Assessment of Mechanical Robustness and Resilience in Urban Water Networks Under the Impact of Climate Change Using Metamodels and Artificial Intelligence

Abstract

Urban water networks play a crucial role in urban functionality. These systems are indispensable elements of urban resilience. However, these long-established systems are facing challenges due to global population growth, urbanization, and climate change. These challenges reduce the mechanical robustness and resilience of these aging systems as an inevitable consequence. This study proposes a metamodeling approach to determine the indices of robustness and resilience from a mechanical point of view by conducting a study on various 3D soil-pipe configurations under the effect of climate change. These configurations are identified on an experimental basis, especially using a central composite design to build the metamodel, whereas a sensitivity analysis will identify the most important inputs. Machine learning models are the alternative predictive modeling tool that can be used instead of computationally heavy traditional simulations. The computational efficiency offered by the modeling approach therefore allows a more rapid evaluation of a larger number of scenarios, even scenarios that are not explicitly represented in the datasets of original mechanical indicators. Thus, this approach eventually finds the critical pipes and provides the mechanical robustness and resilience of the overall water network, thereby supporting the decision-makers in managing these systems.