A data-driven decision support framework for the integrated airport check-in capacity planning problem

The rapid growth in air travel has intensified the need for efficient airport capacity planning, particularly within terminal operations where passenger experience and resource utilization are closely interlinked. Among these operations, the check-in process represents a critical bottleneck, often accounting for the majority of passenger waiting time. Addressing the limitations of physical expansion and the complexity of diverse allocation criteria, this paper introduces an Integrated Check-in Capacity Planning System (ICCPS) that comprehensively supports strategic, tactical, and operational planning decisions. The framework jointly addresses the three core phases of the Check-in Capacity Planning Process (CCPP) by developing: (i) a bi-objective, compact integer programming formulation for sizing and assignment, which incorporates expected queue parameters derived from machine learning-based passenger forecasts and empirically calibrated arrival and service time patterns, combined with an offline, discrete agent-based queue simulator; (ii) an adjacent counter allocation module, solved using a tailored rolling horizon-based heuristic to ensure computational scalability; and (iii) a conflict resolution module to address potential criticalities in capacity-constrained settings and to derive time-varying counter allocations. Extensive computational experiments, based on synthetic instances representative of medium- and large-sized airports, demonstrate the framework’s scalability and effectiveness, with the ability to efficiently generate Pareto-optimal solutions in a matter of minutes using parallelization techniques. Ultimately, a real-world case study at Milan Bergamo Airport (BGY) further validates the practical value of the ICCPS by evaluating: (i) check-in planning over a month-long horizon; (ii) scenarios involving both dedicated and common check-in strategies; and (iii) long-term forecasts of flight growth and demand distributions. The results confirm the system’s capability to minimize expected queuing times and resource usage while providing strategic insights for layout design, scenario evaluation, and day-to-day operational planning.