This site uses cookies which allow us to give you the best browsing experience possible. To find out more, please see our Privacy Policy.


Early damage detection through intelligent noise pattern analysis

Intelligent noise pattern analysis at the hydropower plant Rabenstein

Austria’s leading energy supplier VERBUND relies on the acoustic monitoring system OnCare.Acoustic from Voith. With this digital “ear” in the hydropower plant, the company aims to reduce servicing callouts and damage to equipment and thus save costs.

VERBUND AG is Austria’s leading energy utility and one of the most climate-friendly electricity producers in Europe. The company generates 95 percent of its electricity from renewable hydropower. Because of the longstanding collaboration between the two companies, many of the highly efficient pumped storage plants in the Austrian Alps and run-of-river power stations on all major rivers in Austria are equipped with Voith turbines. In a pilot project, VERBUND is now testing the acoustic monitoring system OnCare.Acoustic.

The hydropower plant’s “ear”

OnCare.Acoustic is an application that supports hydropower plant operators in identifying potential malfunctions, by detecting anomalies relative to the normal acoustic status of the machines. The system uses artificial intelligence methods and is therefore a useful addition to the existing human monitoring of hydropower plants.

“To record the sounds we install microphones at various locations in the power plant. These record every ambient sound, which are then saved for data preprocessing in the Voith BlueBox. The final interpretation of the data is done in the Cloud on Voith's IIoT platform OnCumulus”, explains Jose Manuel Nieto Diaz, Product Manager Condition Monitoring at Voith.

For calibration purposes, the system records all acoustic signals in an initial learning phase. In doing so, it complies with strict data protection guidelines. “In our OnPerformance.Lab in Heidenheim, industry experts use digital data analysis and diagnostic tools to compare the data collected from various power plants. Due to the combination with the respective operating data we can determine precisely which sounds correspond to normal machine behavior”, explains Jörg Lochschmidt, Vice President Digital Hydro at Voith.

Following the learning phase, the system is capable of autonomously recognizing deviations from the typical noise pattern. In the event of an incident, the system automatically sends out a warning and at the same time notifies one of the power plant operator's service technicians. This allows the operator to plan maintenance work effectively and prevent potential turbine damage in good time.

The intelligent noise pattern analysis detects acoustic anomalies in ongoing operation

Reduction of unnecessary servicing callouts

Many of the VERBUND pumped storage plants are located in secluded regions in the middle of the mountains, so they are often unmanned. It is therefore not possible to preclude the risk that problems might remain undetected and result in severe damage. This applies in particular to equipment that is neither continuously monitored nor connected to a condition monitoring system.

“As well as permanent remote monitoring of our plants, regular inspections are also carried out on site in the power plant”, says Bernd Hollauf, VERBUND project manager for the digital power plant, describing the current monitoring process. “A significant improvement seems possible through developments like acoustic monitoring systems combined with artificial intelligence. This is why OnCare.Acoustic will be specifically tested on a Kaplan bulb turbine in our pilot plant in Rabenstein in the Styrian region”.

VERBUND tests the system on a Kaplan bulb turbine in the Rabenstein hydropower plant

The system permanently monitors the condition of the plant, enabling operators to plan maintenance work more efficiently. Maintenance work should ideally only be performed when there is an actual need for it. This not only saves costs but also minimizes the existing residual risk, as malfunctions are detected in good time. “Ultimately, we expect OnCare.Acoustic to bring about a reduction in servicing and maintenance effort”, says Hollauf. “With this monitoring system we will be able to also remotely evaluate the acoustic situation in our pilot power plant in future at any time”.