Long before the “Internet of Things” became a catch-phrase, aircraft engine makers were using data from engines to understand and improve their products. In the decades since, engine monitoring has spread to virtually all aircraft components and has been redefined as health management, diagnostics, prognostics or predictive maintenance.
Buckle up, because this change is accelerating. New jets are spitting out much more data and smarter ways of analysis are here. But all these riches will only benefit airline maintenance if industry segments work cohesively—and there are still big questions about who does what, and how.
Start with the airframe OEMs, which are orchestrating the expansion of health monitoring across the airplane. Mike Hurd, program manager for Airplane Health Solutions at Boeing, notes that each new aircraft not only increases data from more systems but also expands onboard processing. For example, the 737 MAX’s Onboard Network System detects and summarizes system faults for better troubleshooting by technicians and for ground applications.
Although carriers use a variety of systems for analyzing health data, Hurd sees a trend toward OEM systems as data become more plentiful and complex. Boeing’s Airplane Health Management (AHM) now receives inflight data via the Aircraft Communications Addressing and Reporting System (ACARS) when carriers direct it. “Many customers fully exploit advanced diagnostics and prognostics in AHM today.” He calls the benefits “substantial” and says AHM has been designed for multiple models, so airlines can monitor 737NGs, MAXs and other Boeing types with the same system.
With its AiRTHM service, Airbus experts have advised airlines on inflight problems and health trends on A380s and A350s. Now Airbus is embedding this knowledge in software airlines can use themselves. An Expert module enabling airlines to troubleshoot inflight problems should be available in late 2015 for A330s and A380s, and early in 2016 for A320s and A350s. For trend monitoring, Airbus will offer prognostics and risk-management software in late 2015 for A330s and A380s, and next year for other models.
The OEM also wants to make its predictive capabilities more powerful; it is working with EasyJet and has formed a partnership with IBM to aid this effort.
Bombardier is also taking a big leap forward in predictive monitoring. Q400s and CRJs generate little data; pilots must manually download information about problems using ACARS. The CSeries tracks thousands of parameters on major systems, including avionics, flight controls, fly-by-wire, landing gear, braking, environmental control, thrust reversers, engines, electrical systems and APUs; “almost everything on the airplane,” explains Todd Young, vice president of customer services. A maintenance computer automatically downloads reports during flight when any metric hits specified limits. Inflight reports are now made using ACARS, but may come in bigger broadband chunks when the cost permits.
CSeries jets store data for up to 100 flights, but generally download data on landing. This allows carriers to look at data trends to detect deterioration in components and perform proactive maintenance. And OEMs such as Pratt & Whitney can collect and convert downloaded data for analysis by Bombardier, the airline or others.
Bombardier offers conversion files, but Young says not all carriers have the IT systems needed to use them. The company is talking with customers about analysis options, and many react positively to using Bombardier’s analytics.
In CSeries test flights, Bombardier sees gains in reliability from using trend data for proactive maintenance. But Young says the biggest gains come from reduced downtime from failures, because troubleshooting now starts immediately from inflight reports.
Engine makers have been engaged in this effort for some time. Pratt & Whitney’s PurePower engines collect 5,000 parameters continuously, compared with snapshots of a 100 parameters from older engines. Analytics Manager Lynn Fraga says carriers choose how they want this data analyzed, but Pratt is well-suited to do it. In addition to using this data in engine design, Pratt is working with IBM on “Big Data” analytics.
Under any approach, an airline must facilitate access to engine data. “Since my specialty is data analytics, I prefer to get it inflight,” Fraga says. “But there are pros and cons each way. And we can start with what we have today.”
Downloaded data covers full flight profiles, including taxiing, climb, cruise and descent. Fraga says its analysis will increase time on wing, reduce disruptions, improve work scopes and improve the design of upgrades and future engines.
Snecma’s monitoring system will support Leap engines on A320neos and 737 MAXs, generating several megabytes of data per hour over the entire flight. Additional data will help ensure no issues occur in engine starting and in monitoring oil consumption, which is very important in ETOPS flights, says Pierre Schroer, general manager of services business development. Older engines mostly generate data samples during takeoff and cruise.
Schroer believes centralized tools developed by CFM, Snecma and GE can process and analyze the data best. He hopes airlines will download data automatically at gates and says carriers should also download snapshot data via ACARS inflight. Snecma is looking at broadband options for transfer of inflight data, possibly for future engines.
Airlines must make maintenance and flight operations more proactive to fully benefit from Snecma’s recommendations. The company can advise on optimizing MRO and improving engine work scopes, supply chains and fuel efficiency. Snecma has set up a Services Innovation Workshop to help airlines exploit engine data in many areas. The approach is being tested with Luxair, EasyJet and Royal Air Maroc.
GE’s GEnx generates 5,000 data points per second and 0.5 terabyte per flight. “The new engines have exponentially more data, more options in how to use it and what you can do with it,” notes Dave Bartlett, chief technology officer at GE aviation.
