This study investigates the effect of micro-scale geometric and material property uncertainties on the elastic properties and reliability of fibre reinforced composite materials. Composite materials are often designed using conservative design factors to account for a limited understanding of how multi-scale uncertainties effect reliability. Structural reliability analysis can produce more efficient designs, but requires an understanding of how all sources uncertainty effect probability of failure. Previous studies have not considered micro-scale geometrical uncertainties and their combinations in a multi-scale probabilistic-based reliability framework. Thus, this study will investigate the effect of numerous combinations of micro-scale material property and geometric uncertainties on the homogenised elastic properties. Furthermore, to account for the effect in a reliability-based framework, a novel surrogate modelling technique is developed to represent the uncertainties efficiently. The study concluded that the geometrical fibre stacking uncertainty is as influential as the widely investigated constituent material stiffness uncertainties. Consequently, representing the micro-scale geometric uncertainties within the developed multi-scale probabilistic-based framework improves the estimated stiffness. Thus probability of failure is reduced, compared with considering material property uncertainties only. Moreover, the framework clarified and highlighted the importance of representing fibre geometrical stacking uncertainty for a deeper understanding of their effect on composite stiffness properties.
- surrogate model