The 80 eyes PIV

In order to reduce measurement error and expand the dynamic range of PIV, we developed a ellipsoidal polyhedral mirror placed between camera and flow target to capture images of identical particles from $n$ ($=$80 in maximum) different directions. The 3D particle positions were determined from the ensemble average of $_{n}$C$_{\mathrm{2}}$ intersecting points of a pair of line of sight back-projected from a particle found in two images extracted from $n$ images. We hypothesized that the error of measured particle displacement is reduced by a factor of $1/\sqrt n $ thanks to the central limit theorem. A rigid-body rotating flow and a turbulent pipe flow were measured by the above method. In the former measurement, bias error and random error fell into a range of ± 0.02 pixels and 0.025 to 0.05 pixels, respectively, and the random error decreases in proportion to~$1/\sqrt n $. In the latter measurement in which the measured value was compared to DNS, the bias error was reduced, and the random error also decreases in proportion to~$1/\sqrt n $ when the particle displacement was relatively small. When the particle displacement was larger, the random error was difficult to estimate, since a discrepancy of the experiment and DNS results might be dominant.