A Large-Scale Analysis of Corrosion Data in High-Sulfur Natural Gas Purification Systems

Abstract

Corrosion is a significant environmental and safety concern of high-sour natural gas purification systems. However, complex pipeline configurations and corrosion mechanisms frequently lead to ineffective corrosion prevention and control in high-sulfur purification systems. To address this, artificial intelligence and data-driven approaches show significant promise. However, their practical applicability in large industrial purification systems remains insufficiently explored. In this study, a large-scale corrosion database containing 107,255 inspection records collected from 4,258 monitoring locations between 2018 and 2024 was established for a high-sulfur natural gas purification plant in China. A database incorporating time-series machine learning models was established for the rapid analysis and modeling of large corrosion data sets. The results indicate that inspection records are unevenly distributed across pipeline loops and component types, with monitoring data mainly concentrated in several key amine circulation pipelines. Most inspection points are located on straight pipes and elbows, while tees and reducers account for a relatively small proportion. A total of 21,463 valid time-series records were identified for forecasting analysis. Among the evaluated forecasting approaches, the ETS–Theta hybrid model exhibited the broadest applicability and achieved stable predictions in 94% of eligible monitoring locations. The model successfully identified high-corrosion-risk areas such as acid gas (AG), acid water (AW), and low-pressure condensate (LC). The prediction results were further validated through time-series cross-validation, semiempirical corrosion models, field investigations, and corrosion-product characterization. This study provides new insights into the actual corrosion data and the high-risk zones indicated by these data within high-sulfur gas field purification systems. Furthermore, the proposed framework establishes a practical basis for large-scale corrosion data governance, corrosion-risk identification, and intelligent corrosion management in high-sulfur natural gas purification systems.