Physics based algorithm that uses acoustic precursors indicating a Lean Blowout (LBO) is proposed for lean blowout detection in combustors. The proposed technology is presented for a typical multi-nozzle Dry Low Emission (DLE) combustor. Three narrow band dynamic pressure tones, namely LBO (low frequency) tone, high fuel to air (F/A) ratio (hot tone) and low F/A ratio (cold) tones are identified as strong precursors to behavior consistent with combustion instability as LBO event evolves. The likelihood of LBO is computed using a statistical model operating on the RMS value of the LBO tone. Two additional pieces of evidential information are built respectively using the relative change of the RMS values of these three tones and the frequency shift of the high F/A tone. A data fusion algorithm then uses these two evidential signatures to enhance the LBO probability based on the LBO tone. Field test results on GE’s commercial multi-nozzle combustor gas turbines showed that the algorithm is practical and effective giving enough lead-time to take corrective action to avoid a blowout. A closed loop controller that modulates global manifold fuel splits for all the combustors or total fuel flow of individual combustor in a multi-combustor (multi-can) turbine to avoid the incipient blowout upon detection using the method presented in this paper can then be easily designed.

This content is only available via PDF.
You do not currently have access to this content.