Laminate optimization incorporating analysis and model parameter uncertainties for predictable failure

Abstract

Reliability of engineering systems are affected by uncertainties from four sources: inherent variability in the physical variables, modeling errors, estimation errors and human errors. Here the effect of modeling uncertainties on the predictability and robustness of progressive failure in optimized composites is studied. This paper investigates the effects of different modeling deficiencies such as uncertainties in the homogenization models used for stiffness calculations, in micromechanics models used for calculating the ply or constituent level stresses, in the failure criteria used for predicting failure, and in the models used to degrade ply properties after initial failure. Different optimized composite laminate designs are analyzed and compared to illustrate the interaction among model parameter uncertainties and design variability. A composite laminate is optimized in the presence of model uncertainties. It is shown that achieving a predictable failure sequence requires including model uncertainties along with variability in the reliability optimization.