Research team
Expertise
My research activities are focused on theoretical chemistry with a particular focus on the combination of state-of-the-art theoretical chemistry methods with machine learning, aiming to predict the molecular properties of large biological systems based on reactivity. I have wide expertise in modelling large biological systems using hybrid, classical and quantum methods, as well as structural chemistry methods. During my scientific career I have gained a lot of programming experience, and I enjoy coding in the HPC framework.
Polyurethane recycling: Unifying molecular dynamics and process flow simulation for efficient separation and optimization.
Abstract
Polyurethanes (PU) are widely used in mattresses, upholstery, furniture, automotive, construction, and insulation. They are mostly thermoset foams, made by reacting isocyanates (MDI, TDI, or HDI) with polyols. Their thermoset nature limits mechanical recycling, making chemical recycling crucial for circularity. The resulting aromatic molecules, ureas, amines and polyols from depolymerization have varied physicochemical characteristics, impacting separation during recycling. This doctorate aims to use thermodynamic modeling tools to predict the ease of separating depolymerized PU mixtures. Methods include activity coefficient based models (NRTL, UNIFAC, HANSEN) that are accompanied by computational chemistry methods to optimize the models and fill in unknown gaps. Modeling results will inform engineering software for process design, optimizing recycling for recyclers and informing circular design for PU formulators and recyclers. The focus is primarily on predicting interactions between different polyols used in PU, considering monomer composition, degree of branching, molecular weight distribution, and functionality, to facilitate efficient separations. Later, other constituents are covered in more detail as well. The goal is to provide recyclers and formulators with insights for process optimization and improved circularity of PU materials.Researcher(s)
- Promoter: Billen Pieter
- Co-promoter: Cunha Ana
- Co-promoter: Nimmegeers Philippe
- Fellow: de Graaf Christophe
Research team(s)
Project type(s)
- Research Project
Conceptual DFT a novel tool for virtual screening.
Abstract
Computational chemistry methods have proven to be essential tools for drug development. In this research proposal we plan to develop DFT based methods combined with ML, that can accurately predict the activity of drugs in binding sites of viral proteases, by using reactivity indexes. The main goal is to push the limit of virtual screening forward, by introducing DFT reactivity descriptors that can classify possible drug candidates more accurately. Thus, greatly reducing the number of false positives. Therefore, we will use conceptual DFT methods (i.e., reactivity descriptors) to train a neural network that can predict with high accuracy the reactivity between a drug and an active site. This will be first applied to three classes of protease (Serine, Aspartyl, Cysteine) proteins, and later expanded to mutated active sites.Researcher(s)
- Promoter: Cunha Ana
- Fellow: Rostami Pour Kiana
Research team(s)
Project type(s)
- Research Project