Probabilistic Buckling Analysis of Composite and Sandwich Cylindrical Shells

The interest in the characterization of the safety and the robustness of design of thin-walled cylindrical shells has led to an increased attention in the reliability. A probabilistic approach to the buckling behavior was developed in order to achieve reliability based knockdown factor. The Monte Carlo method is applied with the aim to generate a data set of knockdown factor, that is used to obtain the reliability as continuous function. The reliability function allows to estimate the knockdown factor for different reliability level and 95% confidence level. Latin Hypercube sampling with 100 sample size is used to reduce time and computational cost. Three examples regarding the application of the procedure to two laminated composite shells and a sandwich shell are discussed. The sources of the initial imperfections are considered to be the uncertainty of the material properties, the layup, the load, and the geometric imperfections. The influence of the scatter of input parameters is investigated and the results about the buckling response are compared with the NASA guidelines.