Prebuckling, Buckling, and Postbuckling Response of Segmented Circular Composite Cylinders
Discussed is a numerical and experimental characterization of the response of small-scale fiber-reinforced composite cylinders constructed to represent a fuselage design whereby the crown and keel consist of one laminate stacking sequence and the two sides consist of another laminate stacking sequence. This construction is referred to as a segmented cylinder. The response to uniform axial endshortening is discussed. Numerical solutions for the nonlinear prebuckling, buckling, and postbuckling responses are compared to experimental results. Focus is directed at the investigation of two specific cylinder configurations, referred to as axially-stiff and circumferentially-stiff cylinders. Small-scale cylinders, each having a nominal radius of 5 in., were fabricated on a mandrel by splicing adjacent segments together to form 0.5 in. overlaps. Finite-element models of both cylinder configurations, including the overlap regions, are developed using the STAGS finite-element code. Perfectly circular cylinder models are considered, as are models which include the measured geometry of the specimens as an imperfection. Prebuckling predictions show that the segmented cylinder response is characterized by the existence of circumferential displacement, and an axial boundary layer accompanied by circumferential gradients in radial displacement. Experimental measurements, taken with strain gages and displacement transducers, confirm these numerical findings. As the endshortening approaches the critical, or buckling, values, the response of the cylinders is characterized by wrinkling in the axial direction. In the axially-stiff cylinder, the crown and keel segments wrinkle, while in the circumferentially-stiff cylinder the side segments wrinkle. Experimental images taken from Moire interferometry show this response in the circumferentially-stiff cylinder. Four methods are used to predict the buckling values of endshortening and load for both cylinders, and the four values are in good agreement. The experimentally-measured buckling conditions, however, show that the models overpredict buckling values. For the axially-stiff cylinder, the difference could be due to the fact material failure not included in the model plays a role in the cylinder response. For the circumferentially-stiff cylinder, the difference is definitely due to material failure characteristics not included in the model. The predicted postbuckling response of the segmented cylinders is shown to be dominated by the existence of inward dimples in some or all of the segments. For the axially-stiff cylinder, the as-predicted dimpled crown and keel configuration is observed in the experiment but at a load 12 percent below predicted values. For the circumferentially-stiff cylinder material failure in the linear prebuckling range of response triggered buckling that resembled the predicted circumferential rings of dimples, but at a load 31 percent below predictions. Finally, it is shown that the effect of including the measured imperfections in the model has little observable effect on the circumferentially-stiff cylinder. For the axially-stiff cylinder the inclusion of the imperfections is found to effect the transition from buckling to postbuckling, but ultimately has little effect on postbuckling deformations.