Abstract
The growing access to technology combined with the effects of the COVID-19 pandemic has resulted in a significant increase in the demand for e-commerce delivery services. This surge in demand led to several externalities in the last mile of the delivery services, such as illegal parking, cruising for parking, traffic congestion, and missed deliveries. In this paper, we present a multi-mode variable-echelon vehicle routing problem with delivery options (MM-VEVRP-DO) for optimizing parcel deliveries using cargo bikes and lockers. We first present the discrete formulation of the problem which quickly becomes intractable due to the size of real-world networks. We propose a discrete-continuous hybrid model that decomposes the set of customers into regions and estimates the last-mile delivery cost based on asymptotic continuous approximation formulas. We provide a methodology to generate feasible solutions and implement the model on the Chicago transportation network using the POLARIS transportation modeling and simulation framework to obtain realistic estimates of key performance metrics. Numerical results show that the implementation of micro-hubs and lockers has the potential to decrease the number of trucks operating, reduce total operations hours, and improve the delivery system efficiency.