Description

Keywords: plastics recycling, process design, machine learning, crystallization

Background:
The Circular Foam project deals with the recycling process of rigid polyurethane foams (PUR) used for building and appliance insulations. The goal of the Circular Foam project is to change the paradigm by enabling a circular economy of PUR foams thanks to chemical recycling. In this context, effective and reliable separation techniques are the key to successful and sustainable processes. The Energy and Process System Engineering group at ETH (EPSE) has developed a tool to design solvents for extraction processes using genetic algorithms. In this tool, molecules are generated using a genetic algorithm, then the properties of said molecule are calculated either with COSMO-RS or with a pre-trained neural network, and the process performance is evaluated using pinch-based shortcut models for different unit operations. To enable a faster design, EPSE recently developed a machine learning-based property prediction model (ML model) that surpasses COSMO-RS in speed and accuracy. Moreover, the actual shortcut models available for the design of processes are limited to extraction and distillation, thus the addition of other unit operations, e.g., crystallization and filtration, would represent an important contribution for the Circular FOAM project.

Your thesis:
In this thesis, you will expand the current tool for process design. The workload can be broken down into subtasks:

1.     a shortcut model for crystallization, filtration, and solvent recycling will be developed and integrated into the current workflow used for process design. These units will expand the current configuration of extraction and distillation columns.

2.     the thermodynamic properties necessary for the process will be taken either from COSMO, or from a neural network, which will need to be trained for solid-liquid equilibrium properties.

3.     the candidate process will be simulated in a more rigorous way with the aid of a process simulation software AVEVA (Aspen-like), to prove the feasibility and the interoperability with the other project partners of Circular Foam.

About you:
·       Student of mechanical, process, or (bio)chemical engineering, or a comparable subject

·       Good understanding of thermodynamics and interest in coding

·       Independent and goal-oriented working style

·       Above-average grades

·       Programming experience with Python and Matlab recommended

Working at EPSE:
In this master thesis, you have the opportunity to learn about thermodynamic short-cut models, process simulation software (AVEVA), artificial neural networks, and much more. Moreover, you will be part of a young and motivated team of researchers and students.