Prediction of large-scale marine structures under extreme loading conditions (e.g. impact, fire) requires the use of computational tools that are both accurate and efficient. To resolve models at this scale, engineers normally use shell elements in which two of the (in-plane) dimensions are much larger than the third (thickness) one. Herein, we propose a consistent characterization framework for shell-element models commonly used in metal marine structures. The underlying model is based on a three stress-invariant plasticity model that accounts for the effects of stress triaxiality and shear-dominated stress states and accounts for the micromechanical void growth and coalescence that leads to fracture. This framework serves as both a constitutive model as well as a back-end calibration engine which outputs material cards that can be readily used with commercial finite element solvers like LS-DYNA and Abaqus. The calibration framework supports data-driven constitutive models that require engineers to prescribe the Fracture Locus (FL) and Instability Curves (IC) from experimental data, such as the Johnson-Cook model. Moreover, the calibration framework uses the underlying physics based model to computationally generate additional surrogate data that compliments the experimental testing. Thus, it allows the analyst to calibrate a material model based on limited experimental data, hence, reducing the costs associated with data acquisition and processing. A standalone calibration tool (VistaCal) that highlights the benefits of this characterization framework is also presented.