Predictive maintenance is a hot topic in the aerospace sector. Long before the Internet of Things became a subject of conversation, engine manufacturers were using data from their products to analyse them. Since then, engine monitoring has spread to most aircraft components and been redefined into health management, diagnostics, prognostics, or predictive maintenance.
Airlines, OEMs, and maintenance companies are investing in developing predictive maintenance capabilities. This is largely focused on applying advanced analytics tools to large aircraft and maintenance execution datasets to better predict component failures to reduce unscheduled maintenance, resulting in airline operational disruptions.
Although today the sector can perform predictive maintenance much more effectively because new aircraft yield more data, and the tools for exploiting it have developed, ways of thinking and business processes have yet to catch up. “As predictive capabilities mature, the cultural and organisational processes become a limiting factor,” says John Maggiore, director of fleet and maintenance solutions at Boeing Digital Aviation. “Organisations need to move from a reactive to a proactive culture when it comes to maintenance.”
Andrew Starr, professor of maintenance systems at Cranfield University and head of the Through-life Engineering Services Institute, sums up the problem. “The aircraft industry is highly regulated, and highly conservative,” he says. “If we look at more traditional sectors, such as power, they have been confidently condition monitoring as an input to health management for 50 years.
“In many ways we are pleased the aerospace industry is conservative, but when it comes to maintenance it means some of it is predefined, and it takes a strong argument to change that method of maintenance even if it’s not the best way to do it today. This tends to lead to over maintenance and it might include disassembly, which can misalign the machines and components and introduce dirt into assemblies, and the only reason for the disassembly was to check it. If you have a way to determine health then that would avoid that disassembly.”
Provocative maintenance, as it is sometimes aptly called, can also be time consuming, inefficient and costly. EasyJet, which is looking to enhance its predictive capabilities, recognises that one of the biggest problems that airlines want to avoid is too many false alerts, making it difficult to spot real problems.
Aidan Kearney, head of maintenance operations at EasyJet, says: “We are still in the very early stages of predictive maintenance. What that means for us is that, when we are issuing troubleshooting to our line stations, we are relying on them to carry out exactly what the prediction tool is telling us, giving us feedback so we can remove the components and send them to an overhaul facility. Then the vendors come in, and have a big part in it because we may be removing components earlier than anticipated.
“We obviously would not like the component to go out and have a ‘no fault found’, so what we’re trying to do – partly with Airbus – is to identify what the failure cause will be, so we can tell the vendor and guide them towards a specific troubleshooting that they may need to carry out and test.”
With its realtime health monitoring system, Airbus has advised airlines on inflight problems and health trends on the A380 and A350 XWB models. Now Airbus is embedding this knowledge in software that airlines can use themselves. Airbus wants to make its predictive capabilities more powerful, and is working with EasyJet to do this.
“We’ve been working with Airbus for the reason that we’ve got a large fleet – we’ve got 267 aircraft,” says Kearney. “We have a lot of operational data on the aircraft and we’re doing nearly 1,600 flights a day. The operation of that is quite significant. Airbus have all the technical details – they have the build-up of all the warnings and the engineering background behind every component that we are trying to identify.
“The partnership is going very well. We’re not talking about huge teams of people. We’ve a very small working group, we all know each other on a name by name basis. We have regular meetings with Airbus, we have weekly conference calls and we also meet them in Toulouse, and they come to Luton. We all know what we’re working towards, and everyone is very keen to come up with a solution.”
Recently, Rolls-Royce and Microsoft announced that they are collaborating to harness the power of digital technology to transform the aerospace industry. They want to bring together solutions for engineering and cloud computing, including advanced analytics and the Internet of Things, to offer a realtime predictive maintenance service to airlines.
Rolls-Royce has more than 13,000 engines in service on commercial aircraft, and for 20 years it has offered customers engine maintenance services that help keep aircraft available and efficient. Rolls-Royce is using the Microsoft Azure platform to transform how it uses data to better serve its customers.
