Brussels Airport Company
As part of the Airport Operations Plan (AOP), Superlinear’s AI-powered solutions superpower airport operations managers with accurate forecasts for critical processes in the airport. In turn, the airport operations managers communicate these forecasts back to all BAC's partners to ensure the right workforce capacity, maintaining a smooth passenger experience while keeping costs efficient.
Executive summary
Context
BAC faced significant challenges amid changes caused by climate change, COVID-19, and evolving customer needs. BAC launched its Shift 2027 strategy, focusing on three pillars: performing better as a hub, diversifying activities, and embedding sustainability. Implementing AI to reach these pillars is a core part of their strategy.
Goal of the project
BAC aims to evolve to smart connectivity, thinking and acting sustainably and customer-centrically. They partnered with Superlinear and joined the Airport Operations Plan (AOP), co-funded by the European Union, to make airports more efficient and data-driven, focusing on increasing passenger experience and revenue streams.
Solution
Superlinear is building AI-powered solutions on top of BAC's vast data and focusing on five use cases: Check-in, Screening for Departing Passengers, Aid for People with Reduced Mobility (PRMs), Baggage Handling, and Border Control.
Using AI to forecast passenger and luggage volumes, Superlinear helps BAC’s operation managers communicate with its partners to ensure the right workforce capacity, to avoid wait times and increase customer experience. As operations become smoother, passengers' frustrations will be removed, and traveling will become easier.
Case study
The challenge
Staying relevant in a changing aviation industry
Brussels Airport Company (BAC) faced a significant challenge: to stay relevant in the aviation industry amid changes caused by climate change, COVID-19, and evolving customer needs.
Shift 2027 Strategy: focusing on strengths and new tech solutions
BAC responded to this challenge by launching its Shift 2027 strategy in 2022, focusing on its strengths and new tech solutions to create prosperity and well-being.
Three pillars: Performing Better as a Hub, Diversifying Activities, and Embedding Sustainability
As part of the Shift 2027 strategy, BAC aims to evolve to smart connectivity, thinking and acting in a sustainable and customer-centric way. To achieve this, BAC has identified three pillars:
Performing better as a hub, which means enhancing worldwide connectivity as an airport, improving punctuality, efficiency, and reliability. BAC is committed to increasing customer experience for both passengers and cargo customers.
Diversifying its activities, which involves looking for new opportunities and further developing cargo logistics. BAC is exploring how it can maximize the potential of cargo operations to improve revenue streams and customer satisfaction.
Embedding sustainability, which is central to BAC's vision. BAC recognizes that the aviation industry must embrace sustainability to have a future, and BAC is committed to playing a pioneer role in this transition. BAC plans to use new technology like biofuel and AI, as well as improve energy efficiency, to achieve sustainability.
Partnership with Superlinear and Airport Operations Plan
To achieve its goals, BAC has partnered with companies like Superlinear and joined the Airport Operations Plan (AOP), a project the EU co-funded to make airports more efficient and data-driven.
BAC is responsible for setting its own strategy, and it has initiated two key projects: building an Airport Operational Center (APOC) and developing AI-powered solutions on top of its vast data through AOP. With many partners working together in the airport environment, AOP also encourages Collaborative Decision Making, which is essential to support all parties in working smoothly together.
Focusing on strengths, embracing new technology, and collaborating for smart and sustainable connectivity
BAC's desire to stay relevant and set a new benchmark for the aviation industry is at the heart of this challenge. To achieve this, BAC is committed to focusing on its strengths, embracing new technology, and collaborating with partners to achieve smart and sustainable connectivity.
So how is Superlinear helping make this concrete? Read on to find out more.
The briefing
Building AI-powered solutions for operational efficiency at Brussels Airport
Operational efficiency is crucial when it comes to air travel. Ensuring a seamless experience for passengers can be challenging with numerous stakeholders involved in the process. That's where Superlinear comes in - by building AI-powered solutions, we help to optimize operations and enhance the customer experience. This also aligns with Brussels Airport's sustainability goals.
A history of successful collaboration
We have been working with Brussels Airport since 2017, and in 2019, they were one of the first Belgian companies to showcase how they used AI to increase customer experience. Over the years, we have helped them to perform even better as a hub.
The power of data
One of the reasons for our success at Brussels Airport is their rich collection of data. They refer to it as the "new heart of the airport" and have been collecting it for years. The data comes from three streams, and we help to build AI-powered solutions based on that information.
© Brussels Airport Company
The Airport Operations Plan (AOP)
Brussels Airport's Airport Operations Plan (AOP) is a new way of working, focusing on collaborative decision making with all stakeholders, supported by data and predictive computer models. Algorithms and data create forecasts for the (near) future. With these forecasting models, the airport can become more proactive in addressing potential issues before they arise.
“Up to now, we just inform the passengers on the current waiting time at several processes in the airport. Soon we will be able to predict hours in advance the exact waiting time each individual passenger will experience at each step in his/her journey through the airport. Passengers will be reassured that they are right on track to catch their flight. At the same time, our operational partners will have the opportunity to act early-on when longer waiting times are to be expected. A win-win for everybody involved”.
Raphaël Peschi, Machine Learning Team Lead at Superlinear
Collaborating on AOP 2.0
Superlinear has been working on AOP since 2018-2019 (AOP 1.0), and last year we won the public tender for data science for AOP 2.0. Currently, we are working on five use cases to ensure a smooth experience for passengers while remaining cost-efficient.
