Modelos de otimização e métodos de solução para o planejamento da produção e distribuição na indústria de móveis

Abstract

Production and distribution are two key decisions in supply chain management. In order to achieve an effective operational performance along the supply chain, it might be important for these two decisions to be planned in a integrated manner, especially in supply chains with low inventory levels between production and distribution activities. This thesis studies the integration of production, inventory, distribution and routing decisions in the context of furniture companies. Research involving this integration has focus on more theoretical and idealized scenarios and, therefore, analysis on real-world settings are scarce in the related literature. Thus, we contribute to the literature by proposing mathematical optimization models and solution methods to represent and solve the integrated production and distribution planning and scheduling problem in the context of furniture manufacturers. Moreover, we also highlight the advantages of integrating production and distribution planning and scheduling decisions over the current industry practice, wherein both of these decisions are made independently of each other.

The proposed optimization models take into account features rarely considered in the literature, but commonly found in furniture companies, such as producing and stocking multiples components and sub-assemblies (instead of final products), sequence-dependent setups times and costs, different capacity vehicles performing multiples trips over the planning horizon, distribution routes extending over one or more periods, multiple time windows and customers deadlines. As a result, our models properly represent the real situation and can be used to support integrated production and distribution planning and scheduling decisions in furniture companies.

These models, however, are difficult to solve and their use as a production and distribution planning and scheduling tool is limited to moderate sized instances. In order to solve larger problems, we also propose different decomposition-based heuristic methods, which are able to solve larger instances in reasonable computing times. Results show that our heuristics, based on mathematical programming (i.e., matheuristics), are able to produce better solutions than general-purpose optimization solvers in shorter computing times. Exact solution methods, based on Benders decomposition, are also proposed to provide better quality solutions for small instances. Computational evaluation of these approaches shows that they perform as well as a general-purpose optimization solver.

Overall, the approaches here proposed (i.e., mathematical optimization models and solution approaches) can be used in different furniture companies, as they are able to effectively support production and distribution planning and scheduling activities, considering the most relevant and common features found in the furniture manufacturing and delivery processes. Moreover, these approaches can be adapted or extended to be used in other industries with similar production and distribution processes as those found in furniture companies.