(692e) Kinetic Model Building through Simultaneous Optimization from Spectral Data
AIChE Annual Meeting
2023
2023 AIChE Annual Meeting
Catalysis and Reaction Engineering Division
Modeling and Analysis of Chemical Reactors
Wednesday, November 8, 2023 - 1:50pm to 2:10pm
Sampling of a reaction system using spectroscopic techniques can be broadly categorized into either
in situ or ex situ methods. In situ methods are safer and can provide more accurate information.
However, the information collected can be challenging to analyze for multicomponent mixtures,
since the spectra are often composed of highly overlapping signals from various species in the
reaction; these can be difficult to decouple [1]. Common in situ collection methods for kinetic
reaction studies include the use of infrared, ultra-violet, or Raman spectroscopy.
in situ or ex situ methods. In situ methods are safer and can provide more accurate information.
However, the information collected can be challenging to analyze for multicomponent mixtures,
since the spectra are often composed of highly overlapping signals from various species in the
reaction; these can be difficult to decouple [1]. Common in situ collection methods for kinetic
reaction studies include the use of infrared, ultra-violet, or Raman spectroscopy.
A multitude of data analysis approaches exist to decouple the concentration and absorbance pro-
files. These can be broadly categorized into approaches which postulate a kinetic model for the
reaction system, attempt to solve the system without a kinetic model, or a hybrid combination of
the two. Published examples of these problems and their difficulties can be found in works by [1, 2].
A recent alternative method to these modeling approaches is to simultaneously obtain the reaction
kinetic parameters with the curve resolution. Building on a nonlinear programming framework, the
postulated reaction model is considered in the constraints of the optimization problem while also
taking into account noise associated with instrument and model error. To further aid practition-
ers with the rapid advancement of spectral analysis methods, an optimization modeling platform
called KIPET was introduced (KInetic Parameter Estimation Tool. The KIPET approach
is derived from maximum likelihood principles and uses nonlinear programming techniques and
collocation methods to simultaneously solve a proposed reaction system, which has been shown
to outperform traditional approaches like MCR-ALS.
In this work, we use a simultaneous solution strategy to investigate a relevant industrial silanol
reaction. We extend the investigation to consider nonlinear deviations from Beer-Lambertâs Law
that occur due to hydrogen bonding