No abstract

No abstract

No abstract

No abstract

Abstract. For an automatic optimization using Computational Fluid Dynamics, all design iterations of a turbo-machine require the generation of geometry and computational mesh as well as the simulation of the flow and the evaluation of one or more characteristic properties. All steps can be expensive regarding time or manpower. In addition, all meshes of the design iterations differ in total number of cells, percentage of a particular element type, local refinement sections or structured mesh block thickness being summarized as mesh characteristics. An object-oriented framework focused on hydraulic machinery is used to generate different geometry variants. Furthermore, the framework is extended to create hybrid meshes consisting of a structured part surrounding the blades and an unstructured part within the remaining flow channel. The structured-unstructured mesh coupling is realized with the Pyramid Open Method. In order to take care of the boundary layer flow effects along the meridional contour upstream and downstream of the runner blade rows a prismatic layer creation algorithm is also implemented. OpenFOAM software is applied to simulate the flow through the hydraulic machine. Each step in the process chain is implemented as part of the framework. The complete study is done using open source applications, libraries and tools. In terms of performing an automatic optimization procedure an assessment for the numerical behavior of the hybrid mesh in a real-engineering application is desired. To limit the number of optimization parameters and the amount of computation time the sensitivity analysis tool supports the identification of relevant geometry input parameters with significant influence on the flow field. On the one hand it is important to discover the influence of geometry variations on the predicted physical behavior (e.g. flow losses, torque and head) of the simulation results. On the other hand it is investigated if the uncontrollable changes of mesh characteristics or user controlled numerical discretization parameters, such as flux gradients affect the physical quantities on the same level as the geometric variations. The goal is to achieve a dominating geometry influence and a minor effect of mesh characteristics on the predicted flow field solution. For both, a typical Francis and a Propeller turbine runner, the fluctuations and absolute values of physical properties are shown for a number of varying meshes, numerical and geometrical setup parameters.