Otimização da programação de montagens estruturais na Indústria Aeronáutica

Abstract

In this research, the structural assembly schedule of activities in the aeronautical industry is studied. Structural assemblies are the manufacturing stage in which the parts, sub-assemblies and the main structural parts of the airplanes, like wings, stabilizer, front, rear and fuselage, are assembled together, forming the aerostructures. Essentially, the problem studied in this thesis is how to allocate the assembly activities over time and resources, considering the use of assembly fixtures and workers, with different specializations, respecting the production plans and technological constraints at a minimum cost. The objective of this research is to optimize the labor cost of the airplane structural assembly scheduling problem, based on a real case in the aerospace industry. To do that, optimizations approaches based in mathematical models and heuristic methods were developed. Another objective of this research is to contribute to the practice of the operational research, considering that the structural assembly is described in details and it is proposed a model to represent it mathematically in a way that its description could be used instead of idealized problems to motivated future correlated research. In this work, a mixed integer linear programming model is developed to represent the problem, based on resource constrained project scheduling and resource availability models. Computational experiments were performed with real data provided by a Brazilian aerospace company, and the results showed that the proposed model is capable of generating solutions to real size problems. Constructive heuristic methods, specialized local searches, random multistart and mathematical programming heuristics were developed and applied using Java and GAMS/CPLEX, in order to obtain good solutions in acceptable computational times in practice. The results of the computational experiments showed that the heuristics achieve different combinations of computational time and solution quality. In general, the computational times used to generate solutions using heuristic methods are shorter than those used by mathematical models with CPLEX and, sometimes, with very close solution quality. The solutions generated using the mathematical model and CPLEX use a long computational time in some cases, however, they generated the best solutions in all experiments. The results showed that is possible to generate good quality solutions for the aeronautical assembly scheduling using the proposed methods. Structural assembly of airplanes requires several operations, with a complex precedence network and interaction among many productive resources, which makes assembly and scheduling and planning a complex activity to be performed manually by humans, without the support of computer and analytical technology. In addition, this research studies an industry with great economic relevance to Brazil. The development of technology aiming competitive benefits to this industry has the potential to contribute to its development and competitiveness in Brazil and worldwide.