Detailed simulations of the DLR auto-igniting pulsed jet experiment

Abstract

Abstract Numerical simulations of an auto-igniting pulsed jet in a vitiated co-flow experiment by DLR (German Aerospace Center) are conducted by highly-resolved large-eddy simulations using direct chemistry with an augmented reduced mechanism. The experiments consist of two operation modes: continuous injection used for code-verification and pulsed injection utilized for fundamental investigation of auto-ignition dynamics. Initially, reference one-dimensional self-igniting counter-flow flames are investigated. Then, a grid convergence study has been performed. It is shown that even a coarser grid would be sufficient to describe the ignition chemistry since the ignition kernel appears at low velocities and fuel-lean conditions in zones of low scalar dissipation rates. For the statistically steady jet, numerical predictions are in a very good agreement with the experiments, giving confidence in the applied models. For the pulsed jet, all of the predicted ignition delay times and locations are in the range of the experimental observations. Time-resolved statistics reveal that thermochemical properties of the gas in a pulsed jet achieve states that are impossible to reproduce in laminar conditions. For further analysis, hydroxyl and formaldehyde are chosen as a marker for the established flame and for the ignition, respectively. In laminar conditions, these two species are perfectly correlated. However, the unsteady dynamics of the pulsed jet invalidates the correlation between the minor species chemistry prior to ignition. This yields the discrepancy in the auto-ignition delay time and the location of the ignition kernel between different pulses, as the thermochemical state needed for the ignition occurs in a random manner.