Projects
Numerical methods for parametric dynamic optimisation of (bio)chemical processes KU Leuven
This project deals with parametric optimisation of large-scale (bio)chemical processes, i.e., calculating the solution path of an optimisation problem as a function of one or more varying parameters. Typical examples include heat transfer coefficients (which change when fouling occurs), design parameters (e.g., dimensions) and operational parameters (e.g., maximal admissible reactor temperature).
As ...
Highly efficient non-Euclidean optimization and MCMC methods for deep matrix completion KU Leuven
The matrix completion (MC) problem is of interest in many applications, especially in bioinformatics, chemoinformatics, and systems biology. The existing methodologies for dealing with MC problems are based on the linear dependency between the data entries that factorize the matrix into two (or more) low-rank matrices. Moreover, due to the sparsity and scarcity of the entries of the data matrix, MC suffers from unavoidable large ...
Novel methods and 4D-XCT tools for in situ characterisation of materials and their microstructural changes during functional testing. University of Antwerp
Next generation building energy assessment methods towards a carbon neutral building stock KU Leuven
The development of future-proof assessment methods for the energy performance of buildings supporting the transition to low-carbon housing, taking into account the technical social-economic impact at the level of the individual building, the local network and the building stock.
Rethinking 'thinking about your thinking': Identifying the role of metacognitive monitoring in academic learning through cognitive psychology methods in educational research. KU Leuven
Identifying the origin and correlates of individual differences in academic learning is a critical goal of educational sciences. This is because educational research aims to develop learning environments that are optimally tailored to accommodate these individual differences in academic learning, such as arithmetic. Metacognition has been put forward as highly relevant for learning, yet we lack a functional insight into the role of ...
Adjoint-based optimization methods with fluid/kinetic plasma edge codes for nuclear fusion reactors KU Leuven
With a clean, cheap, widely available and virtually inexhaustible fuel, nuclear fusion has the potential to provide a sustainable answer to our energy needs. In order to sustain the "burning" plasma, a careful reactor design is crucial. Both vessel and magnetic fields should shape the outer plasma region in such a way that heat loads do not exceed material limits while Helium ash is sufficiently removed.
Recently automated ...
Massively Parallel and Robust High-Order Methods for Transitional Hypersonic Flow Modelling on Unstructured Grids: Application to Reusable Launcher Stages KU Leuven
During their ascent and descent trajectories, reusable space vehicles travel primarily in the hypersonic regime (5 < Mach < 25) which is characterized by high speeds, strong shock waves, chemical dissociation, radiation, viscous interaction, etc. Additionally, the flow experiences various changes in regimes, i.e. laminar, transitional and turbulent, along the trajectory. Prediction of the onset and extent of the transition from laminar ...
Modeling and simulation of air plasmas using particle methods applied to Air-Breathing Electric Propulsion KU Leuven
An emerging concept called Air-Breathing Electric Propulsion (ABEP) could allow to fly spacecraft in the currently unexploited Very Low Earth Orbit by using air collected from the atmosphere as the propellant for a plasma thruster. A small-scale hypersonic low density facility funded by ESA is going to be commissioned at the von Karman Institute for Fluid Dynamics (VKI) for experimental testing of an ABEP intake-collector system. The stream ...
Enriched conversational XAI methods for healthcare KU Leuven
Despite the rich set of eXplainable Artificial Intelligence (XAI) methods that have been proposed to justify the outcome of Machine Learning (ML) models in healthcare applications, many open challenges remain: most of these efforts are focused on algorithm developers rather than healthcare professionals, who often have little or no knowledge of ML models. The majority of the XAI systems also rely on complex visual representations of models ...