Industrial Data Analytics: An Edge to Cloud Strategy

Jay Allardyce, COO, GE [NYSE:GE]

Jay Allardyce, COO, GE [NYSE:GE]
Adding edge-computing to an industrial cloud environment results in a more holistic architecture that prioritizes real-time insights for better business outcomes

The journey to digitization in industry starts with the cloud, but it’s not a journey that industrial CIOs are embarking on without trepidation. For most of GE’s industrial customers, the focus is on the precision operation of big machines—locomotives, aircraft engines, and power plants. Shifting that focus to cloud platforms and software applications that analyze and extract value from machine-generated data can be hard. In an industry, heads of operations know their machines intimately. By comparison, data can seem abstract and unfamiliar.

"While consumer internet data is generated hourly and daily, machine data is collected by the tens of macroseconds, or faster than the blink of an eye"

It’s ironic that the volume and potential value of data in an industry is overwhelming in comparison to the consumer world. A power plant generates more data in a day than a human active on social media does in a year through tweets, posts and videos. We believe the industrial internet could grow to be twice the size of the consumer internet. While consumer internet data is generated hourly and daily, machine data is collected by the tens of macroseconds, or faster than the blink of an eye. So how do industrial CIOs partner with their peers to manage and lead a move to digital? And how can the cloud accelerate that journey? The answer from a technology perspective lies in formulating an edge-to-cloud strategy that connects operations enterprise-wide and enables real-time analytics and insights to be captured and shared from the machine to boardroom with ease.

Energy’s Big Data Problem

The energy industry is a wonderful example of an industrial sector embracing the cloud and data science to drive operational efficiency and meet the challenges of business model disruption posed by emerging technologies. The shifting mix from traditional to renewal fuels types; the increase of distributed energy resources; changing greenhouse gas emissions standards; and new pressures on grid infrastructure are among the changes that present challenges and opportunities for the industry. So, where does big data come into play?

Check Out: Top 10 Data Analytics Companies

Consider this: a single gas turbine generates more than one petabyte of data every year. Globally, gas turbines generate more than 20 percent of the world’s electricity. In sum, these machines generate more than two exabytes annually. Now expand this math across all generation types, from nuclear and coal to wind and hydroelectric; and the enormous scale of data in the energy industry is clear.

Today, eight percent of the world’s electricity capacity never reaches a customer because of unplanned outages. The World Bank assesses that 75 percent of those outages could be avoided by the effective application of predictive maintenance software. If we can solve that problem, imagine the progress we could make in extending power to the more than 1 billion people in the world who currently are without access to reliable power.

While the big data opportunity for the industry is clear, progress is another matter. Currently, only two percent of the data produced in the industry is analyzed. If we can change that, and expand data science and predictive analytics across the entire industry we could generate $1.3 trillion in value according to the World Economic Forum. Addressing asset reliability in power plants and wind farms alone represents a $387 billion global value.

In order to truly capture this value, the energy industry needs to move from analog technology to digital. The foundation for such a move must be a cloud platform that is capable of capturing and storing data at scale, coupled to software applications that analyze data in real-time and power the machine intelligence and deep learning that enable predictive and prescriptive analytics.

Cloud-Based Analytics Applications

The first step for an industrial cloud strategy is a secure cloud platform that can serve as a basis for developing and deploying industrial applications. The scale and agility afforded by cloud-based industrial applications will enable energy businesses to begin analyzing in real-time the volumes of data currently unused. Why the cloud? We estimate the cost to a typical utility of building its own infrastructure at sufficient scale could exceed $200 million.

Here’s how the cloud can transform the way utilities manage outages. Today, most utilities have operations centers which manage alerts. Typically, those alerts are received a few hours, or maybe a day in advance of an outage. A nimble operations team can use that time to prevent an outage, or will at least hastily schedule necessary maintenance. Worst case, the alert is so severe that it causes a shutdown and millions of dollars in lost productivity.

Now let’s replay that scenario with the cloud, by applying data analytics. Thousands of data points captured from machine sensors, analyzed in real-time, are compared to historical reference data. Software predicts an outage long before it happens. An alert is generated showing that the last time the machine functioned in that way, an outage followed. Instead of getting a 24-hour warning or less, operators are notified ten days in advance of the potential breakdown, and they gain the time needed to fix the problem, avoid an outage and extend the overall life of the machine. That’s how analytics improves machine reliability by percentage points and delivers on that $387 billion value.

Real-Time Analytics at the Edge

However, in the industrial world, the cloud cannot stand alone. Whereas the consumer internet showed us how to generate insights over time from consumer data that can inform—for example—advertising strategies, industrial data is far more perishable. In the industrial world, taking immediate action based on machine intelligence is critical. The delivery of energy through the Electricity Value Network, from generation to transmission and distribution to end user, is successful only when machines are reliable, operations are seamless and the business of energy trading is accurate. A breakdown of the electricity network would certainly be felt differently than a loss of streaming video or the wrong advertisement being served up on Facebook.

And that’s where edge computing comes in. Edge devices and edge analytics collect and analyze data at its source, and take action directly at the source. Extracting data directly at the edge enables key workloads and analytics close to the machine; connecting the edge to a larger cloud environment gives the context. How does that data compare to the rest of your fleet, regions and business overall?

Take a locomotive, for example. A modern locomotive carries more than 200 sensors that collect gigabytes of information, processing over one billion instructions per second. On-board edge computing analyzes data and applies algorithms to run smarter and more efficiently.

Adding edge-computing to an industrial cloud environment results in a more holistic architecture that prioritizes real-time insights for better business outcomes, which should be the real focus of any digital transformation initiative.