On the Structure of Premixed Flames Subjected to Extreme Levels of Turbulence

Abstract

Developing next-generation propulsion and energy production devices that are efficient, cost-effective, and generate little to no harmful emissions will require highly-accurate, robust, yet computationally tractable turbulent combustion models. Models that accurately simulate turbulent premixed combustion problems are particularly important due to the fact that burning in a premixed mode can reduce exhaust emissions. A common tool employed to identify when a particular model might be more appropriate than others is the theoretical Borghi Diagram, which possesses boundaries that are meant to separate various regimes of combustion (i.e. where a particular model is superior to others). However, the derivations of these boundaries are merely based upon intuition and dimensional reasoning, rather than experimental evidence. This thesis aims to provide such evidence; furthermore, it proposes novel approaches to delineating regimes of combustion that are consistent with experimental results.

To this end, high-fidelity flame structure measurements were applied to premixed methane-air Bunsen flames subjected to extreme levels of turbulence. Specifically, 28 cases were studied with turbulence levels (u'/S_{L}) as high as 246, longitudinal integral length scales (L_{x}) as large as 43 mm, and turbulent Karlovitz (Ka_{T}) and Reynolds (Re_{T}) numbers up to 533 and 99,000, respectively. Two techniques were employed to measure the preheat and reaction layer thicknesses of these flames. One consisted of planar laser-induced fluorescence (PLIF) imaging of CH radicals, while the other involved taking the product of simultaneously acquired PLIF images of formaldehyde (CH_{2}O) and hydroxyl (OH) to produce ``overlap-layers." Average preheat layer thicknesses are found to increase with increasing u'/S_{L} and with axial distance from the burner (x/D). In contrast, average reaction layer thicknesses did not vary appreciably with either u'/S_{L} or x/D. The reaction layers are also observed to remain continuous; that is, local extinction events are rarely observed. The results of this study, as well as those from prior investigations, display inconsistencies with predictions made by the theoretical Borghi Diagram. Therefore, a new Measured Regime Diagram is proposed wherein the Klimov-Williams criterion is replaced by a metric that relates the turbulent diffusivity (D_{T}) to the molecular diffusivity within the preheat layer (D*). Specifically, the line defined by D_{T}/D* ~ 180 does a substantially better job of separating thin flamelets from those with broadened preheat yet thin reaction layers (i.e. BP-TR flames). Additionally, the results suggest that the BP-TR regime extends well beyond what was previously theorized since neither broken nor broadened reaction layers were observed under conditions with Karlovitz numbers as high as 533. Overall, these efforts provide tremendous insights into the fundamental properties of extremely turbulent premixed flames. Ultimately, these insights will assist with the development and proper selection of accurate and robust numerical models.