Deterministic and stochastic optimization of dilute acid pretreatment of sugarcane bagasse

Abstract

Dilute acid pretreatment of lignocellulosic biomass depolymerizes xylan into xylose and lignin into acid-soluble lignin (ASL). Xylose is a desired product, while the lignin fraction can be reduced by producing ASL, since lignin is a non-competitive inhibitor for enzymatic hydrolysis. The optimal conditions for depolymerization of xylan into xylose and lignin into ASL differ during acid pretreatment of bagasse. Hence, multi-objective optimization problems (MOOPs) were formulated for a batch reactor considering temperature as the decision variable. Two different MOOP studies, namely maximization of the weighted sum of final xylose and ASL yields, and minimization of weighted deviation of final xylose and ASL yields around a targeted value, were considered. The optimal temperature varied between 143.2 °C and 113.1 °C along the Pareto front for static optimization. The batch time minimization problem reduced batch time by 32% to achieve desired xylose yield. Dynamic optimization provided minimal benefits over static optimization. Stochastic optimization considering expected uncertainties showed that uncertainties in kinetic parameters impacted results selectively. The optimal temperature decreased by 10 °C for ± 15% variability in kinetic parameters for ASL yield maximization, while it changed marginally for xylose yield maximization. Stochastic optimization showed benefits over static optimization for large variability in kinetic parameters.