Optimization-based simultaneous modelling of stoichiometries and kinetics in complex organic reaction system

Efficient understanding and modelling of kinetics in complex organic reaction systems are crucial in the development cycle. The dynamic experimental measurements can be analyzed automatically by the data-driven method in a stepwise modelling approach: after the preliminary screening of candidate stoichiometries, they are formed into different stoichiometric groups, and the reaction data were regressed for each stoichiometric group so that the best fitting kinetic model could be selected. However, the stepwise modelling approach brings about the computational difficulty for complex reaction systems due to the combinatorial nature of forming stoi-chiometric groups. In this article, a novel optimization-based method, simultaneously combining stoichiometry grouping and kinetics fitting, is proposed to build the kinetic model of a homogeneous organic reaction system. Through reformulation of the original nonlinear optimization model, a mixed integer linear programming model is developed to identify the reaction stoichiometries and kinetic parameters with improved efficiency. Two computational examples are presented to demonstrate the accuracy and effectiveness of the simultaneous modelling methodology.