Nigel Jackson, head of the analytics lab at Rolls-Royce, says: “Our collaboration with Microsoft enables Rolls-Royce to concentrate on generating actionable insight without being distracted by the complexities of running a high-availability computer platform. It’s a case of both parties playing to their strengths.
“Microsoft has a strong software product development drive that enables Rolls-Royce to do more with the data it receives, and through rapid service content development we can attract more data through the value it generates for our customers. This in turn feeds service content innovation to generate even more value.”
Harnessing and leveraging big data has become one of the most heavily discussed topics within the industry. Indeed, lack of data is not usually a problem. Carriers already use this data for flight monitoring, as required by regulations, and increasingly for predictive maintenance. However, while OEMs, maintenance firms and operators all have aspects of this data, not all the information has the same relevance.
For example, new airlines may capture hundreds of megabytes of data about workings on a single flight, but operators want help in sifting it. This is a matter of deciding where to focus.
Charles DiSano, director of condition-based maintenance services at Honeywell Aerospace, says: “Gigabytes and terabytes of data are great, potentially holding insights to many unknown or unforeseen problems. This type of data enables broad spectrum analytics, machine learning and other big data techniques. However, we’re not able to make business decisions based on potential alone. We need to see tangible results that hit the bottom line.
“Predictive maintenance data, and maintenance data in general, is prioritised based on need. Some data you want in real time and you’re willing to pay to transmit that via satellite communications. Other data is interesting, but you’re willing to delay the transmittal until you have a lower-cost connection.”
Jackson says: “Modern aircraft are more data-enabled than older aircraft, with an increased number of sensors, higher-rate data capture, larger bulk data storage and higher bandwidth communication with ground-based systems. With such a fat pipeline of data becoming available, the challenge comes in the scale and pace of analysis needed to reliably detect issues when the fundamental equipment designs are inherently more reliable than older aircraft. This affects the content of the services that can be offered.”
Starr adds that today’s problem is the flood of data, and that the data storage is still not free. “Monitoring data fills space in a server or cloud,” he says. “The cloud takes some of your cash to run a data centre – and at that point the companies want to get revenue from that data to justify its existence.
“I’d ask all the owners of that data whether they are brave enough to delete it? The going unofficial answer is that none of them are prepared to do this. So they’ve got to pay for the storage, and then it has to be worth it.”
Maggiore at Boeing Digital Aviation says that the capabilities of predictive maintenance vary from aircraft to aircraft: “Advanced analytics and optimisation have the potential to transform the aviation maintenance domain. The challenge in determining the opportunities involve evaluating each aircraft type and how systems and data collection architectures enable predictive maintenance analytics. The 777X, for example, is much more enabled for predictive maintenance analytics than a regional jet designed in the 1990s.”
While new aircraft may be like flying servers, Jackson at Rolls-Royce argues that predictive maintenance plays a part across the industry. “The aircraft industry is mature and delivers ever increasing performance, lower environmental impact and lower operating costs with every generation of aircraft design,” he says. “Making a good product great is often not enough, though, and being able to constantly innovate to drive lower costs and higher reliability through an aircraft’s service life inevitably leads to the need to provide data and value-adding data services to customers.
“The regional aircraft market is highly competitive, and this is a way to strengthen customer uptake through the promise of lower operating costs through more explicit knowledge of the aircraft condition and more efficient management of maintenance.”
One of the biggest leaps the industry needs to take is recognising the value of data sharing. The industry does not have a culture of sharing data, but the best predictions are based on multiple-airline data. “The final catalyst will be a recognition of the value of data sharing,” says Honeywell’s DiSano. “Predictive maintenance for a product, system, or entire platform will truly flourish in an environment that shares relevant data. Then we’ll see automated solutions from automated data streams enabled by readily available data sources. It’s really exciting to consider the potential.”