Files

Abstract

All hydraulic turbines containing a casing with stay vanes face the potential dynamic problem of stay vane von Kármán vortex shedding. For hydraulic efficiency purposes the stay vanes tend to be relatively slender in the direction normal to the flow thus being flexible in this direction. As a result structural vibrations may be excited by the von Kármán vortices shedding at the trailing edge of the vanes. When the excitation frequency coincides with one of the natural frequencies of the stay vane, resonance occurs, causing vibration and potentially initiating cracks in the vane geometry if the amplitude of the excitation force is sufficient. Although the problem occurs rarely for high head or medium head Francis turbine, it is particularly true for low head turbines such as Kaplan and propeller type turbines. The frequency and amplitude of the von Kármán vortices are highly dependent on the free stream velocity and on the trailing edge profile of the vane. At the design stage, it is important not to match the vortex shedding frequency with the vane natural frequency. The traditional method of determining the vortex shedding frequency is by using the empirical Strouhal number with the given wake thickness and free stream velocity at the vane trailing edge. But this approach is not valid for geometries that are very different from standard geometries and the dependency of the Strouhal number on the flow Reynolds number prevents us from obtaining a good empirical correlation with experimental data. The Strouhal number varies in the range of 0.15 to 0.3, and it is dependent on the vane geometry and the local Reynolds number. The determination of the Strouhal number from the literature is usually for model testing where the flow Reynolds number is usually not higher than 3.0E5, but for prototype size operating condition, the local Reynolds number (based on the stay vane length) may vary from 1.E6 to 5.E7. A CFD methodology for the prediction of the von Kármán vortex shedding frequency using unsteady flow computation has recently been developed at GE Energy, Hydro. An accurate prediction of excitation frequency and exciting forces is paramount in order to prevent damage to the structure in the prototype. The validation of the CFD results were made for both prototype and model sizes. The validation with site measurements from prototype size is important in order to evaluate the accuracy of the prediction in the context of the machine as a whole where there are a lot of unknowns such as: measurement precision, presence or absence of lock-in, actual local incidence angle to stay vane, 3D effects etc. This kind of validation gives us an indication of the level of confidence we can have in the prediction method. However, with prototype frequency measurements at site, it is impossible to validate the CFD code regarding its ability to predict frequencies and flow patterns correctly due to the quite often presence of lock-in phenomena. On the other hand, the study of the stay vane vortex shedding phenomena with a whole turbine assembly in model size is difficult due to limited access for instrumentation. Therefore an experimental investigation of the vortex shedding phenomena of a NACA profile in the EPFL high-speed cavitation tunnel has been initiated in the context of the Hydrodyna project. The reliable high fidelity experimental measurements help to benchmark and to fine tune the CFD tool such as the choice of mesh size, computational time step, type of turbulence model or wall function, etc.

Details

Actions

Preview