涡轮冷却叶片低周疲劳可靠性协同分析

COLLABORATIVE ANALYSIS OF AIR-COOLED TURBINE BLADE LOW-CYCLE FATIGUE RELIABILITY

考虑材料性能和载荷的分散性,运用分布式协同响应面法对涡轮冷却叶片低周疲劳可靠性进行协同分析。运用有限元分析软件,对涡轮冷却叶片进行流-热-固耦合分析;利用人工神经网络构建其平均应力和总应变幅的响应面模型;结合GH4133镍基高温合金的疲劳试验数据,运用线性异方差回归分析方法构建材料寿命响应面模型;以叶片的平均应力和总应变幅作为输入变量,代入到材料寿命响应面模型中得到叶片低周疲劳寿命分布式协同响应面模型;将获得分布式协同响应面模型代替有限元分析模型和材料疲劳试验进行Monte Carlo可靠性分析,得到了不同可靠度下的叶片低周疲劳寿命。与传统响应面方法相比,分布式协同响应面法降低了响应面的非线性程度,提高了分析精度。

Considering the dispersion of the material properties and load,the collaborative analysis of the air-cooled turbine blade low-cycle fatigue reliability was done using the distributed collaborative response surface method.The deterministic analysis of the turbine blade was completed by the finite elements software.The total strain amplitude and mean stress response surface was established using the artificial neural network and the material life response surface using the linear hetero-variance regression method.The blade life response surface was established by collaborating with the two response surfaces.The low-cycle fatigue reliability analysis of the blade was completed using the Monte-Carlo method.Comparing with the traditional response surface method,the distributed collaborative response surface method is more accurate.