Análise de riscos em sistemas de abastecimento de água: uma abordagem via teoria de redes complexas

Abstract

Water Distribution Networks (WDNs), responsible for distributing water in adequate quantity and quality, are susceptible to leaks, pipe ruptures, and contaminant intrusions, negatively impacting hydraulics and water quality. Therefore, it is crucial for decision-makers in various domains, such as network planning and management, to identify critical elements within the network to formulate more effective maintenance strategies, thereby enhancing system performance and preventing failures. With the aim of identifying critical elements in water supply systems, this study seeks to conduct a risk assessment related to events such as leaks, pipe ruptures, and contaminant intrusions. To achieve this objective, the approach employed is the Complex Networks Theory (CNT), an alternative to conventional methods relying on multiple hydraulic simulations through modeling software. It is noteworthy that, as the network's scope increases, the time required to conduct such simulations under traditional approaches also significantly escalates. CNT is a scientific discipline examining network characteristics (connectivity, robustness) to analyze its behavior and dynamics. To conduct this type of analysis, the WDN can be modeled as a complex network, where demand nodes, tanks, and reservoirs are represented as vertices, and edges correspond to pipes, pumps, and valves. In order to assess the importance of pipes and nodes, metrics of robustness, redundancy, centrality, and connectivity are applied to the graphs generated by the mathematical abstraction of WDNs. The three types of failures are modeled using the EPANET software to evaluate their impacts on hydraulic parameters (pressure and flow) and substance spreading in the network. Subsequently, CNT metrics, such as calculating shortest paths between source and water demand vertices to assess the importance of pipes, and vertex centrality metrics to evaluate node influence on the network, are applied. The results of hydraulic simulations are then compared with the outcomes of metric application on graphs. The convergence of results indicates a reduction in resources and computational time required to analyze the risk in water distribution networks for these three types of failures. This study contributes to advancing knowledge by proposing an innovative approach to risk assessment in WDNs, overcoming the limitations of conventional methods. By adopting CNT, more efficient analyses can be conducted, providing valuable insights for the development of targeted maintenance strategies, thus promoting the improvement of water distribution system performance and the prevention of failures.