Applying Realistic Chemistry in Direct Numerical Simulations

Tianfeng LuResearch AssociateMechanical and Aerospace Engineering DepartmentPrinceton UniversityDirect numerical simulations (DNS) of reacting flows invoking realistic chemistry
constitute the ultimate approach to produce results of high fidelity. This would allow for the solution of a broad range of problems from first principle such as fuel utilization, pollutant emissions, climate change, as well as biochemical cycles and human health. However, DNS with detailed chemistry has been computationally unaffordable due to the high dimension of variables and the mandatory fine resolutions that are required both spatially and temporally. Recently, DNS of turbulence reactive flows with realistic hydrocarbon fuels have been successfully carried out on supercomputers in collaboration with the Sandia National Laboratories, following a significant reduction in both the dimension and the stiffness of the involved detailed kinetics. Most of the major difficulties being faced in the reduction process, such as variable elimination from largescale
nonlinear systems and analytic solution of quasi steady state equations, have been plaguing the scientific community for decades. To solve each of these problems satisfactorily, we have developed a suite of new techniques involving graph theory, binary integer programming, singular perturbation, and spectral analysis. In this talk, representative components of these methods will be discussed and their potential impacts on other fields will be outlined.

Location: Stauffer Science Lecture Hall, Rm 102

Audiences: Everyone Is Invited

Contact: April Mundy