Between aircraft generations, there has been an exponential increase in the volume of data automatically transmitted to the ground – be it in real time via radio or satellite data link, or through post-flight downloading at the airport gate.

While an Airbus A320 transmits about 15,000 parameters per flight, the figure is 250,000 for the A380 and 400,000 for the A350.

However, Lufthansa Technik's head of innovation management and product development Henning Jochmann argues that the wave of data is "useless" without targeted analysis.

"You can watch the sky for hours and perhaps see by accident a shooting star. But if you know when and from where to look up at a certain angle, you will see significantly more shooting stars," says Jochmann.

Across the industry, airlines and manufacturers anticipate that increased in-flight monitoring of aircraft systems will result in a shift from traditional, schedule-based maintenance to a more efficient practice of accurately predicting components' service life and replacing equipment in time before failures.

While conventional maintenance schedules tend to be based on conservative averaging of when parts might fail, continuous equipment monitoring can detect potential component performance degradation at an early stage and thus help to avoid disruptions, optimise maintenance planning, and allow greater exploitation of parts' service life.

Today, airlines employ reliability monitoring programmes to detect causes of technical disruptions, in order to find ways of reducing delays and AOG situations. EasyJet, for example, noticed in the past that windscreen cracking was a frequent cause for aircraft becoming unserviceable. The UK budget carrier then started replacing the cockpit windows before their mean time between failures and was able to reduce the number of AOGs, head of maintenance operations Aidan Kearney tells Flightglobal.

Maintenance cost for the windscreens have increased, but the overall effect has been beneficial, thanks to increased aircraft reliability and the avoidance of AOG-related expenses – be they incurred in transferring passengers to alternative flights or hotels, or in encountering unscheduled aircraft downtime.

Kearney says the airline's fleet technical department monitors system performance across all equipment categories and tweaks maintenance schedules to achieve an optimum balance between maximising part life and reducing technical disruptions.

However, he stresses that the scheme is based on retrospective analysis of fleet reliability – it does not give insight in aircraft systems' actual condition or provide prognostics for potential future failures.

Ryanair has implemented several predictive analysis tools – developed by US technology group Teledyne – to address specific technical issues on its Boeing 737 fleet. Using engine performance data routinely transmitted to manufacturer General Electric, Teledyne devised algorithms to alert Ryanair about performance changes in the aircraft's pneumatic system as some of that equipment had frequently caused disruptions, says the airline's operations chief Mick Hickey.

The alerts have enabled the airline to foresee impending component failures and replace the affected parts during regular overnight stops in order to avoid AOGs. Hickey says the analysis tool has been "very successful in terms of managing that [pneumatic system issues] and limited the number of unscheduled removals".

Engines have for decades been equipped with sensors to transmit to the ground key parameters for real-time performance monitoring. Parameters from different sections of the gas path – such as rotation speeds, temperatures, pressures, vibrations and oil contamination levels – not only convey conditions in the engine core but can indicate potential issues in associated systems.

UK manufacturer Rolls-Royce has acquired a bioscience company with a view to adapting analytical techniques developed for diagnosing medical issues through basic parameters – such as body temperature or assessment of blood samples – to improve monitoring of engines. "The analogies were so strong," says the head of the Rolls-Royce's operational service desk Matt Burns, "that the way it is being done in that field [healthcare] could equally be deployed in the way we look after our engines."

Over the last decade, Rolls-Royce reduced the number of engine-related operational disruptions by three-quarters through early detection of technical issues before they could develop into severe problems. The manufacturer argues that the accuracy and insight of its diagnostics is a result of its role as both the equipment's designer and being able to monitor the entire fleet's performance over time. That expertise, the engine maker asserts, is central to being able to detect and find solutions for potential issues that might otherwise not be noticed.

Today, a large amount of data is available about airframe systems and components, too, especially on new-generation types such as the 787 and Airbus A380. Air France-KLM's engineering arm has devised algorithms to alert its maintenance control centre about impending fuel-system failures on its A380 fleet.

This was the first area to which engineers turned their attention for the development of predictive maintenance processes, says Air France KLM Engineering & Maintenance director of innovation James Kornberg. But he adds that the MRO provider pursues projects for other components and aircraft too, including the 787. "Of course we concentrate our efforts on the systems which cause the most technical delays or operational issues," he says, noting that development of algorithms requires "a lot of expertise [and] time".

Lufthansa Technik's Jochmann agrees that the definition and construction of data filters is "clearly" the main challenge in the development of automatic monitoring systems for aircraft equipment.

"In order to correctly interpret the data coming from a system or component, one has to completely understand its purpose and mode of operation," he says. "Only then it is possible to define triggers in the data filters which can recognise and indicate the deterioration of parameters." This, he adds, is only possible with "a high degree of experience and deep technical know-how", both in operational and technological terms.

Rolls-Royce's Burns questions whether the industry now is at a "tipping point", where future development should perhaps be less about further data growth than about improving the analysis of existing information to create more insight. He suggests that bandwidth limitations – much of the engine data is transmitted via VHF radio signals – may in fact have contributed to the available data being used in the most efficient manner.

Airbus vice-president of engineering and maintenance services Philippe Gourdon says the manufacturer is in a "unique and very privileged position" to develop algorithms as its role as aircraft designer provides an immediate view of how the equipment should function and what impact performance issues may have on the overall system. But he suggests that the data analysis demands may go beyond capabilities of aircraft design engineers.

