Detection and Identification of Near-Wall Coherent Structures Through Conditional-Sampling

Abstract

Local detection of coherent structures by simple means performed by Eulerian techniques through single or multiple probes, provide useful information for the near wall turbulence control, in an easy and feasible way. Such techniques are undoubtly uncomplete and suffer from the significant smearing produced by the three dimensionality of the random structure advecting past a sensor. Table 1 recapitulates the mostly used Eulerian schemes. Their commun feature is in their detection function. Indeed, almost all of them are based on the level crossings of fluctuating velocity signals, as for example the u-level, quadrant and to some extent VITA and WAG. One of the well known problems encountered in such techniques is the high sensitivity of the number of identified events to the threshold, even though the conditional averages are more or less “universal” in the inner layer, when they are properly normalized. We will first question this dépendance and show that the level crossing activity may more or less be estimated by supposing that the signal is Gaussian. Based on this observation, we develop in 3.2 a simple formulation for the “ejection” frequency identified by Eulerian techniques and show the utility of this approach by considering WAG (Table 1). It is further argued that the most significant characteristic of the level crossings is the probability density function of their interarrival times, and that these characteristics differ a near wall signal from a normal one. The grouping methods dealing with the classification of the dynamicaly similar events (Tardu, 1995) are subsequently and briefly reviewed. Finaly some results concerning a simplified model linking the bursting mechanism to the shear stress distribution are presented and discussed.