Multivariate Regression Analysis of Bio-Oil Yield Using Literature-Derived Biomass Data

Abstract The distribution of products formed during biomass pyrolysis depends strongly on the physical and chemical properties of the feedstock, yet quantitative comparisons across studies remain limited. This work compiles experimental and modelling data from literature to examine how five commonly reported biomass characteristics (moisture content, lignin content, ash content, higher heating value, and particle size) influence bio-oil yield. Pairwise trends highlight substantial variability in the explanatory strength of these parameters, with higher heating value showing the most consistent positive association with yield. A multivariate regression approach was then applied to assess the combined effects of all variables within a single analytical framework. The resulting model identifies higher heating value as the most informative predictor across the heterogeneous dataset, while the influence of other parameters is more dependent on interactions and covariation among biomass properties. Predicted yields for locally analysed biomass samples align with general tendencies reported in the literature, though these estimates remain provisional due to the limited size and heterogeneity of the compiled data. The study provides an integrative perspective on feedstock quality factors relevant to bio-oil production and underscores the need for experimental validation and expanded datasets to support more robust predictive modelling.