This thesis addresses the issues surrounding the application of the finite element method to analyse composite structure repairs with an emphasis on aircraft applications. A comprehensive literature survey has been carried out for this purpose and the results are presented. A preliminary study and a comparative study of different modelling approaches have been completed. These studies aim to explore and identify the problems in modelling repairso n simplec ompositep anelsw ith speciala ttention given to adhesivem odelling. Three modelling approaches have been considered: Siener's model which is an extension of the traditional plane strain 2D model used for adhesively bonded joints, Bait's model which is a promising new approach and a full 3D model. These studies have shown that these methods are complementary providing a different insight into bonded repairs. They have also highlighted the need for a new modelling approach which will provide an overall view of bonded repairs. Improved modelling approachesh ave been developedf or externallyb onded patch and flush repairs. These models enable the study of adhesive failure as well as composite adherendf ailures.T hesea pproachesh aveb eena ppliedt o real repairs and the predicted results compared to experimental data. Four case studies have been conducted: external bonded patch repairs to composite plates, a scarf joint for bonded repairs, a flat panel repaired with a scarfed patch and a repaired curved panel. These case studies have shown that bonded repairs to composite structures can be analyseds uccessfullyu sing PC-basedc ommercialf inite elementc odes.

This thesis addresses the issues surrounding the application of the finite element method to analyse composite structure repairs with an emphasis on aircraft applications. A comprehensive literature survey has been carried out for this purpose and the results are presented. A preliminary study and a comparative study of different modelling approaches have been completed. These studies aim to explore and identify the problems in modelling repairso n simplec ompositep anelsw ith speciala ttention given to adhesivem odelling. Three modelling approaches have been considered: Siener's model which is an extension of the traditional plane strain 2D model used for adhesively bonded joints, Bait's model which is a promising new approach and a full 3D model. These studies have shown that these methods are complementary providing a different insight into bonded repairs. They have also highlighted the need for a new modelling approach which will provide an overall view of bonded repairs. Improved modelling approachesh ave been developedf or externallyb onded patch and flush repairs. These models enable the study of adhesive failure as well as composite adherendf ailures.T hesea pproachesh aveb eena ppliedt o real repairs and the predicted results compared to experimental data. Four case studies have been conducted: external bonded patch repairs to composite plates, a scarf joint for bonded repairs, a flat panel repaired with a scarfed patch and a repaired curved panel. These case studies have shown that bonded repairs to composite structures can be analyseds uccessfullyu sing PC-basedc ommercialf inite elementc odes.

No description available.

The purpose of this study was to gain a better understanding of the parameters affecting crack growth direction in unidirectional comµosite materials. To achieve this, the effect of anisotropy and biaxial loading on the direction of crack growth in unidirectional off-axis composite materials were investigated. Specific emphasis was placed on defining the crack tip stress field and finding a consistent criterion for predicting the direction of crack growth. Two models are presented to predict the crack tip stress field, an anisotropic elasticity solution and a singular isoparametric finite element formulation. After defining the crack tip stress field, three crack extension direction criteria, the Normal Stress Ratio, the Tensor Polynomial and the Strain Energy Density Criterion, were applied to predict the direction of crack extension. The theoretically predicted crack extension directions were then compared with experimental results. After comparison, it was determined that only the Normal Stress Ratio Criterion correctly predicts the direction of crack extension. / Master of Science

M.S.

published_or_final_version / abstract / Mechanical Engineering / Doctoral / Doctor of Philosophy

No description available.

No description available.

The work in this thesis concerns the estimation of the electromagnetic, elastodynamic and piezoelectric properties of homogenized composite materials (HCMs). A composite may be considered homogeneous if wavelengths are suficiently large in comparison to the size of the particles of each component material. This thesis examines HCMs constructed from two component materials and several methods of estimating the HCMs constitutive properties. Firstly, the Maxwell Garnett estimates and Bergman-Milton bounds on the electromagnetic properties of HCMs are examined. While both are widely used, we re-examine them, for isotropic dielectric HCMs, in light of recent advancements in material manufacture. Secondly, we examine the strong-property-fluctuation theory (SPFT). The SPFT estimate is calculated using iterations upon an initial ansatz, these iterations being dependent on statistical cumulants of the spatial distribution of the particles of the component materials. The zeroth-order SPFT estimate is identical to the first-order and both are taken to be identical to a comparison material. For the second-order estimate a two-point correlation function along with its associated correlation length are used to characterize the component materials' particle distribution. The general framework for the elastodynamic SPFT was established in 1999 by Zhuck and Lakhtakia. Here we further develop the elastodynamic SPFT for orthotropic HCMs, in order to undertake numerical studies. We simplify certain integrals in order to make them amenable to numerical computation. Also, we establish the piezoelectric SPFT for orthorhombic mm2 materials. The general theory is developed first in a manner analogous to the elastodynamic SPFT. We then implement a two-point covariance function, perform similar integral simplifications to those done in the elastodynamic SPFT and carry out numerical experiments. From the numerical studies it is clear that, for both the elastodynamic and piezoelectric HCMs, the lowest-order SPFT estimate is similar to that provided by the corresponding Mori-Tanaka formalism. It is also apparent that the second-order SPFT estimate provides a significant correction to the lowest-order estimate, which reflects dissipative losses due to scattering.

Submitted in fulfillment of the requirements for the degree of Doctor of Engineering: Mechanical Engineering, Durban University of Technology, Durban, South Africa, 2015. / Vacuum Infusion (VI) and Resin Transfer Moulding (RTM) are liquid composite moulding processes used in the manufacture of components from composite materials. The composite material in this case consists of a resin matrix combined with fibre reinforcement. In both moulding processes, a dry reinforcement preform is placed in the mould cavity and a liquid resin is introduced, driven by a pressure differential. Two rigid surfaces are used in RTM to create a fixed mould cavity. In contrast VI implements only one rigid surface and a flexible membrane or vacuum bag to form a non rigid cavity. The flexible cavity in VI influences and differentiates resin flow behaviour from that of RTM. Modelling resin flow enables the velocity, pressure and flow direction to be predicted. Resin flow in the RTM process is understood and modelled using Darcy’s law. However, flow in the VI process is not accurately modelled due to the added complexity introduced as a result of the flexible cavity. In the present work a novel approach was developed to investigate fluid flow in both processes. A unique experimental setup and testing procedure allowed for the direct comparison of fluid flow in RTM and VI. Identical flow parameters, conditions and preform construction were used in the assessment. The comparison isolated the effect of preform thickness variation as a differentiating factor influencing flow. From the experimentation, material behaviour was characterised and used to evaluate flow models for RTM and in particular VI. The model solutions were compared back to corresponding experiments. The pressure distribution behind the flow front, fill time and thickness behaviours were assessed. The pressure distribution / profiles behind the flow front of both VI and RTM were noted to be scalable with flow front progression. The profiles were curved in the VI experiments and linear in the RTM case. All VI models evaluated including the non accumulation based model accurately predicted the pressure distribution and consequently thickness variations in the VI tests. Fill times of the VI experiments were longer than that of the equivalent RTM tests. This behaviour is in contrast to previously interpreted fill time behaviour for the VI process based on VI models. It was also noted that the VI fill times were not only proportional to the square of the fill length, as in the RTM case, but also proportional to the square of the mass present. In addition, no significant accumulation was noted in the VI experiments. / D