The global imperative to transition from fossil fuel-based energy sources to renewable alternatives has become increasingly urgent in the pursuit of sustainable and renewable power generation. This paradigm shift necessitates innovative approaches to manage the intermittency and variability inherent in renewable energy sources. Hydropower plants emerge as pivotal assets in this transition. This doctoral thesis focuses on the off-design operation of hydraulic machines, especially Francis turbines and reversible pump-turbines leveraging variable-speed, to study benefits and limitations of these asset to act as flexible power generation resources. Indeed, their inherent ability to efficiently regulate power, adapt to varying demand, and serve as a stabilizing force within the grid infrastructure, renders them crucial in ensuring sufficiency of regulating reserves in power grids. Here specifically, this work investigates the unique capabilities and advantages offered by variable speed hydraulic machines. Nevertheless, despite the myriad advantages of such a technology, it comes with certain drawbacks that necessitate careful consideration. The original design of hydraulic machines was tailored for steady-state operations, serving as a fundamental component of baseline electricity production. However, the increasing need of ancillary services provision in modern power grids, has introduced transient operational paradigms, altering the usual usage of these machines. The occurrence of transient and off-design operations tends to accelerate the wear and tear, thereby potentially reducing the unit’s lifespan prematurely. For
ensuring safety and power production reliability, accurate lifespan prediction and well-defined maintenance intervals are of importance. The study presented herein centers on wear and tear characterization methodologies, particularly
emphasizing the understanding and forecasting of the fatigue-induced damage on the runner blades during transient operation. Leveraging the reduced-scale model testing facilities at EPFL Technology Platform for Hydraulic Machines through a dedicated experiments, a diverse range of analyses is conducted to comprehensively examine the impact of transients on runner lifespan, facilitated by onboard strain gauge measurements. Two primary testing campaigns were executed
on distinct hydraulic machines, offering insights into both Francis-type pump-turbines and highhead Francis turbines equipped with variable speed capacity. The incorporation of a full-size frequency converter in the power unit presents an opportunity to manipulate the runner’s damage level by modifying the unit’s start-up methodology. The increased flexibility provided by the variable speed unit enables the mitigation of induced damage during startup by implementing a tailored speed trajectory that minimizes stress. The assessment of damage during transient operation reveals that variable speed significantly reduces damage during start-up sequences compared to fixed-speed units. Avoiding speed-no-load conditions proves beneficial. Exploring flow phenomena and fluid-structure interactions during start-up sequences identifies vortical structures and rotating stall, offering insights for understanding the origin of the most damaging operating conditions. A stress forecasting tool, relying on steady-state measurements, is proposed to predict stresses during various transient scenarios, reducing the need for experimental tests. Finally, scaling laws for transposing results from reduced scale models to full-scale machines show good accuracy in dynamic stress transposition for steady state, while challenges in transposing transient sequences are highlighted. The results of this PhD Thesis contribute valuable insights for optimizing operational and maintenance strategies in hydroelectric units, guiding hydraulic machines design improvements, and informing decisions about variable speed technology investments.
This PhD has been supported and framed within the XFLEX HYDRO project, granted by the Europan Commission (Grant agreement No 857832).