A Simulation-Based ALNS Approach for Last-Mile Delivery with Uncertain Parcel Locker Capacities

This paper addresses a last-mile delivery optimization problem that integrates both home delivery and shared parcel locker services, where locker capacities are limited and characterized by stochastic availability. The problem is formulated as a novel variant of the vehicle routing problem with two delivery modes under uncertainty. To solve it, we propose a simulation-based optimization framework that integrates an Adaptive Large Neighborhood Search (ALNS) metaheuristic with Monte Carlo simulation. The ALNS explores high-quality delivery schedules, while the simulation assesses solution robustness against uncertain locker usage. Computational experiments on Solomon benchmark instances demonstrate that the proposed approach consistently outperforms deterministic strategies, delivering superior solution quality and stability. These findings underscore the effectiveness of intelligent optimization for solving complex logistics problems under uncertainty.