A rather thorough and novel buckling analysis of an axially-loaded orthotropic circular cylindrical shell is formulated. The analysis assumes prebuckling rotations are negligible and uses a unique re-defining of the orthotropic material properties in terms of a so-called geometric mean isotropic (GMI) material. Closed-form expressions for the buckling stress in terms of cylinder geometry and orthotropic material properties are presented, the particular closed form depending on the specific character of the orthotropic material relative to the GMI material. With the formulation, the specific character of the buckling deformations - e.g., axisymmetric or nonaxisymmetric, the number of axial and circumferential waves - can be established. By using the maximum radius of curvature of an elliptical cross section in this formulation, the analysis is used to demonstrate the detrimental effects of an elliptical cross section on axial buckling capacity when compared to a circular cross section with the same circumference. Using the circumferentially-varying radius of curvature of an elliptical cross section, the analysis is then further used as the basis for developing two methods for improving the axial buckling capacity of elliptical cylinders. The first approach involves varying the wall thickness of an isotropic elliptical cylinder with circumferential position. Uniformly stable elliptical cross sections which preserve the same critical stress, critical load, or volume of an axially loaded circular cylinder of the same circumference are designed with the formulation. The second approach involves maintaining a uniform wall thickness but varying the orthotropic material properties with circumferential position. This approach is applied to a cylindrical lattice structure where it is assumed that the ribs are dense enough to be able to describe the lattice structure by means of an equivalent homogenized material. The orthotropic properties of the homogenized material are varied by varying the lattice rib angle with circumferential position. Considerable recovery of the axial buckling capacity of the variable-rib-angle design elliptical cylinder compared to the same cylinder constructed in isogrid fashion is demonstrated. In fact, recovery relative to an isogrid circular cylinder of the same circumference is demonstrated. For both approaches confirming finite element models are used to verify the findings. The two different approaches are compared, and finally the two approaches are recognized as special cases of a more general design philosophy. / Ph. D.

In this study, a progressive failure analysis is used to investigate leakage in internally pressurized non-circular composite cylinders. This type of approach accounts for the localized loss of stiffness when material failure occurs at some location in a structure by degrading the local material elastic properties by a certain factor. The manner in which this degradation of material properties takes place depends on the failure modes, which are determined by the application of a failure criterion. The finite-element code STAGS, which has the capability to perform progressive failure analysis using different degradation schemes and failure criteria, is utilized to analyze laboratory scale, graphite-epoxy, elliptical cylinders with quasi-isotropic, circumferentially-stiff, and axially-stiff material orthotropies. The results are divided into two parts. The first part shows that leakage, which is assumed to develop if there is material failure in every layer at some axial and circumferential location within the cylinder, does not occur without failure of fibers. Moreover before fibers begin to fail, only matrix tensile failures, or matrix cracking, takes place, and at least one layer in all three cylinders studied remain uncracked, preventing the formation of a leakage path. That determination is corroborated by the use of different degradation schemes and various failure criteria. Among the degradation schemes investigated are the degradation of different engineering properties, the use of various degradation factors, the recursive or non-recursive degradation of the engineering properties, and the degradation of material properties using different computational approaches. The failure criteria used in the analysis include the noninteractive maximum stress criterion and the interactive Hashin and Tsai-Wu criteria. The second part of the results shows that leakage occurs due to a combination of matrix tensile and compressive, fiber tensile and compressive, and inplane shear failure modes in all three cylinders. Leakage develops after a relatively low amount of fiber damage, at about the same pressure for three material orthotropies, and at approximately the same location. / Master of Science