Numerical assessment of roughness effects on boundary layer development over adjacent blades

Abstract

Surface roughness is a condition that affects the aerodynamic performance of compressor blades. For instance, erosion, corrosion, and fouling, among other phenomena, can cause surface roughness, in compressor blades to increase significantly. Accumulation of these particles, sand, dust and salts alter the blade geometry. In the presence of rough surfaces, an increase in transport of heat, mass and momentum is generally noticed in a turbulent boundary layer, which sparks off considerable interests in the engineering field. Although surface roughness has been widely investigated during the last decades, both experimentally and numerically, the effect of this phenomenon on the aerodynamic performance of compressor blades is not well understood. In fact, many experimental and numerical investigations support the universal velocity defect law, meaning turbulence over rough walls is confined to a thin layer inside the turbulent boundary layer ≈3k-5k where k is the roughness height, while other studies state that the effects of surface roughness propagate to the outer layer. In this investigation, assessment of the numerical prediction of surface roughness effects on the turbulent boundary layer over an NACA 65-410 airfoil in a linear cascade is conducted. The most commonly used turbulence models in aerospace industry, available in the commercial Navier-Stokes code solver package CD-ADAPCO Star CCM+, are investigated. Moreover, Large Eddy Simulation (LES) is performed to obtain detailed flow fields to better understand relevant flow physics of the turbulent boundary layer developing over the rough and smooth airfoil. Results from the validation case were in good agreement with experimental data. Hence, the stall angle of attack was reduced when roughness is added to the blade surface. However, the solver overpredicted the lift coefficient at higher angle of attack. Furthermore, the effects of increased turbulence intensity in decreasing the lift coefficient was captured by the solver when free stream turbulence increased from 0.01% to 6%. All the turbulence models investigated predicted similar flow physics for attached flow conditions with significant deviation from one model to another when the airfoils are near stall angle of attack. The results of LES were compared to RANS and were found to be in good agreement. It was shown that roughness effects penetrate up to the outer edge of the boundary layer as opposed to the belief that roughness is confined to a thin layer in the boundary layer. It is suggested that more investigations are needed to be conducted in RANS turbulence modeling with wall functions as the flow physics captured by the turbulence models differ when the airfoils are near stall angle of attack.