A Flow-Field-Driven Zonal Framework for Mechanism-Based Corrosion-Hotspot Identification in Gas-Field Three-Phase Separators

Internal corrosion in gas-field separators is highly localized because flow, phase distribution, deposits, and water chemistry vary strongly over the vessel wall. This study presents a flow-field-driven zonal framework for corrosion-hotspot screening in horizontal three-phase separators. A finite-volume solver resolves local velocity, pressure, water-holdup, and wall-shear-related screening fields, while temperature and chemistry enter as prescribed operating/scenario inputs rather than independently validated local fields. The wall is divided into ten diagnostic zones using explicit numerical and geometry-based gates, and a gated phenomenological layer combines an engineering CO2-corrosion baseline with scenario-based susceptibility terms for erosion–corrosion, top-of-line condensation, under-deposit corrosion, concentration-cell corrosion, galvanic effects, and sulfate-reducing bacteria (SRB)-related microbiologically influenced corrosion. A revision-stage three-level resolution audit confirms stable bulk phase-placement indicators while revealing appreciable grid sensitivity in the finite-difference wall-shear estimate; wall shear is therefore used only as a relative localization indicator. Under the baseline scenario, the water-outlet and lower sediment-prone regions receive the highest screening attention; for the metering separator, the water-outlet mean screening output is 0.085 mm y−1, and the largest reported part maximum is 0.129 mm y−1. A structured sensitivity analysis shows that absolute magnitudes vary with the scenario factors while the principal vessel-part ranking remains unchanged over the tested perturbations. Two separator-body observations are used as qualitative same-domain consistency evidence. The framework is positioned as a physics-informed screening and inspection-prioritization tool; calibrated absolute-rate prediction requires spatially co-registered inspection histories.