Back to Table of Contents

46991-AC9
Experiments on Strained Premixed Flames in the Distributed Reaction Regime

Alessandro Gomez, Yale University

Highly turbulent premixed flames have been stabilized in a countercurrent configuration. The flames are established by counterflowing a hot combustion product stream from a separate burner to a stream of fresh, premixed reactants. A properly designed and positioned high-blockage plate is used to generate high turbulent Reynolds numbers, preserving flow homogeneity. Stable flames at bulk strain rates up to 1.3 10³ of s^-1 could be stabilized resulting in turbulent Reynolds numbers (Re) up to 1.3 10³ and Karlovitz numbers (Ka) ranging from 1.5 to 400, depending on the strain rate and the flame stoichiometry.

In turbulent premixed combustion, experimental investigation of the non-flamelet regime, often called broken reaction zone regime or distributed reaction zone regime, still remains an open challenge. The non-flamelet regime corresponds to a situation of discontinuous flame sheets locally extinguished by turbulent eddies and is expected in very intense turbulence, at turbulent Karlovitz number greater than 100. Conditions of strong turbulence and chemistry interaction are particularly hard to reproduce in a laboratory-scale burner and rare evidence of the non-flamelet regime under such conditions has been reported to date. Furthermore, in practical combustors and engines, additional factors, such as volumetric heat loss in the vicinity of walls, intense strain rate imposed by the mean convective flow field or local mixing of the reactants with combustion products, can contribute to reduce the heat release rate, which eventually may lead to local extinction by turbulence and possible re-ignition. In this project we demonstrated how highly turbulent flames with real flame effects, such as heat losses and high strain rates, can be established in a compact, bench-top experiment, that is amenable to computational modeling and even direct numerical simulation. The configuration shows a rich phenomenology, which make it the ideal target for model testing.

            Six flames were considered (A-F) and the phenomenological appearance of two of them is shown in Fig. 1 via OH Planar Laser Induced Fluorescence (PLIF). In each picture we notice: a dark region on the top, corresponding to the cold reactant part of the domain in which no OH can be detected; a relatively weak signal diffused throughout the bottom stream, as a result of the fact that under the high temperature of the hot combustion products OH is present in small concentrations; between these two regions, one can notice brighter region(s) in all but the last picture that can be attributed to the OH generated by the turbulent premixed flame. The picture on the left shows a stoichiometric CH₄/air flame at a strain rate of 980 s-1 , Re= 670 and Ka= 1.5. The flame is singly connected and wrinkled. The picture to the rights shows OH PLIF for a flame with the same strain rate and Re, but at lower equivalence ratio F=0.7, which doubles Ka, weakens the flame and causes local extinction. Despite the modest Ka, it should fall in the non-flamelet regime. Yet, the OH evidence of local extinction, suggests otherwise.

            The so-called extended Borghi diagram is used to identify various turbulent combustion regimes, with turbulent velocity and integral length scales normalized with respect to the laminar flame speed and the flame thickness (Fig. 2). Six flames were considered, A-F. Scaling considerations would suggest that  Flames A, B, C and E fall in the flamelet regime, whereas Flame D should belong to the non-flamelet regime. Because of strain and heat losses, also Flames C and F fall outside the flamelet regime in the broken/distributed reaction zone regime, whereas “Flame” E can not even be supported. The fact that Flame E with virtually identical position in the figure as Flame C is not burning is tentatively attributed to differences in CO concentration because of the different stoichiometry and different impact of the incomplete conversion to CO₂ at these high strain rates. Clearly, the extended Borghi diagram is an idealization that does not account for the effect of strain and heat losses, both of which are present in practical systems. The present burner allows for a systematic examination of these effects and should enable us to redraw a modified Borghi diagram accounting for these “realities” and redefining the boundaries of the various regimes.  Also noteworthy is the fact that the achievable experimental conditions for these premixed flames overlap part of the domain of IC engines.