Big data analytics is expanding from engines to all of aviation, as players gear up to compete in an arena viewed as critical to aftermarket success.
But the technology is still evolving and not everybody is sold. “The majority of … players still are exploring ways to benefit from big data … that go beyond simple visualization of aircraft system health,” says Diogenis Papiomytis, director of commercial aviation for market analyst, Frost & Sullivan. (See market sidebar.)
Boeing Global Services (BGS) agrees, at least with the first point. At a recent BGS conference only about 25 percent of its customers said they were satisfied or highly satisfied with the results of their current analytics projects, says Ken Sain, the unit’s vice president of digital aviation and analytics. BGS sees this skepticism as an opportunity for it to drive value through Boeing AnalytX, as it terms its analytics efforts.
Big data plus tools and platforms are necessary but not sufficient to meet customer needs, Sain says. Engineering insight, real-time applications, integration/automation, and prognostics/optimization are also essential, he contends. These six interrelated and mutually reinforcing elements make up the company’s “analytics ecosystem.”
State of the Art
“The emphasis in the industry currently is on generating sufficiently sized data sets, generating a statistical baseline in order to be able to… [detect] trends over time or any exceedances,” says Friedhelm Kappei, head of industrial engineering with MTU Maintenance. “We believe the majority of monitoring programs are in the diagnostic phase and capable of some prescription though the aim of all providers is, of course, to enter the predictive sphere as soon as possible,” he adds.
There’s “still some hesitation” regarding prescriptive maintenance, adds Steve Sword, principal business development manager for commercial systems with Rockwell Collins. The avionics company is working with Airbus to launch a new on-board server/router for Airbus’ SkyWise analytics platform.
Airlines get concerned about the risk of no-fault-founds, he says. Sword thinks carriers are looking for “advisory intelligence so that their maintenance teams can make decisions.”
AFI KLM E&M
AFI KLM E&M recently added an inventory feature to its Prognos platform, which already includes functionality for aircraft, engines, and auxiliary power units (APUs). “Predictive maintenance and big data are at the heart of our strategy,” says Rodolphe Parisot, the MRO’s vice president for digital and innovation.
Parisot cites some tangible results. The MRO “is now capable of predicting a failure 10 to 20 days [in advance],” he says. And for the last 50 removals done following a Prognos alert, “100 percent of the removed parts or systems have been confirmed faulty by the OEM.”
“Big data is expected to transform the industry by enabling proactive, predictive analysis, as opposed to the reactive analysis we have seen to this point,” MTU’s Kappei says. “We are reaching a point where developments could be called ‘prescriptive,’ in that data gathered from operations” – relating to the region used, engine derating, and engine performance, for example – “are connected and help to forecast remaining on-wing time and optimal engine removal points.”
An example of how Boeing’s ecosystem works is an application upgrade that grew out of a consulting project with an airline customer to look at increasing the efficiency and reducing the cost of preventative maintenance checks of flight control surfaces. An overnight maintenance check might involve 30 hours of work for a mechanic team.
The project looked at finding a way – using predictive analytics – to focus inspections on the subset of the fleet that had “a higher probability of having an issue,” Sain says.
The project involved machine learning – training a computer to look for things that might indicate a problem. Boeing examined almost 130 parameters at 8 Hz – 8 times per second – over 60,000 flight hours, resulting in some 220 billion data samples, he says.
The company then worked to “identify candidate features that might be precursors … of a potential problem.” It also looked at “truth data” – what has been found in actual inspections – and traced back to figure out the signatures of a potential problem.
Boeing was able to determine – with 95 percent accuracy – where there was likely to be a problem, Sain says. The computer flagged a potential issue in five cases, he says, “and we confirmed – through direct measurement – that in all five cases [the carrier] needed to take preventative action.” The solution is prescriptive as well as predictive, he says, in that it indicates certain aircraft should be inspected.
Boeing is involved in engine maintenance management, as well, with its Engine Fleet Planning And Costing (EFPAC) application. As of late spring 2018, EFPAC had 11 customers including engine “total-care OEMs,” with more than 6,500 engines enrolled and generating some 10-20 percent savings in engine maintenance costs, according to the company.
One customer, KLM, was able to cut out five engine visits, Boeing says. Although the airline has only had the software for two years, it had already projected maintenance visits, using other means. When the carrier used EFPAC to assess its planning, it was able to optimize the service visit schedule, reducing the number of planned visits by five.
Having the right data is equally important. MTU’s engine trend monitoring (ETM) system, for example, currently works with smaller, more specific data sets culled directly from operations, Kappei explains, adding that the company is closely monitoring industry developments.
