Flexible operation of Francis turbines

Abstract

With the planned cuts in emissions, more of the electric energy will have to come from renewable sources. Two of the largest potential sources are wind- and solar, which have the downside of also being intermittent and non-dispatchable. Since the electricity production must match the consumption, other energy producers in a grid must increase their rate of flexibility to compensate, of which hydropower is well suited to do. However, with more frequent start-stop cycles and high ramping, an increase in fatigue and failures of reaction turbines, such as Francis turbines, are to be expected, so being able to predict how much additional damage this new scheme of operation will cause is of high interest. The primary objective for the thesis is to improve the understanding of how various operation schemes affects the Francis turbines through model experiments and onboard measurements of strain where runner blades are most prone to develop cracks. Focused on presenting both an experimental setup and findings, this study delves into a comprehensive measurement campaign focused on a low specific speed Francis turbine. These experiments make up a critical part of the HydroFlex project, which seeks to validate simulations and gain deeper insights into the reduced lifespan of Francis turbines attributed to increase in fatigue loading resulting from flexible operation. A setup and procedure for the calibration of runner blade mounted strain gauges has been developed, and the results is presented along with proposals for further improvements. This calibration setup consists of both a custom made jig to accurately load the runner blade in a predictable manner, and a numerical counterpart in ANSYS Mechanical to simulate the stress and strain at the location of the strain gauges during the same loads. Onboard measurement of the blade strain during operation in the model Francis turbine test rig at the Waterpower Laboratory (NTNU) is also presented with both some of the challenges encountered and also the final results obtained. The measurements span a wide range of speed- (nED) and discharge factors (QED), and demonstrate the impact of these factors on the resulting dynamic strain. Key challenges have been low sensitivity in the strain gauges first used and a high amount of electrical noise in the measurement chain. High susceptibility to small changes in the water temperature has also been a big challenge, causing the measured mean strain at no load to drift far away from its calibrated offset. Improvements to the strain gauge setup, and suggestions for how to capture mean trends despite temperature drift is discussed. The results shows that while the peak to peak strain increases as the load is increased above the design point, the highest peak to peak occurs at part load. It is also seen how the trailing edge is impacted differently near the hub versus the shroud at off-design operating conditions due to which effects are the main source of the vibrations. The credibility of the results were also strengthened by an FFT analysis of the measured strain, as all the expected peaks were present, and they were orders of magnitude larger than the baseline noise and unexplainable frequencies. Future research regarding onboard strain measurements in model scale Francis turbines should first focus on capturing the mean strain values despite temperature drift, either through compensation or extra procedures during the measurements, or in the post-processing. From there, the foundation for a numerical model or tool to estimate the life time reduction based on operation schemes as the input can be made. This foundation can be further expanded upon and generalised with more measurements of a similar fashion on other runner designs of various specific speeds.