Nowadays, hydropower plants are being used to compensate for the variable power produced by the new fluctuating renewable energy sources, such as wind and solar power, and to stabilise the grid. Consequently, hydraulic turbines are forced to work more often in off-design conditions, far from their best efficiency point. This new operation strategy increases the probability of erosive cavitation and of hydraulic instabilities and pressure fluctuations that increase the risk of fatigue damage and reduce the life expectancy of the units. To monitor erosive cavitation and fatigue damage, acoustic emissions induced by very-high-frequency elastic waves within the solid have been traditionally used. Therefore, acoustic emissions are becoming an important tool for hydraulic turbine failure detection and troubleshooting. In particular, artificial intelligence is a promising signal analysis research hotspot, and it has a great potential in the condition monitoring of hydraulic turbines using acoustic emissions as a key factor in the digitalisation process. In this paper, a brief introduction of acoustic emissions and a description of their main applications are presented. Then, the research works carried out for cavitation and fracture detection using acoustic emissions are summarised, and the different levels of development are compared and discussed. Finally, the role of artificial intelligence is reviewed, and expected directions for future works are suggested.

Cavitation erosion is a major concern in hydraulic systems exposed to strong pressure fluctuations. Well-developed experimental techniques exist for detecting cavitation based on measuring induced noise or vibrations, but additional tools are needed to assess its aggressiveness under operating conditions. This study investigates the capability of acoustic emission (AE) to characterise cavitation aggressiveness during long-duration cloud cavitation. A 50 h erosion test was performed in a closed-loop cavitation tunnel using a Venturi equipped with an aluminium 7075-T6 specimen. Hydraulic conditions were controlled to maintain a constant cavity length, and AE signals were recorded every 10 min during two representative 4 h intervals at 34–38 h and 46–50 h. A new AE-derived power parameter was defined using the amplitude distribution of AE envelope peaks. Both the number of impacts and the power parameter increased markedly from the intermediate to the final interval, consistent with the growth of erosion and increasing surface roughness. Conversely, both quantities decreased systematically within each 2 h test as water temperature increased. Image analysis of selected areas confirmed the progression of pitting between 34 and 50 h. Overall, the findings demonstrate that AE can capture the combined influence of temperature and surface degradation on cavitation aggressiveness, highlighting its potential as a monitoring technique for hydraulic components.

Cavitation erosion is a well-known problem for hydraulic turbines working at off-design conditions which is currently worsening because their operating range needs to be widened to cope with the variability of the new renewable energy sources, such as wind and solar. The use of acoustic emission (AE) measuring ultrasonic vibrations induced by the erosion process is believed to be the most promising technique to monitor cavitation erosion. In this paper, a 50-hour cavitation test performed in a cavitation tunnel is presented. The test was divided in periods of two hours in order to limit the heating of the water generated by the pump operation. During two specific intervals between 34 to 38 h and between 46 to 50 h, respectively, acoustic emission measurements were taken every 10 minutes. At 38 and 50 h, photographs of the erosion were taken which demonstrated a significant increase of erosion. Based on the AE measurements, cavitation impacts were estimated, sorted into classes by amplitude and also using a new “power parameter”. It was seen that the amount of impacts and the mean “power parameter” corresponding to a 2-h’ time interval increased significantly between the first and the second interval. But during the 2 h period within each interval, the number of impacts and the power parameter decreased as the temperature increased.