Adaptive Control of Combustion Instability Based on Dominant Acoustic Modes Reconstruction

Abstract

Strong combustion instability within gas turbines may severely reduce the working range, exacerbate emissions, and shorten the engines' life cycles. A typical and practical control strategy is to introduce a secondary heat release rate perturbation using a high-frequency fuel valve so as to nullify the positive feedback between pressure and heat release rate oscillations. This article presents an observer-based adaptive controller, which is aiming at suppressing time-varying multiple unstable modes without knowing the characteristics of the combustion and actuation systems. The key part of this controller is an observer, which is capable of online estimating the dominant frequencies, amplitude, phases, and delay-free reconstruction of the dominant modes. The procedures for frequency estimation are somewhat similar to those of the Discrete Fourier Transform, but this observer is suitable for real-time computation, and is capable of more accurate and faster frequency estimation using a shorter sample length. The procedures for attenuating multiple unstable modes with unknown characteristics are demonstrated by controlling a “virtual” combustor that is a system identification model (Riley et al., 2004) containing 2 unstable modes with comparable amplitude and not-well-separated frequencies.