New features include: remaining on-wing time prediction, based on performance parameters such as exhaust gas temperature (EGT) margin; automatic diagnosis to identify the root cause of a trend shift; and a “quick fleet analysis tool” to review on-wing deterioration per engine serial number and “shop visit effects.”
ETM can be used across multiple engine platforms. “We can … monitor a customer’s GE90 and V2500 fleet with the same tool,” something that is unusual in the industry, he says.
Shop visits also generate data concerning workscope, hardware condition, repair, scrap rates, build-up, and test runs, Kappei says. When this information is combined with ETM information, the two data streams enhance MTU’s knowledge across the lifecycle so that areas such as scrap rates, module deterioration, and removal dates, become more predictable.
Customers who have been using MTU’s ETM system for a number of engine runs have benefitted from the identification of data patterns unique to them that help in planning shop visits, logistics, and fleet management, leading to parts savings and shorter turn times, he says.
AFI KLM E&M’s Parisot predicts that analytics eventually will become real-time. As of today, however, the focus is on “real-time-enhanced troubleshooting.” Among the challenges are “efficient and cheap, real-time connectivity and improving communication between the aircraft and the ground.”
Engine trend monitoring traditionally aims to identify long-term trend changes, MTU’s Kappei says. For this purpose the aircraft transmits single-snapshot reports in different flight modes, which produces only a small volume of data (about 10 KB per report). Newer developments use continuous data from the whole flight, which includes snapshots from each second. Although this data is still below 1 GB, it exceeds current in-flight data transmission capabilities and therefore is downloaded via a wireless quick access recorder (WQAR) after arrival.
Currently, most engines produce data snapshots at various points in flight, such as take-off, climb, and cruise, adds Jayesh Shanbhag, executive, services digital leader for GE Aviation. These snapshots can include up to 1,000 different measurement parameters and can range in size from 50-200 MB of data per flight, depending on flight time.
Another challenge is that not all aircraft employ the same data format, transmission capabilities, and frequencies, Kappei points out. They are equipped by different hardware and software manufacturers and run hardware and software at different upgrade levels. “These factors require highly flexible databases, data processing, and analysis.”
It is also too expensive at this time to transmit data for the whole flight in real time via ACARS although new technologies might help to reduce the costs to a level that makes real-time transmission a reality, he says.
Last year Pratt & Whitney launched EngineWise, a platform that includes “data analytics and real-time intelligence to help predict and prevent operational disruptions before they occur,” the company says. It has announced EngineWise agreements in connection with aircraft orders involving the selection of the P&W geared turbofan (GTF).
EngineWise has also added new data analytic capabilities for V2500 engines – via the eFAST “data ecosystem” — thanks to an Airbus A321 STC that allows installation of a device that can access, store, and transmit information. The ecosystem includes a “highly secured acquisition, storage and transmission infrastructure that is capable of accessing and recording aircraft and engine full-flight data, generating reports based on recorded data, and offloading data and reports to a remote ground station upon landing,” says Paul Finklestein, director of marketing. One of the key benefits of this system is the flexibility and control the operator has in sharing access to its data with other parties, he adds. “In the near future, we aim to deploy eFAST across additional platforms.”
The hardware component of eFAST includes a “small avionic device that can reside anywhere on the aircraft, but is commonly found in the avionics bay,” Finklestein says. The system has access to 100,000 engine and aircraft parameters and is capable of recording 6,000 parameters on a continuous basis, he says. eFAST provides customers advanced configuration management options without requiring physical access to the aircraft. Data can be sent via Wi-Fi, cellular, or ACARS, allowing customers to better manage the cost of data transmission.
EngineWise promises “cradle to grave product knowledge – the ability to anticipate, prevent, and proactively prescribe to create a planned environment for ourselves and our customers, by having … all critical data at our fingertips,” says Karine Lavoie-Tremblay, associate director for engine business intelligence.
“We capture an enormous amount of data to analyze,” she says, but what is more imperative is to focus on the most important data that helps improve overall reliability.”
Real-time data transmission and analytics are performed today and are critical for maintaining flight safety, she says. But adding more of these capabilities needs to be balanced with the availability of technologies, operational cost, safety, and the ability to react.
She sees “three technical shifts” enabling the expansion of digital capabilities: global enterprise systems; fast, high-volume processing; and the ability to aggregate asynchronous data.”
GE Aviation continues to refine and expand its concept of “digital twins,” digital replicas of engines that allow systems to be modeled and analyzed. “By creating a digital twin, we’re able to gather and analyze more data from engines,” explains Jon Dunsdon, chief technology officer with the Digital Solutions business. “This process allows us to get a clearer idea of how an engine will respond for better predictability.”
GE has applied the concept to improve engine availability and fuel efficiency. It can model the overall health of the engine and can “create an analytic model of the health … specific to that engine because you’re getting the actual data off that engine.”
