Resumo:
In a globalized context, the decision-making process is crucial for the progress of planned activities. This importance has directly reflected in the pursuit of process improvement, enabling the application of system optimization, specifically through the use of Discrete Event Simulation (DES). However, this research field remains underexplored due to the convergence time of algorithms, as real-world problems present numerous objectives that is often conflicting. In this context, this work proposes a method capable of reducing the search space in Multi-objective Simulation Optimization (MOSO) problems and, consequently, the computational time, obtaining high-quality solutions, identifying the best variation limits for each decision variable, and additionally presenting the decision-maker with the best resource allocation. To achieve this, the proposed method combines DES, the Latin Hypercube Design (LHD) method, and Super-efficiency analysis through Data Envelopment Analysis (DEA) with variable returns to scale (VRS). In this method, the complete search space is represented by the LHD matrix, and based on the scenarios generated by the matrix, the DEA – VRS Super-efficiency method is applied, adopting the new limits for the multi-objective problem. The proposal was applied to two study objects: the first encompasses a logistics sector operation, and the second portrays the process of a multinational leader in construction solutions, with both study objects using real data and presenting distinct levels of complexity. Thus, the first study object showed a 70% reduction in the search space and a 17.44% reduction in computational time. For the second object, there was an 89% reduction in the search space and a 28.71% reduction in computational time. In summary, the proposed method proved to be promising in addressing complex MOSO objects and presented significant results.
