Multiphysics Vibroacoustic Simulation in EV Powertrains: A Comprehensive Review

The rapid adoption of electric vehicles has fundamentally altered noise, vibration, and harshness (NVH) requirements, as the absence of internal combustion engine noise exposes previously masked drivetrain excitations. In this context, vibroacoustic simulation has become a key enabler for achieving low-noise electric powertrains while reducing development time and physical prototyping. This review provides a comprehensive overview of multiphysics simulation methodologies applied to EV powertrains, covering the full excitation–response–radiation chain from electromagnetic motor forces and gear meshing dynamics to flexible multibody behavior, structural vibration, and acoustic radiation. The literature is systematically analyzed with respect to modeling approaches, numerical methods, and software workflows used to couple electromagnetic analysis, gear contact mechanics, multibody dynamics, finite element structural models, and acoustic FEM/BEM solvers. Particular attention is given to transmission error, time-varying mesh stiffness, and electromagnetic torque ripple as dominant tonal noise sources, as well as to the role of housing dynamics in sound radiation. The review highlights the strengths and limitations of time-domain and frequency-domain formulations, reduced-order models, and high-fidelity numerical simulations, emphasizing the trade-off between accuracy, computational cost, and practical applicability. Beyond summarizing existing methods, this paper critically discusses current limitations in predictive capability, including insufficient treatment of manufacturing variability, limited system-level validation, and the lack of standardized benchmark datasets. Emerging trends such as stochastic modeling, machine-learning-based surrogate models, and digital twin concepts are identified as promising directions to address these challenges. Overall, the review underscores that effective EV NVH prediction requires a holistic, system-level multiphysics approach in which electromagnetic, mechanical, structural, and acoustic phenomena are considered jointly rather than in isolation. From a knowledge-structuring perspective, the reviewed methodologies establish a clear conceptual mapping between classical NVH theory and electric powertrain–specific eNVH simulation. Fundamental concepts such as excitation–transfer–radiation paths, modal superposition, and frequency-order analysis remain valid, while their dominant sources shift from combustion-related mechanisms to electromagnetic forces and gear meshing phenomena. In this sense, electromagnetic excitation and transmission error can be interpreted as the primary counterparts of traditional engine orders in EV applications, propagated through flexible multibody and structural models toward acoustic radiation. This explicit linkage between established NVH principles and EV-specific excitation mechanisms provides a coherent framework that supports both human understanding and machine-learning-based knowledge extraction of multiphysics eNVH simulation workflows.