SMA Embedded Panel Optimized Through a Genetic Approach

Abstract

The ability of modulating the dynamic response of structural elements may play a fundamental role in terms of noise and vibration propagation and reduction levels. Specifically, controlling some dynamic features, like stiffness and damping, may remarkably extend the working range of a specific component, with consequent integration and efficiency benefits. Smart Materials, combined with innovative design philosophies (i.e., ‘Self-Adaptive Structures’, ‘Self-Repairing Structures’...) gave rise to real possibilities for the implementation of non-conventional solutions. Within the noise and vibration field, a family of strategies, focused on damping (active constrained layer dampers, rheological layers) and stiffness (embedded Shape Memory Alloys, SMA, acting on the stress field) control, is developing, by giving birth to original and efficient solutions. SMA, due to their capability of transmitting large forces and deformations, and producing remarkable stiffness variations, represent good candidates for actuation problems and stiffness control solutions. The idea of using embedded SMA components to affect the structural dynamic response was already considered by several authors; among the others, Diodati and others focused their attention on the prediction of the effects due to the heat activation of SMA wires, embedded within a fiber-glass laminate. SMA induced stress originated significant FRF peaks shift, encouraging the authors to develop an optimization procedure to find out the most efficient placement and orientation of the active elements within the panel, aimed at maximizing the achievable frequency peak shift. In this article, a numerical model already introduced was examined and upgraded to suit the logic of a generic optimization process. The specific connections between the structure and the wires (sliding wires) was realized by proper constraint architecture, able to catch the best the physical nature of the mutual interaction. Then, due to the large amount of the parameters to be identified (in plane angle and location of each wire) and the non-continuous nature of some of them, a genetic optimization approach was picked up and implemented, assuming the peak shifts as the fitness function. The activated is then compared with the non-activated response, in order to estimate the attained performance.