The Dynamics of Turbulent Premixed Flames: Mechanisms and Models for Turbulence-Flame Interaction. Turbulence-Flame Interaction.

Abstract

The use of turbulent premixed combustion in engines has been garnering renewed interest due to its potential to reduce NOx emissions. However there are many aspects of turbulence-flame interaction that must be better understood before such flames can be accurately modeled. The focus of this dissertation is to develop an improved understanding for the manner in which turbulence interacts with a premixed flame in the 'thin flamelet regime'. To do so, two new diagnostics were developed and employed in a turbulent slot Bunsen flame. These diagnostics, Cinema-Stereoscopic Particle Image Velocimetry and Orthogonal-Plane Cinema-Stereoscopic Particle Image Velocimetry, provided temporally resolved velocity and flame surface measurements in two- and three-dimensions with rates of up to 3 kHz and spatial resolutions as low as 280 um.

Using these measurements, the mechanisms with which turbulence generates flame surface area were studied. It was found that the previous concept that flame stretch is characterized by counter-rotating vortex pairs does not accurately describe real turbulence-flame interactions. Analysis of the experimental data showed that the straining of the flame surface is determined by coherent structures of fluid dynamic strain rate, while the wrinkling is caused by vortical structures. Furthermore, it was shown that the canonical vortex pair configuration is not an accurate reflection of the real interaction geometry. Hence, models developed based on this geometry are unlikely to be accurate.

Previous models for the strain rate, curvature stretch rate, and turbulent burning velocity were evaluated. It was found that the previous models did not accurately predict the measured data for a variety of reasons: the assumed interaction geometries did not encompass enough possibilities to describe the possible effects of real turbulence, the turbulence was not properly characterized, and the transport of flame surface area was not always considered. New models therefore were developed that accurately reflect real turbulence-flame interactions and agree with the measured data. These can be implemented in Large Eddy Simulations to provide improved modeling of turbulence-flame interaction.