The Journey from Data to Predictive Maintenance

October 2019

AIN is developing a 14-part series of articles, infographics and videos that speak to Tomorrow's Aviation Challenges. Sponsored by Honeywell Aerospace. Learn more. .

HONEYWELL FORGE FOR AIRLINES: Connected Maintenance & Flight Efficiency

The rise of the Internet of Things and new analytical tools have given business and commercial aircraft operators the means to realize major benefits from the terabytes (TB) of data generated by their aircraft. While engine and airframe manufacturers have been installing various sensors in their products for decades, the few data points these sensors produced have traditionally been used for diagnostics. With today’s aircraft including thousands of sensors—the Airbus A350 has nearly 250,000 of them generating about 2.5 TB of data per day—sifting manually through all that information would be overwhelming.

But data-analysis technologies such as machine learning and artificial intelligence coupled with exceedingly fast processors and distributed processing networks provide the ability to find patterns across massive data fields. The insights provided by analyzing hundreds of flights of the same aircraft or an aggregate of thousands of flights across a fleet can result in reduced maintenance and operational costs.

In a presentation given at the 2018 Maintenance Cost Conference, Joost Groenenboom, principal at the consulting firm ICF International, estimated that digitization of aircraft and flight data could enable airlines to save in excess of $5 billion per year, including $3 billion from aircraft health management and predictive maintenance, $1.7 billion in fuel savings from flight optimization, and $800 million in delay reductions from improved operations processes.

Recently, Honeywell announced that its Honeywell Forge Flight Efficiency solution is expected to reduce fuel costs at Nippon Cargo Airlines and Kuwait Airways by up to 3 percent. According to Honeywell, fuel use can account for 20 to 40 percent of an airline’s operating costs, even single-digit percentage improvements can potentially save airlines tens of millions of dollars each year.

“Airlines today are trying to meet very tight margins while also improving operational efficiency, and Honeywell’s Flight Efficiency solution helps on both fronts,” said John Peterson, vice president and general manager, software and services at Honeywell Connected Enterprise, Aerospace. “We give operators the information they can use to develop, implement, sustain, and measure fuel-efficiency initiatives.”

Predictive maintenance uses advanced data analytics to process the aggregate data from an aircraft or fleet to predict when components will fail based on actual usage and fatigue data. For example, in 2017, data analysts used Honeywell Forge Connected Maintenance software to reduce auxiliary power unit (APU) failures at Hong Kong’s flagship airline Cathay Pacific. The airline was experiencing an average of 242 delays per year due to APU failures, costing an estimated $24 million annually.  

“Combining our fleet experience with [Honeywell’s] expertise managing APUs, it was the logical decision to work with Honeywell on this new predictive maintenance program,” said Philippe Christol, Cathay Pacific tech service manager—powerplants, in a Honeywell Aviation video. “The program has met our expectations…We’ve gone from a reactive to preventative maintenance program, so we’ve been able to eradicate defects before they even happen, and that has been absolutely essential for us.”

Honeywell Forge collected and analyzed a vast amount of data—more than 1.6 million flight records, 327,000 maintenance entries, 44,000 fault messages, and 13,000 APU reports—to create a predictive model capable of spotting parts failing on 61 of Cathay Pacific’s Airbus A330 aircraft up to three days before actual failure. As a result, the airline reduced delays and cancellations by 30 percent, lowered repair times by 90 percent, and experienced improved customer satisfaction.

Aircraft availability is a major issue for both commercial and business aviation operators, which is one reason most airplanes undergo a rigorous preventive-maintenance strategy. Aircraft’s go offline and perfectly good parts are being replaced well before they are estimated to go bad. And yet failures still occur and force aircraft operators into unscheduled maintenance, affecting operations and costs.

“Inoperative APUs affect Cathay Pacific in a number of ways,” said Christol. “[Pilots] may not be able to start the engine quickly, and it’s an issue for cabin comfort. It’s also affecting us from a cost point of view. Managing the defect, rectifying the defect, is also a very costly exercise for us.”

Switching from a preventive to predictive maintenance strategy can reduce the number of “premature” removals, unscheduled maintenance actions, and operational disruptions. Honeywell estimates that using predictive maintenance can reduce premature component removals by 10 to 15 percent while reducing operational disruptions by as much as 35 percent.

Honeywell Forge Connected Maintenance analyzes data from various areas on the aircraft to deliver diagnostic information as well as predictive, prescriptive alerts. While nearly every business or commercial aircraft flying has some Honeywell component on it, the aircraft does not need to be powered by Honeywell engines or avionics to take advantage of the Connected Maintenance analytics platform.

“All OEM components are covered; this is one of the strengths of the Honeywell offering,” said Josh Melin, Honeywell Aerospace product line director. “We have demonstrated success on non-Honeywell components, such as the [Airbus] A320 Environmental Control System, which is not a Honeywell system. As a component OEM with 100+ years of experience designing, manufacturing, and repairing mechanical and electrical components and systems from the nose to the tail of the aircraft, we are uniquely positioned to apply this deep domain expertise along with advanced analytics techniques to create unparalleled value for our customers.”

A nose-to-tail service, Honeywell’s Connected Maintenance solution can gather, synthesize, and analyze data from sensors monitoring navigation, air conditioning, engine operation, and exhaust, fuel storage and usage, pneumatics, hydraulic power, nitrogen generation, landing gear, and the auxiliary power unit to return predictive maintenance alerts and cognitive diagnostics. 

“Honeywell Forge Connected Maintenance doesn’t just show trends,” said Melin. “It gives specific alerts about impending failures and advises you on the best way to fix them. Our algorithms prevent false alarms by sifting through nuisance fault messages coming off your aircraft and alerting [the operator] only when there is a 98 percent probability that a failure is imminent.”

The platform uses wireless data transfer to seamlessly move packets of data, and software to and from the aircraft. In the past, the data generated by AHM systems resulted in files too large to download or analyze in flight. Data would be downloaded by maintenance crews after a flight and sent to a central data-processing center for analysis. Now broadband data services from the cockpit—such as through Honeywell’s JetWave in-flight Wi-Fi service partnered with Inmarsat’s Jet ConneX Ka-Band network—allow large data files to be downloaded and analyzed while the aircraft is still flying.

“The Honeywell Forge Connected Maintenance solution and operational data resides in Honeywell’s secure data warehouse, where it’s cleansed, sorted, and quality-checked to the highest industry standards,” said Melin. “The airline-controlled [data analytics] engine allows access for Honeywell apps, airline-developed apps, and third-party apps. The respective operations departments have controlled access to exactly the right data when they need it most, including access to advanced analytics algorithms, and automatic predictive and prescriptive maintenance notifications to prevent the need for unplanned maintenance. Additionally, customers can utilize our software to explore and analyze their data, plus create, host, and run their own analytics.”

CLICK HERE to Download Honeywell's ebook “From Data to Predictive Analytics” to learn more