Development of a Predictive Computational Model for Corrosion Rate in Nigerian Crude Oil Transmission Pipelines Under Multiphase Flow Conditions

Internal corrosion under multiphase flow conditions remains the dominant cause of pipeline failure and hydrocarbon release in Nigeria's oil transmission network, yet existing mechanistic and empirical corrosion prediction models have not been calibrated against the specific physicochemical properties of Nigerian crude oils, produced waters, and operating conditions, severely limiting their predictive accuracy in Nigerian field applications. This study develops an original predictive computational model for corrosion rate in Nigerian crude oil transmission pipelines under multiphase flow conditions, integrating electrochemical corrosion mechanisms, fluid dynamics modelling, and machine learning ensemble methods. Experimental corrosion rate data are generated from a custom-designed multiphase flow loop operating with Nigerian crude oil, synthetic produced water of controlled ionic composition, and carbon dioxide and hydrogen sulphide partial pressures representative of fields in the Niger Delta and offshore shallow water blocks. Corrosion rates are measured by linear polarisation resistance, electrochemical impedance spectroscopy, and weight loss coupons at varying flow velocities, water cuts, temperatures, and pH levels. The mechanistic model incorporates de Waard-Milliams CO2 corrosion chemistry extended with hydrogen sulphide and organic acid contributions. A deep neural network ensemble trained on the experimental dataset and augmented with published field data provides a data-driven predictive layer. Bayesian uncertainty quantification provides confidence intervals for model predictions. The model is validated against independent field corrosion monitoring data from three operating pipelines in Rivers State and Delta State. An original corrosion prediction software tool is delivered as a practical output alongside the theoretical framework. Keywords: pipeline corrosion, multiphase flow, predictive modelling, crude oil, Niger Delta