关键词: 多目标, 多学科设计优化, 近似模型, 流热耦合, 涡轮冷却叶片

Abstract: Quick multidisciplinary design optimization technique for complicated structures is studied, and the high computational costs and numerical noise issues are successfully resolved for cooling turbine blade design. Based on analytic and feature modeling methods, parametric geometric model is developed. The model can be rebuilt automatically according to the parameters changed, based on secondary development technology of CAD software. Coupled aerodynamic and heat transfer analysis is accomplished to get the result accurately. Pressure on the blade surface and temperature in the blade body obtained from former coupled analysis are specified as boundary conditions for structural analysis. As the grid for structural analysis keeps the same with the one for CFD analysis, these data are applied in one-to-one correspondence by the nodes. Creep life of blade is predicted by empirical formula according the strength result. 18 parameters are taken as design variables at the blade tip and blade root sections. Optimal Latin hypercube sampling method is used to generate the experiment design points, and quick design optimization model is established by using Kriging functions. Thus, a cooling turbine blade multidisciplinary design optimization program, relating to aerodynamics, heat transfer, mechanics, vibration and fatigue life is established. The optimization objective is to minimize the blade temperature, total pressure loss and blade weight, with the constraints of resonance margin, stress, creep life and maximum deformation. Finally, the optimization is accomplished and blade comprehensive performance is improved, significantly.

Key words: Cooling turbine blade, Coupling aerodynamic-thermal analysis, Multidisciplinary design optimization, Multi-objective, Surrogate model