Modification of a turbulent boundary layer by circular cavities

Abstract

It is shown how well-chosen perforations in a wall flow can locally reduce skin friction drag by modifying the generation of bursts in the boundary layer. For this purpose, a detailed hot wire experimental boundary layer investigation of the flow past a perforated plate, complemented with large eddy simulations, is carried out and compared to the smooth baseline. The perforated plate is obtained with an array of flush-mounted circular cavities. These cavities are disposed in a periodic staggered arrangement. For the three tested flow velocities, the momentum thickness based Reynolds number varies from 1830 to 3380 and the cavity diameter and spacing in wall units respectively from 130 to 250 and 587 to 1075, the latter being identical in both spanwise and streamwise directions. The mean velocity profiles evidence a thickening of the viscous sublayer and a decrease of the friction velocity as compared to the smooth wall case. The application of the Variable Interval Time Averaging (VITA) technique highlights an upward shift of the bursts from the wall and an attenuation of the average burst intensity and duration.Spanwise measurements evidence an overall bursts attenuation despite the lack of spanwise uniformity.The 3D mean flow topology arising from the large eddy simulations provides evidences of the qualitative similarities between the current setup and the spanwise wall oscillations.

Evaluation of the local skin friction drag of turbulent boundary layers over cavities

“…From left to right, the figures show a clear reduction in the upward vertical shift at all four measured locations when the Reynolds number increased. This finding agrees with the work by Scarano et al (2022). Scarano et al (2022) also found that the upward vertical shift was more pronounced after the trailing edge of the cavity array, L/d = 125.8, than the one at the array center, L/d = 59.2.…”

“…Figures 7a, 7b, 7d, 7e, 7g, and 7h clearly show an upward vertical shift in the velocity profiles of the turbulent boundary layers over the cavity array in comparison with the ones over a smooth wall at U ∞ = 5.6 and 9.9 m/s at all four measurement locations. This upward vertical shift in turbulent boundary layers over cavities was also found by Silvestri et al (2017) and Scarano et al (2022) and was associated with a reduction in the skin friction drag of turbulent boundary layers using riblets (Choi, 1989). Figures 7c, 7f, 7i, and 7l show that the velocity profiles of the turbulent boundary layers over the cavity array were identical to the one over a smooth wall, which indicates the small effect of the cavity array at U ∞ = 14.2.…”

“…The results show that the turbulent intensity slightly decreased in the near-wall region from the wall to the peak but increased in the logarithmic region. The increase in turbulent intensity in the logarithmic region was also observed by Scarano et al (2022) and Gowree et al (2019) for the turbulent boundary layers over staggered cavities, and by Jafari et al (2022) for the turbulent boundary layers over a porous surface with square cavities. The figures from the left to the right show that the increase was more pronounced when the free stream velocity increased.…”

“…The estimation of the friction velocities by the fitting methods in the work by Silvestri et al (2017), Gowree et al (2019), Scarano et al (2022), and Scarano et al (2023) assumed that turbulent boundary layers over cavities show similar inner scaling to canonical turbulent boundary layers. The turbulent boundary layers over a cavity array may not be canonical, especially in the wall region where the cavity array has a significant effect.…”

“…The error of the estimated friction velocity can be up to 8% at Reynolds numbers of Re τ = 1000 (Wei et al, 2005b). Silvestri et al (2018b) and Scarano et al (2022) used a fitting method proposed by Kendall and Koochesfahani (2008) for estimation of the friction skin drag of turbulence boundary layers over cavities at a Reynolds number of Re θ ≤ 3700. The fitting method used Musker's and Spalding's profiles to combine the wall and logarithmic profiles to generate a universal profile.…”