Self-organized criticality study in natural gas pipeline systems: A system & data science approach

The natural gas pipeline system (NGPS) is a crucial component of the natural gas supply chain. Due to the complexity of its external environment, disturbances, topological and hydraulic characteristics, it is essential to study NGPS disaster mechanisms from a systems science perspective. Self-organized criticality (SOC), a widely used theory in systems science, has been applied to many network systems, but its application to NGPS is still in the early stages. One challenge is the lack of sufficient data on disturbance effects, which limits the study of SOC in NGPS. To address this, we propose a data augmentation model based on the Conditional Tabular Generative Adversarial Network (CTGAN) algorithm to overcome the shortage of gas transportation data. By augmenting practical data from open datasets in various ways, the results show that the CTGAN-based approach can capture key characteristics of gas transportation data, and the data size is from 333 to 18,545. Additionally, the augmented data reveal an atypical form of SOC in NGPS. The results show that there is a local range of gas transportation parameters where the NGPS has SOC characteristics, and the mean value of R2 is 0.92. However, the NGPS do not fit SOC characteristics in the whole range of gas transportation, where the mean value of R2 is 0.45. To further understand this atypical SOC, graph theory and network analysis are applied to divide the affected areas of NGPS, quantifying the disaster mechanism by measuring the propagation scale of disturbances. The results of this study offer a novel perspective by shifting the focus from the global system to specific affected areas, which is particularly useful for analyzing NGPS vulnerability, reliability, and resilience.