Predicting Passenger Flows at Dubai International Airport: An IoT Case Study

The customer in this case study is Dubai International Airport, one of the busiest airports in the world in terms of international passenger traffic. The airport is known for its high volume of transfer traffic, with peaks in passenger volumes occurring throughout the day. The airport's operational departments, in collaboration with the Strategy & Development and IT departments, are responsible for managing passenger flows and ensuring smooth operations. The airport's Airport Operations Control Center (AOCC) uses sensors and cameras to monitor real-time queue information. However, the airport faced a challenge in predicting passenger load profiles and managing resources effectively to prevent overcrowding and long queues.

Dubai International Airport, known for its high volume of transfer traffic, faced a significant challenge in managing passenger flows. The airport experienced peaks in passenger volumes throughout the day, with immigration halls and transfer security checkpoints fluctuating between being completely empty and overcrowded within short periods. While the airport could plan for expected passenger load profiles, changes to flight arrival times could significantly impact the actual passenger load profile. The Airport Operations Control Center (AOCC) had access to real-time queue information from sensors and cameras, but there was a need for advanced passenger load predictions and resource requirements. This would enable operational teams to collaborate with other stakeholders and proactively open more security or immigration lanes, in anticipation of passenger arrivals. The challenge was to predict passenger load profiles and manage resources effectively to prevent overcrowding and long queues.

The airport developed a prototype of a passenger flow prediction tool using Alteryx and Tableau. The tool used a real-time feed from the airport's flight database, enriched with supporting reference tables, to provide the latest information on expected flight arrival times and aircraft parking locations. Two detailed workflows per terminal, PAX Demand and Counter Supply, were created using Alteryx Designer. These workflows combined flight-level data with other reference inputs, cleansed the combined data, processed it into required passenger demand aggregations, optimized the counter requirement, and wrote the output into a database for visualization in Tableau. The workflows were scheduled to run every five minutes to keep the data as near real-time as possible. The output data was used to create visualizations and dashboards in Tableau Desktop, which were then hosted on Tableau Server and shared across the organization and wider airport community. The deployment strategy was to keep the workflows simple and streamlined for easy tweaks and improvements.

• The passenger flow prediction tool provided significant operational benefits. Days after the tool went live, operational teams used the information to move resources from one terminal to another, thereby preventing serious overcrowding and queue build-up. This was not possible before as the information was not available. After a few months of using the tool, the airport observed clear decreases in queue lengths and wait times across many different touchpoints. The tool also proved to be invaluable during the Covid-19 pandemic, when the airport had to introduce PCR test stations, segregate passengers based on origin, and temporarily close terminals. The tool's flexibility and the close collaboration between teams allowed the airport to respond rapidly to changes and release multiple updates per week. The tool has been running successfully for over two years and has given the airport the flexibility to continue to add value in a cost-effective way.