A novel tolerance optimization approach for compressor blades: Incorporating the measured out-of-tolerance error data and aerodynamic performance

With the continuous enhancement of compressor performance, traditional empirical and passive blade tolerance design methods have led to increasingly stringent blade tolerances, resulting in significant manufacturing challenges including reduced processing efficiency, elevated scrap rates, and escalating production costs. These challenges could potentially be addressed by optimizing blade tolerance bands. Therefore, there is an urgent need to develop a tolerance optimization methodology for compressor blades that can determine the optimal or maximum allowable tolerance bands without compromising aerodynamic performance. This paper presents a novel tolerance optimization approach for compressor blades that uniquely integrates both the statistical distribution of measured out-of-tolerance error data and aerodynamic performance constraints. The proposed methodology was applied to optimize three tolerance bands associated with four key errors. Firstly, based on previous research conclusions, key error types were identified from numerous machining errors. Then, the distribution of measured key error data under existing process capabilities was analyzed to determine the tolerance offset coefficient and out-of-tolerance distribution range for each type of key error. Subsequently, a full factorial design was employed to plan Computational Fluid Dynamics (CFD) numerical simulation experiments, conducting simulations for all level combinations of four key errors to obtain total pressure loss coefficient values, thus constructing a dataset linking key errors to aerodynamic performance. The obtained dataset was used to iteratively train a dual hidden layer Backpropagation (BP) neural network, thereby forming a high-precision surrogate model for the relationship between key errors and aerodynamic performance. Finally, based on the constructed surrogate model, the Sparrow Search Algorithm (SSA) was applied to optimize the blade tolerances related to four key errors while maintaining aerodynamic performance. Results demonstrate that this research can reduce the out-of-tolerance rate of blade maximum thickness error by 15.77 %, decrease the out-of-tolerance rates of leading-edge contour maximum and minimum errors by 31.58 % and 26.32 % respectively, and reduce the out-of-tolerance rate of twist angle error by 15.79 %. This research holds significant importance for improving blade manufacturing yield and efficiency, as well as ensuring aeroengine performance.