Learn how Voith’s future IIoT predictive maintenance solution for acoustic plant monitoring, powered by Cloudera, helps hydropower plants detect anomalies based on what it hears.
Voith is a leader in delivering industrial machinery and solutions for power transmission, hydropower generation, and paper making. Voith Digital Solutions offers industrial IoT (IIoT) and automation solutions to help companies improve productivity of their operations. Headquartered in Germany, Voith has around 19,000 employees, sales of €4.3 billion, and locations in more than 60 countries.
Voith’s tremendous footprint in the hydropower market gives its staff unique insights into industry challenges. “We noticed that while many hydropower plants could operate either unmanned or with limited onsite staffing, companies still were sending experienced inspection personnel in every week to check on the condition and operation of equipment,” said Dr. Felix Flemming, Vice President, Incubation, Voith Digital Solutions. “Even a few minutes of downtime for these massive turbines can mean significant costs due to lost productivity. Our question was: How could we automate inspections for these companies so they could learn about potential equipment issues earlier and avoid costly downtimes.”
Their answer: HyGuard™, a future IIoT predictive maintenance solution for acoustic plant monitoring. HyGuard captures, analyzes, and interprets sounds in hydropower plants to diagnose the health of the equipment. “Just like an auto mechanic diagnoses problems based on engine sounds, we analyze the noises coming out of the machinery in the power plants and can provide early warnings based on what we hear,” said Dr. Flemming.
To create HyGuard, Voith Digital Solutions teamed with Cloudera to implement an advanced analytics and machine learning platform for its IIoT data. The solution uses acoustic sensors to capture turbine, generator, and equipment noises. Each plant has multiple power units, which each generate approximately 60 gigabytes (GB) of data per month. “If we connect just 1,000 power units, that’s 60 terabytes of data a month,” said Dr. Flemming.
The Cloudera platform enables Voith to capture, store, analyze, and detect anomalies in the audio, and apply machine learning algorithms. Based on the noise level and type of noise, the platform can detect if there is a problem with the equipment and will eventually predict when this issue might impact plant operations. Operators, who might be stationed hundreds of miles away, can easily view any alerts in the HyGuard dashboard and take the appropriate action—such as dispatching maintenance or scheduling service activity.
Voith Digital Solutions chose to implement the platform in its own cloud—which enables the company to cost-effectively and easily scale to support the needs of each hydropower plant. “We need cloud capabilities because one hydropower plant is not the same as another, that means we need to connect as many plants as possible to create a large enough sample size for machine learning to be applied,” said Dr. Flemming. “This is how we can create value for our individual customers. As we initially had little internal know-how related to big data applications, Cloudera supported us in building up the technology stack for our cloud-based solutions.”
Security was a crucial requirement. “This is a highly protected infrastructure project, so we need to ensure security for our customers’ data and access to their plants,” said Dr. Flemming. “We selected Cloudera because it provides enterprise grade software with security and privacy built into it.”
Voith’s HyGuard solution helps hydropower plant operators continuously monitor the health of their assets and potentially predict turbine, generator, and other equipment issues much earlier and more cost effectively. Earlier detection can help dramatically reduce equipment downtime and maintenance costs, enabling increased energy production and increased profit margins for the asset owner.
“Several years ago, we established a digital agenda that aims to drive forward the Internet of Things in the industries in which we operate, and provide our customers with optimization tools for the operation and service of their systems,” said Dr. Flemming. “Cloudera is an important partner in this work. Cloudera helps us to transform vast amounts of complex data into clear and actionable insights, and bring new, innovative solutions to our customers. Ultimately, the goal is moving from anomaly detection to predictive maintenance and to use the insights gained to provide better equipment for future plants.”
Industrial/Energy and Utilities
Headquarters: Heidenheim an der Brenz, Germany
- Modern Data Platform: Cloudera Enterprise
- Workloads: Data Science and Data Engineering, Operational Database
- Components: Apache HBase, Apache Spark, Apache Kafka
- Predictive maintenance
- Industrial IoT
- Acoustic sensors
- Control System (SCADA)
- Detects anomalies in machine behavior
- Reduces plant downtime with earlier detection of maintenance issues
- Decreases operational costs for increased profit margins
- Expands services to provide customers with actionable insights
Big Data Scale
- 60 GB per power unit per month