An intelligent model to predict the life condition of crude oil pipelines using artificial neural networks

Inefficient scheduling of a pipeline system may lead to severe degradation and substantial economic losses. Earlier studies mostly focussed on corrosion and statistical analysis. This study presents a novel approach for the prediction of life conditions and the classification of metal loss (ML) faults for a group of five pipeline sections of a pipeline system. An intelligent model is developed using artificial neural networks. The historical reports are grouped from the oil and gas industry located in Sudan. The results obtained by a proposed intelligent model are found to be satisfactory based on the highest coefficient of determination (R2) and the lowest mean squared error (MSE) values. The model developed with 12 number of hidden neurons accurately predicted the pipeline condition with an overall R2 value of 0.98148, 0.99359, 0.9943, 0.99336, and 0.99084 for the pipeline sections S1, S2, S3, S4, and S5, respectively. A sensitivity analysis has been carried out to understand the interrelationship between the factors affecting pipeline conditions for all sections of a pipeline system. The remaining useful life for all the sections of the pipeline system has been estimated, and a comparative analysis has been conducted in this work. The significant advantage of the present work is that the developed model can estimate the type of ML due to which the pipeline condition would mostly deteriorate. The deterioration profiles of the selected factors considered for this study have been generated, and the assessment scale has been designed. The proposed approach is more valuable in oil and gas industries to avoid unnecessary inspection costs and to plan the maintenance schedule. This study is progressively worthwhile to organize pipeline inspections and rehabilitation necessities.