Data Reduction Techniques for a Low Pressure Turbine Noise Test

The scope of this paper is to analyze different data reduction techniques that conduce to an accurate prediction of the acoustic field for a Cold Flow Low Pressure Turbine Rig. This noise test has been part of the SILENCE(R) (Significantly lower community exposure to aircraft noise) EU program to analyze and quantify the novel concept of a Highly Loaded Low Pressure Turbine. Noise measurements are in-duct, and taken by means of a rotating device located downstream the Outlet Guide Vane and a set transducers at different axial positions that are optimized for accurate radial mode decomposition. Also, two different stationary transducers are located to allow the cross mode detection computation. For circumferential mode detection, the aim is not only identify the interaction sources that contribute to the tonal energy but also a quantitative split between Tyler & Sofrin modes, steady distortion and non-periodic or random component. A discussion between Least Squares Fit (LSF) and Discrete Fourier Transform (DFT) techniques is presented. Also, a spectral analysis based on the so-called Lomb Periodogram is performed prior to the circumferential mode detection to assess the quality of the numerical technique. Other discussions as auto vs cross mode detection or time width for data reduction are also included. For radial mode detection, the Singular Value Decomposition technique (SVD) has been chosen. General assumptions and some examples are also provided.