As wind power becomes a bigger player in global energy production, maintaining a multitude of rotating behemoths becomes a challenge, especially since wind farms are often located in remote areas. And while it's fairly easy for a turbine technician to check the nacelle components, inspecting the blades is more time-consuming and dangerous. Unmanned aerial vehicles (UAVs) have gained popularity in the wind turbine maintenance world, as they can be outfitted with ultrasound, cameras and thermal imaging, but most of these drones still require a human pilot and they are only used to assess a turbine’s health once or twice a year. Australian upstart Ping Services has introduced an IoT-based product-as-a-service—the Ping Monitor—which continuously and autonomously inspects a turbine as it’s running, constantly checking for potential blade damage. Ping Services CEO Matthew Stead gave me a rundown of the company’s turbine monitoring technology.

Ping Monitor magnetically mounted on a turbine tower. (Image courtesy of Ping Services.)

Ping Monitor

The Ping Monitor is a maintenance-free, battery-powered device that mounts directly to a turbine’s metallic tower via magnets. Its battery is recharged by an integrated solar panel. Once installed, its “Intelligent Listening Sensor” (a microphone attached to an embedded ARM processor) starts recording the sound that the turbine makes as it rotates, spending its first two weeks on the job establishing an acoustic baseline. The device is able to reject ambient sounds (wind, traffic, etc.) that aren’t related to the turbine itself.

The Ping Monitor’s main components. (Image courtesy of Ping Services.)

After determining the turbine’s normal sound patterns, the Ping Monitor starts listening for audible anomalies, which may indicate leading-edge blade damage associated with routine wear and tear, as well as splits, pits, cracks and holes caused by lightning strikes and other sources. Most of the sounds that the Ping Monitor detects are in the human hearing range; the intensity levels differ, depending on the amount of damage. The device is also measuring the variation in sound over time, which helps it perform predictive maintenance. The on-board ARM processor analyzes the sound data and uploads the information to the “Ping Cloud,” allowing wind farm operators to remotely monitor the site on a dashboard.

Ping Monitor Dashboard

The “Damage Indicators” dashboard shows a list of the turbines (identified with a Y number on the left) and the probability that a turbine is damaged. (Image courtesy of Ping Services.)

The “Damage Indicator” graph shows the damage level of an individual turbine over time. (Image courtesy of Ping Services.)

Looking at the two images above, we see that turbine Y7 (top row of the color dashboard) is fine up until June 11, when it starts showing signs of damage, as indicated by the orange boxes in the first image and the graph in the second image going above level 4. The machine dips for a few days and then goes critical (red on the dashboard, level 5 on the graph) beginning on June 19.

Using the alert options, a grid or wind farm operator can be informed when a turbine shows signs of damage above a selected threshold, over a certain period of time, and with various sensitivities and durations.

Various alert options can be selected. (Image courtesy of Ping Services.)

When a potential problem is detected, a drone can be sent to the location to conduct a visual analysis. In this way, the Ping Monitor doesn’t replace the UAVs or technicians; it simply sends them to the turbines that are most likely in need of service. This changes the inspection model from preventative maintenance to predictive “just-in-time” maintenance, which could save money by reducing unneeded inspections and assuring that any questionable turbines are checked as soon as they show signs of deterioration.

Edge and Cloud Computing

In addition to the edge computing that provides the damage probabilities, a cloud-based server runs a more thorough analysis, examining trends over time and comparing data with other turbines of the same make and model, including those located at different sites. This deeper analysis can help the software to differentiate between real problems and phantom issues, as well as help engineers to refine its algorithms accordingly.

Data analysis at the edge and in the cloud. (Image courtesy of Ping Services.)

Ping Monitoring Service

Ping offers its service for an annual fee and an initial setup cost, both of which are dependent on the number of assets being monitored, the communication method (cellular or satellite) and the environmental conditions under which the turbines will operate. Currently, the device is limited to land-based wind farms, but the company has plans to improve weather and saltwater protection for offshore applications. It expects to have a seaworthy product available by late 2020 or early 2021.

Since the first prototype in 2014, the company has tested its device on more than 300 wind turbines, with diagnostic results confirmed by field technicians. At the time of this writing, 25 Ping Monitors are operating in the field and another 150 are on their way to U.S. wind farms. Whether this will significantly reduce long-term operations and maintenance costs remains to be seen, but as turbines get larger and wind farms become more prevalent, self-monitoring systems like these will undoubtedly be a part of the picture. They’ll include an array of inexpensive sensors—measuring stresses, strains, audio and visual parameters—and a combination of edge and cloud computing.

In addition to wind turbine monitors, Ping Services is working on proof-of-concept projects in mining, rail and electric utilities. Sound can be used to measure machine health for a wide range of assets. Bearings, for example, tend to make an obvious noise when they start to fail. As the company verifies the effectiveness of its methods, it will develop additional products and new services.