Dieter Most

Structure of locally quenched highly turbulent lean premixed Flames

The local flame structure of a premixed swirl-stabilized gas turbine burner has been investigated, where “thickened flames” or flames in the “stirred reactor regime” are expected from the Borghi diagram. Simultaneous 2-D OH and temperature measurements show that the flame structure fluctuates between four typical flame modes: (1) flamelet-like burning, (2) modified preheat zone with sharp reaction zone, (3) locally quenched reaction zone, and (4) hot nonreacted holes. In mode (3), evidence for local quenching processes is found with no superequilibrium OH between fresh gas and recirculating burned gas. In about 10% of the obtained images, regions without detectable OH but with temperatures above 1300 K are seen, surrounded by sharp reaction zones with low thermal gradients (mode 4). Although the explanation is not clear yet, significant modifications of the reaction processes due to strain combined with transient effects seem to occur here. From 2-D and 3-D measurements, gradients and flame thickness distributions have been determined, showing strong fluctuation of the thermal gradients, but on average, no significantly broadened instantaneous flame fronts, contrary to the expectation of thick flames. Instead, mean gradients conditioned to the local reaction progress variable c=(T−T 0 )/(T max −T 0 ) =0,50 result in a slightly thinned thermal thickness, and within the preheat zone at c =0.25 a thinning of about 30% is found. For an explanation. strained laminar calculations are helpful, indicating the same trend for lean CH 4 -air mixtures. Obviously, strain effects from medium- and large-scale eddies have much more influence than diffusive effects from entrained small eddies. While the thermal thickness distribution shows large fluctuations, the OH thickness distribution (deduced from the OH ascent) is narrow. Comparing the Oh ascent thickness (similar to the calculated width of the CH peak) with the inner layer Kolmogorov size, a ratio of order 1 is found, when local quenching occurs.

Simultaneous two-dimensional flow velocity and gas temperature measurements by use of a combined particle image velocimetry and filtered Rayleigh scattering technique

For the first time, to the best of our knowledge, two-dimensional instantaneous measurements of the flow velocity and the gas temperature have been performed in a turbulent flame with simultaneous use of particle image velocimetry and planar filtered Rayleigh scattering. These single-shot measurements provide simultaneous information on the local flame structure (curvature and temperature gradients) and on the local flow conditions (vortices, flow divergences, and strain rates). The applicability of the technique is demonstrated in a turbulent lean CH(4)-air V flame.

Direct determination of the turbulent flux by simultaneous application of filtered rayleigh scattering thermometry and particle image velocimetry

In moderately and highly turbulent, premixed CH 4 /air flames, two-dimensional simultaneous measurements of instantaneous velocity, temperature, and density have been made using particle image velocimetry (PIV) and planar filtered Rayleigh scattering thermometry (FRS thermometry). The resulting joint data sets enable the direct determination of the Favre mean turbulent flux (FMTF), which is of high interest for combustion modeling. An evaluation strategy is given for the calculation of the FMTF using this joint temperature-velocity data. Measurements indicate the existence of countergradient transport in the highly turbulent bluff-body-stabilized flame, while in the wire-stabilized flame with moderate turbulence, the FMTF fluctuates around zero. Additionally, the direct FMTF determination was tested against other models of the turbulent flux using either a scalar gradient approach or binary conditioned velocity data, following the thin-flame approach of the Bray-Moss-Libby (BML) model. While only the latter shows good agreement with the direct FMTF determination. In the bluff-body-stabilized flame from the two-dimensional plots of the axial FMTF, regions with either dominant gradient diffusion or dominant countergradient diffusion can be identified within the same flame. This feasibility study outlines the potential of simultaneously applied FRS-PIV techniques for the direct determination of unclosed terms in the Favre mean balance equations.

Lifted reaction zones in premixed turbulent bluff-body stabilized flames

Flame liftoff phenomena are well known in turbulent non-premixed flames but less investigated in turbulent premixed flames. Here, in the special case of flames stabilized by the recirculation of burned gases a particular type of lifted reaction zone can be observed which is not only originated simply by the lack of heat, but by some other mechanisms (e.g., ignition delay). In principle, this premixed flame liftoff either might be based on turbulent effects (e.g., local flame extinction due to high turbulent strain rates, which are especially intensive in the shear flow region near the exit) or are considered to be based on chemical ignition delay processes. To investigate these liftoff mechanisms, experiments have been conducted at bluff-body-stabilized premixed methane/air flames, where flow and flame parameters have been varied systematically over a broad range of exit velocities and stoichiometries. While the liftoff height has been measured with laser-induced OH fluorescence, also the local characteristic turbulent flow and temperature field has been measured to allow correlated data determination for the liftoff height. Three different theoretical model approaches are discussed for the prediction of this lifted reaction zones by comparing local flow, turbulence and reaction parameters with the local burning or non-burning status. It is found that for this burner configuration not only one but two different liftoff criteria must be met. For very lean mixtures the chemically dominated ignition delay is found to be the rate-determining step. For other cases, the liftoff height can be determined by a newly described turbulent mixing dominated model. In contrast to this, a dimensionless turbulent strain rate, often described with a Kovasznay or Karlovitz number, is not a suitable criterion here.