Process Design and Bayesian Optimization of 5-Hydroxymethylfurfural Hydrodeoxygenation

The hydrodeoxygenation (HDO) of 5-hydroxymethylfurfural (HMF) is an essential step in lignocellulosic biomass conversion. In this work, we developed process flowsheet simulation models using the hydrodeoxygenation experimental conditions, yields, and separation requirements. Economic and environmental evaluations are then performed and incorporated with Bayesian optimization to search for reaction conditions that minimize production costs and greenhouse gas emissions. The Bayesian optimization demonstrates that the HDO reaction condition with the lowest production cost has a relatively high reaction temperature, no water content, high H2 pressure, medium HMF loading, and short reaction time. However, the HDO condition with the lowest greenhouse gas emission has a lower temperature but a longer reaction time.