Identifying critical regions in industry infrastructure: A case study of a pipeline network in Kansas, USA

Abstract

In the face of the budget cuts and increased size of industry infrastructure, one of the top priorities for industry infrastructure protection is to identify critical regions by vulnerability analysis. Then, limited resources can be allocated to those critical regions. Unfortunately, difficulties can be observed in existing approaches of vulnerability analysis. Some of them are unavailable due to the insufficient data. Others are susceptible to human biases. Here, we propose an approach to overcome these difficulties based on the location data of failure events. The critical geographic regions are determined by the risk ranking of different candidate regions. Risk is calculated by integrating the probability of the failure event occurring (risk uncertainty) and total failure cost (the severity of failure consequences) in each candidate region. By changing the modeled object from the components to the region where the whole industry infrastructure is located, it collects the rarely failure events which are dispersed in different positions of the industry infrastructure to provide sufficient data, then the probability can be obtained by using a Poisson point process and kernel density estimation. Meanwhile, the application of hypothesis testing avoids the susceptibility of the approach to human biases by verifying the correctness of the assumptions used in the approach. Finally, a case study of this approach is performed on a pipeline network in Kansas, USA. In addition to the validation of the feasibility of our approach, risk uncertainty is proven to be less instructive for identifying critical regions than the severity of failure consequences.