Material Point Method for Modelling Additive Manufacturing

As one of the additive manufacturing technologies, selective laser melting has shown a good potential yet still having technical barriers, such as residual stress management which can determine the manufacturability and integrity of a component. Simulating this process on a full scale also needs to face a large physical scale difference, which makes the current numerical method difficult to adapt. Currently, it seems that finite element modelling is a natural choice to address those challenges in additive manufacturing. However, it is still challenging to overcome the huge gap in time and length scales in a powder-bed fusion process. Furthermore, a requirement of adding new materials is hard to model in a finite element method because of the difficulty of rezoning a finite element mesh, especially in a 3-D complex shape while a material point method can take fewer efforts to add materials than a finite element model and has been employed in simulations such as large deformation and crack failure problems. Those potentially happened in additive manufacturing process are still tough nuts using finite element method. This thesis presents work on using material point method to simulate the selective laser melting process at the full component scale. Using this model, a series of investigations are performed to demonstrate the effects of different scan strategies, boundary conditions and geometrical shapes. This is the first attempt to simulate an additive manufacturing process using the material point method at a full component scale. A ghost point method is developed to imitate the adding of materials in the manufacturing process. This function is controlled by a time step counter with only two results: 1 and 0. In this model, the mapping process of mass and momentum between node and material points are modified and controlled by multiplying this function. As a result, the unopened material points have a mass and momentum of zero during mapping process so that these material points temporarily disappear in the system. Furthermore, combining this method with the coordinate transformation system can achieve a different scanning strategy. Three simplified underlying physics, including residual stress, solid-state phase transformation and thermal strain, are considered and added in material point model. All three sub-models are based on existing literature or data for simplicity and the constitutive model has been modified accordingly. These models can be added directly to the material points to be achieved with the material point to open and close together. From the simulation results, material point program has achieved a good function of each sub-physics. Additionally, an optimization program, which switches on the material points layer by layer, is performed to significantly reduce the computational throughput for simulating residual stress effects in selective laser melting. The material point code is developed from an open source code called MPM3D-f90. This model is a macro-scale analysis which is implemented on desktop PC. Compared to the finite element model, the simulation time is greatly reduced. Therefore, the material point method has the potential to become a powerful tool in simulating selective laser melting.