The determination of chemical reaction networks is generally a limiting step in process development, requiring considerable time, expertise and intellectual effort. Mathematically, finding a reliable model of a set of interacting chemical reactions is equivalent to a network inference problem, thus the problem may be redefined as the identification of the structure and parameters of the kinetic model of a reaction network.
This work follows that of Burnham et al (2008) which aims to identify chemical reaction networks solely from batch process data. The procedure involves postulating a global ODE model structure consistent with single collision event kinetic rate expressions. These represent changes in species concentrations and are directly linked to the underlying networks of the reactions taking place. The model structure is constrained to be a weighted sum of a set of non-linear basis functions which are restricted to be linear in the parameters polynomial models. Then, mathematical and statistical methods are used to iteratively select a subset of the basis functions based on the identification of their corresponding parameters. This work aims to use sensitivity analysis to determine the effect of parameter uncertainty in identifying the subset of basis functions, therefore the structure of the reaction networks. The process is developed using simulated case studies and, finally, a real experimental case is provided to demonstrate the techniques.
It is concluded that parameter uncertainty has a significant impact on successfully determining chemical reaction networks.

Burnham et al(2008) Inference of chemical reaction networks, Chemical Engineering Science, 63, 4, 862-873