(560f) Recent Advances in Kipet | AIChE

(560f) Recent Advances in Kipet

Authors 

McBride, K. - Presenter, Carnegie Mellon University
Biegler, L., Carnegie Mellon University
Garcia-Munoz, S., Eli Lilly and Company
KIPET (Kinetic Parameter Estimation Toolkit) is an open-source toolbox used to determine kinetic parameters from a variety of experimental datasets including spectra and concentrations (Schenk et al. 2019). KIPET distinguishes itself from standard parameter estimation packages in the manner in which the problems are defined. Instead of approaching the deconvolution and parameter fitting as a two-step problem, the simultaneous collocation optimization framework proposed by Chen et al. (2016) has been implemented. This methodology is based on maximum likelihood principles and large-scale nonlinear programming that involves systems of nonlinear differential algebraic equations (DAEs). All these features have been implemented in the Python programming language and the algebraic modeling package Pyomo. The software is available on GitHub and also as packages hosted by PyPi and Anaconda.

In recent years, much work has been done to improve several aspects of KIPET’s capabilities in the areas of variance estimation (Short et al. 2020), parameter selection and estimability analysis (Chen and Biegler 2020), incorporating multiple data sets (Short et al. 2019), improving the integration of mixed data sets, handling unknown absorbing species (Chen et al. 2018), improving the user interface, and much more. One of the most obvious updates is in the user experience. Most of the model formulation and preparation required previously to define the problem (setting up the model, the parameter estimator, the variance estimator, etc.) has been simplified or placed into the background. This makes it much more inviting for practitioners to use KIPET who do not have much programming experience.

The other major feature introduced to KIPET is the improved handling of multiple datasets. Modeling multiple datasets is now as simple as modeling several individual models, all but eliminating the interface presented in Short et al. (2019). In addition to this, a new methodology for parameter fitting using multiple datasets based on nested Schur decompositions has been integrated into KIPET. This works by separating the experiments into individual scenarios represented by various models or substructures thereof. A global set of model parameters is selected and fixed in each individual scenario such that sensitivity information (in the form of the duals) can be obtained. This information is used to construct a reduced Hessian and right-hand side following the nested Schur decomposition strategy. The global parameters are then updated using a trust-region approach or interior point approach, such as IPOPT (Waechter and Biegler 2006), to solve the outer problem. In this manner a more robust approach to parameter fitting with multiple scenarios is achieved. Several examples of this method are presented in this work.

References

[1] Schenk, C., Short, M., Rodriguez, J. S., Thierry, D., Biegler, L. T., García-Muñoz, S., Chen, W. Introducing KIPET: A novel open-source software package for kinetic parameter estimation from experimental datasets including spectra, 2020, Computers and Chemical Engineering, 134, 1 - 16.

[2] Chen, W., Biegler, L. T., García-Muñoz, S., 2016, An Approach for Simultaneous Estimation of Reaction Kinetics and Curve Resolution from Process and Spectral Data, Journal of Chemometrics, 30, 506-522.

[3] Short, M., Biegler, L. T., García-Muñoz, S., Chen, W., 2020, Estimating variances and kinetic parameters from spectra across multiple datasets using KIPET, Chemometrics and Intelligent Laboratory Systems, 203.

[4] Chen, W., Biegler, L. T., 2020, Reduced Hessian based parameter selection and estimation with simultaneous collocation approach, AIChE Journal, DOI: 10.1002/aic.16242

[5] Chen, W., Biegler, L. T., Garcia-Munoz, S., 2018, Kinetic Parameter Estimation based on Spectroscopic Data with Unknown Absorbing Species, AIChE Journal, 64, 3595-3613.

[6] Wächter, A., Biegler, L. T., 2006, On the Implementation of an Interior-Point Filter Line Search Algorithm for Large-Scale Nonlinear Programming,Mathematical Programming, 106(1), 25-57.