Three-dimensional (3-D) turbulent flows are widely observed in many flow configurations. Flow over swept wings, through turbine passages, through curved ducts are a few examples of 3-D flows in practical applications. However, as compared to two-dimensional (2-D) turbulent flows, there are few fundamental studies related to these, mainly because of their inherent complexity. This thesis studies the effect of:
i) a streamwise; and ii) a skewed shear on an upstream 2-D turbulent boundary layer. The wall shear is applied to the main flow by a mechanism housing a moving rubber belt. Four individual cases are investigated.
In the first case, the streamwise shear at the wall is applied along the direction of the freestream. Velocity measurements, using hotwire anemometry, are carried out at various locations upstream, on and downstream of the moving belt. A standard TSI boundary-layer probe is used. Mass transfer measurements are conducted using a naphthalene-coated sublimation plate, placed downstream of the moving surface.
With the freestream velocity maintained at approximately 14.7 m/s, velocity and mass transfer measurements are conducted for four belt velocities 0, 6, 8 and 10 m/s. Velocity measurements above the stationary belt give the location of the origin of the turbulent boundary layer in the absence of laminar and transition regions (virtual origin).
The variation of the shape factor and the skin friction coefficient with Reynolds number compares well with experimental results from other studies. The variation of the streamwise turbulence parameter across the boundary layer shows a local peak in the buffer region. It also suggests an `inner' scaling until the logarithmic region of the boundary layer. With the belt in motion, the streamwise turbulent kinetic energy (TKE) on the moving belt follows an `inner' scaling that is distinctly different from the scaling observed at locations upstream and downstream of the belt. Much lower TKE values are observed over the moving surface. In addition, the virtual origin of the turbulent boundary layer moves closer towards the trailing edge of the belt. Mass transfer measurements with the naphthalene plate show the variation of the Stanton with the Reynolds number. Lower turbulence levels associated with higher belt velocities result in a local minimum in this variation.
In the second part of this research, similar velocity and mass transfer measurements are conducted with the belt translating in an direction opposite to the freestream. Streamwise turbulence levels downstream of the belt show the growth of a second peak in the wake region in addition to the regular peak in the buffer region. This suggests the presence of larger length scale turbulent eddies at locations away from the wall in the boundary layer. However, no difference in the mass transfer results (variation of the Stanton with the Reynolds number) is observed compared to the stationary belt.
In the third case, the belt translates transverse (normal) to the freestream. This gives rise to a three-dimensional turbulent boundary layer on the belt with a strong transverse velocity component near the wall. Downstream of the belt, the flow relaxes back to a nominal two-dimensional turbulent boundary layer. Previous studies have shown lower length scales being associated with these flows. In this study, a slant hotwire probe was used to determine the three mean velocity components across the boundary layer. In this technique, the hotwire probe is rotated about its axis at a particular distance from the wall. The voltage response of this probe shows a minimum value (corresponding to the lowest cooling velocity) at the angular position which corresponds to the direction of the absolute velocity. Therefore, in the freestream, this minimum occurs at zero degree which corresponds to the streamwise velocity component. The mass transfer results show that the transverse motion of the belt results in a higher Stanton number compared to the stationary belt. This result is attributed to higher turbulent stresses and kinetic energy in the boundary layer as observed by other researchers.
In the fourth and final part of this research, the moving section is placed in a cascade wind tunnel, upstream of a row of simulated gas-turbine blades. The relative motion between the stationary blades and the translating belt is envisioned to simulate the relative motion between a row of stator and rotor blades for a single stage in a gasturbine engine. Mass transfer results on the suction surface of the blade suggest that the onset of the secondary flow region near the endwall moves upstream with the belt motion. This is probably due to the lifting of the passage vortex by the inlet skew.
It also appears that in the two-dimensional region of the blade, the transition and reattachment region is prolonged with the belt motion. In comparison, no change in the distribution of the Sherwood number on the pressure surface is observed with the moving belt. In addition, the effect of the moving belt on the heat transfer variation over the endwall surface is observed to be negligible.