This thesis examines the failure behaviour of spot welded connections in components using three different aluminium-silicon coated hot stamped steels, Ductibor^® 500-AS, Ductibor^® 1000- AS, and Usibor^® 1500-AS, with 1.2 mm and 1.6 mm sheet thicknesses. Spot weld connections are first characterized using single spot weld experiments. A novel experiment is then developed to characterize weld failure propagation within groups of welds under predominantly shear loading conditions. The experiments are modelled to evaluate how a calibrated weld failure model from single spot weld test data performs in predicting spot weld group response.

Tensile lap shear and cross tension single spot weld experiments were conducted under quasistatic conditions, according to the AWS D8.9M:​2012 standard, for all materials and thicknesses considered in this work. Recorded force versus crosshead displacement curves and integrated absorbed energy versus crosshead displacement are reported for all single spot weld experiments. The higher strength materials, Ductibor^® 1000-AS and Usibor^® 1500-AS, exhibited brittle weld failure modes and thus absorbed almost no energy following failure initiation. The lap shear experiments showed similar levels of spot weld strength, around 15 kN, for all 1.2 mm specimens and approximately 20-26 kN for the 1.6 mm lap shear specimens. The cross tension experiments showed similar strength for the Ductibor^® 500-AS and Ductibor^® 1000-AS specimens, 7 kN for the 1.2 mm thickness and 12 kN for the 1.6 mm thickness. The peak loads for the Usibor^® 1500- AS cross tension specimens were approximately 50% of the loads for the other two materials. The Ductibor^® 500-AS specimens absorbed the most energy for the lap shear and the cross tension experiments which is attributed to increased parent metal deformation and more ductile weld failure characteristics.

A new mechanical test, termed the Caiman Mode III, was developed to promote shear failure within a group of spot welds in a manner similar to a mode III fracture mechanics specimen. A custom rail design using U-channels, that is fabricated in stages, is selected. The Caiman Mode III experiments are tested under quasi-static and dynamic loading rate conditions to examine the mechanical properties of spot welds in a structure in which the applied load can be shared across multiple spot welds. The Caiman Mode III experiments further showed that the three materials have similar peak loads and that the Ductibor^® 500-AS spot welds have the highest toughness and most absorbed energy of the three materials tested. Spot weld failure timing was recorded to characterize the extent and rate of spot weld failure propagation in the Caiman Mode III experiments. High speed thermal imagery was applied to determine precise spot weld failure times in both the quasi-static and the dynamic Caiman Mode III experiments since it was difficult to identify failure of specific weld from the force-displacement data. The Ductibor^® 500-AS exhibited a slower rate of failure propagation through the weld group compared to the two higher strength alloys.

Two different spot weld material models were considered in numerical simulations of the single spot weld and Caiman weld group experiments. The first weld material model, *MAT_100_DA, is used commonly by industry in car crash simulations, while the second weld material model, *MAT_240, used a cohesive zone approach that enables more direct control over the spot weld post-failure behaviour. Both weld material models were calibrated to the single spot weld experiments with respect to force versus displacement. The *MAT_100_DA model showed no post-failure unloading response for the normal-tensile loading condition, thus under predicting the total absorbed energy. The *MAT_240 model enabled more accurate post-failure unloading for all of the single spot weld conditions tested.

Simulations of the Caiman Mode III experiments were performed to validate the calibrated single spot weld models. The *MAT_240 simulations and the *MAT_100_DA simulations show similar predictions for each material condition except for the Ductibor^® 500-AS 1.6 mm model. The numerical simulations of the Caiman Mode III experiments were able to qualitatively predict the overall behaviour of the Caiman Mode III experiments, including aspects such as high initial load followed by load drops at each sequential weld failure and load drop off, as well as progressive failure propagation through the weld group. However, the simulations showed inconsistent peak force and energy absorption accuracy results when examining the results of different material and thickness simulations. The Caiman Mode III simulation inconsistencies for both material models suggests that the differences between the predictions and the experiments may be from physical inconsistencies between the design and the as-fabricated final specimens. Edge tear out failure was observed in the Caiman Mode III experiments due to 10 mm nugget-to-edge distance while the spot weld models were calibrated from single spot weld experiments with 20 mm weld-edge distance that exhibited button pull out failure instead.