Numerical simulation of metal microstructure evolution is essential for material design and performance optimization. This paper reviews major simulation methods for key evolution mechanisms, including recrystallization, grain growth, slip, twinning, and phase transformation. The reviewed methods are classified into atomistic models, discrete-field models, and continuous-field models. Molecular dynamics (MD) is discussed as an independent atomistic approach, with emphasis on its role in revealing atomic-scale mechanisms, calibrating mesoscale parameters, and bridging atomistic, mesoscale, and continuum simulations. Discrete-field methods, including Monte Carlo, cellular automata, and vertex models, are compared with continuous-field methods, including artificial neural networks, phase field models, finite element methods, and level-set methods. Furthermore, a semi-quantitative evaluation matrix based on accuracy, computational cost, scalability, and applicability is established to clarify the practical trade-offs among different methods. The results show that no single method is universally optimal; instead, method selection should depend on the dominant physical mechanism, target length scale, required accuracy, and available computational resources. This review provides methodological guidance for multiscale microstructure simulation and supports future applications in precision machining, additive manufacturing, and process parameter optimization.
Liu et al. (Fri,) studied this question.