Developing digital twins of infrastructure for risk analysis

Abstract

Infrastructure are vulnerable to natural and anthropogenic hazards. The general steps in the risk analysis of infrastructure include modeling the undamaged network, modeling the hazard, predicting the damage to components, and assessing the functionality of the network given the predicted damage. However, these steps require information about the infrastructure that might not be available. This work develops a general procedure to generate representative and functionally equivalent models (i.e., digital twins) of infrastructure. The proposed procedure is systematic and uses detailed geospatial information that are generally available. By using only these data, representative networks can be generated for most locations of interest. We also develop a general modeling approach to assess the impact of localized damage on infrastructure networks. We consider the uncertainty in the network topology by generating multiple representative network realizations. Finally, we demonstrate the proposed procedures by generating representative wastewater networks and assessing the impact of localized damage on the network for a case study location.