Incentive-aware pickup-station integration for electrified last-mile delivery under time commitments

Parcel distribution is among the most time-sensitive and cost-intensive legs of e-commerce logistics, and electrified fleets increasingly require coordinated routing, charging, and service-commitment decisions in dense urban areas. This paper introduces the Electric Vehicle Routing Problem with Time Windows and Pickup Stations (EVRP-TWPS) to jointly optimize EV routes, partial recharging, Pickup-Station (PS) selection, customer assignment to PSs, and budgeted incentives, subject to deterministic delivery time windows, PS operating hours, and promised pickup-availability deadlines. We first formulate a mixed-integer linear program that captures energy-feasible routing with partial charging, compartment-based PS capacities by parcel size, and an incentive mechanism that bounds total compensation while enforcing minimum acceptance requirements for assigned customers. To solve medium- and large-scale instances, we propose a Feature-Aware Variable Neighborhood Search with Simulated Annealing (FAM-VNS) that integrates coupled home-to-PS mode-switch neighborhoods, EVRP-TWPS-specific feasibility repairs (hours, promises, compartments, and budget), adaptive penalties, and charging re-optimization within a Metropolis acceptance scheme. Computational experiments on adapted instances show that EVRP-TWPS reduces total cost by 39.72% on average for small and medium instances compared with EV-only home delivery, and FAM-VNS attains near-exact quality across small to large instances while finding the 21 new best solutions for related covering-routing problems in the literature. Sensitivity and scenario analyses reveal threshold-driven operating regimes in which pickup activation and fleet-sizing decisions dominate cost through discrete breakpoints. These results provide actionable guidance for designing PS coverage, capacity, and incentive policies that reduce the need for EV deployment while maintaining deterministic service commitments.