On the relationship between image intensity and velocity in a turbulent boundary layer seeded with smoke particles

Simultaneous particle image velocimetry (PIV) and flow visualization measurements were performed in a turbulent boundary layer in an effort to better quantify the relationship between the velocity field and the image intensity typically observed in a classical flow visualization experiment. The freestream flow was lightly seeded with smoke particles to facilitate PIV measurements, whereas the boundary layer was densely seeded with smoke through an upstream slit in the wall to facilitate both PIV and classical flow visualization measurements at Reynolds numbers, Re _θ , ranging from 2,100 to 8,600. Measurements were taken with and without the slit covered as well as with and without smoke injection. The addition of a narrow slit in the wall produces a minor modification of the nominal turbulent boundary layer profile whose effect is reduced with downstream distance. The presence of dense smoke in the boundary layer had a minimal effect on the observed velocity field and the associated proper orthogonal decomposition (POD) modes. Analysis of instantaneous images shows that the edge of the turbulent boundary layer identified from flow visualization images generally matches the edge of the boundary layer determined from velocity and vorticity. The correlation between velocity deficit and smoke intensity was determined to be positive and relatively large (>0.7) indicating a moderate-to-strong relationship between the two. This notion was extended further through the use of a direct correlation approach and a complementary POD/linear stochastic estimation (LSE) approach to estimate the velocity field directly from flow visualization images. This exercise showed that, in many cases, velocity fields estimated from smoke intensity were similar to the actual velocity fields. The complementary POD/LSE approach proved better for these estimations, but not enough to suggest using this technique to approximate velocity measurements from a smoke intensity image. Instead, the correlations further validate the use of flow visualization techniques for determining the edge and large-scale shape of a turbulent boundary layer, specifically when quantitative velocity measurements, such as PIV, are not possible in a given experiment.