Kian is an industrial doctoral candidate researching about hybrid modelling in chemical engineering systems, specifically within production of biofuels.
Hybrid modelling seeks to combine methods from artificial intelligence and machine learning with mechanistic models, which in Kian's particular case corresponds to a reactor model. Hybrid models can be used when certain parts of the mechanistic model are unknown or too difficult to model. This also has certain advantages over a pure data-driven model such as extrapolation capabilities, it requires less data than a pure data-driven model and using the laws of nature, it sets constraints on variables, preventing any implausible results.
The doctoral project aims to use hybrid modelling to create a reactor model for biofuel production. Second-generation biofuel production has exceptionally complex reaction kinetics with a vast number of chemical species and a large number of reactions taking place. Therefore the kinetics can not be accurately represented with the law of mass action, consequently machine learning is used to approximate the reaction kinetics and combining this with the available knowledge about chemical reactors and the laws of nature.
Ask me about