Multiscale modeling of NiTi-based shape memory alloys: From atomic mechanisms and microstructural evolution to macroscopic mechanical behaviors

Shape memory alloys (SMAs) have emerged as a core material system in the field of smart materials and structures due to their unique superelasticity, shape memory effect, and elastocaloric effect derived from thermoelastic martensitic transformation. However, their macroscopic nonlinearity, thermo-mechanical coupling, functional fatigue, and fracture behaviors intrinsically arise from the synergistic interplay of multi-level, non-equilibrium physical mechanisms, ranging from atomic-scale lattice shear and mesoscopic variant self-accommodation to macroscopic transformation band evolution. Single-scale experiments and simulations are insufficient to reveal the complete cross-scale correlations. In this work, the research progress and inherent logic of three core methodologies for the multiscale simulation of NiTi-based SMAs, i.e., molecular dynamics, phase-field, and finite element methods, are systematically synthesized, with the aim of constructing a theoretically closed loop from atomic mechanisms to engineering applications. Furthermore, the critical bottlenecks in current single-scale simulations regarding spatiotemporal resolution, complex mechanism coupling, and computational efficiency are analyzed. The key challenges including cross-scale parameter transfer, machine-learning-based interatomic potentials, and integrated processing-microstructure-property simulation chains are prospectively discussed, providing a systematic theoretical framework and methodological support for transitioning the SMAs from an empirical design to a predictive design.