Smarter maintenance planning with AI

How AI is revolutionising life for engineers and planners

What started as a graduation project has grown into a real AI tool that makes life easier for many KLM employees. Thanks to LiMiTS, aircraft maintenance at the gate now runs much more efficiently which is fantastic for our ground engineers and planners alike. Colleagues Pim and Daan explain how the idea for the tool came about and what problems it solves.

Coming and going

‘Every aircraft arriving at Schiphol gets an inspection and maintenance check at the gate,’ says Daan, product owner in the Big Data team at Engineering & Maintenance - KLM's maintenance branch. 'Customs first check that no prohibited goods have been smuggled in. Then the ground engineers of the Line Maintenance team carry out minor checks and repairs, from software updates to replacing brake pads.

‘Some aircraft come through Schiphol several times a day.’ explains analytics translator Pim who works in the same team as Daan. 'On average, the engineers handle approximately 300 visits per day. That's a significant number, don't you think?'

Time-consuming job

‘Each aircraft has its own to-do list,’ Daan continues. ‘The planning was often tight because although there are sixty of the engineers during the day, not everyone is allowed to take every task, specific skills and licences are involved. In addition, the team on the platform covers considerable distances from one aircraft to another, sometimes they can be on the road for up to half an hour. All these aspects made planning difficult.'

From concept to product

‘We'd been trying to solve this problem for some time,’ says Daan.
'That started with two graduation studies. After this, we knew which direction to take, or in other words, how to approach the issue in terms of IT, data and mathematics. With that information, we were able to work out a proof of concept - this is how you test the feasibility of an idea such as this.

It was a success: with all signals green, we had the go-ahead to further develop the tool. We did this with our Big Data team and in close cooperation with the planners. Eventually, you end up with a minimal viable product: the first variant we actually use.

Predicting flights

Daan explains: 'As KLM increasingly uses real-time data to determine fleet deployment, manual planning is becoming increasingly difficult.

Just think about it, if you've just made a maintenance planning, but aircraft are suddenly assigned to other flights, you have to start all over again. After all, maintenance is not tied to a flight schedule, but to a specific aircraft.'

According to Pim, LiMiTS has solved that problem. 'It allows schedules to be made much shorter in advance. Not only that, it also makes the feasibility of the planning transparent. In other words: are all the required materials present, and does the quantity and type of work match the scheduled engineers?'

Better for everyone

Since LiMiTS went live in June 2024, the initial response has been very positive. The work for Line Maintenance is much better distributed, meaning the ground engineers suffer less from the peak moments. On the other hand, the planners have time to spare. They can now spend this time on difficult jobs that the computers can’t handle as easily. By organising both processes more efficiently, aircraft are available faster and we have fewer delays.

You won't find this anywhere else

Pim finds the most rewarding aspect of working at KLM is being right in the heart of the operation— literally working among the planes. It makes the work feel incredibly real and hands-on. Plus, by working with different departments, you gain a deeper insight into the company as a whole. 'Daan: ‘The fact that KLM has been around for so long also brings challenges. Some of the legacy systems: some are decades old. It can be quite complicated applying new AI techniques to the existing systems, but for fans of complex systems, there's plenty to enjoy here.'