Thispaper describes part of a comprehensive joint-industry project on upheaval buckling.It develops a semiempirical simplified design method and detailed design methods based on a new numerical analysis, and illustrates their application by examples.It assesses alternative design strategies, and the implications of strain-based design.

No abstract

A re-examination of some 240 past experiments on the general elastic buckling of axially loaded stringer stiffened cylinders suggests that the observed buckling behaviours exhibited a degree of imperfection sensitivity considerably higher than that so often ascribed to them. It is demonstrated how the underestimation of these experimental knock-downs may have resulted from the use of theoretical models insufficiently representative of test conditions. By incorporating a more precise modelling of the test shells the present analyses show how, contrary to a widely held belief, the potential knock-down factors for general elastic buckling can approach those of the isotropic cylinder. Furthermore, despite the relatively perfect nature of many of the test shells, the recently described 'reduced stiffness analysis' method provides lower bounds of the observed experimental scatter that are sufficiently reliable to recommend this method as a basis for future design.

Re-examination of some 85 past experiments on the elastic buckling of axially loaded ring stiffened cylinders shows the existence of two distinctive behavioural regimes. Lightly stiffened cylinders, like isotropic cylinders, buckle into non-axisymmetric modes having long axial wavelengths at loads that are sensitive to the precise magnitudes of small initial imperfections. Heavily stiffened cylinders are characterised by snap buckling into axisymmetric modes which, for elastic behaviour at least, show only limited sensitivity to initial imperfections.Each of these characteristics is shown to be predicted by the recently developed 'reduced stiffness analysis' method, which, despite the relatively perfect nature of test specimens, predicts reliabk lower bounds to the experimental scatter. Taken together with the previously demonstrated empirical validity of the reduced stiffness analysis for the prediction of buckling modes and lower bounds to buckling loads for both isotropic and stringer Stiffened cylinders, the present comparisons provide further support to the recommendation that this method be considered as an alternative basis for future design.

No abstract