Data: the cornerstone of aerospace maintenance
Data: the cornerstone of aerospace maintenance
Interview with Bart Claus, Business Development Director at Air France Industries KLM Engineering & Maintenance, the MRO branch of the Air France KLM Group.
Big Data and the evolution of the aerospace sector: is the subject mature or still in its infancy in this sector and its value chain?
It is neither in its infancy nor yet fully mature: in our vision, the initial exploratory phase, without immediate value generation, lies behind us. Data and data analysis subjects are now producing extremely concrete results, particularly for predictive maintenance. But data is still bringing us technological opportunities that we think are very important. For us, maintenance activities are starting to generate the most value: predictive maintenance, optimising scheduled maintenance, optimisation across the supply chain, directly linked to access to the results of data mining. The airlines, our primary customers who benefit from our services, are seeing immediate improvements to their maintenance operations and in their operational performance.
A bit of background: where did the need to make use of data come from and what has moved those operating in the sector forward?
Traditionally, data has always been used for two purposes: flight safety (flight data analysis is a mandatory regulatory activity) and technical monitoring of aircraft, which has developed considerably for engines, but also in certain targeted parts of the aircraft.
“Over recent years it’s been the combination of more access to data, faster computing infrastructure and better data skills which is driving value."
The revolution is that aircraft have started to record more and more data, and to communicate more and more. As a result, the maturity of companies in the aerospace sector has grown and their ability to make use of that data has increased. At the same time, infrastructure and data science skills have both improved dramatically. Over recent years it’s been the combination of more access to data, faster computing infrastructure and better data skills which is driving value. The difference then lies in the ability to combine this data with knowledge of both our maintenance operations and the aircraft itself and its systems. In fact, we systematically set up teams that combine people with knowhow of the aircraft system – traditional maintenance engineering – with others who wear a data science hat. They constantly work hand in hand to combine a systems vision with a data vision, but also keeping the impact on operations in view, which allows the appropriate prediction algorithms to be set up.
“Firstly, we reduce the impact on operations, while a second beneficial effect of this reduction in breakdowns is that we improve aircraft availability, since they are not taken out of service unpredictably.”
What do you think are the levers of opportunity which have emerged as a direct result of the use of data in aerospace?
1 / Dedication to performance
As an airline, the first lever for savings is still the search for operational performance; we might as well start talking about predictive maintenance right now, because it is a perfect example. With our Prognos solution, using data from aircraft and organising maintenance activities now give us fewer technical delays, a reduced number of technical tolerances, and allows maintenance units to work on aircraft even before equipment breaks down. We are now tracking several hundred aircraft. We are experiencing a significant drop off in technical impact. We can anticipate a technical failure and thus reduce the volume of breakdowns by 50 to 70%. So firstly, we reduce the impact on operations, while a second beneficial effect of this reduction in breakdowns is that we improve aircraft availability, since they are not grounded unpredictably. The main beneficiaries, and this is the key point, are the passengers.
2 / The direct impact on stocks of spares and the supply chain
Being able to predict when you will need a part makes it possible to optimise your spare parts inventory and to keep them in fewer different locations. In mechanical terms, reducing inventory has an immediate financial impact. Our systems enable us to anticipate the majority of aircraft breakdowns, so parts can be replaced on an anticipated basis, without disrupting the supply chain.
3 / Repair instead of replace!
Predictive maintenance allows maintenance work to be carried out in time and encourages repairs rather than replacement and purchasing new parts.
What about the AOG (Aircraft On Ground) impact?
Although there is certainly a benefit, it is difficult to quantify the savings. An AOG means a whole compensatory arrangement, making alternative provision for passengers, with a significant cost and a negative impact on customer perceptions.
“Thanks to maintenance, we encourage repair where possible, rather than replacement and purchasing new parts."
However, all too often we approach predictive maintenance topics from an economic perspective. But when we meet with an airline, we also discuss issues of flight safety, operational performance, aircraft availability, avoiding disruption to operations. It is the combination of all these elements which guarantees a real result and gives the service value.
Tell us about the profiles of your teams: are they mainly made up of data scientists? Are they difficult to recruit?
The major investment is still in human resources: we are providing a service involving IT design and data science. So we have recruited many data scientists, some directly into the engineering and maintenance teams. Historically, in the Air France - KLM group, this type of profile already existed: we brought them together, as well as training up a number of complementary traditional engineering profiles. It was never the idea to build an ivory tower for data scientists. They work very closely with our engineers, and they wear two hats, as experts in both data science and traditional airline engineering. Our objective is to position them as close as possible to practical needs on one side and knowledge on the other. That calls for a certain agility, because you have to apply slightly different management methods: ultra-agile development, which directly empowers small teams and others.
