Modeling and Prediction of Corrosion Penetration Rate in Crude Oil Pipelines Using Back Propagation Artificial Neural Network Approach

Abstract

ABSTRACT

Today in oil and gas fields, one of the most important crucial issue problem for any oil and gas industrial is the corrosion penetration rate (CPR)during crude oil transportation processes by pipeline that made of carbon steel.Many parameters have been known to be effective for corrosion control especially in the pipeline transportation process.  These parameters are pH, temperature, pressure and shear stress. 

Several researches have been done with these issues using different methods. In this study, the main issue is to implement back propagation artificial neural network approach to develop a strong and capable model that is able to give an accurate prediction values for CO2 corrosion penetration rate (CPR) under certain operating parameters.

Areliable model is developed to map inputs parameters namely pH, temperature,pressure and shear stresswith the outputs (CPR).The results from this prediction model showed that, with small set of examples, the back propagation network (BPN) was able to adjust its weight coefficients. Which means that, the input generated a proper output. Also, the (BPN) model developed was validate by means of calculating the mean absolute errors (MAE).  The value of (MAE) was 0.00457 mm/y which indicated the accuracy and reliability of the model.