Data‐driven simulation of functional fatigue in shape memory alloy wires

Abstract

Data-driven mechanics offers great potential in engineering applications, where the efficient simulation of complex and, in particular, path-dependent material behavior is often challenging. Within this approach, conventional material models are replaced by data sets containing snapshots of stress, strain and the history of both assumed to be sufficiently accurate representations of the underlying material behavior. Based on these snapshots and on physical admissibility, a distance function is built which is minimized to yield the boundary value problems' solution. The aim of this paper is to apply this framework to accurately simulate the complex behavior of shape memory alloy wires under cyclic loading, under which these materials exhibit a degradation of their features, denoted functional fatigue. The constructed synthetic data sets are enriched by real experimental data, showcasing that the data-driven method is capable of combining both approaches of classic material modeling and modern methods which directly incorporate experimental data.