Computational & Technology Resources
an online resource for computational,
engineering & technology publications
PROCEEDINGS OF THE SIXTH INTERNATIONAL CONFERENCE ON ENGINEERING COMPUTATIONAL TECHNOLOGY
Edited by: M. Papadrakakis and B.H.V. Topping
Development of Multiscale Modeling Techniques for Composite Scarf Joints
Y.W. Kwon1, T.R. Greene1 and S. Bartlett2
1Department of Mechanical & Astronautical Engineering, Naval Postgraduate School, Monterey, California, United States of America
Y.W. Kwon, T.R. Greene, S. Bartlett, "Development of Multiscale Modeling Techniques for Composite Scarf Joints", in M. Papadrakakis, B.H.V. Topping, (Editors), "Proceedings of the Sixth International Conference on Engineering Computational Technology", Civil-Comp Press, Stirlingshire, UK, Paper 42, 2008. doi:10.4203/ccp.89.42
Keywords: composites, scarf joints, multi-scale model, interface strength.
Currently, the maritime industry renewed its interest in applying composite materials for primary structures in ship construction. In order to construct a large composite structure in a modular manner, scarf joints have been used. Scarf joints connect separate sections of a large solid-laminate composite structure while maintaining a constant laminate thickness. This way, a large structure is divided into more easily manufactured sections, with minimal structural losses, and without introducing more complicated joints. In addition, scarf joints can be used for repairing damaged composite sections. However, a drawback with the scarf joint is the reduced strength of the joint when one section is infused onto a previously manufactured section. Therefore, it is critical to determine the joint strength.
In this study, fracture mechanics-based multiscale computational modeling and simulation techniques were developed to predict tensile and compressive failure strength of composite scarf joints to a high degree of accuracy. Global, local, and element scale models were used in order to calculate the energy release rates at the scarf joints. Along with those three scale models, three different material models (isotropic, isotropic with resin interface layer, and orthotropic), two different initial assumed crack models (stepped and tapered), and six variations of failure criteria (mode I, mode II, mixed linear, mixed quadratic, bilinear, and interactive biquadratic) were evaluated to determine the choice of combination which best predicts failure strength in composite scarf joints constructed of E-glass fiber plain woven composites.
The study showed that explicit modeling of the resin layer at the scarf joint, where cracks initiate, was important for accurate prediction of the joint failure strength. In addition, the consideration of the joint interface slope in the fracture model was important especially for compressive joint failure strengths. In terms of the mixed failure criteria for crack propagation, the interactive biquadratic criterion was found to be useful for reliable prediction of joint failure strengths.
As a result, the best model for predicting tensile and compressive failure strength in composite scarf joints was the isotropic material model including a discrete interface resin layer, using a tapered initial assumed crack path and the interactive biquadratic failure criterion. For compressive loading, the fracture mode was the mode II while there was a mixed mode fracture for the tensile loading. Hence, the interactive biquadratic criterion became the pure mode II fracture criterion for the compressive loading. This combination model produced an average magnitude error of 10% across all geometric variations of scarf joints tested under tensile loading. The same model predicted the compressive joint failure strength with an 8% average error when compared to the experimental data.
These modeling techniques can be used to predict composite scarf joint failure strength analytically so as to eliminate or at least minimize expensive and time consuming experimental tests required for design and analysis of composite structures with scarf joints.
purchase the full-text of this paper (price £20)