Model-based run-to-run optimization under uncertainty of biodiesel production

Abstract

A significant source of uncertainty in biodiesel production is the variability of feed composition since the percentage and type of triglycerides varies considerably across different raw materials. Also, due to the complexity of both transesterification and saponification kinetics, first-principles models of biodiesel production typically have built-in errors (structural and parametric uncertainty) which give rise to the need for obtaining relevant data through experimental design in modeling for optimization. A run-to-run optimization strategy which integrates tendency models with Bayesian active learning is proposed. Parameter distributions in a probabilistic model of process performance are re-estimated using data from experiments designed for maximizing information and performance. Results obtained highlight that Bayesian optimal design of experiments using a probabilistic tendency model is effective in achieving the maximum ester content and yield in biodiesel production even though significant uncertainty in feed composition and modeling errors are present.