The aim of this PhD research project was to investigate the static mechanical performance of innovative composite-metal (hybrid) joints strengthened by the presence of protrusions on the metal adherends. This advanced hybrid joining technology uses arrays of macro-scale protrusions (surfi-sculpts) on the surface of the metal part which can penetrate into the composite material during manufacture to achieve a high performance joint in the finished part. This joining technique combines many of the respective advantages of a bonded joint and a fastened joint. Joints formed between an aerospace grade titanium alloy and carbon fibre reinforced composite have been the focus of this study. Firstly, the effects of two composite adherend design parameters on the static mechanical performance of advanced hybrid joints strengthened by surfi-sculpt were investigated experimentally. The design parameters studied were (i) the composite ply orientation and (ii) the composite thickness. Single lap joints were manufactured and tested. The results from this part of the investigation showed that the performance of the hybrid joints was strongly affected by these design parameters. An optimum combination of these parameters was then selected for the next part of the study. Secondly, the effect of the surfi-sculpt protrusion density on joint performance was also investigated experimentally. Single lap joints were again manufactured and tested, and Digital image correlation analysis was used to measure the growth of the debond along the joint overlap during the tests. The results from this part of the investigation showed that the protrusions resist the initial, unstable failure mechanism that was observed in the control joints (without protrusions) and convert it into stable growth. With increasing protrusion density, the debonding propagation rate decreased and the failure mode changed from debonding to intra-laminar failure of the composite and fracture of the metallic protrusions in a mixed shear and bending mode. Increasing the protrusion density was shown to significantly increase the ultimate failure load, joint extension and hence absorbed energy. Thirdly, the effect of inserting disbonds between composite and Titanium adherends during manufacture, extending from either end of the joint overlap, was investigated as a way to study damage tolerance of the hybrid joints. Joints with four different initial debonding ratios (the ratio of disbond area to overlap area) were investigated and for each, two different protrusion densities were studied. Control joints without protrusions were again tested for comparison. The results showed that protrusions were able to maintain the strength and damage tolerance of the hybrid joints. Indeed, by increasing the protrusion density the ultimate failure load of the joints with a disbond was improved. By measuring the propagation of debonding it was shown that increasing protrusion density decreased the debonding rate from both ends of the overlap. Finally, the performance of the adherends and hybrid joints has been studied using finite element models. The models were three-dimensional, implicit, finite element models incorporating interface failure using surface-based behaviour and composite failure using a ‘User Defined Field’ subroutine in Abaqus. Three different joints were modelled: firstly the control (reference joint) without protrusions and two joints with different protrusion densities. The models were able to reproduce the experimental load-displacement traces with good accuracy. The numerical result for the reference joint shows that the resin rich zone in the joint at the end of the composite adherend only affects the first debonding damage location but has negligible effect on the following damage propagation and joint loading capacity. The numerical results for the joints with protrusions reproduce all the critical features of the experimental data: the initial linear behaviour and following softening behaviour; the knee points in the softening stage reflecting the pulling-out of metal protrusions from the composite adherend and finally the transition point (representing joint failure) shown in the load-displacement curve for the joint with the highest protrusion density.

The objective of this thesis is to propose a new model for particle dispersion and clustering for use within an (unsteady)-Reynolds Averaged Navier-Stokes ((u)RANS) computational framework. The need for an improved model stems from industrial requirements to address certain limitations of the currently used models. Namely, low predicted particle entrance into recirculation zones for particles with large Stokes numbers and unrealistically spatial and temporal uniform predicted particle concentrations. Abstract The literature review presented within this thesis examines the various computational tools available for modeling the Lagrangian phase and identifies Kinematic Simulations (KS) as potentially capable of reproducing accurate Lagrangian statistics and particle clustering across a range of physical scales while at the same time requiring a modest increase of computational resources relative to the presently used methods. Abstract The thesis proposes a combination of (u)RANS and KS in a coupled Eulerian-Lagrangian framework. The (u)RANS calculations will be responsible for modeling the large coherent motions while the KS will be employed to model the effects of all the other scales, that are represented statistically in the (u)RANS context, on particle motion. In other words, the representation of the velocity field within the 'eddies' will be simulated by tracking a particle through an isotropic turbulent field constructed with the aid of KS. The extent of scales and the energy content of the isotropic field modeled by KS at every instance is determined by the local properties of the under-resolved 'eddy' as determined by the Eulerian framework. Abstract The proposed model is evaluated on an axisymmetric sudden expansion test case through comparisons with experimental results, LES calculations as well as RANS simulations employing the current industry standard dispersion model. Improved overall performance was observed with significant differences between the particle trajectories computed with the proposed model and those with a model widely used in industry. This last point is of particular significance as one of the limitations of the currently used models was the high degree of spatial uniformity in the predicted particle distribution.

In this thesis, particle formation in reacting flows is investigated experimentally. Two separate systems are considered. First, silica particle synthesis is characterized in a cold, turbulent jet doped with trace amounts of silane gas that issues into a vitiated co-flow. Additionally, soot formation is characterized in a laminar ethylene diffusion flame. Laser diagnostic techniques are the cornerstone of this work and make it possible to perform measurements with minimal disruption to the system. In both scenarios, elastic light scattering (ELS) and OH-PLIF are employed to obtain experimental signals that contain information about temperature, particle formation, OH concentration and other physical quantities. Additionally, line-of-sight extinction is used in the soot-forming system to recover integrated soot volume fraction profiles and multi-angle light scattering (MALS) is demonstrated on the silica-forming system to obtain in-situ information about particle size. Laser-based datasets are supplemented by probe measurements, including temperature profiles measured using radiation-corrected thermocouples and TEM analysis of particle samples obtained by location-specific thermophoretic sampling. Fully-defined numerical models, available in both scenarios, are validated following an unconventional approach based on the comparison of “predicted signals” with experimentally-obtained signals, as a means to avoid introducing additional assumptions. Satisfactory agreement is found, even though some discrepancies remain concerning silica particle formation, which are likely related to uncertainties in precursor chemistry and nucleation. Nevertheless, this is one of a few joint numerical and experimental studies that address particle formation under turbulent conditions. Regarding soot formation in laminar flames, a consistent underprediction of soot formation on the centreline is identified, which is believed to be a limitation of the acetylene-based model. In summary, this work uses optical diagnostic techniques to generate extensive datasets of particle-forming reacting jets, making a major contribution towards the validation of computational tools to predict particle formation in turbulent reacting flows as well as soot formation in laminar flames.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.