High spatial resolution PIV experiments are performed in the x-y plane at two different spanwise positions to compare the turbulent boundary layers over smooth and rough forward-facing-steps as well as the rough step's two coarse scale-resolution approximations. The Reynolds number based on the step's mean height, Reh, is 3450 and the ratio of the boundary layer thickness to the step's height is δ/h = 8. The roughness topography on the top surface of the rough step is replicated from a realistic turbine blade and is intrinsically three-dimensional and highly irregular. The surface topographies of the coarse scale-resolution approximations of the rough step are obtained from a multi-resolutional analysis using discrete wavelet transform(DWT).
Mean flow structures, Reynolds normal and shear stresses, quadrant analysis of instantaneous ejections and sweeps, spanwise vorticity and characteristics of the coherent spanwise vortices are first compared between flows over the smooth and rough forward-facing steps. The results illustrate that the rough surface conditions on the top surface of the forward-facing step tend to weaken the
separated flows from the sharp edge. The comparison of the results at two different measurement positions indicates that the slope of the roughness profile immediately after the step's edge plays an important role in affecting the flow. This is probably due to the local pressure gradients induced by the roughness topography. A relatively strong favorable pressure gradient may prevent the generation of a recirculation region downstream of the step.
The effects of roughness length scales on the forward-facing turbulent flow are explored by comparing the characteristics of the flow over the rough step and its two approximations, A6 and A4, which contain about 44% and 88% of the roughness energy, respectively. It is observed that the coarse-scale roughness only slightly distorts the downstream recirculation region. For the
higher resolution step approximation A4, the flow characteristics are significantly modified at measurement position P2 compared with the case of step A6. However, A4 still cannot reproduce the flow over the step of full roughness studied herein. On the other hand, a much smaller difference exists in the flow characteristics
between A4 and full surface at position P1 than at P2, suggesting that the high resolution approximation may be needed only in the part of the roughness topography that has a larger impact on the forward-facing turbulent flow.