Another key is providing context, which Bartlett calls tagging. “What part of the flight [is it]? What are the operating characteristics of the flight? Who is flying it? How are they flying it? The more context, the more questions you can answer.”
To exploit Big Data from engines, GE has built its Predix platform. “It’s the first platform to handle this type, volume and velocity of data securely,” he says. GE has used Predix internally, launching it for external customers in September.
Predix collects and stores data, and it runs apps. A catalog of GE analytic apps will be available, and users can load their own apps. Different apps will work together on Predix. “It’s the industrial Internet,” Bartlett says. “You only pay for what you need when you need it.”
Major avionics OEMs are also active in data exploitation. Rockwell Collins makes information management systems that take output from central maintenance computers on 787s and A350s and determine how to transfer it to the ground, explains Information Strategy Director Joel Otto.
The company also builds avionics systems that generate data. Inflight diagnostics can sometimes spot interface problems that are harder to locate when an avionics box is sitting on the ground. Otto expects airlines to choose a mix of in-house, OEM and MRO analytics based on MRO philosophy, repair difficulty and the effect of reliability on operations.
For downlinking data, Rockwell has an ACARS network and resells GX Express broadband. It is testing SwiftBroadband for operational data with Hawaiian Airlines. Otto is confident of gains down the road: “We are just at the pointy end of the spear in harvesting Big Data.”
Honeywell makes central maintenance computers to interpret data, routers to collect and send it, plus many systems—such as APUs, brakes and avionics—that generate data. It has long experience with trend monitoring on 2,500 APUs.
Marketing Vice President Carl Esposito says Honeywell can combine data on mechanical and avionics components to draw richer insights. “We can correlate data; for example, the flight management system knows a lot about how aircraft are operated, and that helps analyze other components.”
Honeywell is now testing GX Ka-Band for operational data. Data storage should pose no problems. “Data storage is a problem solved long ago,” Esposito says.
Southwest Airlines expects to do much more predictive maintenance with the 737 MAX, according to Kent Horton, director of engineering. The MAX will generate orders of magnitude more data than the NG, providing data and fault reports from many systems, including enhanced ground-proximity warning systems, flight controls, flight management, GPS multimode receivers, APUs and engines. Sensor data flows to the onboard network system, which stores data and does simple processing, turning data into intelligence and recommended actions.
There are several options for downloading. One is ACARS, now used for NGs. Higher-volume MAX data can be downloaded at gates via Wi-Fi or cell networks. In the long-term, Southwest is looking at Ku-broadband downloads during flight.
Data analysis will initially be done by the airline and suppliers like Boeing and GE. “We still have the option to go to a third party for this service,” Horton says. He sees benefits in independent review of OEM analysis with “another set of eyes.” Southwest outsources most component maintenance under both flight-hour and time-and-material contracts. Even in flight-hour deals, there are logistics and service costs for failures. And maintenance recommendations must be integrated into overall maintenance plans, so interaction is needed between an airline and MRO advisers.
Horton expects to analyze trends in component performance and perform needed removals during scheduled maintenance, before failures, reducing service disruptions. Southwest has done this with valves and actuators on NGs. Doing it well takes time. “It’s usually 12-18 months before you hit full stride, and you still learn after that,” he says.
One challenge is managing internal processes for a mixed fleet of NGs and MAXs. “You want to standardize processes as much as possible,” Horton notes. Southwest may consider retrofitting NGs to make them “near capable” with the MAX for data monitoring.
Horton advises airlines with similar goals to remain agile, avoid rigid plans and be willing to change programs. “If you go with a third party, don’t get locked in, you need a flexible scope,” he says.
Air France Industries-KLM Engineering & Maintenance is moving into predictive maintenance, first with its A380s, next with customers’ A380s, then with 787s and other models, says Director of Innovation James Kornberg.
AFI-KLM E&M has developed and tested the techniques on Air France’s A380s for three years. It downloads data by Wi-Fi at gates and has IT specialists look for trends and develop algorithms to forecast failures. When a failure is predicted, the report goes to the maintenance control center.
The MRO is starting with the A380’s most critical systems. Kornberg says algorithms have forecast six out of seven failures so far. The company works with aircraft and component OEMs, but does not simply use their prognostic tools.
For other airlines, AFI-KLM E&M could analyze data and forecast problems under a component-support agreement or a pure advisory service. Kornberg believes the support approach is best.
He also plans to exploit 787 data from Air France-KLM’s operations, and sees experience with A380s aiding predictive maintenance on 787s. As for the A320neo and 737MAX, he says, “The principle is the same. Once you store the data properly, why not?”
The same principles apply to older aircraft but are harder to implement with much less data. Predictive maintenance is also harder for carriers with smaller fleets, since scale helps with both forecast accuracy and managing IT investment.
Broadband data inflight would help, Kornberg says. Cost is still a broadband issue, but storage of Big Data is not; AFI-KLM E&M stores its data in data lakes.