Turbulent burning velocity is a crucial factor in describing how fast premixed flame can burn, while it is still not well-predictable under various turbulence levels and flame conditions. A clearer definition of turbulent burning velocity based on the reaction rate will help understand the mechanism of turbulent premixed flames since current definitions of turbulent burning velocities may not be fully consistent in different configurations of burners and will lead to confusion and unsolvable discrepancies between different measurements. A clean review of previous experimental investigations, filtered by the type of burner configurations and the re-calculated flame characteristics, was firstly conducted to illustrate the existing problem of predicting turbulent burning velocities. The current issues of experimentally measured datasets consist of two parts, the significant discrepancies of turbulent burning velocities under comparable conditions and the inequality of velocity enhancement and area enhancement based on flamelet assumptions (’missing area’).

The definition of turbulent burning velocities was revisited and re-clarified to avoid systematic error based on the conservation equation of Favre-averaged progress of reaction across the flame brush in different turbulent premixed flames (stationary flames and spherical expanding flames). Turbulent consumption speeds s_c were provided at any arbitrary isocontour of ¯c, and connected with local displacement speeds sT in both Bunsen and spherical expanding flames. sc,0 = m˙ /(ρ_uA0) was found to be equal to sT,0 (displacement speed at the leading edge of the flame brush, iso-contour of ¯ c = 0) in Bunsen flames as the most consistency definition of turbulent burning velocities, while there was a non-negligible discrepancy between sc,0 and sT,0 in spherical expanding flames.

The transportation of fuel/mixture by convection, turbulent flux and molecular diffusion was theoretically analysed and experimentally measured. Relevant experimental validation was conducted on a turbulent piloted Bunsen flame by the 2D laser diagnostic. A number density method was defined to detect flame edges from particle Mie scattering images so that a high-speed PIV system could provide both velocity field and instantaneous flame edges with an acceptable uncertainty. By that, turbulent flux and molecular diffusion could be evaluated across the flame brush with well-extrapolations onto the leading edge and trailing edge. Through modelling the reaction term by the BML model, the balance of the conservation of ˜ c was estimated with uncertainty analysis, although the closure was not fully realised across the entire flame brush. One explanation was located in the filtering effect of mean reaction rates that DNS results had validated.

Another approach to resolving the issue of missing area enhancement was to evaluate the 3D effect of measurements of flame surface area and the discrepancy between actual 3D flame surface density (FSD) Σ and conventional 2D FSD estimations. By converting the definition of 3D flame surface density into an angle form, 3D FSD Σω were measured by a cross-technique that measured flame edges on two orthogonal planes and evaluated the ensemble-averaged normal angle of the 3D flame surface. Such 3D FSD was measured at different heights of a turbulent Bunsen flame and was compared with conventional 2D FSD and corrected 2D FSD measured on the vertical plane. A discrepancy of 20%∼30% between 2D and 3D FSD was observed, and resources of this discrepancy were also presented.

To acquire more 3D information about flame surfaces, a scanning method based on an acoustic-optic-deflector was embedded into the cross-technique to conduct the scanning process with the cross-planar measurements simultaneously. Half of the turbulent Bunsen flames were scanned and reconstructed to evaluate global flame surface density Σg A/V. Two approaches measured local flame surface densities, angles measured from two cross planes Σω and reconstructed flame surface near the intersection line of two cross planes ΣA/V. Principal curvatures κ1, κ2 and mean curvatures κm were also measured via the combined technique to evaluate the effect of flame surface curvatures on the laminar burning velocities, which could give a reasonable compensation on the non-closure of the conservation of ˜ c . Overall, this dissertation has endeavoured to resolve problems of defining and measuring turbulent burning velocities under low turbulence levels as instructional methods for further investigation of turbulent flames with even higher turbulence levels.