No description available.

No description available.

Cast aluminium alloy components are being increasingly used in transport applications where they may experience cyclic in-service loading. A quantitative prediction of fatigue performance is required during the component design stage to ensure components have the necessary lifespan at minimum weight. A multiscale, through process modelling methodology was developed to calculate fatigue life of cast aluminium alloy components. This technique first predicts the microstructure and porosity formation during casting. These features are then tracked through the subsequent processing steps of heat treatment and finish machining where the residual stresses are predicted. Finally, all of this informafion is passed to a model for the prediction of the component's in-service performance. The final fatigue behaviour is then predicted as a function of the microstructural features, the residual stress state, and the cyclic in-service loading. To test this methodology, the fatigue life of an A356- T6 automotive wheel was predicted and then validated experimentally. Prior authors have found that pores dominate fatigue life in cast A356-T6 if their size is larger than the secondary dendrite arm spacing. The pore size distribution (and secondary dendrite arm spacing) in the A356 wheel formed during casting, the first processing step, was predicted using model-based consfitutive equations run within a validated macroscopic heat flow model of the process. These results were validated using x-ray microtomography. During heat treatment, the second processing step, large residual stresses evolve in the wheel during quenching. These stresses were predicted using a two-stage thermal stress model. The results were found to be sensitive to the flow stress data of the A356 alloy. Therefore, the inelastic behaviour in the as-solutionised condition was measured as a function of temperature and strain rate. Using the measured data significantly improved residual stress predictions. The release of residual stress during the third processing step, machining, was then determined. The influence of both microstructural features and residual stress state was incorporated into the in-service model for final fatigue life prediction. This infiuence was quantified using x-ray microtomography of interrupted fatigue test specimens. Local stress concentration analysis was performed to determine the effect of 3D pore characteristics upon fatigue damage evolution. Applying the full multiscale, through process model to the A356-T6 wheel, the location of fatigue crack initiation and fatigue life were accurately predicted. Fatigue life was most influenced by applied loads, followed by pore size and then residual stresses.

This research has investigated the modelling and optimisation of the tilt casting process of Titanium Aluminides (TiAl). This study is carried out in parallel with the experimental research undertaken in IRC at the University of Birmingham. They propose to use tilt casting inside a vacuum chamber and attempt to combine this tilt casting process with Induction Skull Melting (ISM). A totally novel process is developing for investment casting, which is suitable for casting gamma TiAl. As it is known, gamma TiAl alloys has great properties including low density, high specific yield strength, high specific stiffness, good oxidation resistance and good creep resistance at high temperature [Clemens -2000][Appel et at. -2000]. A worldwide research effort has been made to develop gamma TiAl because it can offer a great potential for reducing the weight of high performance components and also engine of power generation gas turbine. Titanium alloys are very reactive at molten condition, and so, they are melted in an ISM crucible in order to avoid crucible contamination. There is still a big challenge to produce a long blade, up to 40 cm, due to the low superheat provided by the Induction Skull Melting (ISM) furnace which is widely used to melt the alloys. Here computational simulation has been seen important to predict the casting defects and to help optimise the experimental process. Computational modelling for the casting process involves a range of interactions of physical phenomena such as heat transfer, free surface fluid flow, solidification and so on. A number of free surface modelling techniques are applied to simulate the interface between the molten metal entering the mould in the filling phase, and the gas escaping. The CFD code PHYSICA developed in the University of Greenwich is used to simulate the above physical phenomena and to simulate the fluid flow both within the rotating mould cavity/crucible assembly and in the porous mould wall (including vents). Modelling the mould in a finite volume method is cumbersome, so an alternative 3D/1D coupled transient heat transfer model has been developed in this study. It is based on the fact that the mould filling for titanium aluminide (TiAl) is carried out during a few seconds and the thermal conductivity of the mould material is very low. Heat can be assumed to transfer mainly in a direction perpendicular to the mould wall ID. ID transient heat transfer model is governed by ID heat conduction equation in the mould part where the coordinates of each defined cell centre were calculated rather than meshing them. The coupling method between ID and 3D model is presented. The model is then validated using two simple geometries which describe two similar states in the mould filling as test cases. It has been applied to model short thin and long blades, especially to obtain accurate thermal boundaries. Comparisons with experiments have also been done. Across the presentation of the results, the factors affect the quality of the casting in the mould filling have been discussed. This thesis also presents a novel Counter Diffusion Method which was developed with suggestions from my supervisors as a corrective mechanism to counter numerical diffusion. This is a novel method to discretise the free surface equation fully implicitly in a fast, efficient way without numerical diffusion. Validation of the novel method was undertaken against the classical collapsing column experiment. The results showed that they are in good agreement. Then the method has been used to model a long thin blade for TiAl. A huge reduction in computational time is seen when the geometry is complex and massive amount of mesh cells are generated. That greatly speeds up the simulations. Solidification is modeled during the cooling which is following the filling stage. Gap formation between metal and mould is covered and the effects of the gap and gap size are presented by the application of model on a long twisted turbine blade.