Using narrow-gap Metal Active Gas (NG-MAG) arc welding, the low-temperature impact toughness and microstructures of the 316LN welded joints have been researched and characterized by Charpy impact test, optical microscopy, scanning electron microscopy and transmission electron microscopy. The results show that the low temperature toughness in the as-welded state is 132J and the toughness of the post-weld heat treatment was 112J. There are many dimples in the impact fracture, which indicated plastic fracture. The ferrite is located at the austenite grain boundary and exhibits a strip shape with a width of about 400 nm.

Mechanical scratches that penetrate the protective layers expose the steel substrate and establish a zinc–steel galvanic couple beneath an electrolyte film. This study develops a mechanism-oriented multiphysics model of scratch-width-dependent galvanic protection loss in hot-dip galvanized (HDG) steel enclosures. The model couples Butler–Volmer interfacial kinetics with equilibrium potentials related through the Nernst framework, Nernst–Planck transport of Zn²⁺, OH⁻, Na⁺, Cl⁻, and dissolved O₂, electrolyte charge conservation, reaction-derived boundary fluxes, a simplified corrosion-product deposition and transport-resistance treatment, and level-set tracking of interface evolution. Numerical field cases at scratch widths of 1, 3, and 5 mm show that widening the scratch increases the exposed-steel cathodic demand and lengthens the ionic-current path from the zinc edges to the scratch center. The resulting ohmic drop reduces the protective current and permits local iron dissolution when the center can no longer be maintained at a sufficiently negative potential. NSS morphology, cross-sectional SEM/EDS, corrosion-depth trends, and XRD observations show qualitative consistency with this mechanism. Under the investigated NSS conditions, a marked descriptive change occurs between the experimentally sampled widths of 2 and 3 mm, with 3 mm representing the first sampled condition showing pronounced protection loss. Because replicate-level corrosion-depth data and independent electrochemical measurements are unavailable, no statistical significance or quantitative model validation is claimed.