No abstract

In this paper, innovative investigation of the flame dynamics of an industrial gas turbine combustor using high-speed imaging and flame luminosity measurement has been reported. Both digital and optical integration methods have been applied to study the flame dynamics of an unstable combustion mode, which has a fixed peak frequency around 105 Hz. It is found that the light emission of the seeded combustion possesses the same dominant peak frequency as the acoustic emission. On the other hand the natural flame light emission failed to match the spectrum of the acoustic emission due to signal drop-out, which is a common problem for direct imaging of unseeded flame. The paper has demonstrated that the transition from stable to unstable combustion could be captured successfully by high-speed imaging using a pre-trigger mode. Three stages of transition from stable combustion mode to unstable mode have been identified. In stage one, the flame luminosity and fluctuation increased gradually, which could be assumed as the incipiency of the instability. In stage two, there was a sudden and sharp burst in flame luminosity, which indicates that the combustion process was becoming very intense in the middle of transition. In the third stage, the drastic flame luminosity dissipated slightly and oscillated with constant amplitude and the combustion locked to unstable combustion mode. The two applied techniques have been cross-checked with reference to the unique peak frequency observed.

No abstract

No abstract

Stereoscopic imaging and reconstruction technique is introduced to reconstruct the flame structures that are subject to acoustic excitation. The laminar diffusion flame under investigation was created in a cylindrical tube and excited by a loudspeaker. Stereo images were taken at a shutter speed of 1/10000 th second using a 'stereo camera' and the depth cue of the flame structures along the camera viewing direction were then computed using machine vision methodology. By visualizing the computed three-dimensional flame models, as well as judging the corresponding attribute such as surface gradient, the understanding of the flame and acoustic wave interaction could be improved.