Contribuições para o balanceamento de uma linha de manufatura robotizada por meio de programação matemática

Abstract

The automotive productive systems are generally based on production lines oriented to the product, a flow shop structure. The main objective of this productive system is an high throughput (productive \cadence") or, in a complementary way, the lowest cycle time for production. However, high throughput is usually associated with an elevated specificity of the line’s equipment and a high implementation cost. This cost and specificity forces the involved professionals to develop new ways to work around the system rigidity, allowing the production line to produce as many products as possible. This masters thesis presents a methodology capable of aiding the process of introducing a new vehicle model in a production line already in operation. Specifically, the focus is on the balancing of robots that perform spot welding tasks. In the case study, 500 weld spots were distributed in thirty-four robots, allocated to eight workstations. In the balancing process, adaptations were done in a Mixed-Integer Linear Programming mathematical model available in the literature. The final model needed to be able to consider the inherent aspects of the line layout, e.g., the symmetry between the robots and the vehicles, besides the inherent aspects of the individual capabilities of each robot. In addition to the initial weld spot balancing problem, a parallel project, which was seeking to update eight of the thirtyfour robots, showed the necessity of elaborating a second variant of the mathematical model. This variant was able to make a temporary distribution of the weld spots that were originally assigned to the eight robots under upgrade, keeping, at the same time, unaltered the weld spots distribution from the other 26 robots. In that way, the application of all weld spots assigned to the production line during the period in which line upgrade works was guarantied. The feasibility of the proposed answers were evaluated by system experts. Thus, a methodology for aiding the operational decision-making was proposed. This methodology suggests mathematically optimized results for the distribution of weld spots between the robots of the productive system, respecting the practical conditions of the production line under analysis.