Gourdon says Airbus is employing mathematicians who can define complex algorithms that involve multitudes of different parameters – but who might not be experts in aircraft design. "For me," he says, "the biggest challenge is to be able to bridge these two activities... capitalising on our [design] knowledge of the aircraft with more sophisticated analysis relying on larger amounts of data."

New-generation aircraft may provide huge amounts of data that allow detailed analysis of their systems performance, but the vast majority of today's fleet does not supply such streams – and this will likely remain the case, as the systems architecture of the re-engined 737 Max and A320neo families will not be changed from their predecessor models.

Individual components and subsystems may be monitored, as in Ryanair's case. Lufthansa Technik has developed surveillance algorithms for certain components, such as angle-of-attack sensors and main landing gear door actuators on A320s. Air France Industries KLM Engineering & Maintenance is pursuing projects to employ predictive maintenance techniques on legacy aircraft, including 747s, says Kornberg.

More data is available on existing aircraft than is being used for predictive maintenance analyses, Gourdon asserts. But he says the challenge is to centrally access all available information. While systems of new-generation types like the 787 and A380 are interconnected through an integrated computer network– with information circulating throughout the aircraft and available for extraction at various points – the architecture of the 737 and A320 comprises distinct, separate subsystems.

Airbus is working on a narrowbody modification aimed at collecting available information, Gourdon indicates. The equipment is derived from computers used to process information during flight tests and can "drastically increase" the amount of data that could be employed for predictive maintenance purposes, he says.

EasyJet is among the airlines that are co-operating with Airbus on that project. Kearney says the aim is to implement a "very basic" initial system in 2016 that can monitor components in certain equipment categories, such as flight controls, landing gear or the pneumatic system: "If we can find some gains in those chapters initially, then you can build on it and add [functionalities] to it as time goes on."

If equipment monitoring and failure prognostics were to be rolled out across other aircraft systems, Kearney argues, it would be a "game changer". He adds: "Where we are now… It is very much reactive. We are waiting for a crew to give us a tech-log entry on an issue, we are waiting for a scheduled task to find a defect, or we are waiting for a failure." But in future, he says, "we want to be able to say to an engineer: in 72 hours, we predict this component to fail, we recommend replacement."

This would allow affected aircraft to be directed to hangar locations where maintenance staff and spare parts are in place to resolve issues. Kornberg sees prognostic tools making maintenance operations much more efficient as tasks can be targeted to resolve specific issues rather than to search for potential defects.

In-flight performance data will not be the single one source for evaluations, says Lynn Fraga, analytics manager at Pratt & Whitney's engine services division. Maintenance data will also play a central role. "It's not going to be enough that we can see data [showing] x percent of turbine blades [being] scrapped," she says. "I want to know how much [margin] did those blades fail by against the limits, but I also want to know: by much did they pass?"

That data can be used to improve component design on future equipment. But Fraga says the analyses will also put into perspective how maintenance efforts relate to equipment deterioration and, ultimately, how engines perform on wing. That information, in turn, can be used to tailor maintenance service to customer requirements. "You can do limited workscopes that allow you to best run your business based on your operation," says Pratt & Whitney engines services vice-president Eva Azoulay.

"Each airline is different," she adds. "This is the key in this – the more we learn, the better are our algorithms."

Azoulay argues that engine maintenance can be tailored around airlines' specific circumstances, such as environmental severity or particular operational requirements like, for example, airports where pilots need to select full rather than de-rated thrust for take-off. "When we talk about analytics, it is not just about trend monitoring and health monitoring," she says. "It's the integration of all the various data sources."

There is disagreement over whether or not analysis tools need to be customised to operators' individual circumstances. Some manufacturers say data algorithms for their equipment will be standardised, but that conclusions and advice drawn from the analysis will take into account airlines' distinct requirements. However, GE executive director of fleet support Vijayant Singh foresees analysis tools being customised for different operators: "I think every engine and every customer needs to have their own predictive modelling."

This, he adds, will increase the effectiveness of analytics and reduce the number of alerts triggering maintenance action, while equipment might still be in serviceable condition for airlines in question.

For Singh, the validity and accuracy of maintenance prognostics depends on a "marriage" between a "physics-based model" of the processes in an engine and a "statistics-based model" of how technical parameters translate into reliability and time on wing. He argues that devising algorithms to take into account airline's individual operating conditions will be a crucial challenge in the development of predictive maintenance processes. "We are not there yet," Singh says. "It will be an evolution that will be part of our normal operating rhythm."

The expansion of predictive maintenance shows that collection and analysis of operational data is becoming central to fleet technical management. Manufacturers are in a privileged position in this race, as they have visibility on entire fleets of aircraft and engine types over long time periods, plus control over the equipment's engineering data.

In the view of Azoulay, third-party MRO providers are not being disadvantaged as they can have access to customer airlines' aircraft health monitoring data and use records from their own maintenance operations for analyses to improve services. She argues that staying competitive in this field is about investment and building strategic partnerships: "We are investing in technology; we are investing in the analyses; we are partnering with universities and third parties like IBM that have [relevant] capabilities. There is nothing that prevents a third-party [MRO specialist] to do that as well."

Employment of predictive maintenance and increased service customisation can lead to greater fleet technical management complexity, Airbus's Gourdon acknowledges. But even if some airlines stick with convention schedule-based maintenance, others with extensive engineering and IT expertise may embrace the increased liberty.

Source: Cirium Dashboard