Pipeline failure evaluation and prediction using failure probability and neural network based on measured data

Abstract

The chemical corrosion of metals in large industries such as oil and gas is a fundamental and costly problem. Gas transmission and distribution pipes and the other structures submerged in the soil and in an electrolyte, according to the existing conditions and according to the metallurgical structure, are corroded, and after a period of work, they disrupt an active system and process and lead to loss. The worst corrosion that occurs for metals embedded in the soil is where there are stray electric currents. Based on this, the cathodic protection of metal pipes is known as the most effective protection method to prevent the corrosion of structures buried in the ground, which is widely used to protect the corrosion distribution and transmission pipes of gas, oil, and water. In gas networks, current and voltage measurements for cathodic protection are carried out and recorded in specific periods according to the standards approved by the National Gas Company. The effect of stray currents on the obtained results is significant. The reason for this is that the available data is recorded as a time series, and as a result, the critical value of this time series will significantly impact the remaining life of the gas pipelines. Therefore, the purpose of this article is to investigate the stray currents effect on failure rate using normal probability distribution. In the following, the estimation of the remaining useful life of gas pipelines under cathodic protection is obtained using neural networks and compared with the results of the failure probability to check the accuracy of the results. According to the data history of the equipment, the amount of failure and the remaining useful life of the gas pipelines will be obtained.