Neural network modelling of high pressure CO2 corrosion in pipeline steels

The effect of carbon dioxide (CO2) corrosion on pipelines is of great relevance to the petroleum as well as the Carbon Capture and Storage (CCS) industries. CO2 corrosion is responsible for lost production as it brings about the gradual degradation of pipe internals with time. The cost of general corrosion is said to be between 3 to 5% of an industrialised nation's gross domestic product (Schmitt et al., 2009; Popoola et al., 2013). In the U.S., the cost of corrosion in the production and manufacturing sector was $34.4 billion in 2014, with the oil and gas industry accounting for more than half (Abbas, 2016). The use of neural networks (NN) as an analytic tool for corrosion data has been established however the aim of this paper is to characterise selected Matlab transfer and training functions, and assess their degree of suitability for CO2 corrosion rate prediction. Assessments of the training functions include the evaluation of the correlation coefficient (R-2-value) and determination of a cumulative absolute error to indicate the level of precision and the extent of model accuracy. A NN model is developed for predicting CO2 corrosion at high partial pressures by considering the results of the various tests and analyses on the given Matlab functions. The results showed that the model is reliable with all test results falling within the 95% confidence limits. Leave-One-Out Cross-Validation (LOOCV) was implemented as a means for carrying out an additional assessment on model performance as well as for model selection from possible alternatives. (C) 2018 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.