Axial compressors are susceptible to uncertainties during their manufacturing and operation, resulting in reduced efficiency and performance dispersion. However, uncertainty quantification and robust design of compressors remains challenging due to the complexity of structure and internal flow. In this study, an automated framework for uncertainty quantification and robustness optimization of micro axial compressors is proposed. Ten geometrical uncertainties are propagated for the nominal design point and two off-design points, i. e., near stall and choke conditions, respectively. The main objective of this paper is to optimize the aerodynamic robustness performance at these operating points. The sparse grid-based probabilistic collocation method is used to propagate these uncertainties, and a multi-objective genetic algorithm is employed to perform robust optimization based on a novel constructed surrogate model.
The results show that the optimal configuration achieves an improvement in aerodynamic robustness and mean performance across the entire characteristic map, with greater improvement at the design working point than at the off-design points. At the design working point, the mean isentropic efficiency and pressure ratio of the optimal configuration increase by 0.6% and 0.5%, respectively, while the standard deviation of isentropic efficiency, pressure ratio, and mass flow rate decreases by 32.4%, 41.2%, and 25.1%, respectively. This optimization framework proves to be both feasible and efficient and can be applied to aerodynamic robust optimization of turbomachinery. In the future, we will apply this framework to different aspects of the gas turbine life cycle to model and analyze uncertainties of larger orders of magnitude.
Keywords
Micro transonic compressor
Geometric uncertainties
Uncertainty quantification
Surrogate model
Aerodynamic robustness optimization
Multiple working points
