MODEL-BASED INFERENCE OF FLAME TRANSFER MATRICES FROM ACOUSTIC MEASUREMENTS IN AN AERO-ENGINE TEST RIG

Flame dynamics in the form of a flame transfer matrix (FTM) is not directly measurable in a test rig, but must be deduced from transfer matrix measurements of the combustion system. The burner-flame transfer matrix (BFTM) approach for estimating the FTM is based on local pressure signals from two microphone arrays located upstream and downstream of the combustor. It combines acoustic transfer matrix measurements in non-reacting and reacting conditions, where the latter implicitly includes the flame dynamics. A simple matrix operation then yields the FTM. However, this approach assumes that there is loss-free wave propagation at a constant speed of sound with no change in cross-sectional area between the microphone locations and the burner/flame. The present work demonstrates the limitations of these assumptions when applied to a test rig with complex features, such as effusion cooling, bypass annulus, and downstream end contraction. To remedy the shortcomings of the BFTM approach, this work proposes a novel method to infer the FTM for complex combustors by combining reactive transfer matrix measurements of the entire combustor with an accurate low-order thermoacoustic network model (LOM) of the test rig. This generalized method reduces to the BFTM approach as a special case. In this work, the Rolls-Royce Scaled Acoustic Rig for Low Emission Technology (SCARLET) operated under realistic engine conditions (T_in ≈ 825 K, p_in ≈ 25 bar, kerosene) is used to analyze the capabilities of the proposed model-based inference method and the limitations of the BFTM approach. In a first step, a LOM based on the geometry and operating point of SCARLET is formulated using a generic FTM. This generic model is used to visualize the limitations of the BFTM approach in terms of various physical and geometrical parameters. Finally, experimental measurement data is used to deduce the FTM of SCARLET using the proposed approach.