Prediction of the rupture pressure of a corroded pipeline by ANN

Laboratory of Modeling and Simulation of Intelligent Industrial Systems, ENSET MOHAMMEDIA, Hassan II University, Morocco

^* Corresponding author: yassine.elkiri-etu@etu.univh2c.ma

Abstract

This work highlights the use of artificial intelligence (AI) in fracture mechanics, in particular to solve complex problems such as the fracture of a corroded pipeline subjected to both internal pressure and axial compressive stress. The paper describes the use of artificial neural networks (ANN), a popular technique in AI, to replace the empirical expression of the DNV method with a system of simple equations based on weights and biases. The use of ANN avoids problems associated with theory, such as assumptions, boundary conditions and exact modeling. The choice of neural network structure was made on the basis of the required accuracy, measured by indicators such as the coefficient of determination R² and the root mean square error MSE during the validation phase. The results obtained with this neural network model were satisfactory, showing good linear correlation with the target values and low divergence during the validation phase. The implementation of this model in computer applications facilitates the prediction of failure pressure without requiring in-depth expertise in finite element analysis.