Residual Stress Evolution in Machined Supermartensitic Steel under Corrosive Exposure: A Combined Experimental-Machine Learning Study

Abstract

Residual stresses and surface finish play an important role in the performance of materials used in the oil and gas industry, especially under combined mechanical loading and exposure to corrosive environments. This study investigates the residual stresses induced by grinding and polishing in supermartensitic steel (S13Cr) and their evolution during constant tensile loading in NaCl and H 2 S media. Residual stresses were measured by x-ray diffraction, and surface roughness and topography were characterized. Grinding produced higher compressive residual stresses, reaching approximately − 340 MPa, and rougher surfaces, with Ra = 0.64 µm. Polishing resulted in lower compressive residual stresses of − 200 MPa and smoother surfaces with Ra = 0.14 µm. The skewness parameter indicated that the polished surface (Rsk = − 8.95) is dominated by valleys, whereas the ground surface (Rsk = − 0.43) presents a more symmetric profile. Machine learning models captured the main stress evolution trends, with GPR showing the lowest test RMSE (22 MPa) and maximum error (35 MPa). The results highlight the combined effect of finishing, corrosive exposure and loading on residual stress evolution in S13Cr steel, and suggest that grinding provides a surface condition with greater resistance to corrosion due to its more favorable residual stress state.