Minimização do custo de antecipação e atraso para o problema de sequenciamento de uma máquina com tempo de preparação dependente da sequência: aplicação em uma usina siderúrgica

• Ano de defesa: 2006
• Autor(a) principal: Leandro de Morais Bustamante
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Problema de atraso e antecipação; Programação linear inteira mista; Tempo de preparação dependente da sequência; Sequenciamento da produção de uma máquina; Programação linear; Engenharia de produção; Usinas siderurgicas Automação
• Link de acesso: http://hdl.handle.net/1843/NVEA-72CN7W

In this dissertation was studied the one-machine scheduling problem, with the objective of minimizing the sumo f earliness and tardiness costs, with sequence dependent setup time. The operational planning problem of rolling production in a steel plant, which was the inspiration and environment to this research, consists of determining a best sequence to produce a set of production planning orders. Each order has a due date. If the sequence programming the order produces after the due date, there is tardiness cist, otherwise, if the order is programmed before the due date, there is earliness cost. To present the problem were developed Mixed Integer Linear Programming models. Two models were proposed, using the Mannes [16] and Wagner1s [24] formulations, dealing with setup times like a problem restriction. However, from the initial conclusions in idleness situations, was understood the necessity of dealing with setup costs also, when was developed another model, This difficulties of the modeling of the problem studied, from Mannes and Wagners definitions, are showed, such as the comparison about the performance when it is used to resolve the some problem. The models were implemented using the language MathProg and optimizing package GLPK 4.8. The modeling proposed and the result analysis, even that limited to problems with reduced size, allow evaluate the application for models in the studied case and understand the problem and its solutions, generating knowledge that can contribute to development of suitable heuristics to solve real problems.