A Study On Robust Optimization Of Highly Loaded Compressor Blade-End Considering Fine-Scale Geometry Deformations

The rapid development of the aviation industry makes the modern aero-engine performance become higher and higher.As we know,compressor is one of the core part of aero-engine and how to design the highly loaded and efficient compressor is a key problem.During the design process of compressors,the fine-scale geometry deformation from maching error,mechanical vibration and manual operation error plays a significant role in determining the flow details near the endwall,even the overall performance.Understanding the effect of these uncertainty factors on compressor performance and building a robust optimization platform are important basis for the design of the highly loaded and efficient compressor.The development model of the endwall boundary layer displacement thickness in the multi-stage environment was firstly built and the simulation results derived from the different operating points agrees well with the prediction values.The blade-end treatment flow control technique which consists of end-bend,end-dihedral and end-sweep was then deployed to eliminate the aerodynamic losses near the endwall region and the overall compressor performance was improved accordingly.However,the peak efficiency gains caused by blade-bend treatment which were published over the past almost 30 years shows that the effects of the blade-end flow control are not as obvious as before.This is due to the wide use of three-dimensional CFD models in the original design process during which the complicated flow caused by the endwall boundary layer has been taken into account.The new blade-end treatment based on the robust optimization method is more promising in the further design process.The effect from fine-scale geometry deformations,i.e.,the uncertain tip clearance size,the uncertain wall surface roughness,the uncertain leading edge shape and the uncertain blade root fillet,on the compressor performance was then investigated.The overall compressor performance is degraded with the increase of clearance size.The axial distributions of tip clearance significantly influences the performance of compressor and the compressor with non-uniform tip clearance shows better aerodynamic performance than that with uniform clearance size.Surface roughness setting on the blade,hub or shroud results in a degradation of compressor performance and the blade roughness leads to the most substantial loss.The transition process from the laminar boundary layer to the full-developed turbulent boundary layer is “delayed” when the leading edge shape is changed from the circle to ellipse shape and the total pressure losses can be reduced with the increase of the leading edge ellipse ratio.The fillet near the blade root improves the overall performance when the fillet radius is smaller than the coming boundary layer displacement thickness.However,the fillet always degrades the compressor performance when the fillet size is larger than the boundary layer displacement thickness.Thereafter a UQ platform was built with Dakota in conjunction with ANSYS CFX to quantify the uncertainties effect on the performance of NASA Rotor37.These uncertainties come from the random inlet total pressure,inlet turbulence intensity,blade surface roughness,endwall surface roughness,blade thickness and tip clearance size.Polynomial chaos expansion method shows obvious advantages on reducing the computing time under the precondition of a higher guaranteed precision when compared with the traditional Monte Carlo method.The sensitivity analysis results show that the blade thickness and the blade surface roughness play the most significant role in determining the compressor peak efficiency and choked mass flow rate,sequentially followed by the factors of the inlet turbulence intensity,the endwall surface roughness and the tip clearance size.The inlet pressure shows slight influence on the compressor characteristics.However,as for the surge margin,the inlet pressure shows a much more important effect while the inlet turbulence intensity is deemed not as significant as that for peak efficiency and choked mass flow rate.The built UQ platform successfully predicted the probability distributions of compressor peak efficiency with considering the influence induced by the uncertain clearance size and blade surface roughness.The NAGA-II+PCE+Kriging robust optimization design framework was finally built and four optimization strategies were proposed in this paper.Strategy 1 aims to improve the tranditional blade-end treatment based on optimization method and the retrofitted blade features end-sweeps and end-dihedrals near the blade root and features end-sweeps and end-bends near the blade tip.Strategy 2 investagtes the influence of wall roughness on the optimization results of axial compressor blade-ends.The result shows that the roughness effect can be regarded as an additional factor and be considered in the end of the design process for single-stage compressors.The optimum distribution of clearance size is obtained by employing a surrogate model in conjunction with a multi-objective genetic algorithm for optimization in Strategy 3.The Pareto optimal front shows that the axial compressor with well-designed distributions of tip-clearance size gives 0.64% improvement in peak efficiency and 48.18% enhancement in surge margin.The 88% improvement in the adiabatic efficiency is attributed to the reduction of the clearance size,whereas 30% enhancement in the stall margin is driven by changes in the tip-gap slope when comparing with the prototype performance.The tip clearance uncertainties inherent in an axial compressor are quantified to determine their effect on performance in Strategy 4.The robust optimization results show that the mean value of the adiabatic efficiency and total pressure ratio is rather higher while the standard deviation of the adiabatic efficiency and total pressure are decreased considerably when compared to that of prototype.The major aerodynamic loading distribution of optimized airfoil is substantially behind than the initial sample which validates that the blade with major loading behind is more non-sensitive to the uncertain tip clearance shape,which is meaningful to the compressor designers with intension of high potential design to manufacture.