High Reynolds number turbulent pipe flow

Fully developed turbulent pipe is studied in this thesis. Streamwise and wall-normal turbulence components are measured using a crossed hot-wire probe. In the process, a new calibration method for the crossed hot-wire probe is developed, and the binormal cooling error for hot-wire measurement, which is caused by cooling in the direction normal to the hot-wire measurement plane, is studied and found to be the major error contributor for both mean velocity and turbulence intensity measurements using a crossed-wire probe.; The new calibration scheme utilizes the fact that the total stress in a fully developed turbulent pipe flow is defined by the streamwise pressure gradient, so directional sensibility calibration could be done by recording the crossed hot-wire signals against a known shear stress distribution. This information, when combined with mean velocity calibration against a Pitot tube measurement, provide a full calibration for crossed hot-wire probes. The new calibration method is especially convenient for pipe and channel flow measurements. For other measurements, the calibration could be done by using a simple pipe apparatus as the calibration device.; Streamwise and wall-normal turbulence components are measured over a Reynolds number range from 1.1 x 105 to 9.8 x 10 6. Similarity arguments are studied for turbulence intensity and spectra. The most relevant physical assumption for the 'similarity' is Townsend's distinction between 'active' and 'inactive' motions. Perry's attached eddy hypothesis, which is based on Townsend's work, offers a more detailed physical model and provides extensive quantitative prediction, is also reviewed and discussed in the context of these new measurements.; For the wall-normal turbulence intensity, a constant region in u'rms is found for the region 200 ≤ y+ ≤ 0.1R+ in inner and outer scaling for Reynolds numbers up to 1.0 x 106. An increase in u'rms is observed closer to the wall at about y + ∼ 100, and is suggestive of the first maxima of streamwise turbulence intensity profile reported in Morrison et al. (2004). This result is new, and is not expected from Townsend's 'active' and 'inactive' motion distinction and Perry's attached eddy model, may signal the existence of interaction between 'active' and 'inactive' motions, which will lead to the incomplete similarity argued in Morrison et al. (2004).; The wall-normal spectra are carefully examined, especially in the mean flow overlap region (where the logarithmic law applies). Collapse is found for the energy containing part with inner scaling, but for the low wave number region, a y/R dependence is observed which also indicates an influence from the outer flow. (Abstract shortened by UMI.)...