Algebraic surrogate-based flexibility analysis of process units with complicating process constraints

Abstract

Flexibility analyses are widespread in chemical engineering to quantify allowed deviations from nominal conditions. Standard approaches to perform flexibility analysis can be hard to apply if process constraints are difficult to handle, as it happens in bioprocesses with dynamic constraints. Here, focusing on the computation of the traditional flexibility index in problems with complicating constraints, we apply symbolic regression to build algebraic expressions of the said complicating constraints, simplifying the flexibility analysis of complex process models by enabling the application of state-of-the-art deterministic solvers. Our approach is applied to ethanol production in fed-batch operation mode and a chromatographic process. The performance is assessed in terms of model building time, predictive accuracy of the model, and the time required to solve the flexibility formulations. Overall, our approach, which focuses on computing the original flexibility index proposed in the literature, provides an alternative way to analyse the flexibility of processes entailing complicating constraints.