3 Presentations

• 
  01:30 PM - 01:55 PM

  A Two-Echelon Location Routing Problem for last-mile delivery

  • Fernando R. Senna, presenter, Federal University of São Carlos and CIRRELT
  • Reinaldo Morabito, Federal University of São Carlos
  • Pedro Munari, Federal University of São Carlos
  • Leandro C. Coelho, Université Laval

  The increasing complexity of last-mile delivery systems creates new challenges and opportunities for practice and research. In this work, we study a two-echelon location routing problem for last-mile delivery with time windows considering a scheme in which vehicles perform routes over some customers while the remaining ones are visited by carriers on foot (or by bike). By effectively creating clusters, the parking locations of the vehicles serve as depots for the second echelon routes that are executed by deliverymen. They may serve customers in parallel, reducing the overall service time and the number of vehicles needed. This problem reflects applications in congested urban areas where the difficulty of finding parking locations and the proximity of customers make it interesting to include more than one deliveryman per vehicle. The problem is an extension of the Vehicle Routing Problem with Time Windows and Multiple Deliverymen, in which decisions on customer clustering and deliverymen routes are not included. We present a formulation for the problem, propose novel sets of valid inequalities for this formulation, and elaborate on exact solution methods. We discuss the cost structure of different instances and assess the performance of the proposed approaches using new and existing instances.
• 
  01:55 PM - 02:20 PM

  Hybrid Modelling for Optimizing Automated Parcel Locker Network Design: A Case Study in Pamplona, Spain.

  • Luis Cadarso, University King Juan Carlos of Madrid
  • Adrián Serrano-Hernández, Public University of Navarre
  • Sergio Martinez-Abad, Public University of Navarre
  • Javier Faulin, presenter, Public University of Navarre

  In recent years, last-mile distribution in urban areas has been facing a number of challenges such as limited delivery time windows, increased demand for fast delivery options, and the issue of no-show customers. To address these challenges, the adoption of automated parcel locker (APL) systems has emerged as a promising solution. Our work examines the usage patterns of APLs in the city of Pamplona (Spain) and proposes a hybrid model approach of simulation-optimization for designing an optimized APL network in the aforementioned city.
  Thus, we use agent-based modeling in order to estimate the future demand for APLs, taking into account various socio-economic factors such as population size and e-commerce growth rates. Additionally, we apply a dynamic optimization model to identify the optimal locations for APLs, while minimizing operational and service costs. Our hybrid methodology predicts an increase in the number of eShoppers and five times more of APLs in the next three years.
  These results highlight the potential benefits of simulation and optimization tools in promoting the adoption of APLs as an effective last-mile distribution strategy. By leveraging our proposed approach, stakeholders can achieve significant cost savings while improving service quality and customer satisfaction.
• 
  02:20 PM - 02:45 PM

  A freight-on-transit service network design model for same-day deliveries operations

  • Fernando Zingler, presenter, Concordia University
  • Navneet Vidyarthi, John Molson School of Business, Concordia University
  • Ivan Contreras, Concordia University

  Freight-on-transit (FOT) or co-modality has been increasingly used to integrate last-mile deliveries with passenger transportation, where the spare capacity of transit vehicles is used to move courier, express, and parcel deliveries. We develop a tactical planning/service network design model to select the optimal transit routes for routing freight on passenger transportation during off-peak hours for same-day delivery operations. The proposed model maximizes the total profit for routing commodities and accounts for the vehicles involved in this operation's capacity for transshipments between the hubs and the transit route over time. One of the common approaches consists of discretizing the planning time and then restricting the decisions to those time points. However, this approach leads to very large formulations intractable in practice. We formulate the problem over a time-expanded network and propose a Dynamic Discretization Discovery (DDD) algorithm, where a cluster of profitable demands when allocated into the transit pre-determined schedules. We test three different clustering techniques based on the delivery windows to discretize the network and enhance the computational efficiency of the algorithm. Preliminary results over small networks confirm the efficiency of this approach.