Real-time monitoring of dynamical state changes in staged combustion

We describe techniques for diagnosing the state of coal-fired utility burners using dynamic characteristics of the output of optical flame scanner signals. The analysis techniques are optimized for targeting dynamical features associated with nonlinear instabilities that develop as burner operating parameters are changed. Various specific instability indicators are used, including shifts in probability distributions, temporal asymmetry and coarsed-grained descriptions of unstable periodicities. We show that careful application of such methods can accurately characterize a range of different flame states. Specifically, transitions through bifurcation points between attached and lifted flames are targeted, giving insight into causes of instability such as stoichiometry or feed and flow variations. We demonstrate results from the application of these methods to utility-scale staged pulverized-coal burners in a real-time software package. The Flame Doctor™ burner-monitoring software is presently undergoing plantimplementation trials in a program sponsored by the Electric Power Research Institute and participating utilities. In a practical application, we show how real-time monitoring and intervention can significantly mitigate adverse combustion conditions.