多源不确定性下叶盘结构疲劳可靠性分析与优化设计

Fatigue Reliability Analysis and Optimization Design of Turbine Blade Disks under Multi-Source Uncertainties

关键词:  疲劳可靠性  叶盘结构  不确定性  随机有限元  优化设计

Key words:  Fatigue reliability  Turbine blade disk  Uncertainty  Stochastic finite element method  Optimal design

摘要:

为了更好地满足叶盘结构的设计需求，针对叶盘结构疲劳寿命分散和几何参数众多等问题，基于随机有限元法构建了融合多源不确定性的叶盘结构疲劳可靠性分析与优化设计框架。首先，采用应力敏感因子分析筛选叶盘结构关键尺寸；然后，将关键尺寸、材料属性及载荷定义为随机变量，基于其分布规律进行拉丁超立方抽样以开展随机有限元分析和寿命预测；最后，结合寿命预测结果，分别运用插值法、概率累积疲劳寿命法以及Kriging代理模型进行了叶盘结构可靠性优化设计。结果显示，经上述三种方法优化后，叶盘结构中值疲劳寿命分别提升了27%，1.4%和108%。其中，基于Kriging代理模型建立的优化方法效果最佳，显著提升了叶盘结构的使役可靠性。

Abstract:

Fatigue life scatter caused by material variability， load variation and geometrical uncertainty must be considered in engineering design to enable the structural integrity of turbine blade disk. Accordingly， a framework for fatigue reliability analysis and optimal design， coupling with multi-source uncertainties， was developed based on stochastic finite element method. Firstly， the critical dimensions were determined in view of the stress altering factor. Then， the critical dimensions， material property and load were defined as random variables， and the Latin hypercube sampling approach was employed to provide datasets for the following stochastic finite element simulation. Based on the predicted lifetime after simulated data processing， three reliability optimization strategies were raised by employing interpolation method， probabilistic cumulative fatigue life method and Kriging surrogate model， respectively. Results show that the median fatigue lifetime of the turbine blade disk respectively increases 27%， 1.4%， and 108% using three optimization methods. Particularly， the optimization model based on the Kriging surrogate model works best， by which the turbine bladed disk’s reliability is improved significantly.