Welded aluminium connections - Behaviour and numerical modelling for large-scale simulations

Abstract

This PhD thesis explores the behaviour and modelling of welded aluminium connections. It is divided into four parts: Parts 1 and 2 have been published in peer-reviewed journals, while Part 3 has been submitted for potential publication. Part 4 is a technical report that documents supplementary investigations related to this PhD project, not covered in Parts 1-3. Each part is an individual study and can be read independently. The thesis begins with a synopsis that introduces the work, followed by a presentation of its objectives, scope, and research methodology. The synopsis concludes with summaries of each part, key conclusions, and suggestions for further research. Part 1 presents a combined experimental and numerical study on the tensile ductility of aluminium alloys subjected to heat treatments representative of welding. The experimental part includes specimen preparation and uniaxial tensile testing. A Gleeble machine generates heat treatments with high peak temperatures and heating rates that mimic those from a welding process. Following this, uniaxial tension tests are conducted at low and high strain rates. The results highlight the effects of these heat treatments on strength, strain-rate sensitivity, and fracture behaviour. The numerical part of this study examines two widely used damage models: the Cockcroft-Latham fracture criterion and the Gurson-Tvergaard model. Material models are calibrated and a parameter study is conducted based on the tension tests. A case study is also performed to emphasise the significance of the damage parameters within an idealised heat-affected zone and under plane-strain tension. The main objective is to improve the modelling of ductile fracture for aluminium alloys exposed to welding-like heat treatments. Part 2 presents a study on the behaviour and modelling of simple cross-welded connections. This study focuses on large-scale analyses, where the weak zones are represented with a few shell elements. A shell-element modelling framework suitable for large-scale analyses is thus proposed, which accounts for geometrical instability, thinning, and ductile fracture. Calibration and validation of the proposed modelling framework are performed using cross-weld tension tests. The test campaign includes tensile testing of two Al-Mg-Si alloys (AA6082 and AA6060) and two widely used welding techniques (MIG and FSW). The main objective is accurately capturing the overall weld and heat-affected zone (HAZ) behaviour under cross-weld tensile loading with shell-element simulations. Two approaches are investigated where the size of the HAZ is either fully modelled or lumped into one row of elements. The shell-element simulations of the cross-weld tension tests, with several element sizes larger than the plate thickness, show reasonable agreement with the experimental results. Part 3 investigates the testing and modelling of welded aluminium joints at the component level. A novel test campaign examines the behaviour of a welded T-joint under quasi-static and impact loading. The deformation is confined to the heat-affected zones (HAZ), which corresponded well with the low hardness values of the HAZ compared to the base and weld materials. Higher force levels were observed in the impact tests compared to the quasi-static tests, indicating that the T-joint under impact loading experienced inertia and strain-rate effects. This study’s numerical part considers the HAZ modelling using shell elements. A virtual calibration procedure is proposed to establish model parameters applicable in large-scale analyses based on the output from a welding simulation and a microstructure-based model. The virtual calibration procedure is benchmarked against the welded T-joint and a cross-weld tensile case, where the base and HAZ materials’ hardness, yield stress, and work hardening are captured reasonably well. To this end, shell-element simulations of the T-joint tests are conducted, and the results resemble those of the experiments to a great extent. Part 4 of this PhD thesis is a technical report documenting experimental studies carried out as part of this project, which were not included in Parts 1-3. A substantial part of the T-joint test campaign was omitted from Part 3 to satisfy the journal’s page limit. Further, a weld start and stop activity is documented, where cross-weld tensile testing was conducted with specimens having either a weld start or stop. Single-lap joints have also been tested as a part of this PhD project, where the weld was subjected to unfavourable loading conditions. A common observation for all tests reported herein is that fracture initiation occurred within the weld, in contrast to Parts 1-3, where it occurred within the HAZ.