Development of a 3D laser imaging system and its application in studies of turbulent flame structure

The necessity for increasing the efficiency of combustion engines whilst simultaneously reducing the emissions of various pollutants is widely accepted. With reference specifically to premixed spark ignition (SI) engines, the turbulence/flame interaction is of particular relevance as it not only determines the maximum possible burning rate in such an engine, but also if flames are likely to extinguish (quench). The present study describes a three-dimensional laser imaging technique which has been developed to study turbulent premixed flames in a fan-stirred combustion bomb. This has, for the first time, allowed the full 3D structure of flames developing at engine-like conditions to be analysed without the assumptions required in previous works using 2D flame imaging techniques. Methane/air mixtures at a starting temperature and pressure of 298K and 0.1 MPa respectively and an equivalence ratio with respect to stoichiometry of 0.6 were centrally ignited using a laser ignition system at various root-mean-square (RMS) turbulence velocities, where for each explosion, a 3D image of the growing flame was captured at various instants in time. The flame images were quantitatively characterised to yield flame surface area, burned gas volume, reaction progress variable and flame surface density. From the flame surface areas and burned gas volumes, turbulent burning velocities were directly measured from surface area ratios and compared with correlations of turbulent burning velocities made previously at Leeds using 2D flame imaging techniques. This tested the assumption that flame structure observed with 2D imaging techniques gave a good representation of the overall 3D flame structure. An excellent agreement was found between the datasets, indicating that such assumptions were valid under the conditions employed in the present work. The surface area ratio data obtained in the present work were also compared with the predictions of direct numerical simulation (DNS) studies conducted at Cambridge University, where it was found that the experimentally derived results typically exhibited slightly higher flame surface area ratios. This was attributed to the combustion chamber geometry in the DNS study and differences in mixture properties between the two studies, with more work at similar starting conditions required to elucidate the exact cause of the differences. Reaction progress variable was calculated for the 2D laser sheet images obtained in the present work and their 3D counterparts. Although a good agreement was seen where the central laser sheet image in a 3D reconstruction was analysed, disparities were observed when laser sheet images at a fixed position in the combustion bomb was used at high root mean square turbulence velocities. This indicates that the use of fixed position laser sheet imaging techniques, such as particle image velocimetry, are of limited use for highly turbulent flame analysis.