Robust Decision Trees

Session date: 24 February 2025

Summary:

In this talk, we address a real-world problem in city logistics: the Two-Echelon Multi-Trip Vehicle Routing Problem with Satellite Synchronization (2E-MVRP-SS). The problem involves coordinating a fleet of light electric freight vehicles (LEFVs) operating at the street level with a fleet of vessels operating at the water level. The objective is to minimize total logistics costs while serving customer requests with time windows and transshipping goods between LEFVs and vessels at satellite locations. These satellites have limited capacity and no storage, necessitating precise synchronization of operations in both space and time.

To tackle this problem, we propose two models: a joint mixed-integer linear programming (MILP) model and a logic-based Benders decomposition (LBBD) model. The LBBD model demonstrates greater robustness for larger instances, significantly improving solution quality and computational efficiency, with an 18.7% reduction in total travel time. We also investigate the impact of cost allocations between two service providers and satellite locations on overall system performance. Furthermore, we evaluate various service design alternatives, such as single echelon systems using only trucks, only LEFVs, and two-echelon systems using only storage, and synchronization. The key contribution of this work lies in the development of efficient models for solving the 2E-MVRP-SS, offering valuable insights into the trade-offs between different system configurations and operational parameters. This research is highly relevant to the fields of scheduling and optimization, providing practical solutions for advancing sustainable urban logistics.

Relevant papers

A two-echelon multi-trip vehicle routing problem with synchronization for an integrated water- and land-based transportation system

Cigdem Karademir, Breno A. Beirigo, and Bilge Atasoy

European Journal of Operational Research, Jun 2025

Abs HTML

This study focuses on two-echelon synchronized logistics problems in the context of integrated water- and land-based transportation (IWLT) systems. The aim is to meet the increasing demand in city logistics as a result of the growth in transport activities, including parcel delivery, food delivery, and waste collection. We propose two models, a novel mixed integer linear joint model, and a logic-based Benders’ decomposition (LBBD) model, for a two-echelon problem under realistic settings such as multi-trips, time windows, and synchronization at the satellites with no storage and limited resource capacities. The objective is to optimize transfers and satellite assignments, thereby reducing overall logistics costs for street vehicles and vessels. Computational experiments demonstrate that the LBBD model is more robust in terms of solution quality and solution time on average while the added value of the LBBD is more evident when solving large-scale instances with 100 customers, reducing the overall costs by 10.6% on average and significantly reducing the fleet costs on both networks. Furthermore, we assess the effect of changing cost parameters and satellite locations in the proposed IWLT system–analyzing system behavior and suggesting potential improvements–and evaluate several system alternatives in city logistics–consisting of different transportation network designs (single- and two-echelon), vehicle types, and operational constraints. On average, the proposed two-echelon IWLT system reduces the number of kilometers traveled by vehicles at street level by ranging from 20% to 30% compared to a typical single-echelon service design that relies solely on trucks.