A Stent is a small expandable tube used to treat various clinical conditions like narrowed or weakened arteries in the body. The design process of such a device has to achieve a goal of withstanding a cyclic pulsative loads for a period of ten years. This requirement for proper functioning under high cycle fatigue condition may be very difficult to attain especially since the design should take into account vast operational conditions due to variability in patient`s artery geometry and stiffness and due to variability in device manufacturing and surgical placement. Therefore, a repetitive cyclic complex task of CAD geometry redesign and successive finite element analysis is required as a crucial step in product development. In a single design verification cycle the product analyst has to vary several geometrical design factors of the stent, prepare an adequate computational mesh and to further define the dominant physics phenomena in a finite element analysis (FEA). Then, relying on the analysis result, he has to estimate the output performance variables like radial forces during deployment and various safety factors. This unavoidable manual optimization process may consume a significant portion of the device development period. n this work we have implemented an automatic design optimization process to stent design. As a first stage we have used an automatic design exploration method that robustically varies several geometric factors. We have then used a non-parametric geometric optimization approach that smoothly reshaped the stent geometry in order to further reduce the maximum stresses and to achieve stress homogenization. As a final stage, an extensive reliability analysis procedure has been used to account for variability in artery dimensions. An approximated surface model was constructed using radial basis functions. This approximated model allowed to be used very efficiently in a statistical reliability assessment using many Monte Carlo simulations.
