Abordagens robustas para o problema de sequenciamento flow shop cross-docking

Abstract

Globalization and the rapid growth of e-commerce have helped change purchase goods and service habits, increasing the complexity of logistics environments. Cross-docking is a logistics solution aiming to move products directly from di erent suppliers or manufacturers and consolidate them to common final delivery destinations without long-term storage. This strategy allows for fast deliveries and reduced warehousing and transportation costs while increasing logistical service. The success of the strategy depends on an e cient transshipment operation. This work undertakes a study of truck scheduling in a cross-docking center. The specific goals involve developing models, al- gorithms, and resolution methods for the truck scheduling problem in cross-docking centers, adding aspects of uncertainty. The problem is first modeled as a two-machine flow shop scheduling problem with precedence constraints to minimize the makespan and later generalized it to the parallel-dock case. We propose a hybrid method based on a Lagrangian relaxation technique through the volume algorithm. Using information from the Lagrangian multipliers, constructive heuristics with local search procedures generate good feasible solutions. With a series of cuts, the methodology finds tight bounds for small and large instance sizes, outperforming current results. To approx- imate our approach to the real cross-docking operation, we incorporate uncertainty in truck arrival times. A rescheduling approach is provided, and a novel algorithm for solving multi-dock truck scheduling problems is proposed under truck arrival time uncertainty. We discuss two optimization problems, the problem of minimizing the Makespan (Problem Cmax) and the problem of minimizing the Total Weighted Com- pletion Time (Problem WC). Extensive experimentation allows us to compare three methodologies and show that the rescheduling methodology can support managers in their daily cross-docking operations, e ciently handling dynamic and uncertain data, making good decisions quickly.