Mitigating airport congestion through variable message signs and model predictive control algorithms

Airports use Variable Message Signs (VMS) to manage congestion by directing traffic between ramps. However, the VMS deployment remains manual and heuristic, which often limits its effectiveness or even exacerbates congestion. While optimisation algorithms exist for highways and urban roads, no existing algorithm accounts for airport-specific dynamics, leaving a gap in automated VMS optimisation. This research introduces the first set of Model Predictive Control (MPC) algorithms designed to optimise VMS airport curbside management. These MPCs are tailored to airport dynamics, incorporating real-time conditions, passenger demand, anticipatory VMS activation, and optimal curb management. We evaluated the MPCs using a microsimulation model of the Seattle-Tacoma International Airport, covering about 40 miles of roadways and airport facilities. By comparing the performance of MPCs to a baseline without VMS, we found that they effectively reduced airport congestion, as measured by multiple metrics, including the number of vehicles aiming to park, average vehicle speed at terminal ramps, duration and length of queues formed at those ramps, and delays vehicles experience while accessing the terminal. The MPCs improved parking flows by 1.3–17.0% and speeds by 4.0–4.7%, while reducing queue duration by nearly an hour, queue length by 10.9–15.4%, and delays by 12.5–17.2%. The introduced MPCs offer airports adaptable options, with one MPC using speed and another using flow. These MPCs increase the efficiency of curbs, reduce congestion, and improve passenger experience by moving beyond the current heuristic-based management approaches. This study pioneers airport-tailored MPCs, translating optimal VMS management guidelines into practice with airport-specific parameters.