A novel fracture calibration method is presented for material models applied in finite element analysis of pipeline steels exposed to running ductile fracture. The calibration is based on the drop-weight tear test which is commonly applied for qualification of pipeline steels. The method is applied on three L450 steels with low, medium and high impact toughness. The calibrated fracture models are used in a numerical analysis of a full-scale fracture propagation test where the crack-driving force stems from a CO2-rich mixture that initially is in a dense phase. The results from the simulations are compared with experimental results.
Running ductile fracture is an event that must be considered in design of a pipeline. The problem was thoroughly studied at the Battelle Memorial Institute during the 1960s and 70s, and this resulted in the semi-empirical two-curve method, see Maxey (1974), which is applied in several codes and recommended practices, e.g. (ISO/TC 265, 2016, DNV-GL, 2017). However, for pipelines made of modern high-strength steels, pipelines transporting rich gases, and in particular pipelines transporting dense liquid-phase carbon dioxide, the two-curve method has proven to be inaccurate and non-conservative (Cosham et al., 2014, Leis, 2015, Biagio et al., 2017, Michal et al., 2018). To address and overcome this problem, numerical methods have been developed, e.g. (O'Donoghue et al., 1991, Shim et al., 2008, Meleddu et al., 2014, Nakai et al., 2016, Botros et al., 2018). Most numerical approaches, however, focus on natural gas and seldomly include backfill. Over the last years, SINTEF has developed a numerical tool that couples a Finite Element (FE) solver which calculates the response of the pipe and the backfill with a Computational Fluid Dynamics (CFD) solver that calculates the fluids decompression and the pressure distribution in the cross-section (Nordhagen et al., 2012, Aursand et al., 2016, Nordhagen et al., 2017). In our model we have chosen to discretize the pipe segments using shell elements with an approximate in-plane size equal to the wall thickness. With such a coarse discretization, it is impossible to capture the complex phenomenon occurring close to the crack-tip of a propagating fracture, such as tunneling and shear banding. However, the element size is sufficient to capture the local necking phenomenon and can give a good estimate of the material behavior at this length scale. To accomplish this, a good calibration routine for the material model, including fracture, is needed. In previous studies a calibration routine based on the Charpy V-notch test and a virtual uniaxial tensile test have been applied to calibrate the fracture criterion (Nordhagen et al., 2017, Gruben et al., 2018), but this approach predicted the crack to propagate faster than what was observed in the experiments for a high toughness steel, see Gruben et al. (2018). In contrast to the Charpy test with a 10x10mm cross section, the drop-weight tear test (DWTT) usually applies the full thickness of a pipe section. Due to the relatively large size of the specimen, the DWTT gives a good indication of the state of the material prior to fracture in a running fracture event in a pipe. In this study we present a hybrid experimental-numerical methodology for calibration of a propagating ductile fracture model for relatively large shell elements based on the DWTT set-up.