Calibration of the ductile damage model and comparison with the GTN model for high-toughness API X65 steel

Abstract

In this study, the material used was an API X65 sour steel. This steel exhibits high ductility, with a total fracture strain greater than 100% in a standard tensile test. Due to its high ductility and toughness, this steel is widely used in the oil and gas industry for pipeline applications. The Charpy V-Notch (CVN) test was selected for this study due to its wide applicability, low cost, and simplicity. In this test, the specimen is loaded under three-point bending. Several models for modeling ductile failure are available in the literature, the most commonly used being the Gurson-Tveergard-Needleman (GTN). In this work, the Ductile Damage (DD) model was applied to numerically reproduce the CVN test response for the X65 steel. This model is simpler than the GTN, requiring only two parameters: one describing the onset of damage and the other describing the damage evolution. This makes it an attractive choice for practical applications. The main issue of the DD model is determining the fracture onset to obtain its first parameter, particularly when used in tensile test, since fracture initiates at the center of the specimen. A methodology was proposed to calibrate the DD model using seven modified tensile geometries, with different stress triaxiality, to determine the first parameter. Subsequently, the second parameter was calibrated by varying its value until the curve was reproduced. The calibration using the GTN model was also performed to serve as a benchmark for comparison, since it is widely applied to ductile materials. The calibration results obtained with both models for the CVN geometry were compared. The reproduction of the CVN test curve was slightly better for the DD model than for the GTN. The maximum error for both models occurred around 10 mm of hammer displacement; for the GTN it was 13.7% and for the DD it was 7.7%. Additionally, the fracture morphology of the CVN specimen was also better described using the DD model than with the GTN model. In terms of fracture morphology, the remaining ligament exhibited the largest discrepancies. For the GTN, the smallest error was 11.95% and for the DD it was 7.86%. Considering the absorbed energy, the errors were smaller, 3% for the GTN and 1% for the DD. Overall, the results demonstrate the strong potential of the DD model and the proposed calibration methodology.