A Conceptual Model for Urban Interdependent Technical and Social Infrastructure Systems

Infrastructure systems underpin every aspect of urban life. As they become ever-increasingly interdependent, malfunction of some infrastructure facilities would lead to large scale cascading failure and might have catastrophic impact on urban life. In both academia and industry, continuous efforts have been made to develop appropriate modeling and simulation approaches, in order to better understand infrastructure interdependencies and analyze infrastructure performance after disruptions. However, unfortunately, though of great importance to urban resilience, social infrastructure (also known as soft infrastructure) systems, such as health care and emergency management systems, are still excluded from most of the models from a technical point of view. The study presented in this paper adopts the viewpoint of ‘system of systems’ and emphasizes the integration of technical and social infrastructure systems. A conceptual model is thus proposed to describe respective roles of and dependencies between the two types of infrastructure systems. Concretely, three major parts, namely technical infrastructure systems, social infrastructure systems, and non-infrastructure factors, are arranged in a hierarchy structure. Technical infrastructure systems play a fundamental role in urban function, while social infrastructure systems coordinate with each other through the service provided by technical ones. To validate this model structure, this study introduces the Bayesian network structure learning and adopts a likelihood score based structure learning method. Using open data from such as bureau of statistics and utility companies from 2447 zip code areas in California, U.S., this study selects 11 variables representing some components in the three major parts in the model. The score based structure learning method is carried out using the training set with size of 2,447 to discover the underlying dependency relationships among the 11 variables. The structure learning result shows to be consistent with the expected structure in the proposed conceptual model.