Freight-on-Transit introduces the innovative approach of blending freight with passenger services in city transport systems. It encourages freight companies to adopt greener transport options through government policies. Companies can opt for direct (via trucks) or combined (using trucks and trains) methods depending on cost considerations in this mixed transport model. The decision involves selecting road transport or a mix of road and rail options for moving goods. This study presents several groundbreaking developments we are exploring, including the simultaneous planning and scheduling for freight and passenger rail services. We evaluate the impact of policies like fuel taxes and subsidies for trains on making freight transport more environmentally friendly. The issue is framed as a Stackelberg game, with a leader-follower dynamic between the regulatory body and the freight company. The authority imposes a fuel tax for road usage and offers subsidies for the combined use of freight and passenger rail services. In response, the freight company, as the follower, organizes shipments in the most economically beneficial way. "Direct" shipments are done entirely by road, whereas "indirect" ones incorporate rail services into the logistics chain.

Sustainable city logistics planning focuses on multi-tier and multi-modal transportation with efficient consolidation and vehicle types suited for each tier. Integrating logistics services of different providers and shared transportation of multi-directional demand flows are key strategies to reduce congestion. This work introduces a detailed mathematical description of a day-before planning problem in two-tier multi-modal city logistics with on-time synchronization, where no storage exists at handover locations. While this enables the use of existing resources like supermarket parking lots in the distribution process, the requirement of delivery vehicles to meet for synchronized activities is a challenge. The planning approach is based on two-tier scheduled service network design, where transportation services with routes, departure time windows, and capacities are given, and waiting time policies exist for customer and handover locations. Demands involve inbound, outbound, and inner-city commodity flows. The goal is to select services, including a schedule for each of them, and allocate the demands such that both operating costs and waiting times are minimized.
We present a two-step construction matheuristic, where the search space is reduced by fixing a subset of binary selection variables for services based on solutions obtained for a relaxation of the model. Since the adherence of capacity restrictions of handover locations and the precise determination of vehicle waiting times constitute the main drivers of model complexity, these aspects are simplified or neglected thereby. In a broad computational study, the simple construction matheuristic validates the accurate representation of structural and temporal requirements and demonstrates the applicability of the model to obtain near-optimal solutions for medium-size instances. Further, we explore computational effects of varying instance parameters and discuss dependencies within the solution’s structure. Interesting methodological and managerial insights are drawn and conclusions on future research directions are provided.

This work aims to study and develop a decision-support system for planning the distribution of freight packed in containers arriving from the sea to cities built around a port. The problem is approached from the perspective of an urban mobility manager who aims to create a strategic plan for delivering freight from the port to customers located within the city, while managing existing transportation resources represented by firms located on the outskirts of the city. Containers cannot be opened at the port and cannot reach destinations within the city due to local regulations; they must be unpacked at intermodal facilities, and the freight must be delivered to customers by a set of city-freighters. A location-network design problem is examined to achieve this goal: a set of containers carrying pallets needs transportation from the port to facility locations, where pallets are unpacked from the containers, loaded onto vehicles allowed to transit on city streets, and dispatched to their final destinations. This problem involves selecting facilities and vehicles, assigning containers to facilities, determining paths for chosen vehicles, and managing pallet flows. An ALNS-based metaheuristic has been proposed; an initial solution is built and modified at each iteration of the algorithm: a destroy operator is selected to de-assign a certain number of containers from their corresponding satellites. These containers are then reassigned to any open satellite by a repair operator. Some proposed destroy operators remove containers after an explicit opening or closing of a satellite. At each iteration of the algorithm, after the container assignments to satellites have been changed, a multi-depot vehicle routing problem is solved to determine the routes of the city-freighters that will deliver the pallets to customers. The proposed ALNS meta-heuristic algorithm is compared to a heuristic algorithm which is based on decomposition by satellite.

In this presentation, we introduce the stochastic two-echelon multicommodity location routing problem with stochastic and correlated demands. We propose a two-stage stochastic programming formulation where the first stage involves decisions on the design of second-echelon facilities, and the second stage consists of recourse decisions addressing the distribution of observed demands. The objective is to minimize the combined costs of first-stage design decisions and the expected total routing costs incurred in the second stage. To solve the resulting stochastic optimization model, we develop a progressive hedging metaheuristic enhanced with algorithmic improvements to accelerate the solution space exploration. These enhancements include: 1) the use of population structures to derive diverse solutions for the scenario subproblems solved during the search process; 2) alternative strategies for defining reference solutions that efficiently guide the solution-consensus search; and 3) a reset procedure to mitigate the risk of the method becoming trapped in local optima. We validate the efficiency and effectiveness of our strategies through extensive computational experiments, assessing their ability to generate high-quality solutions under various problem settings and demand correlation settings.