Dynamic Bayesian network model to study under-deposit corrosion

This paper develops theoretical and mechanistic aspects of under-deposit corrosion (UDC) into a dynamic model and compares UDC progression and mitigation with current industrial practices. A Dynamic Bayesian Network (DBN) model is developed to understand different risk factors and their interdependencies in UDC and how the interaction of these risk factors leads to asset failure due to UDC. The DBN uses available UDC theory and transforms it into a probabilistic framework. The study compares DBN results from the general theory of UDC (based on the theory and mechanism of UDC) to industrial practices. The corrosion mechanism is represented in a Bayesian probabilistic framework involving solid deposits, flow velocity, operating pressure, under-deposit galvanic cell, chloride, pH, partial pressure of CO2, mono-ethylene glycol, and operating temperature. The Monte Carlo simulation (MCS) is used to characterize the stochastic properties of UDC. The proposed model assesses the asset failure probability over five years of continuous operation. DBN results show that they are consistent with failure probabilities data reported by the industry. Results also reveal that the pipeline wall thickness, outer pipe diameter, tensile strength, and operating pressure are critical contributors to UDC rate and asset failure likelihood. The results of this study are crucial for the inspection and maintenance schedule of pipelines affected by UDC.