Shape memory alloys (SMAs) are functional materials that can recover their initial shape in response to thermal and mechanical stimuli, driven by reversible martensitic phase transformations. SMAs have been widely used in a range of engineering applications, including aerospace, biomedical devices, automobiles, and structures with damping. This systematic review summarizes advances in computational methods and constitutive modeling for shape memory alloys, collected from IEEE, Scopus, and Web of Science from 2016 to 2025. The review covers fundamental functional properties of shape memory alloys, including the shape memory effect (one-way, two-way, and all-round shape memory effect), superelasticity, and high-damping shape memory alloys, along with their underlying phase transformation mechanisms. The review provides a critical analysis of constitutive modeling approaches at three distinct scales: microscopic, mesoscopic, and macroscopic phenomenological models for engineering design applications. Additionally, recent advances in additive manufacturing techniques, gradient microstructure engineering, and multi-scale computational frameworks that couple atomistic simulations with finite element analyses are highlighted. NiTi-based alloys are the most studied material phase, while other emerging SMAs, including Fe-based, Ti-Nb, and high-entropy alloys, are also discussed in the review. Despite substantial progress in the field, several critical challenges persist, including prohibitive computational costs of microscale models, limited integration between experimental validation and simulation predictions across different materials, inadequate understanding of fatigue damage mechanisms under cyclic loading, and the cross‑alloy transferability of unified constitutive frameworks for shape memory alloy modeling. This review primarily focuses on computational modeling of the shape memory effect (SME) in shape memory alloys, while also addressing closely related behaviors such as superelasticity and high-damping response, which share common transformation mechanisms and modeling frameworks.
Fu et al. (Fri,) studied this question.