Search

Selective laser melting is an attractive technology, enabling the manufacture of customised, complex metallic designs, with minimal wastage. However, uptake by industry is currently impeded by several technical barriers, such as the control of residual stress, which have a detrimental effect on the manufacturability and integrity of a component. Indirectly, these impose severe design restrictions and reduce the reliability of components, driving up costs. This thesis documents work on investigating the generation of residual stresses created in the selective laser melting process by the use of a finite element thermo-mechanical model. The thermo-mechanical model incorporated an adaptive meshing strategy which was used in conjunction with the use of high performance computing facilities. These together significantly increased the computational throughput for simulating selective laser melting of a single layer. Additionally, a volumetric hatching method was created to generate the laser scan vectors used in the process, with the ability to both simulate and manufacture on selective laser melting machines. A number of studies were performed to better understand the effect of laser scan strategy on the generation of residual stress in selective laser melting. Using this model, a series of investigations were performed to understand the effect of scan strategy and scan area size on the generation of residual stress in this process. Further studies were also performed to investigate the role of laser parameters, geometry, and support structures in selective laser melting and their effect on the generation of residual stress. The studies showed a complex interaction between transient thermal history and the build-up of residual stress has been observed in two conventional laser scan strategies (unidirectional and alternating) investigated. The temperature gradient mechanism was discovered for the creation of residual stress and the scan area size had an effect on the temperature sustained within the region. The parametric study of the laser parameters showed that an increase in laser scan speed increased the melt pool aspect ratio, and increase in laser power increased the melt pool width. The parametric thermo-mechanical analysis revealed that the laser scan speed had the most influence on the magnitude and anisotropy of the residual stresses generated. Varying the hatch distance had little effect on the maximum magnitude of residual stresses generated, but decreasing the hatch distance significantly increased the level of yielding that occurred. A study of the geometrical effect on scan strategy revealed the importance of the thermal history on the transverse stresses generated, influenced by the arrangement of scan vectors. The higher magnitude longitudinal stresses had predictable behaviour; only dependent on the scan vector length and not the thermal history generated by the choice of laser scan geometry. It was shown that the laser scan strategy becomes less important for scan vector length beyond the typical 5 mm island sizes. From the study of the support structures, it was found the insulating properties of the metal powder used in selective laser melting provide a significant thermal resistance for the dissipation of heat, and caused uniform overheating across the scanned region. In particular, the analysis showed localised overheating using support structures, which affected the melt pool geometry, and the residual stresses generated due to resistance against dissipating heat. Additionally, lattice structures such as the double gyroid showed localised overheating occurs using repeated exposures of short scan vectors. Suitable scan strategies therefore need to be developed to account for support structures. A multi-scale methodology was developed by combining information from the meso -scale obtained from the thermo-mechanical model. This model was used to predict the mechanical response of amacro -scale part. This approach used the assumption that meso -scale regions in island scan strategies behave independently from each other. This assumption was verified by comparing with a thermo-mechanical analysis. This multi-scale method was applied to a 3D structure and also to a complex 2D geometrical shape. Performing the multi-scale analyses has verified that the proposed technique of superposition of meso-scale stress fields at the macro -scale is a valid technique. The main strengths of the proposed multi-scale method is the decoupling of the meso and macro scale analyses. This has the benefit of reducing computational cost of the macro -scale analysis because it is independent of the complexity of the meso -scale analysis, and only requires performing once. These strengths translate into large computational time savings and also great flexibility in the physics incorporated at each scale.

Rural-urban migration and land abandonment can increase secondary forest coverage, which is predicted to alleviate harvesting pressure on vulnerable primary forest species. However, rural depopulation could threaten the conservation and development paradigm of sustainable resource use within inhabited reserves. This thesis examines 1) ruralurban migration in eight roadless regions of the central Brazilian Amazon, and 2) tests assumptions regarding the value of secondary forests for subsistence hunting in a partly deforested landscape of eastern Amazonia. Within central Amazonia, rural populations were clustered and growing within 300 km of towns whereas river headwaters have been largely abandoned. Living in remote upstream locations was costly, despite an abundance of natural resources. Upstream settlements were isolated from urban centres and were smaller, subject to costly trade exchange, and generally lacked schools and healthcare. The lack of education was the principal motive for ongoing out-migration from remote headwaters. Results show that interview data, spatial analysis and censuses of human populations can be used for temporal monitoring of vulnerable species across broad spatial scales. Within the eastern Amazon, research showed that secondary forests do not offer a panacea for over-hunting of primary forest wildlife. Hunters with access to both primary and successional habitats preferred to hunt in primary forest, probably because of good visibility, predictable prey distributions and high availability of other non-timber forest products. The availability of large areas of secondary forests failed to buffer hunting pressure in primary forest, as several primary forest species continued to be overhunted. Secondary forests alone are unlikely to support sustainable hunting across the Amazon as they may provide just 2% of the required protein intake of rural smallholders. This thesis concludes that failing to consider the diverse factors influencing spatial decision-making by rural people undermines the potential contribution of non-timber resource use to conserving tropical forests.