Curvic couplings in aero-engine rotors experience complex, spatially varying loads that can trigger multi-site fatigue failures with strong statistical dependencies. This study proposes a concise yet comprehensive reliability assessment framework that integrates a modified Smith-Watson-Topper (SWT) fatigue model with Bayesian-network calibration and Copula-based dependence modeling. The structural characteristics and dominant failure modes are analyzed to formulate a mechanism-informed life model, whose parameters are rigorously calibrated by fusing material characterization, sub-component, and component-like levels test data through a Bayesian network. To characterize correlation among multiple failure sites, a Copula-based reliability method is developed, enabling realistic prediction of joint failure probabilities. Application to a full-scale curvic couplings demonstrates that the proposed framework more accurately captures correlation-driven reliability evolution than traditional independent-site or cumulative-damage approaches. The results validate the effectiveness and engineering practicality of combining Bayesian-network-calibrated life modeling with Copula-based multi-site reliability assessment for curvic couplings.
WANG et al. (Wed,) studied this question.