A Flight-Wave, Hub-Aware Construction Heuristic for the Airport Dial-a-Ride Problem

Airport shuttle operations in major tourism destinations such as Antalya have a distinctive spatial and temporal structure. Passenger demand is concentrated between a central airport and hotels dispersed across coastal tourism areas, with service times closely tied to flight schedules. This creates a particularly clear clustering signal, as many transfer requests fall within short arrival and departure waves. General-purpose clustering and construction heuristics do not explicitly exploit this structure and may leave difficult requests unserved or generate weak initial solutions for subsequent improvement. This paper proposes a flight-wave, hub-aware construction heuristic for the deterministic, minimum-distance routing of reservation-based airport transfers. The method clusters requests by flight wave alone, dropping the spatial term of the classical spatio-temporal clustering distance. A hub-aware pairing operator then chains each inbound airport-to-region tour with a temporally later outbound tour on the same vehicle, eliminating empty repositioning through the hub. The constructed routes warm-start an off-the-shelf Guided Local Search under a fixed time budget. We evaluate the method on 60 stratified operating days drawn from a full year of real two-hub operations in Antalya, Türkiye. The heuristic serves every request on every instance. Compared to the baselines, its plans are 11.1% shorter than regret insertion and 18.8% shorter than Local Cheapest Insertion, as well as 6.8% shorter than Solomon-I1 on the instances it handles. The advantage grows with demand and vanishes on the quietest days. On every small sub-instance that an exact solver could close, the heuristic matches the proven optimum.