Summary: | A methodology of material characterisation and finite element model discretisation is presented for spot welded boron steel sheets, with the aim of predicting failure during quasi-static loading. The predicted load-displacement curves from the Finite Element model are compared with experimentally measured curves for lap-shear and cross-tension weld destructive geometries, and serve as model validation. During spot welding, the weld and surrounding material are exposed to a wide range of temperatures, from the melting point at the weld centre to room temperature in the base material. As a consequence, the weld exhibits varying microstructures with corresponding varying material properties which have a profound influence on its load bearing capacity and failure strength as a whole. In addition, boron steel spot welds exhibit unique hardness profiles, with high hardness values in the nugget and outlying base material, and a sudden drop in the area between these regions. This sudden decrease in material properties leads to further difficulties in modelling the failure of boron steel welds. The weld process inherently produces localised residual strains which also need to be accounted for in the model simulation, together with significant plastic strain redistributions resulting from the mechanical loading of the spot weld to its ultimate failure. The initial residual strains were measured in weld samples using neutron diffraction and were subsequently input into the FEA models. This thesis aims to quantify the varying material constitutive behaviour throughout the weld, required for the failure prediction. In particular, the following constitutive properties were extracted: the stress-strain response of certain weld regions, failure loci consisting of fracture strain versus stress state for the corresponding regions, and the residual stress distribution through the weld. Due to the small size of the weld, cutting test specimens directly from the weld is unachievable. To overcome this problem, specific weld and heat affected zone micro-structures were recreated onto practical tensile specimens through use of a Gleeble thermo-mechanical physical simulator. These specimens were subjected to the same thermal histories as specific points in the actual weld. From these tensile specimens, stress-strain curves relating to specific weld regions could be obtained. In a similar fashion, three additional destructive specimen sets were created to obtain failure loci. These failure loci give fracture strain as a function of stress state: specifically shear, uniaxial and plane-strain states. Due to the practical limitations in the accuracy of the Gleeble technique, deviations from the target microstructures were expected in the Gleeble samples. To gauge the extent of these deviations, a method of extracting reference material properties directly from the weld was required. Instrumented indentation offers such a solution, where the load and displacement of the indenter are measured and run through an algorithm to calculate the yield strength of the indented locations. These yield strengths are then compared with the yield from the Gleeble stress-strain curves to gauge the accuracy with which the weld microstructures were recreated. This technique serves to quantify the deviation of the Gleeble microstructures from the target material microstructures. It is common practice to discretise the weld into a small number of bulk regions during the design process, with material constitutive behaviour assigned to these discretised parts. In the new methodology, the extracted material constitutive behaviour is modelled as a continuously varying function of the distance from the weld centre. By performing appropriate interpolation, the data may be finely or roughly discretised. The data at a certain distance from the weld centre may then be assigned to the corresponding element in the finite element model. This means one may discretise the model by choosing the level of data interpolation refinement. The following results were observed in the thesis: • Residual strain distributions of boron steel spot welds, which have not been measured before, were presented. Clear correlations between hardness and residual stress distributions were seen. • A new application of instrumented indentation was attempted by verifying the accuracy of heat treated samples with respect to their target microstructures by comparing yield strengths. • The boron steel HAZ was characterised in a finer level of detail than seen in other literature works. • Through physical simulation, stress - strain and failure loci corresponding to certain HAZ areas were successfully extracted and used to model weld failure. • A new method of finite element model discretisation was presented, where material properties may be input as a relatively smooth function through the length of the model.
|