Feature-based analysis of a turbulent boundary layer under spanwise wall oscillation

Spanwise wall oscillations alter the organization of low-speed streaks and ejections in turbulent boundary layers, eventually leading to skin friction drag reduction. Such flow regimes are represented by pointwise statistics or spatial correlation. This work attempts to quantify the systematic distortions of the dominant turbulent structures by feature-analysis, intended to overcome the dispersion observed in pointwise statistics and correlation functions. Furthermore, data from tomographic particle image velocimetry are employed to clarify the mechanism that inhibits hairpin auto-generation, as described in Kempaiah et al. ["3-dimensional particle image velocimetry based evaluation of turbulent skin-friction reduction by spanwise wall oscillation,"Phys. Fluids 32(8), 085111 (2020)]. Based on the instantaneous distribution of Reynolds stresses, a specific spatial template is defined for low-speed streaks and flow ejections. Events corresponding to this template are collected and parametrized with their occurrence, geometrical properties (length and orientation), and dynamics (intensity). The approach is compared with most practiced statistical analysis to explain the significance of the features extracted by the detection algorithm in relation to the drag reduction mechanism. Data comparing stationary and oscillating wall in a drag-reducing regime (A+osc = 100, T+osc = 100) are investigated in the near-wall region (y+ < 100). Ejections and low-speed streaks systematically exhibit a positive pitch, supporting the hypothesis that only the rear region, close to the wall, is affected by the wall motion. A side-tilt of elongated ejection events is observed past the phase of maximum oscillation velocity, which is hypothesized to inhibit hairpin auto-generation. The latter indicates a phase dependence of the side-tilt in the oscillating regime. The results also indicate that low-speed streaks and ejection events are reduced by approximately 10% and 15%, respectively, compared with the stationary wall, further consolidating the mechanism of rapid lateral distortion being responsible for the different organizations of the turbulent structures in the near-wall region.