为了获得不同冲击孔径(IA),冲击孔流向间距(IFD)和冲击孔展向间距(ISD)耦合作用对回流燃烧室小弯管冲击冷却特性及结构热应力的影响,开展了数值计算(CFD)及有限元分析(FEA).选择试验设计(DOE)中的最优拉丁超立方(OptLHD)采样确定了设计空间中的样本点,构建了高精度径向基神经网络模型,并基于改进非劣类(NSGA-Ⅱ)算法对综合冷却效率,壁温分布不均匀系数以及壁面最大热应力进行了多目标寻优,结果表明:综合冷却效率,壁温分布不均匀系数和壁面最大热应力随流向展向间距比、流向间距孔径比和展向间距孔径比的增大而减小;通过多目标NSGA-Ⅱ算法获得了小弯管冲击冷却结构Pareto前沿的3个目标函数值的范围为壁面最大热应力不大于5 MPa,综合冷却效率不小于0.66,壁温分布不均匀系数不大于0.16;小弯管冲击冷却综合最优结构的组合为:冲击孔径为0.94 mm,冲击孔流向间距为4.04 mm,冲击孔展向间距为5.45 mm.
Multi-objective optimization of impingement cooling of concave wall based on NSGA-Ⅱ algorithm
In order to obtain the influences of different impingement aperture(IA),impingement spacing of flow direction of impingement hole(IFD),spacing of span direction of impingement hole(ISD)coupling effect on impingement cooling characteristics and structural thermal stress of concave wall in reverse flow combustor,CFD calculation and FEA analysis were carried out.Opt LHD in DOE was selected to determine the sample points in the design space,and a high-precision RBFNN was constructed.Based on NSGA-Ⅱ algorithm,multi-objective optimization was carried out for comprehensive cooling efficiency,non-uniform coefficient of wall temperature distribution and maximum wall thermal stress.The results showed that comprehensive cooling efficiency,non-uniform coefficient of wall temperature distribution and maximum wall thermal stress decreased with the increase of ratio of IFD to ISD,ratio of IFD to IA and ratio of ISD to IA.Through multi-objective NSGA-Ⅱ algorithm,the value range of the three objective functions of the Pareto front of concave wall impingement cooling structure was obtained,i.e.:maximum wall thermal stress was not greater than 5 MPa,comprehensive cooling efficiency was not less than 0.66,and non-uniform coefficient of wall temperature distribution was not greater than 0.16.According to the combination of the optimal structure of concave wall impingement cooling:IA was equal to 0.94 mm,IFD was equal to 4.04 mm,and ISD was equal to 5.45 mm.
reverse flow combustorconcave wallimpingement coolingNSGA-Ⅱ algorithmmulti-objective optimizationPareto front
NETL
NSTL
万方数据
