SUMMARY
While lean, premixed combustion can reduce nitrogen oxide (NOx)emissions from both aircraft and land based gas turbines,there is a greater risk of lean blowout (LBO). Active control techniques are therefore sought which can stabilize a lean flame and prevent LBO. This involves developing detection methods to assess flame attributes and gauge LBO probability, actuation strategies to stabilize the flame, and control logic to complete the feedback loop. The present work has resulted in the development of flame detection, dynamic modeling, margin estimation, actuation, and control techniques. Optical and acoustic methods have been developed or improved to detect two blowout precursor types – localized extinction pockets and brief, flame shape modulations. The latter behavior is studied in the context of swirling flow dynamics. The understanding of dynamic behavior allows the formulation of an empirical model that determines when flame modulations will occur. This model is then applied to detect both precursor types from the same sensor. Margin estimation methods have been developed that use the frequency of each precursor type to determine the probability of blowout. It has been shown that the flame’s blowout limit can be extended by redistributing fuel from the annular swirler to the central, premixed pilot. However, due to the emissions penalty, the control algorithm must command actuation only to maintain margin. Both rule-based and lead-lag control architectures have been developed and validated.