CSE Community Seminar | February 21, 2025

Presenter Sili Deng, Associate Professor, Department of Mechanical Engineering, MIT

Talk Title Scientific Machine Learning for Kinetic Modeling and Uncertainty Quantification

Abstract

The combustion science is at the intersection of chemical kinetics and fluid dynamics, making this discipline physically rich and intellectually challenging. Applying tools developed in other disciplines to combustion research could provide us new perspectives and enable new discoveries. I will demonstrate our recent work on developing scientific machine learning tools for kinetic modeling and uncertainty quantification.  Specifically, I will discuss the Chemical Reaction Neural Network approach to identify reaction pathways and simultaneously quantify kinetic parameters from data without any prior knowledge of the chemical system but bounded by fundamental physics, as well as an efficient kinetic uncertainty quantification framework for turbulent combustion simulations.