Developing a Validated FEA Model of Thin Nitinol Wire in Bending Using Results From Digital Image Correlation and Load-Deflection Testing

Thursday, May 23, 2013: 16:45
Congress Hall 1 (OREA Pryamida Hotel)
Mr. David R Bow , University of Strathclyde, Glasgow, Scotland
Mr. Robbie Brodie , University of Strathclyde, Glasgow, Scotland
Prof. David Nash , University of Strathclyde, Glasgow, Scotland
Dr. William Dempster , University of Strathclyde, Glasgow, Scotland
In order to design ring stent medical devices using a finite element approach, a validated material model is crucial in representing the mechanical behaviour of the system. Such a validated approach can be achieved by simulating the candidate material, Nitinol, in bending and comparing with physical Digital Image Correlation (DIC) analysis and load-deflection bend test results.

A geometrically representative FEA model was created using non-linear approach with large deflections taking account of the thin flexible wire in the 3-point bend, 4-point bend and free bend test setups, with appropriate boundary conditions and contact conditions.

A simplified superelastic material model, using the ABAQUS UMAT model based on Auricchio & Taylor’s work, was then assigned using parameters derived from basic uniaxial tensile and compressive testing of the candidate Nitinol wire. The simulated bending results were then compared against the DIC test results for inner and outer surface strains and neutral axis position, as well as the load-deflection in bending results.  Subsequently, an investigation into improving the performance of the numerical model, which uses a scaling factor to derive the compressive behaviour from the tensile, was undertaken by varying the material model input parameters.  The potential to improve results by creating independent tensile and compressive UMAT material models for the elements which are predicted to be in superelastic tension and superelastic compression respectively was also investigated.  It is shown that the limitation of deriving the majority of the compressive behaviour from the tensile characteristics has a considerable influence on the accuracy of load deflection results in bending, and these can be improved by independently modelling compressive and tensile superelastic regions.