Progressive damage analysis of laminated dovetail specimens and calibration for damage models based on Bayesian inversion

Abstract

Composite fan blade is the preferred alternative for the fan stage of most advanced high bypass ratio turbofan engines. The dovetail part bears a significant centrifugal load, which is one of the key points of the multi-level “building block” approach to demonstrate compliance for special airworthiness regulation. However, due to a large number of ply drop-off layers inside, fabricating tolerances, as well as in-situ effects, a high uncertainty usually occurs in obtaining the as-manufactured damage variables. In this paper, monotonous static tensile tests are conducted on these specimens with dovetail element feature. Reaction forces are measured and combined with FE model, where Puck and Hashin failure criteria are employed to capture damage initiation and evolution of damage modes, along with deterministic cohesive zone materials to simulate two interfacial delamination cracks. Posterior probability distributions of in-plane damage variables are updated through constructing Gaussian process metamodels to surrogate forward models and likelihood functions in Bayesian inversion. The calibration results indicate that reaction forces locate within the posterior predictive distributions. Matrix tensile damage dominates the neck region. Material property degradation model shows a better performance in simulating progressive damage process over the continuum damage model. This study could promote the understanding of progressive damage mechanisms and ultimate strength prediction for dovetail elements.