Machine Learning Prediction of Corrosion in Nano-Silica Epoxy Coatings for Nigerian Oilfield Applications
Domaine:
environment and energyType de record:
paper
Créateur:
B. E. C.
Éditeur:
SPE
Hôte:
Abstract
Corrosive failures of pipeline coating incur billions of dollars to Nigeria's oil and gas industry annually, but there is no predictive tool for nano-silica epoxy systems in real Niger Delta environments. This study bridges that gap. Five machine learning algorithms were trained and compared on a synthetic dataset of 10,000 scenarios — ranging in temperature to 80°C, chloride concentrations up to 35,000 ppm, and exposure to H2S. Gradient Boosting showed the best results (R2 = 0.924, RMSE = 42.5 mA/cm2); cross-validation confirmed no overfitting. Temperature, coating thickness, and nano-silica content were found to be the three most predominant predictors. The outcome is a usable predictive model providing a data-grounded basis for coating selection and maintenance planning which the current literature lacks.
Disclosure: Synthetic data generation and grammar assistance were provided using Claude AI, per SPE disclosure guidelines.
Similaires
Forecasting the strength of micro/nano silica in cementitious matrix by machine learning approachesImproved machine learning model for vehicle price prediction in the Nigerian economyInvestigating Machine Learning Applications in the Prediction of Occupational Injuries in South African National ParksA Machine Learning‑Based Business Analytics Framework for Customer Churn Prediction in Nigerian Retail SMEsGreen Nano Materials: Using Organic Nano Materials in Air Purification Techniques and ApplicationsApplications of Machine Learning in the Longevity Industry