© Brussels Airport Company
Five use cases across the main three data streams:
Check-in
Screening for departing passengers
Aid for People with Reduced Mobility (PRMs)
Baggage in general and at check-in in particular
Border control (three of the five border control locations in the airport)
We use AI to forecast the amount of passengers and luggage pieces expected at different processes in the airport in advance. This information is critical for the airport to communicate with its partners to ensure the right workforce capacity.
Creating a standardized way of working at BAC
We are also creating a standardized way of working at the airport to bring models to production. The project is forecasted to last two years.
The following section will present three of the five use cases we are currently working on.
The work
AOP use cases: Prioritization and business involvement
In the context of Airport Operations (AOP), BAC has identified five use cases to focus on, which were chosen based on operational costs and feasibility. The goal is to provide maximum return on investment within a reasonable timeframe. To achieve this, BAC worked in sprints, involving business stakeholders throughout the process.
Superlinear supported BAC in making these strategic choices and prioritising use cases. As an AI company, we aim to help our clients’ management in decision-making, not just executing projects.
Managing multiple parties at BAC
BAC has various companies involved in getting a plane in the air. Each Business Line (a passenger process) has a designated "Business Line Owner" responsible for managing the relationship with all partners involved in that Business Line. These Owners provide information on expected passenger numbers and necessary staff to their respective companies. With AI forecasting tools, the Business Line Owners can now offer more accurate data to BAC's partners.
Using AI to forecast passenger and luggage volume
One of the critical uses of AI in AOP is to forecast the number of passengers and luggage expected at different points in the airport in advance. This information allows BAC to communicate with its partners to ensure the right workforce capacity, maintaining a smooth passenger experience while keeping costs efficient. The objective is to have just the right amount of capacity at each point in time: a perfect balance between being cost-efficient and providing the best services to the passenger. The main forecasts are done one month in advance.
Use Case 1: Check-in project
In this project, BAC aimed to predict the number of passengers arriving at check-in every 30 minutes. By analyzing data from previous passenger surveys, we built an ML model that takes into account different behaviors for different types of flights.
“One of the challenges we faced was forecasting the moment people start queuing at check-in, but by combining imperfect ground truths, we were able to build a good target for our model. Our forecast achieved a mean absolute percentage error of ~30% per interval of 30 minutes, providing a more accurate forecast than the previous "worst-case scenario" approach.”
Raphael Peschi, Machine Learning Engineer and Team Lead
Use Case 2: PRM - Person Reduced Mobility project
Every day, hundreds of passengers with reduced mobility need assistance navigating the airport, which requires enough staff from BAC partners to assist them from arrival to departure.
As you can imagine, providing this kind of end-to-end guidance is quite labor intensive. On top of that the service is provided for free, to be able to do that the airport needs to go about it in an efficient way. This means having a good idea of the amount of PRMs to foresee the right amount of staff and avoid breaking the bank.
“As an input to the staff planning we forecast the amount of PRMs on each flight. This level of granularity is needed to know the amount of staff needed and to make sure that each passenger can be helped by an agent. We're currently achieving a mean average percentage error of 14% on a day level.”
Victor Hutse, Solution Lead
Use Case 3: Baggage Project
In this project, BAC aimed to accurately predict the number of bags per flight, especially for transfer flights where time is critical. In January 2023, the average number of departing bags processed daily was 17150, of which 5796 were “transfer” bags. The airport can avoid bottlenecks and improve its performance by making a more accurate bag forecast.
“During the project we had regular conversations with the end users of the forecast we were building, to make sure that what we were predicting corresponded with the numbers they need in day-to-day business. The project is still ongoing but we are now able to predict the amount of bags on each flight with a mean average percentage error per flight of ~28%. Beside that we are also working on predictions of when luggage will be at different points in the airport, allowing the people in daily operations to know when certain bottlenecks could form and enabling them to take mitigating actions.”
Robbe De Sutter, Machine Learning Engineer
Impact
Superlinear has been working on these use cases since May 2022, and they will soon be put into production, generating a real impact for the airport. Our data forecasting tools are directly feeding into BAC's ambition of being one of the best-performing hubs in the world, as well as into their sustainability goals. As operations become smoother, passengers' frustrations will be removed, and traveling will become easier.
The future
As evidenced above, we have just started contributing to the AOP project. We are looking forward to the next few months, with the first models in production.
“Step-by-step, we are building on the airport of the future, where data and AI play a crucial role in supporting our operations. Only in this way can we reach maximal operational efficiency and serve an ever-increasing number of passengers with the best experience throughout their journey.”
Victor Hutse, Solution Lead at Superlinear
How can other companies benefit from forecasting with AI?
Demand forecasting is a very hot topic at the moment. At Superlinear, we have started a project to help us gather all of our knowledge on forecasting in one framework: the Connected Supply Chain Optimiser (CSCO).
“The situation over the last couple of years has demonstrated that it is difficult for industries to keep up with rapid changes in supply and demand. The existing systems in place split the supply chain process into small parts and lack a global vision. With the CSCO, we are creating a flexible framework to help companies model their entire operations and produce a near-optimal planning that gets updated based on the latest available demand forecast.”
Raphaël Peschi, Machine Learning Team Lead at Superlinear.
Interested to learn more about forecasting with AI? Contact us here.
Contact Us