Investigation of Stiffening and Curvature Effects on the Residual Strength of Composite Stiffened Panels with Large Transverse Notches

Enjuto, P., Walker, T.H., Lobo, M., Cregger, S.E., and Wanthal, S.P.

2018 AIAA/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference, AIAA SciTech Forum, (AIAA 2018-2251), 8 – 12 January 2018, Kissimmee, Florida.

Design of robust aircraft structure requires consideration of the load-carrying capability with large damage. Large notches, typically introduced as machined cracks (aka “notches”) severing a single skin bay and a central stiffening member, are often used to conservatively address the wide range of possible large damage scenarios. The objective of the current effort was to develop more generalized and rapid analysis methods addressing large-notch residual strength of stiffened panels to support preliminary design activities.

Predictive methods addressing the large-notch capability of metallic structure generally involve factoring the fracture capability of a flat, unstiffened panel using factors that address the configuration variables of the final structure (e.g., curvature, stiffening, etc.). These factors have been determined using a combination of numerical methods and experimental evidence.

Residual strength predictions of composite large-notch configurations are more challenging due to the complexity of the damage mechanisms and resulting trajectories, as well as the additional layup and stacking-sequence variables. As a result, progressive damage finite element (FE) analyses are often used for these predictions. Specifically, Boeing has successfully used laminate-level cohesive-zone models (CZM) with prescribed self-similar growth to address translaminar damage growth in the skin and stiffeners, together with cohesive or VCCT approaches to address skin/stiffener disbonding. This methodology has involved explicit modeling of the entire stiffener cross-sectional geometry.

The complexity and computational intensity of these strategies preclude their direct usage in preliminary design studies, where thousands of configurations may be evaluated. Instead, this methodology is used to analyze a range of configurations over the design space of interest, and response surface equations are developed to predict the residual strength. These response surfaces have been specific to a stiffener cross-sectional shape (i.e., hats, blades, I’s, bulbs, etc.), loading (i.e., longitudinal tension, longitudinal compression), and a set of design-variable limits. Parametric FE models are often developed and used in support of the response surface development to provide efficient, error-free model generation across a range of design geometries.

The response surfaces can be developed using standard response surface methods (Reference 1) that use linear combinations of polynomial terms addressing the main effects (i.e., design variables) and their interactions. One by-product of this approach is that rapid changes in the predicted residual strength can occur in the response surface beyond the limits of the underlying data, severely restricting their application to new design spaces. The resulting response-surface equations, while providing a good representation of the response over the design space of interest, provide little insight into the physical responses and their interactions.

The objective of the current effort was to develop more generalized and rapid analysis methods addressing large-notch residual strength of stiffened panels to support preliminary design activities. More specifically, the effort was limited to addressing the effects of curvature, central severed stiffener, and the first adjacent intact stiffener on the residual strength of panels with bonded or cocured stiffeners, subjected to uniaxial compression loading and exhibiting self-similar damage growth. By developing an improved understanding of key trends over broad ranges of the design variables, appropriate functional forms can be selected for use in the response surfaces, thereby ensuring applicability of the surfaces over a wide range of design variables typical of preliminary design.