Uncertainty and Sensitivity Analyses of the Simulated Seawater—Freshwater Mixing Zones in Steady-State Coastal Aquifers

The uncertainty and sensitivity of predicted positions and thicknesses of seawater-freshwater mixing zones with respect to uncertainties of saturated hydraulic conductivity, porosity, molecular diffusivity, longitudinal and transverse dispersivities were investigated in both head-control and flux-control inland boundary systems. It shows that uncertainties and sensitivities of predicted results vary in different boundary systems. With the same designed matrix of uncertain factors in simulation experiments, the variance of predicted positions and thickness in the flux-control system is much larger than that predicted in the head-control system. In a head-control system, the most sensitive factors for the predicted position of the mixing zone are inland freshwater head and transverse dispersivity. However, the predicted position of the mixing zone is more sensitive to saturated hydraulic conductivity in a flux-control system. In a head-control system, the most sensitive factors for the predicted thickness of the mixing zone include transverse dispersivity, molecular diffusivity, porosity, and longitudinal dispersivity, but the predicted thickness is more sensitive to the saturated hydraulic conductivity in a flux-control system. These findings improve our understandings for the development of seawater-freshwater mixing zone during seawater intrusion processes, and give technical support for groundwater resource management in coastal aquifers.

混流式水轮机转轮设计变量耦合强度分析

在开展混流式水轮机转轮的多学科优化时通常需要进行设计变量的学科耦合强度分析,以便为建立简洁、高效的多学科优化求解策略提供依据。该文提出了一种基于全局相对灵敏度的转轮设计变量耦合强度分析方法,该方法基于参数化全三维反问题设计理论实现对转轮几何的参数化控制,并通过改进后的MorrisOAT法进行各学科目标函数对转轮设计变量的全局相对灵敏度计算,然后以各设计变量的全局相对灵敏度集合为论域,引入模糊隶属度函数量化计算设计变量对各学科目标函数的隶属度。最后,以各设计变量的隶属度值为基础,提出了设计变量的耦合强度判定准则,为设计变量的耦合程度属性确定提供了参考。采用该文所提出的分析方法对某混流式水轮机模型转轮的设计变量开展了耦合强度分析,并根据变量的耦合强度分析结果开展了转轮的优化设计,优化后的转轮不仅使得水轮机在3个优化工况下的水力效率分别提高0.2%、0.82%和1.2%,同时叶片的空化和强度性能也得到改善,该结果表明该文所提出的混流式水轮机转轮设计变量耦合强度分析方法能可靠有效的界定各设计变量的耦合程度属性,从而验证了方法的可行性。