Identification of Model Parameter for the Simulation of SMA Structures Using Full Field Measurements

Abstract

With the design of new devices with complex geometry and to take advantage of their large recoverable strains, shape memory alloys components (SMA) are increasingly subjected to multiaxial loadings. The development process of SMA devices requires the prediction of their thermomechanical response, where the calibration of the material parameters for the numerical model is an important step. In this work, the parameters of a phenomenological model are extracted from multiaxial and heterogeneous tests carried out on specimens with the same thermomechanical loading history. Finite element analysis enables the computation of numerical strain fields using a thermodynamical constitutive model for shape memory alloys previously implemented in a finite element code. The strain fields computed numerically are compared with experimental ones obtained by DIC to find the model parameters which best matches experimental measurements using a newly developed parallelized mixed genetic/gradient-based optimization algorithm. These numerical simulations are carried out in parallel in a supercomputer to reduce the time necessary to identify the set of identified parameters. The major features of this new algorithm is its ability to identify material parameters of the thermomechanical behavior of shape memory alloys from full-field measurements for various loading conditions (different temperatures, multiaxial behavior, heterogeneous test configurations). It is demonstrated that model parameters for the simulation of SMA structures are thus obtained based on a reduced number of heterogeneous tests at different temperatures.