Modelo fuzzy para análise de riscos em projetos de edificações

Abstract

The civil construction is inserted in a complex and dynamic environment, where a wide variety of factors, often unknown, are influential on the goals of organizations and projects. This environment becomes even more complex due to the need to associate fast and accurate responses in an extremely competitive market. To reach this demand, tools capable of providing these solutions in this scenario have been incorporated into the management model, with the intention of assisting the manager in the decision-making. These tools should be able to create systems that identify and analyze these factors, allowing the decision maker trace paths to avoid risks and promote opportunities. As a result, risk management has been integrated into project management, enabling the effectiveness of the procedures that concern this issue. Many tools that assist in risk management have been developed in the literature and applied in real cases. In order to deal with the uncertain and incomplete data present in this scenario, where natural language is also one of the most used variables in the environmental characterization, several of these studies have been incorporating the fuzzy logic. This study aims to develop a model that adopts fuzzy logic to analyze the risk factors in terms of the most important criteria for the objectives characterization of a project, which are quality, cost and time. In a first moment it will be detailed discussed the risk management in the construction sector, allowing to present the sector, and its necessity of the use of this tool. Forwards, are presented the fundamentals of fuzzy logic that will be applied in the methodology. Then it is proposed a fuzzy inference system based on Mamdani, where grouping functions were elaborated for each of the considered criteria and for the output risk, then rules that make the relationship between these variables were built. The model incorporates this method of inference in the analysis of a case, to return values of risks that should assist the manager in the decision-making. To elaborate a discussion around the model and to evaluate its validity two scenarios are created and simulated through a sensitivity analysis. As the closure of this work, succinct conclusions are made on points discussed and perspectives are composed for this study.