Impact of aleatory and epistemic uncertainties on thermal risk and production assessment: Application to the hydrogenation of levulinic acid and butyl levulinate

The hydrogenation of levulinates, derivatives from cellulose hydrolysis, leads to the platform molecule γ-valerolactone (GVL). While prior investigations indicate the propensity for thermal runaways in this reaction, a comprehensive sensitivity analysis has been conspicuously absent in the existing literature. This study endeavors to scrutinize the influence of input parameters on both thermal risks and the production rate during GVL synthesis. Distinguished by its dual focus on sensitivity analyses, this manuscript also contrasts the implications of aleatory and epistemic uncertainties on thermal risk and production metrics. Aleatory uncertainties arise from the inherent variability of initial conditions, while epistemic uncertainties emanate from incomplete knowledge of the kinetic model. Employing Sobol global sensitivity analysis on the kinetic model, we delineate the hierarchical importance of various parameters. This analysis incorporates a quantification of uncertainty, and algorithmic approaches are proffered. The findings reveal that the initial temperature, the initial mass of the Ru/C catalyst, and hydrogen pressure are pivotal factors that are directly proportional to thermal risk and production efficiency. The implications of uncertainty on these variables are also discussed.