An Onion Peeling Reconstruction of the Spatial Characteristics of Entropy Waves in a Model Gas Turbine Combustor

Entropy noise affects thermoacoustic stability in lean pre-mixed gas turbine combustion chambers. It is defined as acoustic noise that is emitted at the first turbine stage due to the acceleration of entropy waves that are advected from the reaction zone in the combustor to the turbine inlet. These non-isentropic temperature waves are caused by equivalence ratio fluctuations which are inherently present in a technically premixed combustion system. To experimentally study the generation and transport of entropy waves, an estimation of the spatial distribution of the entropy spots is highly valuable as it allows the accurate determination of the cross-section averaged entropy, which is the relevant quantity for the formation mechanism of entropy noise at the turbine stage. In this work, a time-of-flight based temperature measurement method is applied to a circular combustion test rig equipped with a premixed swirl-stabilized combustor. Downstream of the burner, an electric spark discharge is employed to generate a narrow acoustic pulse which is detected with a circumferentially arranged microphone array. The measured time of flight of the acoustic signal corresponds to the line-integrated inverse of the speed of sound between the acoustic source and each microphone. By modulating a share of the injected gaseous fuel, equivalence ratio fluctuations are generated upstream of the reaction zone and consequently entropy spots are advected through the axial measurement plane. The spark discharge is triggered at distinct phase angles of the entropy oscillation, thus allowing a time resolved-analysis of the thermo-acoustic phenomenon. Estimating the spatial temperature distribution from the measured line integrated inverse speed of sounds requires tomographic reconstruction. A Tikhonov regularized Onion Peeling is employed to deduce radial temperature profiles. To increase the number of independent data, the spark location is radially traversed, which enhances the resolution of the reconstructed temperature field. A phantom study is conducted, which allows the assessment of the capabilities of the reconstruction algorithm. By means of the reconstructed radial entropy field, spatially resolved entropy waves are measured and their amplitudes and phases are extracted. The characteristics of the entropy waves measured in this way correspond well to former studies.