This work extends the species sensitivity method of model reduction known as Alternate Species Elimination (ASE) to a stochastic version. The new Stochastic Species Elimination (SSE) approach allows for a linear reduction in the number of species retained in the course of reduction. It improves the computational cost and offers flexibility to the user in terminating the reduction process when an acceptable model size is attained. Larger chemical kinetic models, such as the recent literature model of n-octanol, are approached with the SSE method coupled with multiple species sampling. This further allows for a faster model reduction process. These modified approaches are applied to the reduction of selected chemical kinetic models based on ignition simulations: the n-heptane model by Mehl et al. (654 species, 5258 reactions), reduced using the SSE method (293 species, 2792 reactions) and the ASE method (245 species, 2405 reactions); the iso-octane model by Mehl et al. (874 species, 7522 reactions), reduced to an SSE version (315 species, 3037 reactions) and an ASE version (306 species, 2732 reactions); and the n-octanol model by Cai et al. (1281 species, 5537 reactions), with a reduced SSE version (450 species, 2532 reactions). Resulting skeletal models are shown to adequately predict ignition delay times as well as flame propagation when compared to the predictions of the detailed models. Burning velocity predictions are well-captured even though the reduction is based on ignition delay simulations.
Chemical Kinetic Models of Early Diagenesis
