Advance Prediction Method of Failure Consequence for Natural Gas Pipeline Soil Corrosion Leakage

Run-time failure consequences play a key role in the risk assessment and layout planning of natural gas pipelines. However, previous work on the failure consequence predictions mostly focused on established pipeline networks. To overcome the hysteresis of the forecasting results, we present a model for calculating the failure consequence (CF), which is based on an inherent relationship between the soil corrosion grade and the economic classification of the failure consequence. The predictive model of the failure consequence (PF) is developed using a neural network (NN). The presented examples demonstrate the efficiency of the CF and PF based on the entropy weight method and three forecasting methods. Although the prediction accuracy of this model can only reach 20%, it proves the feasibility of a new approach to predict the consequences of failure in the planning stage. As a result, a new scheme and the flow process provide accurate advanced predictions for the pipeline corrosion failure consequence, and the method can be expanded to third-party damage and geological disasters.