Digital twinning also can move actions from unscheduled to scheduled maintenance, reduce the inspection burden, and increase flight efficiency.
More specifically, creating a digital twin helps airlines fly the aircraft better, Dunsdon says. The approach can provide insights on the most efficient way to climb. “You don’t want to climb too aggressively and deteriorate the engine,” he says. Or you can look at “things like taxi[ing] on one engine, when you operate the APU… .”
The company has “continued to expand on these time-series models,” incorporating maintenance and operational data “to bring more context and fidelity … and have another step change in our results.”
We continue to use data analytics to drive more accurate, prognostic fleet monitoring, says Bill Dwyer, general manager services marketing. “Our Fleet Support facility monitors some 36,000 engines, and processes more than 100 million records per year.” Moreover, the OEM’s digital capability “enables us to analyze that data to issue about 14,000 customer notifications per year with 90 percent accuracy,” he adds. “That’s an incredible productivity driver for our customers, because we’re giving them advance notice of issues, enabling them to proactively schedule maintenance and to avoid costly operational disruptions.”
Rockwell Collins and SkyWise
Airbus is working to get more data off its airplanes, partnering with Rockwell Collins to get the avionics company’s latest server/router on its A320s as both a forward-fit and retrofit.
FOMAX – for flight operations and maintenance exchanger – combines maintenance and flight operations functions in a single box, says Rockwell Collins’ Sword.
Capable of uploading up to 12 GB per day, FOMAX provides two air-to-ground channels – a direct connection to Airbus’ SkyWise analytics platform, hosted in the Airbus cloud, and a direct connection to the airline, via two 4G cellular radios. There is also broadband satellite connectivity, but this is used primarily for flight operations, he says. “At this point there is no real-time parameter streaming in flight.”
The way the box is being configured, it’s going to collect significantly less data up front, Sword says, because the amount of data it collects is driven by what you can do on the ground. “Until the algorithms can make sense of that much data, there’s [no] point in collecting it and downloading it.”
The promise of Skywise is predictive, Sword says. If it detects an anomaly, it suggests to the operator that there is some root cause worth investigating. He cites Airbus’ public statement that in early trials with launch customers SkyWise was able to demonstrate a 30 percent reduction in operational interruptions, using predictive analytics.
GE Aviation’s Accelerator in Washington, D. C. Launches to Drive Efficiency
In October GE Aviation’s opened an Accelerator to provide a place in the D.C., Virginia and Maryland area for hands-on collaboration with federal and defense customers to achieve mission-critical outcomes. The accelerator is an innovation space focused on using data and analytics with idea generation, incubation, and development for products and services.
GE Aviation Chief Digital Officer John Mansfield officially opened the accelerator with Colin Parris, vice president of Software Research at GE’s Global Research Center and Tony Mathis, GE Aviation president of Military Systems. Technology partners participating in the event included Microsoft, Teradata, Hewlett Packard Enterprise, Intel, CACI, and VION.
“GE Aviation’s Accelerator in Washington, D.C. is home to software developers, architects, data scientists and domain experts with specific backgrounds in analytics, maintenance, and engines,” said Mansfield. “Building on the strong partnership we have with our customers will allow us to continue to share both our physics and digital-based capabilities, improving their asset availability, efficiency and operations.”
Through the partnership, data scientists, domain experts, software developers, and solution architects from GE will work together with federal and defense customers to distill some of the 10 billion data points produced by the defense sector annually into solutions that can achieve condition-based maintenance and connect disparate data sources.
“GE Aviation’s Accelerator is where digital and physical technologies come together in powerful ways to improve industrial operations of all kinds,” said Parris. “We’re combining data, AI and our deep industry domain expertise into digital twins that provide unique insights and solutions to drive better mission-based outcomes for the military and for industry.”
“Digital transformation is a strategic imperative for aviation to stay competitive in today’s ever-changing world,” said Mathis. “We’re partnering with customers to bring real and tangible outcomes leading to increased readiness, affordability, mission effectiveness and operator safety.”
GE Aviation’s Accelerator is located at the Warner Building, 1299 Pennsylvania Avenue, NW, 9th floor, Washington, D.C. The collaboration space is comprised of data scientists, engineers and UI/UX designers in the Delware/Maryland/Virginia area.
GE Aviation’s Accelerator in Washington, D.C., is their second collaboration space in the United States. The first one is in Austin, Texas, the headquarters of the Digital Solutions business. GE has similar spaces in Dubai and Munich.
GE Aviation says more than 300 unique airlines, OEMs and business jet operators covering more than 10,000 aircraft are GE Aviation’s Digital Solutions customers for services such as flight and fuel analytics, navigation services, operations management, and planning and recovery.