New approaches for the two-level facility location problem: models and algorithms

• Autor(a) principal: Paganini Barcellos de Oliveira
• Data de Publicação: 2020
• Tipo de documento: Tese
• Idioma: eng
• Título da fonte: Repositório Institucional da UFMG
• Texto Completo: http://hdl.handle.net/1843/33071; https://orcid.org/0000-0002-0639-8844
• topic: Two-level facility location problems; Benders decomposition method; GRASP; Optimization of large-scale logistics systems; Engenharia de produção; Método de decomposição; Algoritmos; Logística empresarial

This thesis presents exact and heuristic algorithms to solve different variants of the two-level facility location problem; variants which were less studied in the literature up till now. This is a topic of great relevance within the field of Large Scale System Optimization and with an ample presence in several logistics systems. Generally speaking, the problem consists of selecting a subset of points from a set of candidate sites to act either as a supply or as an transshipment point to serve customers geographically scattered at minimal installation and transportation costs. In the studied variants, the service or distribution network is hierarchical, being composed of a first level with facilities that supply customer demands via a second tier having transshipment points. The three studied variants of the problem are: (i) the version in which the facilities and transshipment points are assumed uncapacitated; (ii) the case in which customers’ demand varies over a discretized planning horizon, leading to a multi-period or dynamic location problem; (iii) and the alternative that considers additional costs arising from the effects of congestion derived from the delay of accumulated flow in the facilities and transfer points. All investigated variants consider two possible types of interconnection between the first and second levels. In the single allocation, a transshipment point can interact with only one facility; while, in the multiple assignment, a transshipment point can connect with many facilities. Finally, as one of the greatest challenges of these problems is their combinatorial nature, mathematical models and specialized methods based on the Benders decomposition method and GRASP were proposed and assessed both in terms of computational running time and obtained solution quality for each of the studied variants.