An Experimental Investigation of Premixed Combustion in Extreme Turbulence

Abstract

This work has explored various aspects of high Reynolds number combustion that have received much previous speculation. A new high-Reynolds number premixed Bunsen burner, called Hi-Pilot, was designed to produce turbulence intensities in the extreme range of turbulence. The burner was modified several times in order to prevent boundary layer separation in the nozzle, and a large co-flow was designed that was capable of maintaining reactions over the entire flame surface. Velocity and turbulence characteristics were measured using a combination of Laser Doppler Velocimetry (LDV) and Particle Image Velocimetry (PIV). Flame structure was studied using a combination of formaldehyde (CH2O), hydroxyl (OH), and the CH radical. Planar Laser Induced Fluorescence (PLIF). The spatial Overlap of formaldehyde and OH PLIF qualitatively measures the reaction rate between formaldehyde molecules and OH radicals, and is a measure of the reaction layers of the flame. CH PLIF provides an alternative measure of the reaction zone, and was measured to compare with the Overlap PLIF results. Reaction layers are the full-width at half-maximum of the Overlap or CH PLIF signal, and extinction events were defined as regions where the PLIF signal drops below this threshold. Preheat structures were measured using formaldehyde PLIF, and are defined as beginning at 35% of the local maximum PLIF signal, and continue up to the leading edge of the reaction layer. Previous predictions of regime diagram boundaries were tested at the largest values of turbulent Reynolds number to date. The Overlap and CH PLIF diagnostics allowed extensive testing of the predicted broken reaction zones boundary of Peters. Measurements indicated that all run conditions are in the Broadened Preheat - Thin Reaction layers regime, but several conditions are expected to display a broken reaction zone structure. Therefore the work shows that Peters’s predicted boundary is not correct, and therefore a Karlovitz number of 100 is not a valid criteria for broken reactions in the Bunsen geometry. Several measures of the turbulent burning velocity, including the global consumption speed and the extent of flamelet wrinkling, were measured at these conditions. Reaction layers for the burning velocity measurements were provided by the OH PLIF. The measurements showed that the global consumption speed continues to increase for all levels of turbulence intensity u'=SL. In contrast, the flame surface wrinkling rapidly increases the flame surface area for u'=SL < 10, but the flame surface area does not increase further at larger turbulence intensities. This indicates that the flame is not in the laminar flamelet regime, and the consumption rate per unit of flame surface area must be increased. The turbulent diffusivity is thought to be the mechanism enhancing the consumption rate, which is a scenario first hypothesized by Damköhler. The flame structure and burning velocity measurements motivated the measurements of the evolution of turbulence through regions of very thick preheat layers. This measurement utilized simultaneous PIV and formaldehyde PLIF in order to obtain conditioned statistics of the turbulence as a function of eta, the distance from the reaction layer. Together, the results tell a consistent story, and deepen our understanding of premixed combustion at large turbulent Reynolds number.