Uma comparação computacional dos algoritmos de Benders e o Dual-Ascent no problema multi-período de localização de facilidades

• Ano de defesa: 2016
• Autor(a) principal: Joab Symon Costa Santos
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Localização de Facilidades Multi Período; Dual-Ascent; Benders; Algoritmos; Engenharia de produção
• Link de acesso: http://hdl.handle.net/1843/BUBD-AFDNKC

The study of facility location is a very important factor for competitive edge of many companies working with shipping goods consumption and services on a demand network. To dene of the best locations for facilities aimed at minimal cost is an essential factor for greater and science in use of resources. When considering a long-term horizon, the location of facilities may change due to changes in customer demands. Thus, this work presents a model Location Multi Period facilities where is an accomplished balance between the best locations, considering dierent time periods, according to implementation costs and closing reallocation possible certain facilities. The study comprises the implementation of Benders decomposition method approaches pareto-optimal cuts andimprovement through convex combination, and Dual-Ascent method for a model adapted for multi-location facilities period. The computational experiments were conducted with data from municipalities in the state of Minas Gerais to simulate the denitionof the best places to install a network of distribution centers considering the variation of the population spot. The Benders simple algorithms and Pareto-optimal Cortesdid not achieve satisfactory results for larger instances, while improving the convex combination was dominant in most the tested instances. Dual Ascent algorithm was the most aware and, solving instances in less computational time. It was concluded that the variation of the spot populacinal for sample period multi-location facilities can become more interesting when there is a greater variation in demand within the periods.