As e-retailers increasingly compete by offering consumer-centric services such as expedited deliveries, urban freight systems face mounting pressure to adopt sustainable last-mile distribution practices. While existing research has explored the performance of various distribution strategies under static planning and network design scenarios, their effectiveness under the dynamic and stochastic conditions characteristic of modern e-commerce remains insufficiently understood.
To address this gap, this study introduces a Dynamic-Stochastic Last-Mile Network Design (DS-LMND) problem, formulated as a Multi-Echelon Capacitated Location Routing Problem with Time Windows (ME-C-LRP-TW). The proposed approach utilizes a Monte Carlo simulation–optimization framework integrated with an Adaptive Large Neighborhood Search (ALNS) metaheuristic to solve the problem efficiently.
The framework is used to configure different distribution structures and simulate their last-mile operations under uncertainty. For each configuration, the study evaluates performance across the three pillars of sustainability: economic viability, environmental efficiency, and social equity. Moreover, it quantifies the impact of demand uncertainty on sustainability outcomes using two key metrics: the Value of Information (VI) and the Coefficient of Variation (CV).
Through this approach, the study:
a) Assesses the effectiveness of conventional last-mile strategies,
b) Demonstrates the competitiveness of electric delivery vehicles,
c) Evaluates the potential of crowdsourced delivery services,
d) Strengthens the case for consolidation-based multi-echelon logistics networks,
e) Establishes the viability of customer pickup models, and
f) Explores the role of drones and delivery robots in meeting dynamic, time-sensitive demand.
This research contributes a comprehensive, data-driven framework for evaluating and designing resilient and sustainable last-mile delivery systems under realistic, fluctuating demand conditions.