Degree Name

Doctor of Philosophy


Department of Materials Engineering


The effect of varying welding conditions on multipass weld metal has been investigated in all-weld test plates produced by flux cored arc (FCA) welding. The principal welding conditions varied for the test samples in the current research program were heat input (between approximately 1 and 2.5 kJ/mm), welding technique (stringer or weave beads), shielding gas, welding position (flat or vertical) and consumable type.

Mechanical testing of weld metal included tensile testing and Charpy-vee-notch impact tests over a range of temperatures from -60 °C to 40 °C. The tensile testing results showed that yield and tensile strength generally decreased with increasing heat input and that the changes were of the order of 100 MPa. Impact properties showed more varied behaviour depending on consumable type.

The effect of varying welding conditions on some less commonly studied microstructural factors has also been investigated, such as proportions of reheated and double-reheated regions and mean free path (MFP) in various distinctive microstructural zones. Using the simplifying assumption that the structural gradient the heat affected zone (HAZ) of reheated weld metal can be characterized in terms of three regions: the grain coarsened, GC; grain refined, GR and intercritical, IC heat affected regions; a second overlapping weld bead creates double-reheated weld metal structures which can be defined in terms of 9 sub-zones: GC-GC, GC-GR, GC-IC, GRGC, GR-GR, GR-IC, IC-GC, IC-GR, IC-IC. Linear and areal measurement techniques have been used on cross-sections of seventeen test plates to determine the volume fractions of solidified weld metal (SW), GC, GR and IC sub-zones of the reheated weld metal and the double-reheated zone. The hardness variation in the sub-zones and the influence of welding conditions has also been investigated in the present research.

The mean free path was determined for SW, GC, GR and IC regions and the overall average mean free path of ferrite in each weld was also calculated to provide a characteristic length of the totality of the structural microconstituents in the weld metal. Strong relationships were found between mean free path and heat input and also with the strength of the weld metal. In contrast, toughness correlated poorly with overall mean free path.

Impact toughness is one of the most important mechanical properties of welds. Assessment and understanding of the impact properties of welds is central to the avoidance of catastrophic failure of welded steel structures. Neural networks analysis has been used for the purpose of assessing which factors are beneficial to the low temperature impact properties of steel welds in order to obtain practical guidance as to how impact properties might be improved in FCA welds.

Four major data fields were involved in the training of the neural networks: chemical composition, microstructure, non-metallic inclusions and welding condition. The backpropagation algorithm was used for the analysis. Sensitivity analysis of the neural network model revealed that the characteristics of the non-metallic inclusion distribution exerted only a small direct influence on toughness. In contrast, chemical composition and microstructure were indicated to be very important. In particular the toughness is predicted to increase with increasing % acicular ferrite and % reaustenitised region and these factors in turn are sensitive to the selected FCAW process and the welding conditions.