Correlation analysis of direct numerical simulation data of turbulent non-premixed flames

Mathematical modeling of turbulent reacting flows is and will remain a big challenge. A calculation from first principles is computationally prohibitive, and several statistical modexls have been developed to simplify the numerical simulation of turbulent flames. The problem of these approaches is that they remain models that can only be improved if information about the coupling of turbulence, chemical reaction, and molecular transport is available. Besides experimental investigations, direct numerical simulation (DNS) is one of the tools able to supply such information. One major problem of DNS is the overwhelming amount of data that is obtained from each calculation, and methods are needed that allow to extract the information that one is really interested in. It is known that in reacting flows, the chemical kinetics does not access the whole composition space, but it is confined to low-dimensional subspaces, the so-called intrinsic low-dimensional manifolds (ILDM). Thus, the fast chemical processes introduce correlations between the species concentrations. These correlations can be extracted from the DNS data. In this work, we present DNS calculations of a turbulent non-premixed hydrogen-air flame and a method that allows an analysis of the data and a determination of the local ILDMs. The results reveal interesting information on the coupling of the chemical kinetics with the turbulent processes. However, the method described is not limited to this specific example. It can be used for the analysis of DNS of other flames (premixed, other fuels) and even to analyze data obtained from experimental investigations.