A virtual environment for resilient infrastructure modeling and design

Abstract

This thesis considers the interoperability of recent modeling efforts that apply constrained optimization (combined with representations of system function and management) to assess and improve the operational resilience of critical infrastructure (CI) systems to disruptive events. We implement these mathematical models using the Pyomo optimization package, which is built on top of the Python programming language. This computational environment provides advantages for data preprocessing and postprocessing, including convenient and efficient methods for manipulating CI network data. Moreover, the object-oriented nature of Pyomo creates a natural means for representing interdependent CI systems. Specifically, the model for each CI system can be implemented as its own object, and the combined model can be implemented as another object built from its dependent components. This allows for increased flexibility and extensibility beyond previous implementations. We manage the inputs and outputs of the models in a way to be able to compare them across studies, obtaining insight on their performance, interactions, and effectiveness. This thesis supports a broader effort to build a repository of functional CI models enabled from a geospatial user interface and connected to a common, backend simulation engine.