Optimization Design of a 2.5 Stage Highly Loaded Axial Compressor with a Bezier Surface Modeling Method

Abstract

Due to the complexity of the internal flow field of compressors, the aerodynamic design and optimization of a highly loaded axial compressor with high performance still have three problems, which are rich engineering design experience, high dimensions, and time-consuming calculations. To overcome these three problems, this paper takes an engineering-designed 2.5-stage highly loaded axial flow compressor as an example to introduce the design process and the adopted design philosophies. Then, this paper verifies the numerical method of computational fluid dynamics. A new Bezier surface modeling method for the entire suction surface and pressure surface of blades is developed, and the multi-island genetic algorithm is directly used for further optimization. Only 32 optimization variables are used to optimize the rotors and stators of the compressor, which greatly overcome the problem of high dimensions, time-consuming calculations, and smooth blade surfaces. After optimization, compared with the original compressor, the peak efficiency is still improved by 0.12%, and the stall margin is increased by 2.69%. The increase in peak efficiency is mainly due to the rotors. Compared with the original compressor, for the second-stage rotor, the adiabatic efficiency is improved by about 0.4%, which is mainly due to the decreases of total pressure losses in the range of above 30% of the span height and 10%–30% of the chord length. Besides, for the original compressor, due to deterioration of the flow field near the tip region of the second-stage stator, the large low-speed region eventually evolves from corner separation into corner stall with three-dimensional space spiral backflow. For the optimized compressor, the main reason for the increased stall margin is that the flow field of the second-stage stator with a span height above 50% is improved, and the separation area and three-dimensional space spiral backflow are reduced.