Session Q02: Turbulent Combustion IV

How Fast is the Flame: Displacement vs. Burning Speeds of Fast, Highly Compressible Turbulent Premixed Flames

A comparative analysis of numerically and experimentally determined speeds of fast, highly compressible premixed turbulent flames is presented. Analysis focuses on direct numerical simulations of the Turbulent Shock Tube (TST) facility developed at the University of Central Florida. The TST generates flames, which exhibit highly time-dependent evolution with significant flame acceleration, overpressures, and turbulence generated within the flame brush. Flame speed, a defining characteristic, is generally determined in experiments through measurement of the flame displacement speed, S_D, relative to the upstream flow. It is not currently clear how accurately S_D represents the burning speed, S_T, of a turbulent flame, which is the proper measure of fuel consumption and heat release rate. Here we investigate the differences between numerically calculated S_D and S_T, as well as S_D experimentally determined in the TST. Results show considerable differences between S_D and S_T, which become particularly pronounced as the flame evolution grows increasingly unsteady. This study aims to develop the understanding of various flame-speed measures in fast, highly unsteady turbulent combustion regimes directly relevant to novel combustion systems such as scramjets and detonation-based engines.   

Analysis of Highly-Turbulent Premixed Flames Using a Retrospective Lagrangian Approach

A retrospective (i.e., backwards in time) Lagrangian analysis is used to examine premixed flame structure and dynamics for highly turbulent reactant mixtures. The computational cost of such a retrospective analysis can be enormous due to the associated data storage requirements, but this approach enables the relatively straightforward study of the dynamical origins of transient flame phenomena such as ignition, extinction, and pocket formation. In this talk, we present results from a retrospective Lagrangian analysis performed on data from a direct numerical simulation (DNS) of premixed n-dodecane combustion in an unconfined domain. The turbulence intensity in the DNS is quite high, with Karlovitz numbers of O(10²-10³), resulting in substantial thermochemical complexity. In particular, we show that fuel consumption and temperature rise within fluid parcels are frequently non-monotonic, resulting in frequent cooling events even after the temperature within a parcel has risen to 1400K or more, and we identify the dynamical origins of this non-monotonicity. Complex transient phenomena such as auto- and forced-ignition, extinction, and pocket formation are also identified, and the retrospective Lagrangian analysis is used to examine the dynamics leading to such events.   

Experimental-Numerical Comparison of Premixed Turbulent Flame Kernels in Expanding Supersonic Channel Flow

It was recently shown using experiments that premixed supersonic flame kernels exposed to a mean acceleration develop a vortex ring motion due to baroclinic torque (Ochs et al., FTC 2018, https://doi.org/10.1007/s10494-018-9947-x). One major conclusion is that, when compared to well established low-speed spherical flame studies, the growth of supersonic kernels is enhanced due to the vortex ring motion. However, the previous work utilized a line of sight Schlieren measurement to assess flame growth and it is unclear to what extent the observations would change if the full three-dimensional flame topology was available. In order to answer this question, extensive Large Eddy Simulations of supersonic premixed flame kernels are performed. Numerical validation is demonstrated by comparing the experimental flame speed to a numerical flame speed calculated using numerical Schlieren and processed in a similar fashion to the previous experimental data. Reasonable agreement is found across all cases. It is shown that the vortex ring motion has a non-negligible effect on the late time flame topology, which has important implications on turbulent flame speed scaling in supersonic expanding flows.   

Transient laser-induced ignition kernels in supersonic flow

Simulation predictions with quantified uncertainty are made for ignition of an axisymmetric slot jet on a centerbody injecting hydrogen into a M=2.7 crossflow that is seeded by the optical breakdown of a focused laser (LIB). The LIB dissociates the local mixture into elemental species, generates vorticity, and heats the gas above 10⁴ K, activating reactions within ≈10^-9 s and producing the ignition kernel. The flow near the kernel is complex: turbulent, compressible, separating boundary layers, non-premixed fuel and oxidizers, and velocities induced by the LIB exceeding 10³ m/s. As a prediction testbed, blind comparisons are made with corresponding experiments, and they focus particularly on the kernel's position, size, and orientation based on the light intensity emitted by excited hydroxyl radical. In the simulations, this comparison entails using a detailed kinetics model and simulating the same data-acquisition characteristics (e.g. frame rate and exposure time). Depending on the breakdown site, the kernel orients and convects in a primary direction with the flow. The initially sub-millimeter-scale kernels are three-dimensional, elongate along their principal axis, and grow by over a factor of ten as ignition ensues.   

On burning rate enhancement in spherically expanding turbulent flame

The mechanism responsible for the enhancement of the burning rate in turbulent premixed flames is an active topic of research. The dependence of the burning rate on the turbulence intensity is well documented through a large number of experiments, although the scatter is high even at low turbulence levels. In the present study, we aim to explain this dependence with data from three Direct Numerical Simulations (DNS) of spherically expanding turbulent flames at different Reynolds numbers. First, the ratio of turbulent surface area to that corresponding to the mean radius is decomposed into factors involving the flame brush thickness and the surface density function. Then a detailed governing equation for the flame brush thickness is derived systematically starting from the surface density transport equation. It is observed that the flame brush evolves due to four different mechanisms, one of which is the turbulent transport. However, the gradients in the mean velocity field, dependence of propagation velocity on mean curvature, and the flame stretch term also contribute to the evolution of the brush. Our analysis proves that the burning rate enhancement in turbulent flames depends to a large extent on the configuration, which may explain the large scatter in experimental data.