Abstract

A turbulent boundary layer was subjected to adverse pressure gradients of different strengths and the effects this had on the pressure distribution, wall shear stress, mean velocity and perturbation velocity moment profiles were studied. Measurements were made using Particle Image Velocimetry (PIV) and the results of these examinations were compared to the results obtained from conventional measurements using static pressure taps, a Preston tube and a hot-wire anemometer. Three different cases were examined: 1. Complete separation, 2. Separation bubble and 3. Attached boundary layer subjected to a strong adverse pressure gradient (APG).

In the first case that was studied, the adverse pressure gradient was so strong and persistent that there was no reattachment in the studied region. This case was used to evaluate some empirical methods for the simulation of boundary layer development, based on the integral momentum equation [5], [10], which were found to predict the boundary layer growth rather well. However, all the examined methods indicated separation too far upstream in the separated case that was studied, and the method proposed by Stratford [30] also falsely indicated that there was separation in the least rapidly expanded flow studied.

In the second case where there was a limited area of reversed flow - a so-called separation bubble ( PIV was used to study the flow in the reattachment region, while the separation line was kept fixed with a separation trip. It was found that the shape factor H decreased from its maximum value of around 15 to 3.1-3.4 at reattachment, which is higher than the value of H=2.85 reported by Alving and Fernholz [1], but lower than the value of 4.0 reported by Kline [15].

In case 3, a thorough examination of a non-equilibrium boundary layer subjected to a strong APG was made. It was found that while the near-wall peak in urms disappeared as the adverse pressure gradient "worked" on the boundary layer, sharp near-wall peaks developed in both the skewness and the flatness of the streamwise velocity. A good qualitative agreement with earlier examinations was generally found. In contrast to some of these examinations, a logarithmic sub-layer was found even far downstream at H=2.4. Instead of disappearing, the logarithmic region seemed to move inward - at H=2.4 it stretched from around 10 to 40 y+. When the velocities in this region were used to calculate wall shear stresses, fair agreement with measurements using Preston tubes was found.

Apart from its appealing ability to study the 2D structure of a separated region, PIV was regrettably also found to suffer from a few drawbacks, most notably its low spatial and temporal resolution and its tendency towards giving a discrete velocity distribution.

Back to the main Turbulent boundary layer separation page.

Jonas Gustavsson