Solution Mapping for Numerically Efficient Implementation of Chemical Kinetics
Thursday, October 14, 1999 - 1:30pm - 2:15pm
Michael Frenklach (University of California, Berkeley)
The use of chemical kinetics often requires repeated solution of differential equations describing the kinetic model. In the approach of solution mapping, the model responses are parameterized as simple algebraic functions of model variables through performed a priori or in situ computer experiments. The computer experiments are arranged by factorial designs that possess certain optimal properties. By choosing appropriately the model responses and model variables, the developed algebraic representations can effectively replace the differential-equation solver in various applications of chemical kinetic modeling, such as model optimization and reactive-flow simulations. Utilization of modern data structure architectures facilitates numerically efficient implementation of the technique.