RoundRobin Scientific Aim

Shape memory alloys (SMA) show highly nonlinear hysteretic stress-strain temperature responses due to martensitic phase transformations reversible up to about 5% tensile strain while reaching stresses of about 1GPa in thousands of cycles. This yields very large energy which can be utilized in smart structures for actuation, damping, impact absorption or simply for designing highly elastic (superelastic) members of engineering structures. They have been around for over five decades, and since that, many mechanics models capable of simulation of their response have been developed and published in the literature over the last two decades.

Nevertheless, when one tries to use these models to design or safely control smart structures, many common problems arise. While trying to identify the reasons for those problems and their best solutions within the S3T community involved in the smart structure design and control, it was very difficult to achieve a mutual agreement between individual teams developing various SMA models. The models and teams involved in the activity are best known by the names of their original proposers (Patoor-Ben Zineb, Aurichio, Lexcellent, Brinson, Lagoudas, Favier-Rio, Sittner-Sedlak).

Following 5 key issues were identified:

1. Functional responses of SMA elements are thermomechanical hysteretic in nature, which requires the SMA models to predict great variety of responses to simultaneously varying stress (strain) and temperature conditions. It is not sufficient to simulate mechanical response at constant temperature.
2. Frequently multiple very different deformation mechanisms proceed in a thermomechanically loaded SMA elements depending on the stress, strain and temperature which makes parameterisation and formulation of constitutive equation very complicated.
3. SMA models must be able to capture partial (internal) cycle responses of SMA elements in thermal, mechanical, thermomechanical loads originating from energy dissipation and related hysteresis and leading to path and history dependence of the simulated responses.
4. Modelling mechanical responses of SMA elements in which strain (stress) gradients exist remains to be a problem.
5. There is still a lack of good experimental data for general multiaxial loading conditions. This is true in particular for the cases of non-proportional loading, where phenomena as coupling between tension and torsion, path memory or ratchetting upon cyclic loading naturally appear in the absence of dislocation slip and must be captured by the constitutive equation.

Hence it was decided to perform the Roundrobin SMA modeling activity as a benchmark for the available SMA models which would test their ability to deal with the above 5 key issues simply by providing suitably selected set of experimental dataset to all participating teams, asking them to use their models to perform appropriate simulations and mutually compare the results. The goal was to map how these SMA models approach the problem and identify the common weak points and their consequences for the quality of simulations i.e. not to compare the mutually very different SMA models. The emphasis has been put on the SMA model development.

A single well defined SMA element exposed to simple thermomechanical loads in tension, torsion, combined tension/torsion has been selected for the benchmark dataset. In order to minimize the instabilities due to plastic deformation interfering with martensitic transformation upon cyclic loading a very strong thin superelastic medical grade NiTi wire was used in the tests.