in both communities. In close collaboration with the Institute of Theoretical Informatics (ITI) and the Institute for Micro Process Engineering (IMVT) it is planned to advance Chemical Engineering, in particular the design of unit operations, leveraging the capabilities of novel machine learning tools and state-of-the-art simulations. For this purpose, a new CFD-code based on the OpenFOAM framework will be developed for simulations of various microfluidic systems such as heat-exchangers, microevaporators and microreactors with possibilities of geometry modification. The simulations generate the main dataset for datadriven geometry optimization with machine-learning algorithms developed at ITI, after which the optimal designs will be experimentally tested using metal 3D-printer fabrication at the IMVT facilities.