The hype phrase today is “Big Data.” It is a phrase Mark Bünger, Research Director for Lux Research, believes can have a number of definitions depending on who is doing the defining. Bünger was the lead analyst on a report that Lux published in April 2015 that surveyed both the technology as it is today and how various industries are adapting to the opportunities—and perils—in trying to exploit it.

Avoiding a precise definition in an interview with Aerospace & Defense Technology, Bünger explained Big Data is about raw data, databases, and speed. Companies now have access to a growing number of sensors attached to infrastructure and assets, from oil derricks to aircraft engines. These sensors provide vast volumes of data in high velocity. The data is variable—think spreadsheets to Twitter tweets. A major finding of its study is that, while information industries like banking and media know how to derive benefit from vast data sets, there is risk for industries that Lux terms “material-centric.”

There is one material-centric industry that is gaining advantage in collecting, storing, and using data vast enough to qualify as Big Data. “The aircraft industry is pretty far along and other industries should follow its lead,” he said.

This is especially true for engines over airframes, according to Bünger. “The [operating] conditions in engines are dynamic, with thousands of parameters, sometimes every millisecond. With airframes, the data rate is lower because data is only interesting during takeoff, some during flight, and then landing.”

Mark Bunger from Lux Research notes that while the Velocity, Volume, and Variety definition of Big Data works for companies in IT, finance, and media, a better practical concept for industries like aircraft manufacturing and maintenance is around operational realities.
He also noted that it’s not just the data for engines during flight that is worth tracking. Tools and parts in support of maintenance and repair operations are also useful. These require more advanced data capturing techniques, since the information is less structured even while using them efficiently is important. All of it becomes useful to the right company that knows how to use it.

“Big Data is a technology that is going to be very impactful for us,” said Larry Volz, Chief Information Officer and Vice President for Pratt & Whitney. He too was wary about putting a precise definition on the term. ”It is more about using the information we already have from our products and processes in a different way, moving from a descriptive analytical look at current information into being more predictive.”

Predictive analytics—predicting future behavior and actions—is really what it is all about, according to Volz.

Moving From Reacting to Predicting

The key to predictive analytics is modeling all of that data statistically to gain new insights. The challenge and opportunity is in the nature of the data. The data store available to Pratt & Whitney today is indeed both vast and varied.

Pratt & Whitney plans to capture 50 times more data on its newest engines, such as the geared turbofan PurePower PW1100G-JM engine shown here, compared to previous models.
“We take data during manufacturing from our ERP/SAP system, we collect data during the build and overhaul operations, from customer service, our global field representatives, and warranty reporting system. We combine that with data from our on-wing engine health monitoring system, the Advanced Diagnostics & Engine Management, or ADEM,” Volz said.

He also stressed that the company has all along been working with large data sets, such as simulation data used by design engineers to predict engine power output, NVH, and fuel burn. Now, by applying statistical models, they extend predictive capabilities to in-service use. He relates that they accurately predict unplanned engine events that could cause a delay or interruption in a flight.

Although there were pockets of usage, as Volz described it, in 2014 the company decided to invest heavily with a company-wide initiative, working with IBM. He believes that incorporating outside expertise was important. “You need that expertise in areas such as statistical data modeling that a company like ours does not necessarily possess,” he explained. “We now have 90-95% confidence in the statistical models, based on our legacy fleet of engines.”

According to Larry Volz, CIO and VP for Pratt & Whitney, predictive analytics is what big data is all about.
What does he view as the enabling technologies to make this true in 2015? “There is a convergence of three major factors,” he answered.

First is P&W’s prior investment in a global enterprise system that captures all of the product and process data needed. “That foundation is critical,” he said. Second is the vast increase in the industry’s high-speed computing infrastructure that can store and process these huge data sets at what Volz describes as a reasonable cost. Third is the ability today to merge both synchronous and asynchronous data and use them in predictive models—not just tables of numbers (synchronous), but word documents from customers and comments from field service representatives (asynchronous). “We can take sentiment data from customers and roll that into our statistical models,” he said.

P&W is expanding this predictive analytic capability beyond engines. It will deliver its eFAST on all of Bombardier’s CSeries aircraft systems to provide realtime monitoring of all critical aircraft systems, not just engines. The eFAST system will be the infrastructure unit used to perform data transmissions from the CSeries aircraft’s onboard Health Management Unit to the ground.

Aircraft Industry on the Cusp

On May 12, 2015, Pratt & Whitney announced it was delivering its eFAST to facilitate remote troubleshooting on all of Bombardier’s CSeries aircraft systems.
IBM, Pratt & Whitney’s partner in expanding its monitoring and predictive analytic capabilities, certainly sees a future in the field. IBM has invested more than $30 billion to build its capabilities in using data better. The investments include R&D, more than 30 acquisitions, and new business units for Analytics, Watson, and Internet of Things (IOT). But a focus on the data itself, however “big” it might be, is not the primary focus of its efforts, according to Jerry Kurtz, North America Leader, Big Data & Analytics at IBM Global Business Services.

“The focus of our energy is making use of all available data, internal or external, structured or unstructured, to provide insights not available before,” he explained. “It is all about moving beyond the past and present by predicting the future. We want to help companies like Pratt & Whitney predict problems before they happen.”

IBM’s view of the ultimate evolution of predictive analytics is using the data stores available today in a cognitive manner.
His experience with companies like Pratt & Whitney among many others led him to stress three important fundamentals that companies need to address in getting value from predictive analytics: business capability, information foundation, and organizational governance. “Business capability is about identifying specific use cases that provide specific value, such as the case with Pratt & Whitney predicting future engine events” through modeling, he explained.

Information foundation is the investment needed to make the right data accessible to provide useful predictions. “This is more expensive and actually provides less actual value than developing a specific use case, but it is required,” he remarked.

Finally, he also stressed the third factor— an organization ready to move into a new world of predictive, prescriptive, and cognitive use of data. Questions around positions and roles need to be answered. “What is the role of the CIO or Chief Data Officer? Does the organization need new skills? Do you insource, outsource, or co-source these capabilities?” he asked rhetorically.

He also believes that the aircraft industry, more than some, is ready to significantly expand its capabilities quickly. If industries like finance, insurance, and banking have significantly moved up a capability curve in taking advantage of analytics, aircraft will follow quickly. Why?

The three pillars of developing a predictive analytical capability are equally important, according to IBM.
“They have spent the last few decades getting their house in order in basic transactions,” he explained. This means installing the latest generation of ERP as well as getting structured processes and data in place. “This is really hard to do, but most companies [in aircraft manufacturing] are beyond that now,” he said, making them ready to exploit data effectively for predicting and anticipating future events rather than reacting to events as they occur. The future will include tying this knowledge back into future designs, making a closed-loop system.

“I have seen fads in my 25-year career, and this is no fad,” said Kurtz. “Everybody in aerospace is dipping their toes in the water. Everyone in aerospace manufacturing is working with us (or others) on some number of projects.”

He thinks that in just a few years companies will have scalable, on-demand, cloud-based solutions with hundreds of models running in production, with predictive dashboards helping managers not only understand how engines or aircraft are currently operating, but how to intervene best to keep them that way for the future.