Machine Learning for Optimal Material Design in CO2 Capture

This proposal aims to accelerate the discovery of materials in Chemical looping combustion (CLC),
which is an emerging technology that requires lower energy expenditure to capture CO2 from fossil fuels.
A machine learning (ML) framework that only requires a limited amount of data for training will be
developed. Atomistic simulations will be conducted to predict the energetics of the CLC system for
different set of materials. Results from the atomistic simulations will be used to train the ML method,
which will use Meta-learning and domain adaptation techniques to predict promising materials using a
Bayesian optimization framework. Experimental testing will be conducted to validate the predictions
from the proposed ML framework. Syngas on a perovskite-based material will be used as the system to
develop, test, and validate the proposed ML algorithm.

Faculty Supervisor:

Luis Ricardez-Sandoval;Luis Ricardez Sandoval;David Simakov;Pascal Poupart

Student:

Partner:

Bank of Montreal

Discipline:

Engineering

Sector:

Finance and Insurance

University:

University of Waterloo

Program: