| description abstract | Modern gas turbine combustors operating in leanpremixed mode are prone to thermoacoustic instabilities. In annular combustion chambers, usually azimuthal acoustic modes are the critical ones interacting with the flame. In case of constructive interference, high amplitude oscillations might result. In this paper, the azimuthal acoustic field of a fullscale engine is investigated in detail. The analyses are based on measurements in a fullscale gas turbine, analytical models to derive the system dynamics, as well as simulations performed with an inhouse 3d nonlinear network model. It is shown that the network model is able to reproduce the behavior observed in the engine. Spectra, linear growth rates, as well as the statistics of the system's dynamics can be predicted. A previously introduced algorithm is used to extract linear growth rates from engine and model time domain data. The method's accuracy is confirmed by comparison of the routine's results to analytically determined growth rates from the network model. The network model is also used to derive a burner staging configuration, resulting in the decrease of linear growth rate and thus an increase of engine operation regime; model predictions are verified by fullscale engine measurements. A thorough investigation of the azimuthal modes statistics is performed. Additionally, the network model is used to show that an unfavorable flame temperature distribution with an amplitude of merely 1% of the mean flame temperature can change the azimuthal mode from dominantly rotating to dominantly standing. This is predicted by the network model that only takes into account flame fluctuations in axial direction. | |
