Michael T. Klein
Dean, College of Engineering
Catalysis and Reaction Engineering; Automated Kinetic Modeling; Hydrocarbon
in Supercritical Fluids.
| Molecule-Based Modeling of Complex Reaction Systems
Our work focuses on reaction systems of such enormous complexity that they have historically defied fundamental analysis. In this, we are developing the Chemical Engineering science that will allow the fundamentals of chemical kinetics, and the supporting sciences of Physical, Organic, Analytical and Computational Chemistry, to replace the traditional Chemistry-free global lumps as the basis for kinetics models for complex systems. This work has recently taken the tangible form of software (the Kinetic Modeler's Toolbox, KMT) for the automated modeling of the structure and kinetics of complex systems.
Our approach has been to delineate and reduce the essential elements of complexity in these reaction systems. It begins with molecular structure building software that uses Monte Carlo simulation techniques to assemble a molecular representation of complex feedstocks from analytical information, e.g., H/C SIMDIS, NMR, etc. This realization of a feedstock as a set of molecular structures is the first step in the use of quantitative structure-property relationships. Reactivity is an especially important property, and it can be phrased in the terms of the quantitative structure/reactivity correlations (QSRC) developed in our experimental work, which has shown how apparently independent rate constants can be constrained by a QSRC. This allows prediction of rate constants given a molecule's or intermediate's structure using computational quantum chemistry. This also reduces the number of potentially adjustable parameters in detailed kinetic models by more than 90%.
The KMT software has automated the process of building these detailed kinetic models. Exploiting Monte Carlo and graph theory techniques, we can build reaction models containing thousands of species in 1000 CPU seconds or less. This incredible model building speed has changed the serial model building-model use paradigm to a new parallel approach, where a model builder can produce an updated optimal model in seconds. The thus-constructed models react the molecularly explicit feedstock using QSRC's for kinetic parameters.
This approach is being applied in the development of models for FCC, hydrocracking, hydroisomerization, catalytic reforming, oxidation, pyrolysis and alkylation chemistries.
Klein, M.T., Broadbelt, L. J. and S. M. Stark, Computer Generated Pyrolysis Modeling: On-the-Fly Generation of Species, Reactions, and Rates,
I&EC Research, 33, 790-799 (1994)
Klein, M.T., Watson, B.A. and R. H. Harding, Mechanistic Modeling
of n-Heptane Cracking on HZSM-5,
Klein, M.T. and S. C. Korre,
Klein, M.T., Broadbelt, L. J. and D. H. Grittman, Computer Generated
Pyrolysis Models: Applications of Monte Carlo Simulation, Graph Theory,
and Quantum Mechanic,
Klein, M.T. and S. C. Korre,