Presentation
ALChEMY (mAchine Learning for ComplEx MultiphYsics problems) aims to develop a unified modelling approach, to predict the behavior of low-emission combustion devices using a combination of experimental and computational methods, tied together by machine learning techniques. High-fidelity experimental data and simulations for complex multi-scale and multi-physics reacting systems will be processed using advanced machine learning approaches, with the objective of developing reduced-order models able to mimic the behavior of a real system without the computational burden associated to full-scale simulation. The methodology will be demonstrated for the case of a MILD combustion system, for which the world first digital twin will be developed. Moreover, we anticipate large application of the proposed methodology beyond combustion, in process industry, material science and manufacturing processes.
Promoters
- Alessandro Parente, Brussels School of Engineering
- Nedunchezhian Swaminathan, Department of Engineering, Cambridge
Thanks to the Foundation’s support, Golnoush Ghiasi has been hired as a postdoctoral researcher to work on this project.
Publications
Machine Learning and Its Application to Reacting Flows (Springer 2023) > read more.
“Data-driven models and digital twins for sustainable combustion technologies”, Perspective (Volume 27, Issue 4, April 19, 2024) > read more.