A non-local GTN model with two length scales calibrated for a low alloy steel

This study applies a non-local integral Gurson-Tvergaard-Needleman model, incorporating two characteristic material length scales as regularization parameters, linked to shear deformation and hydrostatic tension, to simulate fracture in low alloy A533B. The model accounts for void growth, nucleation, and coalescence, including a shear-enhanced mechanism. Experiments under various stress triaxialities were used to assess model performance, including: uniaxial tensile test, fracture test, and simple shear test. The aim is to predict both global load-deformation responses and fracture modes using a single, physically motivated parameter set. While some parameters are well-established in the literature and linked to micro-structural observations and micro-mechanical analysis, others are less studied and require calibration against experiments. The calibration process is challenging due to nonlinear parameter interactions causing complex dependencies. A trade-off between competing effects was frequently observed. To address this, three parameter sets were investigated. Set 1 used physically motivated values and successfully reproduced the cup-cone fracture in round smooth bar, but underestimated the strains to failure for round smooth bar and modified Iosipescu specimens. Set 2 employed phenomenological values to closely match all the global load-deformation responses, but failed to replicate the expected fracture mode. Set 3 slightly adjusted parameters in Set 1, achieving improved failure strain predictions and a more accurate cup-cone pattern. • A non-local GTN model with two length scales applied to steel A533B • Length scales distinguish dilatational and deviatoric damage mechanisms • The model predicts load responses and failure modes in tension and shear • Transition from normal to shear failure is captured, e.g., cup-cone fractures • Three parameter sets show trade-offs between accuracy and physical basis