Efficient modeling of imperfections for buckling analysis of composite cylindrical shells

Unstiffened composite cylindrical shells show a large scatter in the load levels that the structure can withstand before buckling occurs. Such scatter is greatly influenced by the unavoidable imperfections of the structure, introduced during the fabrication phase. It is thus of key importance to be able to accurately model such imperfections in a numerical computation, in order to recreate and predict the scatter shown in experimental buckling tests. The imperfections can be analyzed by means of random fields, inferring their statistical properties from available measurements. In this manuscript, evolutionary spectra are used to derive random fields of surface and material imperfections of cylindrical shells. A procedure based on a moving window averaging technique is proposed in order to accurately capture the variation of material properties due to imperfect thickness and laminate manufacturing. Finally, Monte Carlo simulations of compression and torsional buckling of cylinders are carried out to show the combined effect of surface and thickness imperfections in the scatter of the buckling limit load.