PARAMETER CALIBRATION FOR CONTINUUM DAMAGE MECHANICS MODELS TO SIMULATE DUCTILE FRACTURE OF HIGH STRENGTH PIPELINE STEELS

摘要

The ability to arrest a running crack is one of the key features in the safe design of pipeline systems. In the industry design codes, the crack arrest properties of a pipeline should meet two requirements: crack propagation has to occur in a ductile fashion, and enough energy should be dissipated during propagation. While the first criterion is assessed by the Battelle Drop Weight Tear Test (BDWTT) at low temperatures, the latter requirement is converted into a lower bound for the impact energy absorbed during a Charpy V-notch (CVN) impact test. However, the introduction of high strength pipelines steels (X70 and beyond) has revealed that the commonly used relations based on BDWTT and CVN no longer hold. For such scenarios, Continuum Damage Mechanics (CDM) models provide promising potential to obtain a more profound understanding of the mechanisms that govern ductile crack propagation in high strength pipeline steels. In recent years, different types of CDM models have been used to simulate ductile fracture of pipeline steels. This paper focuses on two of these models, i.e. the Gurson-Tvergaard-Needleman (GTN) model and the Modified Bai-Wierzbicki (MBW) model. The GTN model is based on the computation of void growth according to Rice and Tracey, and account for the local softening of the material due to void nucleation, growth and subsequent coalescence. The MBW model is a fully coupled damage model, where the yield surface depends on both the stress triaxiality and the Lode angle. Although both models can predict ductile fracture propagation, their widespread application in pipeline design is hampered by the large number of input parameters to be calibrated. The GTN model requires 10 input parameters, i.e. 3 Tvergaard damage parameters, 4 porosity! parameters and 3 parameters to describe void nucleation. Whereas the Modified Mohr-Coulomb model originally proposed by Bai and Wierzbicki uses merely 2 parameters, the extended MBW model requires no less than 18 parameters to be calibrated: 11 plasticity parameters (5 stress + 3 strain rate + 3 temperature) and 7 damage parameters (4 initiation + 1 propagation + 2 failure). In this paper, different numerical/experimental strategies to calibrate these parameter sets are reviewed and compared. Sensitivity analyses are performed to assess the influence of the different input parameters on the model predictions. For both GTN and MBW models, the robustness and uniqueness of the calibrated parameter sets is investigated. Recommendations on optimum parameter values are derived, with special emphasis on high strength pipeline steels.