Scale model testing can improve the cost-effectiveness of composite structures by reducing the reliance on full size component testing. Use of scale models requires the relationship be known between the responses of the small scale model and full size component. This relationship may be predicted by dimensional analysis or through mechanics formulations. The presence of physical constraints may prevent the complete reproduction of all responses in small scale models. Scaling relationships may not be available at the level necessary to predict all scaled responses. Investigations of the scalability of composite structures are needed in order to evaluate the reliability of small scale model predictions of the responses of full size components.
The scaling of the responses of graphite-epoxy laminated composite Z-section stiffeners subjected to uniaxial, compressive loading has been evaluated. The response regimes investigated are prebuckling, initial local buckling, postbuckling and crippling. A mechanistic approach to scaling has been used, in which the scalability of the responses has been judged relative to governing mechanics models. A linked-plate analytical model has been obtained which predicts the buckling loads, and from which two nondimensional load parameters have been obtained. The finite element method has been used for prediction of the buckling and postbuckling responses. The analytical and numerical analyses were used to define an experimental program involving fifty-two specimens of seventeen basic geometrical configurations and three stacking sequences.
The buckling, postbuckling and crippling responses were largely determined by the flange-to-web width ratio and both the absolute and relative values of the bending stiffnesses. Buckling loads increased with decreasing flange width and the laminate orthotropy ratio, and increasing flange-to-web corner radii and laminate thickness. The postbuckling load range was the greatest for specimens having wider flanges, but the failure stresses were greatest among the narrower specimens. The crippling mechanisms included flange free edge delamination at both nodal and anti-nodal axial positions, material crushing in the flange-to-web corner at nodal axial positions, and ply splitting in the flange-to-web corner at anti-nodal axial locations.
The constraint of the potted end supports of the experimental specimens was not scaled. The effect of displacements within the end supports was manifested by lower prebuckling axial stiffnesses than predicted based on the gage length properties alone. This phenomenon required a post-test adjustment to the data in order to permit comparisons of the experimental and finite element predictions of the response of the gage length on an equivalent basis. Once corrected, the prebuckling stiffnesses were generally observed to have scaled.
One of the nondimensional load parameters normalized the buckling loads for specimens of various web widths only. The second parameter normalized the buckling loads for all of the geometric and material variables contained in the model. This parameter also normalized the postbuckling loads, and is, therefore, a general nondimensional parameter for the buckling and postbuckling responses of the Z-section stiffeners. No scale effects were observed in the buckling response. The quality of the postbuckling load predictions degraded with the width of the postbuckling load range. It was not determined whether genuine scale effects were present in the postbuckling response or whether the observed error was a result of inadequate modelling of structural and material nonlinearities or other effects such as damage development in the specimens.
Good correlation between experimental and finite element predictions of the out-of-plane displacements and load-axis strains has been demonstrated. Predicted local material strain development has been related to the structural deformation characteristics. Consideration of individual strain values, however, could not predict which of several competing failure modes would determine the actual crippling response. Neither could the strain data provide any quantitative prediction of the crippling loads. Thus, the determination of strength scale effects is hindered by the complex structural-material interaction and the lack of a mechanics-based interactive failure model. / Ph. D.
| Identifer | oai:union.ndltd.org:VTETD/oai:vtechworks.lib.vt.edu:10919/39973 |
| Date | 19 October 2005 |
| Creators | Wieland, Todd M. |
| Contributors | Engineering Mechanics, Morton, John, Griffin, O. Hayden Jr., Johnson, Eric R., Mook, Dean T., Singh, Mahendra |
| Publisher | Virginia Tech |
| Source Sets | Virginia Tech Theses and Dissertation |
| Language | English |
| Detected Language | English |
| Type | Dissertation, Text |
| Format | vii, 202 leaves, BTD, application/pdf, application/pdf |
| Rights | In Copyright, http://rightsstatements.org/vocab/InC/1.0/ |
| Relation | OCLC# 25611952, LD5655.V856_1991.W562.pdf |
Page generated in 0.0024 seconds