Probabilistic analysis for the pre-stressed vibratory characteristics of composite fan blades based on a forward uncertainty propagation

Abstract

An investigation is conducted to ascertain the feasibility of using integrated probabilistic design to evaluate structural integrity of composite fan blades used in advanced large bypass ratio turbofan engines. A detailed probabilistic analysis is conducted based on multiscale uncertainty propagation to quantify the failure probabilities of design criteria against the static damage initiation and pre-stressed vibration resonance margins. Due to the geometry and laminate design complexity, this represents an extremely intricate but novel application for probabilistic analysis. Non-intrusive generalized polynomial chaos expansion (PCE), adaptive Kriging Monte Carlo simulation (AK-MCS), and field metamodel are established. A suitable compromise among affordable computational cost, prediction accuracy, and convenient implementation is fulfilled. For field quantities, the explained variance is evaluated for modal shapes by synthetic random field discretization. Uncertainty of ply stresses is evaluated through field statistical measures. It shows that fiber orientation explains 81.1% variance of ply stress field and covers a great layer extent, whereas correlation of ply thicknesses are rather small and has a slight influence on field variance. At last, static damage and resonance frequency design are fully illustrated from a probability insight. Results show the modal stress field is a non-Gaussian and non-stationary random process. Failure probability is 1.38 × 10⁻⁴ for static damage initiation, and the first vibration mode has the maximum resonance failure probability, which could help the engineer to understand the potential risks of structural design for composite fan blades.