Reduced Order Models in Fluid Dynamics: from modal decompositions to machine learning
Tid: Ti 2022-05-24 kl 10.30 - 11.30
Plats: Faxén, Teknikringen 8
Föreläsare: Soledad Le Clainche (Universidad Politécnica de Madrid)
Halting climate change is one of the main challenges of our society. For mitigating its effects, it is necessary to search for different alternatives and to develop new technologies able to reproduce the atmospheric pollution. The field of knowledge of fluid mechanics has multiple applications that should be studied in detail to develop novel eco-friendly technologies. Generally, complex flows (i.e., transitional or turbulent flows, noisy, multi-scale, et cetera) are highly relevant in industrial applications. Hence, researchers are continuously searching different strategies to model and predict the flow behaviour. This is a very challenging task due to the high complexity of these flows when solving realistic systems. As a possible solution, we propose to develop reduced order models (ROMs) based on physical principles using (i) modal decompositions (i.e., singular value decomposition – SVD, higher order dynamic mode decomposition – HODMD, principal component analysis -PCA) and (ii) machine learning tools (neural networks) combined with the previous decompositions. In this work, these techniques will be applied to solve several industrial and academic problems in fluid dynamics, and new strategies for developing efficient and accurate ROMs will be presented.
Dr. Soledad Le Clainche is Associate Professor in the Department of Applied Mathematics at the School of Aerospace Engineering of the Universidad Politécnica de Madrid (UPM). She received three Masters of Science: in Mechanical Engineering by UPCT, in Aerospace Engineering by UPM and in Fluid Mechanics by Von Karman Institute. In 2013 she completed her PhD in Aerospace Engineering at UPM. Her research focusses on computational fluid dynamics, machine learning, reduced order modelling, and in the development of novel tools for data analysis enabling the detection of spatio-temporal patterns. In addition, she has contributed to the fields of flow control, global stability analysis, analysis of flow structures in complex flows (transitional and turbulent) using data-driven methods, and prediction of temporal patterns using machine learning and soft computing methods.