Torrefaction is a remarkable technology in biomass-to-energy. However, biomass has several disadvantages, including hydrophilic properties, higher moisture, lower heating value, and heterogeneous properties. Many conventional approaches, such as kinetic analysis, process modeling, and computational fluid dynamics, have been used to explain torrefaction performance and characteristics. However, they may be insufficient in actual applications because of providing only some specific solutions. Machine learning (ML) and statistical approaches are powerful tools for analyzing and predicting torrefaction outcomes and even optimizing the thermal process for its utilization. This state-of-the-art review aims to present ML-assisted torrefaction. Artificial neural networks, multivariate adaptive regression splines, decision tree, support vector machine, and other methods in the literature are discussed. Statistical approaches (SAs) for torrefaction, including Taguchi, response surface methodology, and analysis of variance, are also reviewed. Overall, this review has provided valuable insights into torrefaction optimization, which is conducive to biomass upgrading for achieving net zero.
All Science Journal Classification (ASJC) codes