Monitoring of oscillatory characteristics of pulverized coal flames through image processing and spectral analysis

This paper presents the monitoring of the oscillatory characteristics of pulverized coal flames using image processing and spectral analysis techniques. The instrumentation system employed in this investigation is an integral part of a multifunctional flame monitoring system, being capable of monitoring the oscillatory frequency of a flame on a two-dimensional and concurrent basis. A quantitative flicker frequency is defined as the power-density-weighted mean frequency over the spectral range to represent the oscillatory characteristics of a specific region of the flame. Digital filtering techniques incorporating direct gray-level thresholding and wavelet shrinkage algorithms are employed to reduce background noise from flame images and white noise from the resulting flame frequency signal. A series of tests was undertaken on an industrial-scale coal-fired combustion test facility (CTF) under a range of operating conditions. Relationships between the measured flame oscillatory frequency and the process data including emissions are identified. Results obtained demonstrate that the flame oscillatory frequency responds in predictable ways to the effects of operating conditions on the dynamic nature of the flame.