Influence of Noise Characteristics on Thermoacoustic Instability Growth Rate Estimates

Thermoacoustic oscillations in gas turbine combustors lead to sub-optimal performance and structural failures. Passive acoustic damping devices are employed to suppress these oscillations. Estimating system parameters that characterize the instability is required for designing these devices. Noise is an inherent feature of combustors and is beneficial in predicting instability via analysis of noise-induced dynamics. Previous reports on noise-induced dynamics model noise as an additive Gaussian white noise. However, experimental studies have indicated that noise, in particular from turbulent flame, is low-pass in character. In this work, we experimentally investigate the influence of Ornstein-Unhlenbeck colored noise on system identification using a prototypical thermoacoustic system exhibiting subcritical Hopf bifurcation. We investigate the effects of noise color and intensity on estimation of growth rates and show limitations of the white noise approximation in stable, bistable and linearly unstable regions. We find that estimated growth rates exhibit a deviation of 0–30% from true values with increasing noise color and intensity. We find that colored noise model, based on known statistics of combustion noise, predicts system growth rates significantly better than white noise model.