Optimal control and stochastic modelling of acid pre-treatment of bagasse

Abstract

Abstract Acid pre-treatment is a preferred pre-treatment process for ethanol production from bagasse, a lignocellulosic feedstock. Better control of the acid pre-treatment is desired to improve yields and thereby its economic feasibility. This work applies the theory of optimal control for developing control strategies for a batch acid pre-treatment reactor. The Pontryagin's maximum principle in combination with the numerical method of steepest descent of Hamiltonian is used to solve the optimal control problem. Three different types of optimal control problems, namely, maximum concentration problem (MCP), minimum time problem (MTP), and maximum productivity problem (MPP), have been studied. The MCP results showed the significant improvement in xylose yield (around 9.23%) for 300 minutes of batch time and MTP reported the significant reduction in batch time (around 28%) to achieve 0.2 gram xylose concentration. However, pre-treatment is influenced by various sources of uncertainties. Therefore, it is essential to optimize the acid pre-treatment process after considering these variabilities. As a step towards solving the stochastic optimal control problem, this work has developed the stochastic model for the acid pre-treatment. Variability in feedstock composition has been considered, and stochastic model has been developed to capture time dependent and independent uncertainty in feedstock composition.