A Probabilistic Approach with Built-in Uncertainty Quantification for the Calibration of a Superelastic Constitutive Model from Full-field Strain Data

Abstract

Superelastic Nitinol is a widely used material to manufacture implantable medical devices. Simulations are extensively used in the design, optimization, and durability assessment of Nitinol implants. The constitutive response of superelastic materials is non-linear, anisotropic, and asymmetric in tension vs. compression. The existing methods to determine the superelastic material properties used in finite element simulations typically do not account for some of these complexities. The goal of this work is to introduce a method to determine the superelastic material constitutive properties using full-field surface strain measurements. We propose using digital image correlation to obtain surface strain fields during tensile testing of specially designed diamond specimens. The material properties are determined by minimizing a difference metric defined in terms of the local strain difference and the global load difference between the experiment and trial simulations. We also propose a metric in terms of the local strain difference between experimental measurement and simulation prediction to report the credibility of the material properties fitted using this method. The philosophy behind this method and the proposed metric to quantify the accuracy of material property fitting reflects the growing consensus in the implantable device community that the simulation results must be accompanied by a quantification of their credibility.