Xiaojia Guo - Practice Based Paper 2

London Heathrow Airport Uses Real-Time Analytics for Improving Operations

Xiaojia Guo (Robert H. Smith School of Business, UMD), Yael Grushka-Cockayne (Harvard Business School), Bert De Reyck (UCL School of Management)

Improving airport collaborative decision-making is at the heart of airport operations centers (APOCs), recently established in several major European airports. In this paper, we describe a project commissioned by Eurocontrol, the European organization in charge of the safety and seamless flow of European air traffic. In Guo et al. (2019), we describe the system that we developed for Heathrow to better estimate passenger connection times, which takes advantage of the integration and data sharing that Heathrow’s APOC offers. In this paper, we describe the process of how we chose the subject of the pilot, namely the improvement of transfer passenger flows through the airport, and how we helped Heathrow move from their existing legacy system for managing passenger flows to an advanced machine learning based approach using real-time inputs. The system, which is now in operation at Heathrow, allows for predicting which passengers are likely to miss their connecting flights, reducing the likelihood that departures will incur delays while waiting for delayed passengers. This can be done by offloading passengers in advance, by expediting passengers through the airport, or by modifying the departure times of aircraft in advance. By aggregating estimated passenger arrival time at various points throughout the airport, the system also improves passenger experience at immigration and security desks, by enabling modifications to staffing levels in advance of expected surges in arrivals. The nine-stage framework we present here can support the development and implementation of other real-time data-driven systems. To the best of our knowledge, the proposed system is the first to use machine learning to model passenger flows in an airport

(INFORMS Journal on Applied Analytics (2020). Finalist of 2019 Daniel H. Wagner Prize for Excellence in Operations Research Practice.)

Figure 1. The Nine-Stage Approach. We used a nine-stage approach for developing and implementing a real-time data-driven collaborative decision-making system for improving the connecting passenger flows through Heathrow airport.

Figure 2. Predictive System Architecture. The IDAHO system provides both flight level and passenger level information in real-time. Flight level information includes a flight’s actual and scheduled arrival time, aircraft body type, etc. Passenger level information is extracted from the Passenger Transfer Message (PTM) files sent by the airline. The PTM files list passengers who are known to be transferring to another flight.