Direct Assessment of the Accuracy of Stereo PIV in Turbulent Channel Flow

Abstract

Stereo particle-image velocimetry (PIV) has become a widely-used method for studying complex flows because it allows one to acquire instantaneous, three-component velocity data on a planar domain with high spatial resolution. However, the accuracy of such measurements must be carefully evaluated before stereo PIV data can be faithfully used in the development of sophisticated turbulence models, assessment of appropriate computational boundary conditions, and in the validation of advanced computations. To this end, the accuracy of stereo PIV is assessed directly in an actual turbulent environment: two-dimensional turbulent channel flow. This flow is a challenging test of stereo PIV because the turbulent velocity fluctuations are quite small compared to the mean (typically less than ten percent of the mean velocity) and strong velocity gradients exist in the near-wall region. Measurements are made in the streamwise–wall-normal plane along the channel’s spanwise centerline using both stereoscopic and conventional 2-D PIV. A large ensemble of statistically independent velocity realizations are acquired with each method at a friction Reynolds number Reτ = u*h/ν = 934. Single-point statistics are computed from the experimental data and compared to statistics determined from a direct numerical simulation (DNS) of turbulent channel flow at a nearly-identical friction Reynolds number of 940 [5]. Excellent agreement is found in the outer region of the flow (y/h > 0.15, where h is the half-height of the channel). For y/h < 0.15, both the conventional and stereo PIV results differ from the DNS data. These differences are most-likely a manifestation of errors associated with strong velocity gradients and intense turbulent events present in this region of the flow.