Issue 78

G. Palomba et alii, Fracture and Structural Integrity, 78 (2026) 233-247; DOI: 10.3221/IGF-ESIS.78.14

Table 2: Summary of the Lopez-Fatemi and Kamaya methods for hardness-mechanical properties correlation. In the equations reported in Tab. 1, ε is the true strain, σ the true stress, E the elastic modulus, and n the strain-hardening exponent. Both methods were applied to AISI 316L non-welded specimens to generate theoretical true stress–strain curves. The two models were calibrated using experimental data on the base material, in order to select the most appropriate value for the plastic strain at the yielding point ε py . Values commonly found in literature do not refer to stainless steel, hence the analyzed case required a calibration on experimental results. In addition, comparison with experimental tensile data obtained from standard tests on a representative specimen as shown in Fig. 9, allowed the identification of differences between the two approaches. The Lopez model produced slightly higher strain-hardening at low strain, while the Kamaya model better represented the transition into the plastic regime.

Figure 9: Comparison between Lopez & Fatemi and Kamaya models for non-welded specimens.

Both analytical mathematical models were evaluated against the full-field experimental curve. The Kamaya framework demonstrates a strong global mathematical correlation with the experimental baseline curve, achieving a Coefficient of Determination (R 2 ) of 0.947 and a Root Mean Squared Error (RMSE) of 63.86 MPa. Although the model exhibits a slightly stiffer elasto-plastic transition within the intermediate strain range (up to0.20), it precisely tracks the material's true strain-hardening rate up to failure without predicting unphysical ductility. The Lopez & Fatemi analytical model yielded a lower correlation coefficient (R 2 = 0.933) and a similar RMSE, equal to 63.05. Crucially, as observed in the full-range stress-strain plot, the Lopez model overestimates plastic strain progression at moderate deformation levels and dramatically diverges past a true strain of 0.20, projecting an unphysical uniform elongation. The results obtained from the application of Kamaya’s model were therefore preferred for the subsequent steps of the analysis, which included the extension of the models to welded specimens. In particular, the values obtained from hardness measurements were used to retrieve true stress-strain curves of the different welding areas, hence material properties variation across the joint were mapped. As reported in Fig. 10, the HAZ curve was found to closely resemble the BM curve due to similar hardness values.

243