Integrated Multiaxial Experimentation and Constitutive Modeling

摘要

Modern plasticity models contain numerous parameters that no longer correlate directly to measurements, leading to a lack of uniqueness during parameter identification. This problem is exacerbated when using only uniaxial test data to populate a three-dimensional model. Parameter identification typically is performed after all experiments are completed, and experiments using different loading conditions are seldom conducted for validation. Experimental techniques and computational methods for parameter identification are sufficiently advanced to permit real-time integration of these processes. This work develops a methodology for integrating multiaxial experimentation with constitutive parameter calibration and validation. The integrated strategy provides a closed-loop autonomous experimental approach to parameter identification. A continuous identification process guides the experiment to improve correlation across the entire axial-torsional test domain. Upon completion of the interactive test, constitutive parameters are available immediately for use in finite element simulations of more complex geometries. The autonomous methodology is demonstrated through both analytical and physical experiments on Ti-6Al-4V. The proposed approach defines a framework for parameter identification based on complete coverage of the stress and strain spaces of interest, thereby providing greater model fidelity for simulations involving multiaxial stress states and cyclic loading.