Finite Element Model Updating for Rotating Machinery: Methods, Applications, and Future Directions—A Review

Abstract

Finite element model updating (FEMU) provides a physics-based approach for reducing discrepancies between numerical models and measured responses by estimating uncertain physical parameters. Its application to rotating machinery is challenging because rotor dynamics are strongly affected by rotational speed, bearing and support properties, nonlinear interactions, operating conditions, measurement limitations, and parameter correlation. This review examines FEMU methods and applications for rotating machinery, distinguishing direct FEMU studies from nonlinear, uncertainty, surrogate, and AI-based studies that primarily provide enabling techniques. Direct matrix correction, sensitivity-based, optimization-based, surrogate-assisted, Bayesian, AI-driven, and hybrid approaches are compared in terms of physical interpretability, identifiability, computational demand, uncertainty treatment, and validation. The reviewed literature indicates that FEMU is most mature for rotor–bearing calibration, bearing and support parameter identification, and operational-response-based updating, whereas experimentally validated inverse estimation of nonlinear and compound-fault parameters remains limited. Based on these findings, a lifecycle-oriented FEMU framework and research roadmap are proposed, emphasizing multi-condition identifiability, uncertainty-aware updating, computational efficiency, independent validation, and governed synchronization. Surrogate and AI-assisted estimation should remain connected to validated high-fidelity physical models and defined operating domains for credible condition assessment and digital-twin applications.