Siemens Worldwide

The Magazine


The Magazine

Vibration diagnostics

No typical workday: The wind power data analysts

At the Siemens Remote Diagnostic Center in Brande, Denmark, data analysts predict turbine faults based on vibration anomalies. Who are the people working in this new business, and what is their working day like?

It’s a grey and rainy late-summer day as I arrive at the Siemens wind power center in Brande, a small town on the Jutland peninsula in Denmark. The train coming up from Copenhagen passes arrays of onshore turbines, placed in a largely flat landscape with low rolling hills, that catch the steady wind sweeping in from the North Sea coast. This land is clearly well suited for wind power. A group of Danish technology pioneers began operations here in 1980. When Bonus Energy was acquired by Siemens in 2004, the wind service headquarters was established in this location; in September 2014, the Remote Diagnostic Center (RDC) was completed.

In Brande, over 130 experts keep watch over nearly 10,000 Siemens wind turbines worldwide, analyzing the steady flow of data to detect minute irregularities that might indicate impending failures. The remote monitoring and diagnostics of wind turbines is part of the Siemens Digital Services, powered by the Sinalytics platform. What sort of people are these data analysts, I wonder, who have come to work in rural Jutland in a new profession that is more usually associated with big Silicon Valley names such as Google? And what do they do all day?

The Siemens Remote Diagnostic Center (RDC) at Brande, Denmark was inaugurated in September 2014.

The anti-nerd

“We don’t really have a ‘normal’ schedule here, and I can’t tell you what a typical workday would be like.” Bo Roemer-Odgaard, head of the vibration diagnostics team, does not look like a nerdish hipster at all. A sturdy-looking, down-to-earth former marine engineer and first officer with the Maersk shipping conglomerate, he spent more than seven years at sea before starting work in the wind industry as a gearbox coordinator. During his first year, he became involved with the vibration diagnostics team, which he subsequently joined.

While turbine monitoring continues around the clock, 365 days a year, much of Roemer-Odgaard’s work involves meeting with customers and identifying their needs, as well as developing the system and refining models to take those needs into account. This is done by harnessing the expertise of the diagnostics team, which includes mechanics specialists with proficiency in turbines as well as data cracks. Some of them hold, or are currently working on, PhDs in arcane topics such as signals processing, deep learning and neural networks, and machine learning.

Hitting the sweet spot

“This mix of people is extremely important,” says Roemer-Odgaard. Hierarchies are flat at the RDC; analysts and engineers exchange knowledge and work closely together. A large wall-mounted LED screen shows the operational states of the Siemens turbines being monitored, reflecting the information derived from 300 sensors attached to each of them. But the head of the vibration diagnostics team has a caveat: “You can go awfully wrong if you only look at the data.” The Sinalytics platform connects domain knowledge with high-quality data and analytics, which Roemer-Odgaard refers to as “hitting the sweet spot”, that helps the team achieve a detection hit rate of more than 99 percent for drive-train damage including, but not limited to, main bearing damage and gear-tooth cracks. The team performs roughly 30,000 manual vibration analyses every year to insure the high hit rate.

The status display helps analysts Morten Kaiser and Mads Lauridsen monitor the health of Siemens turbines around the globe.

Digitalization of this human expertise is progressing rapidly. A new computational framework, Pythia, will help the analyst team dig even deeper into the data, including access to historic datasets and use of advanced diagnostics. “Right now, we are only scratching the surface of the data,” says Roemer-Odgaard. Since 1998, the goal has been to detect both slow- and fast-developing faults as quickly as possible, before they deteriorate. With Pythia, the framework containing algorithms named after the Oracle of Delphi, the team hopes to be able to predict likely problems at the time the turbine is commissioned and accurately forecast its operational lifetime.

Right now, we are only scratching the surface of the data.
Bo Roemer-Odgaard, Head of Vibration Diagnostics, RDC Brande

Faith in wind

Due to his cross-cutting approach, the one thing Roemer-Odgaard can’t really predict is next week’s work schedule. Any given day might involve a visit to the turbine workshop to check new components or optimize the placement of sensors with the designers, who meet with the vibration experts regularly for exchanges about design and risk. Or he may sit down with the service engineers to hear about new trends and developments or failures that the diagnostics people haven’t heard about yet. Or he may work with the chief engineer in charge of condition measurement to optimize the model with even better information.

There is one thing that unites all of the people working up here in windy Jutland, says Roemer-Odgaard: “You’ll find a great deal of idealism among the people working here. We’re all dedicated to renewables, especially wind power, and believe in the need for a true energy transition.” And then, of course, there’s the shared excitement of working in a new, uncharted field and pioneering a new profession on the cutting edge of the digital revolution.

Christopher Findlay, journalist based in Zurich
Picture credits: Lars Moeller, Siemens AG