Thermoacoustic Modeling of a Gas Turbine Using Transfer Functions Measured Under Full Engine Pressure

Abstract

Thermoacoustic transfer functions of a full-scale gas turbine burner operating under full engine pressure have been measured. The excitation of the high-pressure test facility was done using a siren that modulated a part of the combustion airflow. Pulsation probes have been used to record the acoustic response of the system to this excitation. In addition, the flame’s luminescence response was measured by multiple photomultiplier probes and a light spectrometer. Three techniques to obtain the thermoacoustic transfer function are proposed and employed: two acoustic-optical techniques and a purely acoustic technique. The first acoustical-optical technique uses one single optical signal capturing the chemiluminescence intensity of the flame as a measure for the heat release in the flame. This technique only works if heat release fluctuations in the flame have only one generic source, e.g., equivalence ratio or mass flow fluctuations. The second acoustic-optical technique makes use of the different response of the flame’s luminescence at different optical wavelengths bands to acoustic excitation. It also works, if the heat release fluctuations have two contributions, e.g., equivalence ratio and mass flow fluctuation. For the purely acoustic technique, a new method was developed in order to obtain the flame transfer function, burner transfer function, and flame source term from only three pressure transducer signals. The purely acoustic method could be validated by the results obtained from the acoustic-optical techniques. The acoustic and acoustic-optical methods have been compared and a discussion on the benefits and limitations of each is given. The measured transfer functions have been implemented into a nonlinear, three-dimensional, time domain network model of a gas turbine with an annular combustion chamber. The predicted pulsation behavior shows a good agreement with pulsation measurements on a field gas turbine.

Simulation of the flame describing function of a turbulent premixed flame using an open-source LES solver

“…Experimentally, several studies [14,15,5,6,16,11,17,18,19,20] have been conducted to determine the flame models needed for stability analysis. Recent examples of this include Noiray et al [9], who experimentally measured the FDF of an unconfined laminar burner; on combining with an acoustic model they obtained limit cycle predictions in excellent agreement with experiments.…”

“…Palies et al [21] extended this to a turbulent premixed swirling flame, while Silva et al [22] combined a measured FDF for a premixed swirled combustor with a Helmholtz solver, predicting the limit cycle amplitude to within reasonable agreement with experimental data. Note that this approach has been used not only at lab scale, but also for real gas turbines operated at high pressure [20].…”

Experimental investigation of combustion instabilities in lean swirl-stabilized partially-premixed flames in single- and multiple-burner setup

“…However, since these methods have to be specifically adjusted to a given burner system, it would be preferable to avoid combustion instabilities by an adequate combustor design. Common approaches for the prediction of unstable combustor modes imply the determination of combustor transfer matrices and flame transfer functions [15][16][17][18][19][20][21][22][23] or acoustic analysis. [24][25][26] These approaches have proven to be able to predict unstable modes, but they are not yet able to account for all nonlinear effects induced by burner-burner interactions in multiple-burner systems, since they are mostly based on studies performed on single-burner combustors.…”

Combining a Helmholtz solver with the flame describing function to assess combustion instability in a premixed swirled combustor

“…HereQ andū denote a mean flame power and velocity introduced for scaling purposes. The FTF can be either measured or modeled theoretically or numerically (Ducruix et al, 2000;Külsheimer and Büchner, 2002;Truffin and Poinsot, 2005;Durox et al, 2009;Huber and Polifke, 2009;Kim et al, 2010;Schuermans et al, 2011;Duchaine et al, 2011;Tay-Wo-Chong et al, 2012;Palies et al, 2011b). In a second step, this function can be combined to a model of the system acoustics which is often obtained analytically (Dowling and Stow, 2003;Sattelmayer and Polifke, 2003;Poinsot and Veynante, 2005).…”