Public Thesis Defense of Luisa CASTELLANOS - IMMC
sst |
Advancing surrogate modeling of combustion chemical kinetics using latent-space dynamics
Thursday April 2, 4pm - Auditorium BARB92 - Place Sainte-Barbe 1, 1348 Louvain-la-Neuve
With the ongoing climate crisis, it is key to re-design the energetic sector, as well as the hard-to-abate industries. In such tasks, it is important to consider the role of reacting flows for industrial applications. This produces issues for efficient design of technologies, principally due to difficulties in obtaining a-priori measurements of performance, and the high cost and limitations when it comes to gathering experimental data. Recently, Machine Learning techniques have emerged as an efficient solution to such aspects through surrogate modelling, however, it is important to keep in mind that there is still work to do regarding efficient and correct implementation of Machine Learning techniques for combustion science.
In this work, advances are made while encouraging the use of sparse datasets, improving in interpretable machine learning, discussing the implementation of Time-lag Autoencoders for the analysis and reduction of chemical kinetics, allowing to obtain non-linear reduced representations which are physically interpretable. To emphasize contexts with great lack of data, and easing the optimization process, the technique of Gappy-Autoencoder is introduced, enabling to analyze chemical mechanisms and reduce dimensionality in high-sparse data contexts, while keeping physically interpretable reductions. Additionally, the concept of Gradient-based Clustering is explored, which aims to group sets of solutions for chemical kinetics ODEs considering their time derivatives behavior. The meaning of such technique becomes more insightful when considered the gradients of a Combustion Progress Variable, a monotonic function that measure the development of combustion, and represents its dynamics. Lastly, this work also analyses some core cases of time integrators development, principally for the case of Neural Networks. Such experiments, lead to a reconsideration in the typical training paradigms in the literature and provides insights in better practices for surrogate models’ development. .
Jury members
Prof. Aude Simar (UCLouvain) Supervisor
Prof. Alessandro Parente (ULB) Supervisor
Prof. Paul Fisette (UCLouvain) Chairperson
Prof. Estelle Massart (UCLouvain) Secretary
Prof. Axel Coussement (ULB)
Prof. Salvatore Iavarone (CentraleSupelec)
Prof. Nguyen Anh Khoa Doan (TU Delft, The Netherlands)