No abstract

With the need for flexible operation, Francis runners are exposed to various operating conditions beyond the traditional operating range. These machines have to be designed for long time Part Load Operation in order to meet hydraulic and structural requirements. Deep Part Load Operation (DPL) mainly consists of stochastic loads on the blades. Whereas calculation methods for the dynamic stress at full load operation are well established, calculation of the dynamics during low load operation are in the focus of current research to increase the prediction accuracy. An unsteady flow simulation has been performed for a deep part load operating condition with the goal to derive the time-dependent pressure distribution in the runner. Based on the results, a FEA simulation with a 360° model is performed for multiple time steps to calculate the structural behavior. For a subsequent comparison to a strain gauge measurement, relevant characteristics with regard to material fatigue are identified. In addition, the calculation time in relation to the accuracy of the main characteristics of the flow induced stresses will be discussed.

Today, hydropower plants received increased attention as they have the ability to stabilize the electrical grid. However, this is accompanied by more start-stop sequences as well as operation in off-design conditions. One big task is to develop improved runner designs that can resist the increased pressure fluctuations. In the scope of this study, a Francis turbine is simulated for a deep part load operating point with the use of the SAS model and the more recently developed SBES model with ANSYS CFX. Both turbulence models are hybrid RANS-LES models. The simulation results are compared against model test data from a measurement at high cavitation number. For the evaluation of the pressure fluctuations comparison is furthermore made against experimental data at low cavitation number. All in all, SAS and SBES model predict similar flow characteristics. Both, the velocity profiles in the PIV measurement plane and the predicted pressure fluctuations, show only minor differences. The comparison against experimental data shows that the amplitude of the broadband frequency spectrum, which results from the occurrence of vortices, is overestimated at the investigated locations on the runner blades.

No abstract

No abstract