Airport security workforce planning

Abstract

Uncontrollable passenger arrivals cause frequent situations where passenger demands exceed the available capacity. This becomes particularly apparent within the airport’s bottlenecks — such as the security hall — where long queues can build up, leading to delays, which translates to unhappy passengers and airlines. Planning for and quickly adapting to changes in passenger arrivals to the security hall and staff availability is key to keeping a consistent flow of passengers and avoiding a build-up of queues.

This problem can be addressed at different stages in the roster planning timeline. At the strategic planning stage, the chief staff scheduler (CSS) and HR department can change contracts and adapt the workforce structure (e.g. via recruitment) best to suit the expected demand for the upcoming season. At the tactical planning stage, the problem is two-fold. On the one hand, a fast and flexible roster tool is needed to adapt quickly to changing work regulations and union demands. On the other hand, this is the last time the CSS can freely allocate workforce members to the schedule. Hence, they need to use this chance to build flexibilities into the roster that can help adapt to changes in demand at the subsequent operational planning stage.

A solver-independent modelling approach using Mixed Integer Programming and Constraint Programming is proposed as a fast and flexible Roster Tool to support the CSS’s decisions and help with union discussions. Based on this tool, a Simheuristic is derived to allocate flexibility to the roster at the tactical planning stage. The Simheuristic combines a metaheuristic (Simulated Annealing, Genetic Algorithm and Adaptive Large Neighborhood Search (ALNS)) to adjust the flexibility metrics in the roster with Monte-Carlo Simulation of demand patterns and staff absences that need to be addressed. The proposed Simheuristic for tuning flexibility metrics is the first of its kind in rostering literature. It shows significantly better results than simply adjusting the aggregation of the demand forecast into a model week.

Furthermore, a Simheuristic consisting of ALNS to adjust the contracts and Monte-Carlo Simulation to test the feasibility of the resulting rosters is proposed to address the strategic planning problem of finding the best contractual combination for the next season. The experiments suggest that different demand patterns require different workforce structures, which implies that the airport should revisit its contracts every season to best address expected changes in passenger demands.

Finally, a case study of Edinburgh Airport’s security staff scheduling problem and the development of a suitable rostering approach is presented.