Automated Computational Thermochemistry for Butane Oxidation
Sarah Elliott, University of Georgia
The fidelity of combustion modeling can be drastically improved with large-scale implementation of high-level computational chemical kinetics. Fundamental to these simulations is the accurate computation of thermochemical properties, such as enthalpies of formation, entropies and heat capacities. We have developed the Predictive Automated Computational Thermochemistry (PACT) package on ANL HPC machines and on NERSC to generate thermochemical data for the combustion of an arbitrary fuel. The task demands an automated procedure because not only is manual generation of the reaction mechanisms for these simulations difficult and error-prone, but the number of species and computations can easily grow to beyond tens of thousands, even for the combustion of simple fuels.
The PACT procedure begins with user specification of the fuel and combustion conditions with which the package produces a list of combustion-relevant species. For each species in this list, the procedure determines an internal coordinate z-matrix description of its structure, the optimal torsional configuration via Monte Carlo sampling and subsequent optimizations, key rovibrational properties, and high-level estimates of the electronic and zero-point energies via arbitrarily defined composite methods. Partition functions are generated from this data which are then converted to thermodynamic properties and finally to NASA polynomial representations. The end product is an automatically generated, JSON-formatted database of electronic structure results and thermochemical data prepared for use in combustion simulations.
To illustrate the package's utility and performance, PACT was used to generate thermochemical data for the combustion of n-butane. This fuel presents most of the key challenges that must be surmounted in the consideration of larger fuels and is the smallest alkane to provide low-temperature oxidation chemistry that closely mimics that of larger alkanes. The accuracy of data generated with PACT is further displayed by benchmarking against the ANL0 database.
Abstract Author(s): Sarah N. Elliott, Murat Keceli, Carlo Cavallotti, Yuri Georgievskii, William H. Green, Stephen J. Klippenstein