Machine Learning for Chemical Looping Combustion

This chapter describes the application of various machine learning (ML) algorithms/techniques to CLC. Recently machine learning (ML), as a branch of artificial intelligence, is increasingly being applied to various areas of science and engineering such as combustion, carbon capture and sequestration, automation, games, process control, and medical diagnoses among many others. In the last two decades, the field of chemical looping combustion has been rapidly advancing, resulting in a large amount of data obtained from experiments and numerical simulations at multiple spatiotemporal scales, which presents the potential for applying ML techniques to extract underlying information from the CLC phenomena with high level of complexities. ML algorithms are mainly classified into supervised learning, unsupervised learning, and semi-supervised learning according to the learning methods. This chapter presents an overview of widely used ML algorithms and introduces three examples of application of ML to chemical looping field, namely, the investigation of flow characteristics in circulating fluidized bed riser, estimation of the performance of an oxygen carrier, and allocation to optimization of a moving bed reactor. The application of ML should accelerate the development of chemical looping combustion by offering a useful tool for reactor design and optimization as well as the design of new potential oxygen carriers.