(299b) Simultaneous Estimation of Kinetic Parameters and Curve Resolution of Spectral Data -  Applications and Extensions | AIChE

(299b) Simultaneous Estimation of Kinetic Parameters and Curve Resolution of Spectral Data -  Applications and Extensions

Authors 

Garcia-Munoz, S. - Presenter, Eli Lilly and Company
Biegler, L., Carnegie Mellon University
Chen, W., Carnegie Mellon University
In the context of studying reaction kinetics, nowadays a common practice is to use spectroscopic instrumentation to indirectly monitor the concentration of the species in the reaction. Recently, Chen et al1 proposed an approach based on maximum likelihood principles for simultaneous estimation of reaction kinetics and curve resolution. This work presents further learnings and developments in the application of this technique.

Three cases studies will be presented highlighting first the criticality of pre-estimation steps, specifically: i) the importance of the pre-processing algorithm applied to the spectral data prior to the estimation, and ii) wavelength selection where a nonlinear programming (NLP) formulation is proposed as a means to efficient wavelength selection. We also present an application of mixed integer NLP (MINLP) approaches to identify which active species can be observable in the spectra (since this information is not always available). Finally, we present an Estimability Analysis (EA) that identifies the appropriate subset of model parameters to be estimated from the available spectral data. Overall, the simultaneous technique presented by Chen et al. is shown to be a reliable and flexible method to deal with the complexities of spectral data and system dynamics in a reaction system.

References

1. Chen W, Biegler LT, Muñoz SG. An approach for simultaneous estimation of reaction kinetics and curve resolution from process and spectral data. Journal of Chemometrics. 2016;30(9):506-522.