A Data-Driven Approach for Baggage Handling Operations at Airports

Before each flight departs, baggage has to be loaded into containers, which are then forwarded to the airplane. Planning the loading process consists of setting the start times for the loading process and depletion of the baggage storage as well as assigning handling facilities and workers. Flight delays and uncertain arrival times of passengers at the check-in counters require plans that are adjusted dynamically every few minutes and, hence, an efficient planning procedure. We propose a model formulation and a solution procedure that utilize historical flight data to generate reliable plans in a rolling planning fashion, allowing problem parameters to be updated in each reoptimization. To increase the tractability of the problem, we employ a column generation–based heuristic in which new schedules and work profiles are generated in subproblems, which are solved as dynamic programs. In a computational study, we demonstrate the robust performance of the proposed procedure based on real-world data from a major European airport. The results show that (i) the procedure outperforms both a constructive heuristic that mimics human decision making and a meta heuristic (tabu search) and (ii) being able to dynamically (re)allocate baggage handlers leads to improved solutions with considerably fewer left bags.