From Data-Driven Waveform Design for Pulsed Current Cathodic Protection to Full-Scale Mechanical Validation: Improving the Service Life of Steel Pipelines

• GA–GPR framework created an optimal pulsed-current cathodic protection waveform. • Evolutionary GAN augmented finite-difference data for robust machine learning. • Optimized PCCP lowered pit depth and density versus conventional direct current CP. • XPS showed higher Fe²⁺/Fe³⁺ ratio and denser carbonate film with optimized waveform. • Burst tests confirmed higher pressure capacity and improved pipeline integrity. Disbonded coatings form crevice-like environments in which conventional direct-current cathodic protection (DCCP) may not provide uniform protection, thereby promoting localized corrosion and compromising pipeline integrity. This study designs a pulsed-current cathodic protection (PCCP) waveform to enhance protection beneath disbonded coatings while preserving mechanical performance. A finite difference (FD) model was used to resolve transient potential and oxygen-transport behavior under various PCCP conditions. The resulting simulation dataset was used to train a Gaussian-process regression (GPR) model (with evolutionary generative adversarial network (EGAN)-assisted data augmentation), which was embedded in a genetic algorithm (GA) to optimize the PCCP waveform. The optimized PCCP regime was validated in large-scale soil-burial experiments on coated steel pipeline sections and benchmarked against DCCP. Surface-chemistry and pit-morphology assessments, together with full-scale hydrostatic burst tests, confirmed that waveform-optimized PCCP markedly reduced localized corrosion and better retained integrity. In particular, pit populations became sparser and shallower, the effective pit length decreased from 157.32 mm (DCCP) to 3.82 mm, and burst-pressure capacity was only decreased by a 0.15% over the tests period as opposed to a 13.12% decrement for the DCCP-protected section.