Simulation of Cathodic Protection of Buried Steel Pipeline Under Coating Disbondments

Cathodic protection and coatings are common methods for protecting buried metal structures, including pipelines, in soil and water. According to standards such as NACE SP-0169 and ISO 15589-1, the structures in a well-designed cathodic protection system must be polarized to potentials below −0.85 V versus SHE (or −0.78 V versus SCE) to the maximum of −1.2 V versus copper/copper sulfate electrode. In such a system, the presence of the disbondment locations may result in localized corrosion such as crevice corrosion, pitting corrosion, and stress corrosion cracking. The most important factors affecting coating disbondment, especially at locations with higher solution resistivity, are potential gradient and current distribution, electrolyte resistivity, defect geometry, redox potential, and pH. The increase in the capabilities of available computer hardware and software has resulted in a rapid expansion of modeling techniques in various fields to solve complex systems such as the equations controlling corrosion processes, avoiding the economic and practical constraints of experimental studies. Several methods have been used for the simulation of such systems, the most important of which include finite difference method (FDM), boundary element method (BEM), finite volume method (FVM), and finite element method (FEM). Among these, the finite element method (FEM) has shown the most promise due to its versatility, accuracy, and high efficiency for complex problems. Numerous studies have attempted to simulate corrosion conditions in cathodic protection systems using various methods and their results are often comparable to and validated by the experimental data. However, to this day, a comprehensive model for these processes has not been introduced, which is, in part, a result of the complexity of corrosion systems. The recent advances in simulation methods and mathematical solutions as well as the access to the recent AI technology, using cloud computing to train machine-learning algorithms with a high degree of efficiency promises an increase in studies in this area and might result in a comprehensive model for these systems.