The Société de transport de Montréal (STM), a public transport agency in Montreal (Quebec, Canada), has committed to electrifying its entire vehicle fleet as part of its broader sustainability initiatives. This research addresses a critical challenge that arises from this commitment: optimizing the deployment and dispatching of electric service vehicles assigned to operations supervisors to ensure minimal disruption to the bus network. The problem is formulated as a mixed-integer linear program that minimizes the number of deployed vehicles and total response time using a lexicographic approach. The effectiveness of our approach is evaluated through computational experiments using the commercial CPLEX optimization solver. This research provides the STM with a valuable strategic and operational decision-support tool, aimed at optimizing the size and deployment of its new electric vehicle fleet, therefore supporting its sustainability objectives.

The growth of e-commerce has led to a significant increase in parcel deliveries. In 2021, e-commerce accounted for 13.4% of retail trade, and it is projected to grow by 3% each year. To meet the demand for more efficient parcel delivery practices, it is essential to improve the organization of parcel collection (first mile), large-scale parcel transportation between different regions (middle mile), and final delivery to the customer (last mile). This is crucial not only for reducing the environmental impact but also for mitigating the disturbances caused by such activities.In this study, we focus on the middle mile part. The studied problem deals with the parcel delivery between different regional distribution centers with possibility to use the cross-dock. The deliveries are performed directly or through a route. In the context of this research, the objective is to minimize the total distance travelled per parcel for all deliveries. The primary challenge is naturally presented as a vehicle routing problem. To address the complexity of accounting for all the unique specifities of the studied problem, a MIP model is developed. The problem being NP-hard, we show that solving instances with commercial solver (CPLEX) reaches its limits very quickly even for small-size problems.