|dc.description.abstract||Welded joints are considered to be one of the most critical locations in structural components from a fatigue perspective because of high stress concentrations near the weld toe region, the presence of tensile residual stresses, and defects from the welding process. New welding techniques like friction stir welding (FSW) and post-weld treatment technologies like high-frequency mechanical impact (HFMI) treatment have a strong potential to improve the fatigue performance of welded joints. However, it is essential to carefully examine the effectiveness and limitations of these new welding techniques and treatment technologies to ensure their reliable fatigue performance in service. Often, new technology is not employed until it has been proven to be reliable through years of performance under real-life service conditions. The design of welded joints fabricated using new technologies poses another challenge for structural engineers, which is, how to design a component for which there are no design codes or code-specified quality control criteria. In the absence of such design codes, designers often refer to non-compulsory guidelines, which may only be applicable to components fabricated with older technologies. This can result in overly conservative designs. In the absence of specific quality control criteria for components fabricated with new welding technologies like FSW, existing design codes usually recommend quality control criteria based on “best practice” rather than relating defect size to fatigue performance. Against this background, this thesis aims to study FSW joints and HFMI treated joints, from a fracture mechanics perspective, which will contribute to the development of performance-based design provisions and quality control criteria for welds employing these technologies.
FSW joints have been found to have better fatigue performance than arc welded joints. While the tolerance window for the FSW process is wide, there is a possibility of having defects in these joints, which can severely affect the fatigue performance. In this study, a comprehensive testing program was carried out to study the fatigue performance of FSW joints with intentionally introduced defects including angular misalignment, toe flash, lack of penetration or “kissing bond”, and wormhole defects. As fatigue testing becomes time-consuming and expensive, numerical modelling and simulation provide complementary ways to assess the effects of parameter variations on fatigue performance. With this in mind, a previously-developed strain-based fracture mechanics (SBFM) model is improved and extended in this thesis to study the fatigue behavior of FSW aluminum joints.
In its previous form, the employed SBFM model was capable of performing a one-dimensional (1D) crack propagation analysis. For each crack size, the crack shape was allowed to evolve using a pre-defined crack shape evolution function. In the current work, the existing 1D model was first programmed in MATLAB and then improvements in the existing model related to failure criteria were made. Subsequently, the model was extended to perform 2D fracture mechanics analysis. This improved 2D SBFM model is applied to assess the fatigue behaviour of HFMI treated A514 steel and 5083 aluminum welds (welded using metal inert gas welding process). Fatigue tests of as-welded and HFMI treated specimens were carried out to validate the prediction capability of the 2D SBFM model. A comprehensive material testing program was also carried out to estimate the input parameters required by the 2D model. With inputs obtained from material tests, the 2D model shows a reasonably good agreement between the fatigue life obtained from the model and the experiments. A sensitivity analysis is performed with the 2D model to identify the most important parameters, which affect the behaviour of HFMI treated welds.
Following the deterministic SBFM analysis of FSW and HFMI treated joints, the 2D SBFM model is extended to a probabilistic framework to obtain probabilistic stress-life curves (i.e. curves associated with a specific survival probability). To do this, statistical distributions of the input parameters are first defined. The resulting probabilistic stress-life curves are then compared with the available design curves and the differences are highlighted.
The presented probabilistic analysis demonstrates how the 2D SBFM model can serve as a useful analytical tool for developing quality control guidelines and reliability-based design curves for the HFMI treatment technology, which is applicable to a broad range of materials (e.g. various grades of steel and aluminum), scales, and cyclic loading conditions, beyond what can be practically investigated in a purely experimental program.||en