Turbulence Intensity and Length Scale Measurements in Multistage Compressors

Abstract

This paper describes the measurement and processing of turbulence data from multistage low-speed compressors. Measurements were made at the same relative positions in three four-stage compressors, each having different levels of the design stage loading coefficient. A new method of data processing to calculate turbulence intensities and integral length scales is outlined. Using this method, integral length scales have been measured in turbomachinery flows for the first time. It is shown how the turbulence intensity and integral length scale vary with position in the blade passage, with changing flow coefficient and with the value of the design stage loading coefficient. The results have been used to specify representative inlet conditions for experimental rigs and to improve the application of CFD turbulence models to turbomachinery modeling.

On the Setup of a Test Bench for Predicting Laminar-to-Turbulent Transition on a Flat Plate

“…In order to perform this reduction, the measured signal is FFT (fast fourier transformed) and the periodic components (like blade passing frequency and its harmonics) of the signal are "chopped" [16]. Only the non-periodic stochastic parts of the signal are modeled by a single-point turbulence model (like or ) [17].…”

“…The chopped and filtered spectrum was then processed with an inverse FFT to get a time signal again [16].…”

“…Camp and Shin [16] who described in detail how to process the measured signal in order to get the necessary values. Also Axelsson et al [17,18], and Craft [19] discussed different length scales and measurements of turbulence intensity.…”

“…Several overviews have been published in the last years like Refs. [16,17,31,32] and in the following the main outcomes will be discussed.…”

“…In order to determine the chopped FFT spectrum which was used within the signal processing for obtaining the turbulence intensity is multiplied by its conjugated complex part and then transformed back with an inverse-FFT. The time signal obtained this way is then autocorrelated [16].…”