Skip to content

Latest commit

 

History

History
31 lines (19 loc) · 3.21 KB

README.md

File metadata and controls

31 lines (19 loc) · 3.21 KB

Nitinol Design Concepts

Welcome SMST 2017 attendees! You can find Confluent presentations from the conference at the nitinol.com reference library.

This project includes tutorials and examples related to design and simulation of superelastic nitinol components. This content is provided by Confluent Medical Technologies as a resource for our customers, industry and community. If you want to know more about the background, or how you can contribute, read about this project.

The material here includes a deeper dive into topics covered in our Nitinol University courses, as well as research supporting scientific presentations and publications. The Design Tutorial series provides an introduction to design and simulation of a nitinol component, following methods that are commonly applied in the medical device industry. The SMST-17 Volumetric Methods series ventures into more advanced topics related to volumetric and probabilistic methods for assessing fatigue durability, first presented in May 2017 at the SMST conference.

Design Tutorial

This first series follows each step in the design and analysis of a realistic (but non-proprietary) laser cut nitinol component, from designing the geometry using Solidworks to shape setting and fatigue cycling using Abaqus.

  • Design | Create a 3D model of a laser cut Open Frame component using Solidworks.
  • Mechanical Properties | Perform tensile testing to characterize mechanical properties for simulation.
  • Shape Setting | Expand the laser cut component into a complex expanded shape using Abaqus finite element analysis.
  • Fatigue Simulation | Apply fatigue loading conditions using Abaqus.

SMST-17 Volumetric Methods

Building on the foundation of the example design and simulation material above, these next topics explore new methodologies and analysis approaches presented in "Volume weighted probabilistic methods for nitinol lifetime prediction" at SMST-2017 in San Diego. You can find a recording of the talk on YouTube. The slides and animations are here on GitHub.

Cite this project

DOI