Periodic maintenance of aircraft can be made much cheaper by using sensors and artificial intelligence. This is stated in the European research project ReMAP, by TU Delft and KLM, among others. Instead of at fixed times, aircraft are only serviced if the condition of the parts gives rise to it.
Aircraft maintenance now takes place according to fixed schedules: An aircraft enters the hangar for maintenance and is not available for the airline to fly at that time. Difficult.
A fixed maintenance plan also means that a part is not always replaced when it is actually needed, but faster. So good parts are constantly being thrown out, which is a cost item.
For safety reasons, airlines obviously need to be conservative about how long they let parts last. “But the quality of the parts is getting better and better, so we’re leaving room there,” said Paul Chun, Vice President Technology Hub KLM Engineering & Maintenance last Friday during a tour for journalists through KLM’s maintenance hangars.
In the four-year European research project ReMAP, thirteen groups led by TU Delft researched a new approach to aircraft maintenance: condition-based maintenance. The results were presented yesterday at a symposium in Delft.
‘We have succeeded in modeling the entire maintenance process of different aircraft fleets,’ says TU Delft ReMAP project manager Bruno Santos in a press release.
“In the future, it will make it possible to transform current aircraft maintenance based on fixed time intervals (and maintenance due to defects) into ongoing health monitoring of systems. Parts are replaced exactly as needed, which prevents spillage. ‘
In the future, the construction of an aircraft will be full of tiny sensors that continuously measure things and send this data to a central point. It analyzes software based on the artificial intelligence (AI) data and tries to detect trends in them that indicate concrete damage or wear on a part. Or material fatigue, a problem that turns out to be much slower.
As a first step in the ReMAP project, an open IT platform has been developed under the leadership of project partner ATOS, on which AI developers can test their forecasting or planning algorithms based on actual airline operational data. This open approach promotes the development of innovative solutions from third parties.
Airlines do not want to share data
Particular emphasis was placed on the confidentiality of data from different airlines. ‘On the one hand, you need to combine data from different airlines, otherwise the AI can not learn fast enough. But airlines do not want to share their data with each other because too much information that is sensitive to competition can be inferred, ‘said Floris Freeman, Research Lead Condition-Based Maintenance at KLM.
The engineers in the ReMAP project solved this by keeping data from different airlines on their own server and only sharing the AI models on the IT platform. That way, the AI can be better trained, while the airlines do not have to be afraid that competitors may sneak into their data.
What should this research lead to? The dot on the horizon is that in a few years, all airlines will probably operate a system that monitors the state of the entire fleet, a Integrated Fleet Health Management (IFHM) system. Only when this indicates that it is necessary does a plane enter the hangar for maintenance.
It is expected that thanks to this predictive maintenance (“predictive maintenance”) can save a lot of money. The participants in ReMAP aim for a common saving in European aviation of 700 million euros per year.
In addition, passengers will also benefit because the new maintenance scheme will ensure better accessibility of aircraft; passengers suffer less from delays or cancellations due to an unexpected flight error.
If you found this article interesting, you can subscribe to our weekly newsletter for free.