AP1000核主泵模型泵导叶水力设计与数值优化

Hydraulic design and numerical optimization of guide vane of AP1000 model nuclear main pump

采用基础水力设计方法,利用模型泵叶轮外参数设计AP1000核主泵直导叶;基于CFD的数值计算与分析结果,针对所设计的直导叶内部流场流动损失较大的问题,提出2种不同的优化设计方案:第1种直接在直导叶的基础上进行优化设计;第2种采用叶轮与导叶配合设计的方法,将直导叶转化为扭曲空间导叶,并对其进行优化.在空间导叶的优化中也考虑了2种方案,对叶型进行参数化造型,然后生成样本数据库,选择叶轮加导叶结构的水力效率作为优化的目标函数,利用人工神经网络和遗传算法对不同空间导叶叶型进行多次优化改进,最终获得与叶轮水力性能匹配优良的核主泵导叶.最后讨论了不同导叶叶片数对水力效率的影响,从流动的角度对导叶叶片数的选取提供了理论指导和参考.

Based on self-design of impeller of AP1000 nuclear main pump, the upright guide vane was designed using foundational hydraulic methods. In order to reduce the hydraulic loss in the vane flow passage, two different schemes were proposed to optimize the vane design after analyzing the CFD com- putational results. The first one was to directly optimize the upright guide vanes design. And the other was to redesign the upright guide vanes as twisted spatial type to match up the impeller and subse- quently to optimize the vanes. In the space guide vanes design and optimization process, two different solutions were adopted to parameterize the model, and the sample database were generated. With the hydraulic efficiency of impeller and guide vane being selected as the optimization objective, ANN and genetic algorithms were used to repeatedly optimize the blade profile of space guide vanes, and finally the guide vanes were obtained with better hydraulic efficiency matching with the impeller. The influ- ence of different numbers of guide vane on the hydraulic efficiency was also discussed to provide the theoretical guidance and reference in hydraulic performance.