Compressor cascades with service inevitably suffer from blade fouling or erosion, which modifies blade geometries and brings sufficient damage to compressor performance. However, very few studies were cast upon the impact of degraded blade surface morphologies on the compressor aerodynamic performance. In this study, a novel geometry parameterization method is proposed to describe the non-uniform, multi-scale and stochastic geometric features of blade surface morphologies in fouling and erosion conditions. Through computational fluid dynamics (CFD) simulations, the influences of coverage, dirty size and density of dirty units on the cascade performance and flow field are numerically investigated. Finally, uncertainty analyses are conducted to quantify the cascade performance variations due to geometric uncertainties of dirty surface morphologies. To alleviate the computational load, a relatively new dimension reduction technique, i.e., the active subspace method is used to reduce the original eight-dimensional uncertainty quantification problem to a one-dimensional one. The CFD results show that at the high flow rate, the cascade performance is gradually worsened as the dirty surface morphology extends from the blade leading edge to the trailing edge on the suction surface. However, at the low flow rate, the cascade performance degradations tend to be alleviated once the dirty surface morphology enters the region of the separation bubble on the suction surface. At both operating points, the cascade total pressure loss almost increases monotonically with either larger or densified dirty units distributed on the blade suction surface. The probabilistic distributions from uncertainty analyses show that the cascade performance is more sensitive to the dirty surface morphology at low flow rates than at high flow rates. The present study can help understand the loss mechanisms within the 2D cascade in fouling and erosion conditions, which is beneficial for the enhancement of 3D compressor performance.

This content is only available via PDF.
You do not currently have access to this content.