This research presents a quantitative simulation framework for dynamic recrystallization (DRX) during ultrasonic nanocrystal surface modification (UNSM) of WE54 Mg alloy. Isothermal hot compression tests were conducted up to a true strain of 0.7 across temperatures of 350 ℃ to 550 ℃ and strain rates of 10−3 to 101 s⁻¹ . The developed strain-compensated constitutive model yielded flow stress predictions with deviations of less than 5% from experimental data, establishing a robust critical-to-peak strain relationship of εc = 0.77 εp. A mathematical DRX kinetic model was integrated with finite element method (FEM) and Monte Carlo (MC) simulations to predict microstructural evolution. The FEM utilized an optimal mesh size of 0.4 and a mass scaling factor of 1000, ensuring the kinetic energy remained strictly below 10% of the internal energy to maintain quasi-static fidelity. The UNSM treatment, applying a 20 kHz frequency, 30 µm amplitude, and 15 N impact stress, drove microstructural evolution governed by a determined hardening constant, k₁, of 2.8 × 10⁻⁹ m−1. The coupled FEM-MC algorithm accurately predicted the final average grain sizes, exhibiting minimal error margins ranging from 1.5% to 11% when compared to experimental measurements across varying strain levels. Furthermore, phase characterization confirmed that Mg24Y5 and Mg41Nd5 precipitates govern the DRX kinetics via Zener pinning. The novelty of this study lies in the establishment of a multiscale architecture that captures the unique microstructural response of RE-containing Mg alloys by coupling FEM-derived stored energy fields with probabilistic MC grain evolution. This hybridization successfully bridges the gap between macroscopic processing parameters and mesoscopic gradient architectures, enabling the predictive design of surface nanocrystalline layers in complex precipitation-hardened systems.

Keywords