Effect of Strain Rate on Nitinol Constitutive Modeling in the Clinically Relevant Strain Range

The stress-strain response of nitinol has been shown to be affected by the rate of loading. Nevertheless, most nitinol medical device stress analyses utilize material properties obtained from low strain-rate tensile testing. The purpose of this study was to evaluate the effect of cyclic strain rate on nitinol stress-strain behavior in the clinically relevant strain range, and to analyze the effect of incorporating observed variations in a finite element model of an apex specimen. Nitinol wire specimens with a one inch gage length were tested at a range of strain rates in both air and a water bath, both at 37 degrees C. The wire specimens were initially pulled to 6% strain and unloaded to 2% strain at a strain rate of 0.03 s(-1). Following this initial pull and unload, the specimens were cycled between 2% and 2.5% up to five times at strain rates of 0.3% s(-1), 0.7% s(-1), 1.3% s(-1), and 1.6% s(-1) (corresponding test frequencies of 0.6 Hz to 3.2 Hz). Three specimens were tested at each strain rate. Two finite element models of an apex specimen were then analyzed. The first model had uniform material properties fit to the slow strain rate results. The second model incorporated the observed strain-rate induced stress-strain variations by directly varying the material properties based on the strain distribution calculated in the uniform model. The resulting peak strain amplitudes under the same applied displacements were analyzed to evaluate the effect of the incorporated strain-rate behavior. The experimental testing demonstrated decreased stiffness with increasing strain rate for the wire specimens tested in the water bath, while no change in stiffness was observed when testing in air. When the observed stiffness decrease was incorporated into the apex specimen finite element model, an 11% increase in strain amplitude was observed due to strain localization in the softer regions. These results suggest that strain levels at in vivo loading rates may be higher than those predicted by finite element models that utilize constitutive models derived from slow strain-rate tensile tests.