Aeroderivative gas turbine lean blowout (LBO) detection and isolation using acoustic precursors

This paper studies an online algorithm that can effectively detect a complete blowout (LBO) and a partial or incipient blowout (IBO) as well as isolate them in one specific ring by using only one acoustic sensor. The outputs can guide the successive blowout avoidance control logic by taking appropriate corrective action on the ring causing blowout or being blowout. The proposed algorithm is based on the RMS value of a tone around 15 Hz to 30 Hz that was identified as a precursor to blowout. Spectral signatures in other dominant tones were also identified as indications to assist ring IBO isolation. Its performance was successfully demonstrated with field data of GE LM2500 and LM6000PD machine.