“Aerospace has the benefit of standing out in the business world, in tandem with an enthusiastic environment. It is a real powerhouse in terms of attraction for these human resources who, above and beyond their trade of data science, are fascinated and engaged."
As for their attractiveness, it’s true that data scientists have the attention of the market at the moment. These are gold-plated profiles. But aerospace has the benefit of standing out in the business world, in tandem with an enthusiastic environment. It is a real powerhouse in terms of attraction for these human resources who, above and beyond their trade of data science, are fascinated and engaged. It's up to us to give them the challenges they need!
Do you see new players coming into the market as the data boom continues?
We routinely work with a network of innovation and "pure players" (often start-ups), particularly concerning technology and innovation. This works well because we combine their knowledge with ours by providing context, understanding needs and impacts. I also think that by making it easier to access data – a theme taken up by IATA – we could see even more players and solutions emerging.
What is the data for predictive maintenance like?
As far as big data is concerned in aerospace, flight data is not enough; it has to be tied in with aircraft maintenance data, and even with data gathered from equipment on the test bench. It is combining this data that allows us to set up services: we cannot have predictive maintenance without any failure data, for instance. We are also beginning to cross-reference this data with external sources: weather, volcanic dust, external events. This is where the greatest potential lies: succeeding in combining data and contextualising it for each client company. The real challenge is to switch from a tightly statistical approach to one which is more targeted, and to replace traditional preventive maintenance policies with individually targeted policies.
Which systems contribute the most data and how is that data collected?
Engines represent the highest economic cost for an aircraft: they have the longest history of data use. The systems we operate today are targeted to have an operational impact on the company's business. To make predictions, we will be using huge quantities of data which will be retrieved on the ground. Recent aircraft are equipped with communication systems that allow data to be downloaded to our servers as soon as the aircraft has landed and cut its engines. Some aircraft from older generations have systems without the communication facility. However, data from the flight recorders, already analysed for flight safety purposes, can be used. Some equipment, for example, is fitted with PCMCIA cards which are obsolete: as aircraft have a lifespan of 25 to 30 years, technology can pass them by.
In your opinion, are new risk factors emerging from the uses of data?
Obviously this is a matter of cyber risks, with the caveat that the type of data we handle is not really subject to these risks. Our use of the data takes place downstream. Certainly, the data we handle can be economically important at the end of the chain because it has value and must be protected. For that reason, we set up the appropriate infrastructure, which we constantly test and strengthen, just as any digital business that deals with data does. I think that this is a risk which is inherent to the use of data in the broadest sense. The more we expand the use and storage of data (previously we didn't retain all the data, but now we keep far more), the greater the risk of malicious misuse of data. One possibly underappreciated risk is the creation of a dominant position. The emergence of proprietary data formats is now blocking some initiatives for making use of data.
Do you think that using data could help reduce aerospace’s carbon footprint?
For flight operations, using optimisation solutions based on big data can reduce fuel consumption by between 2 and 5%. Predictive maintenance also encourages repair rather than replacement and the purchasing of new parts, which often need to be carried from the other side of the world.
What is certain is the air transport industry’s general commitment to sustainable development objectives. Air France and KLM in particular made several major announcements in 2019, including the target of a 50% reduction in our CO2 emissions by 2030. We are also taking concrete measures, such as eco-piloting, carbon offsetting on domestic flights, on-board sorting of waste, avoiding single-use plastics and reducing our noise footprint. Prognos, our predictive maintenance solution, was among the very first products in the world to receive the Solar Impulse Foundation’s “Efficient Solution” label, which promotes environmentally beneficial solutions.
“What is certain is the air transport industry’s general commitment to sustainable development objectives.”
What innovations are still to come in this area?
Artificial intelligence, which increases the capacity to use data, is now starting to bring results. In the field of maintenance, many new technologies are emerging: the emergence of drones, robots, and more, which are beginning to be used in aerospace maintenance to automatically classify defects, perform inspections, and so on. Likewise, augmented reality is beginning to be used. But there is one factor common to all of these: the need for big data! There is no augmented reality without big data, no automated use of fault data from a drone-mounted camera without using data analytics.
And all this also relies on a competitive environment: proper competition remains a key asset for a real innovation dynamic.