Burst detection with single-point velocity measurements

Abstract

An evaluation of the effectiveness of the VITA, Quadrant, TPAV, U -level, Positive slope, and VITA with slope burst-detection algorithms has been done by making direct comparisons with flow visualization. Measurements were made in a water channel using an X-type hot-film probe located in the near-wall region. Individual ejections from bursts which contacted the probe were identified using dye flow visualization. The effectiveness of each of the detection algorithms was found to be highly dependent on the operational parameters, i.e. threshold levels and averaging or window times. These parameters were adjusted so that the number of events detected by each of the algorithms corresponded to the number of ejections identified by flow visualization, while the probability of a false detection was minimized. Comparing the detection algorithm using these optimum parameter settings, the Quadrant technique was found to have the greatest reliability with a high probability of detecting the ejections and a low probability of false detections. Furthermore, it was found that the ejections detected by the Quadrant technique could be grouped into bursts by analysing the probability distribution of the time between ejections.