Nonlinear Interdependency Measures for Infrastructure Systems: Case Study of Biofuel Infrastructure Development

Abstract

As infrastructure systems become more interconnected and interdependent, the function of one system influences the function of other related systems, which increases their vulnerability to cascading failures. Therefore, quantifying systems’ interdependence is important for assessing their performance. This study aims to enhance current methods, which quantify system interdependency as the linear association between subsystems, by applying measures for nonlinear dependency. Considering nonlinearity is challenging due to the large number of possible association patterns. As such, the aim of this paper is to address this knowledge gap by suggesting two different methods for quantifying the nonlinear associations between interdependent systems: distance correlation (DCOR) and randomized dependence coefficient (RDC). The suggested measures are tested using a case study of a biofuel infrastructure system composed of several interconnected subsystems. The system is modeled as an optimization problem, seeking the best decision variables that maximize the system profit. The results show that relying on linear association measures, as done in former studies, could be inadequate for quantifying interdependency. For example, the relationship between interdependency and system reliability was explored to show how conclusions may change when accounting for nonlinear interdependency.