Infrastructure interdependency modeling for cascading failure assessment

Abstract

As modern societies have become increasingly interdependent, trends toward complex large-scale networked systems are evident in every aspect of our lives. Yet, we are not ready to understand the interde-pendencies among systems, which result in what is termed unanticipated cascading failures. Such is a phe-nomenon in which failure in one infrastructure leads to series of failures in the others, generating a larger im-pact than expected originally. In an attempt to prevent unanticipated cascading failures, two types of infra-structure dependency are proposed in this paper: 1) dependency associated with the degradation of its capability, and 2) dependency related to the increase in demand it has to fulfill. They are named, respectively, capability-driven and demand-driven dependencies. Previous literature tended to address only one of these two dependencies, capability-driven dependency, even though disasters often emerge as a consequence of both dependencies. Thus, the purpose of this paper is to illuminate the benefit of modeling both dependencies by mathematically describing how these dependencies collectively contribute to the propagation of inoperability among infrastructures. The inoperability input-output model (IIM) enabled such modeling since it is figured out that demand-driven and supply-driven IIM show the propagation of inoperability through demand-driven and capability-driven dependency, respectively. However, the result of demand-driven and supply-driven IIM are viewed with a slightly different per-spective. In this paper, an infrastructure is viewed as a function serves to meet its demand with its capability. That is, if the demand it has to fulfill is greater than its capability, then it is inoperable. Hence, the level of inoperability propagated through both dependencies is determined by the difference between the demand- and supply-driven inoperability rather than the sum of them. As an illustration purpose, the propagation of inoperability am...