A laser-doppler anemometer investigation into fully pulsed jet flows with an examination of velocity bias errors

Abstract

The velocity fields of steady and fully pulsed axisymmetric air jets exhausting into still air have been investigated to determine why significant differences occur in their flow characteristics. Previous experimental work on periodic jet flows has been limited, in particular, by the difficulty in making accurate measurements in highly turbulent flows. The measurements in this thesis have been performed using a laser-Doppler anemometer (LDA) which has the potential of producing very accurate results in these flow-fields. LDA measurement is a relatively new technique with many potential error sources. Therefore, the first part of the work is devoted to the investigation of errors associated with the LDA. It was found that a number of important error sources had not been adequately covered in the literature auid would have to be studied to avoid contamination of measurement results. The investigation included clock bias, resolution bias, fringe bias, and statistical velocity bias. Statistical velocity bias, which is a result of inherently biased sampling of LDA, was found to be a complicated problem and has been investigated by many authors in the past. A major original investigation was made into the velocity bias associated with sample and hold processors and controlled processors.It has been widely assumed that sample and hold processors may provide bias free results if the ratio of Taylor microscale to measurement time scale is sufficiently high, typically greater than five. This work shows the appropriate flow time scale is, in fact, the integral time scale of the flow. A model for the sample and hold processor based on the integral time scale is put forward with experimental verification. This model can in many cases predict and correct velocity bias if it exists. A theoretical model for the controlled processor is also presented which shows that results free of bias can be obtained if both the ratio of integral time scale to measurement time scale (integral scale data density) and the ratio of sampling time to the measurement time scale (normalized sample interval) are greater than five. Further, it is shown that for any integral scale data density the controlled processor will not produce any less bias than a sample and hold processor and no more bias than the unweighted processor. Experimental data confirming the theoretical results are presented and are shown to be contrary to at least one model available in the literature.An investigation of velocity bias in periodic flows showed that ensemble averaging of the measured velocity significantly reduces the amount of bias and the statistical scatter. When the period of cycling of the flow was greater than fifty times the measurement time scale the sample and hold processor in conjunction with ensemble averaging produced no bias. At data rates below this, the velocity bias errors were, in most cases, small enough to be disregarded. The investigation of the pulsed jet has shown that the mechanisms of mixing in low Strouhal number flows is very similar to steady jet flows despite the very different appearance of the flow field. Four distinct regions in the flow field were observed. These were the flow development region, the pulse dominated region, the high turbulence steady flow region, and the steady jet region. The rate of growth of the jet width was found to be proportional to the centreline volume averaged intrinsic turbulence intensity which was found to be approximately proportional to the peak normalized Reynolds stress in the plane at the same axial location in the jet. Higher frequencies of pulsation were found to transfer energy to the intrinsic turbulence at a greater rate, producing faster initial growth rates in jet width and volume flow. The position of maximum growth occurred at the boundary between the pulse dominated region and the high turbulence steady flow region and its distance from the jet exit was found to be inversely proportional to the square root of the Strouhal number.With improvements in measurement techniques, the momentum at any axial location in the pulsed jet measured, was shown to remain constant. Jet momentum was then shown to be the appropriate means of normalizing the mean axial velocity decay and the volume flow such that meaningful comparisons can be made between jets with differing time varying exit velocity conditions.