Case study: Indonesian Natural Gas Pipeline Infrastructur

Natural gas transmission pipeline infrastructure owned by the state-owned enterprise in Indonesia was inspected using Intelligent Risk Assessment and Inspection (RAI). 354 piping, 19 filters, 23 vessels, 7 pig launchers, 7 pig receivers, 1 slug catcher, and 1 sump tank are assessed in this study. Due to a high number of infrastructure and the works of assessments is qualitative in nature, the assessments is prone to human biases and errors making the result may vary. This study aims to automate risk assessment and inspection more intelligent using computer algorithm, and can reduce dependency on human judgment. The use of some machine learning algorithms in risk assessment is demonstrated by using several types of classification algorithms i.e., logistic regression (LR), support vector machines (SVM), k-nearest neighbours (k-NN), random forests (RF), and decision tree (DT). The results reveal that the machine learning application could predict the risk levels, optimize inspection plans, and reduce inspection costs, and with random forests being the highest performing classifier.