Effect of start and stop cycles on hydropower plants: modeling the deterioration of the equipment to evaluate the cycling cost

The objective of this research is to model the degradation of hydroelectric equipment in order to estimate the cost of starts and stops. After the deregulation of the energy market, hydroelectric power plants are now operating with up to 6-7 cycles per day, which can be considered as three times what they were designed for. This mode of operation can be very damaging to the equipment of hydro plants. Until now, very few methods have been proposed to estimate the additional cost of this mode of operation, they are mainly based on economic or technico-economic analyses, without looking at the cause of the degradation, which makes them auditable. Thus, the work of this thesis focuses on the design of a methodology that relies on auditable methods to estimate the cost of starts and stops. These methods are representative (by material) and reproducible (can be applied by different operators) based on explicit assumptions and validated by EDF experts. In a first step, a study has been carried out by collecting experts' opinions and analyzing historical data in order to determine which components of the power plant are the most impacted by start and stops and which types of degradation they generate. We then proposed three different degradation estimation methods, each adapted to a particular type of equipment, based on the main degradation mechanisms or phenomena and on the available data. The first method proposes an estimation of the damage of turbine runners subjected to polycyclic fatigue using in situ deformation measurements. The damage estimated during start and stop sequences is compared to the damage estimated during stabilized operation sequences, thus providing a cost in equivalent hours of stabilized operation. On each sequence of recordings, a Rainflow analysis is used to extract the numbers and magnitudes of recorded stress cycles, which are then converted to damage using a Woehler curve and accumulated using the Palmgren-Miner rule to obtain the damage rate. The second method aims at estimating the reduction of the generator life. First, we compare the estimated lifetimes on two populations of power plants, one including cycled and one not cycled, in order to confirm the negative effect of cycling on the lifetimes of the plants. We then quantify this loss in lifetime using an accelerated aging model expressing an acceleration factor as a function of temperature cycling characteristics, and we determine the characteristics of the temperature cycles generated by the start-stops from measurements recorded using the EDF online monitoring tool. Finally, a third method estimates the additional cost of starts and stops by taking into account the degradation of equipment such as the shut-off valve and the circuit breaker, based on feedback, databases and expert judgement. (author)