Programação de tarefas para máquinas paralelas não relacionadas utilizando métodos heurísticos considerando a manutenção de ferramentas

Abstract

Currently, companies face a highly competitive environment, which forces them to seek optimization of their production systems to meet market demands. This study examines task scheduling in unrelated parallel machines, considering the need for tool maintenance. To address this, four heuristic methods were developed and applied, each tackling different criteria for task allocation and production optimization. The instances used were adapted from a widely recognized database in the literature, ensuring representativeness and variation in terms of the number of tasks and machines. The four heuristic methods, each with different approaches to task allocation and sequencing, were evaluated based on metrics such as makespan, delays, and the number of maintenance operations. The results indicate that Constructive Method 4 (CM4) achieved the lowest average makespan (137.43), followed by Methods 3, 2, and 1, with increases of 8.84%, 12.90%, and 19.44%, respectively. On the other hand, Constructive Method 1 (CM1) obtained the lowest average number of maintenance operations (15.25) and the lowest total delay (958.87), whereas CM4 presented the highest total delay (5055.14), with an 81.03% deterioration compared to CM1. These results highlight that, although CM4 is more efficient in terms of makespan, it significantly compromises performance concerning delays and maintenance. This study contributes to the field of production planning and control by providing a relevant comparison of heuristics applied in complex industrial environments.