In this study, we employ a method based on Bayesian statistics to determine the rate of acoustic decay from dynamic pressure measurements inside a combustor. It is common, that lean premixed flames tend to drive thermoacoustic instabilities at specific eigenfrequencies. Hence, the dissipation of acoustic energy inside the combustor, its absorption at the boundaries and its transfer over the in- and outlets must always exceed the acoustic excitation from the flame to avoid pulsations. Quantitative measures for the level of stability are of high technical relevance. In that context the occurring eigenfrequencies and their damping rates are important indicators for the stability margin of gas turbine combustors. A modular swirl burner is investigated in an atmospheric single burner test rig under lean premixed conditions. For the experimental determination of the damping rates, a siren is used to externally excite resonant frequencies of the combustion system. After interrupting the forcing abruptly, time series of the decaying signals are recorded by dynamic pressure sensors inside the combustion chamber. For the analysis of this data, an algorithm based on a Bayesian network approach, which uses a Gibbs Sampler is employed. Probability distributions of frequencies and decay rates are obtained with the Markov-Chain Monte-Carlo (MCMC) method. For the investigated configuration, the influence of the acoustic boundary conditions and the preheat temperature on eigenmodes and damping rates is evaluated. Finally, the results are compared to a network model of the test rig. With that approach, the Open-Loop Gain (OLG) is evaluated for the frequency range of interest. Eigenfrequencies as well as their corresponding damping rates are obtained from Nyquist analysis.

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