Modelling of grain formation during selective laser melting via a finite element-cellular automata method

• Main Author: Ong, Gerard Zi Quan
• Other Authors: Zhou Kun
• Format: Final Year Project
• Language: English
• Published: 2019
• Subjects: DRNTU::Engineering::Manufacturing; DRNTU::Engineering::Mathematics and analysis::Simulations
• Online Access: http://hdl.handle.net/10356/78591
• Institution: Nanyang Technological University

Additive manufacturing (AM) is a process which builds materials layer by layer to form a three dimensional (3D) structure and it has applications in a wide of industries such as aerospace, automotive, marine and biomedical. There are various forms of AM processes which include stereolithography, direct energy deposition, selective laser melting, fused deposition modelling and binder jetting. In selective laser melting, a laser beam is directed onto a layer of powder to form an area of molten metal known as the melt pool. During the cooling and solidification stage, nucleation occurs within the melt pool, which grows and extends into grains. The size and number of grains varies at different locations along the melt pool due to the different thermal gradients and solidification rates. The study of grain formation is important as it affects the mechanical properties of the manufactured structure, such as its deformation mechanism and fracture toughness. Simulation tools have been used to study the process of selective laser melting, in order to optimise the process parameters and resultant mechanical properties. Thermal analysis of the melt pool can be done using finite element method, computational fluid dynamics or discrete element method, while the solidification process can be modelled by cellular automata, phase field or monte-carlo techniques. In this report, the finite element-cellular automata (CA-FE) method was used to study the process of selective laser melting and to correlate and predict the effects of process parameters on the resultant grain structure. A detailed formulation of the 2D CA-FE model and algorithms was presented and subsequently, the validation of the model was conducted on various experimental results from literature. The simulated grain structures have shown that the 2D model was mostly able to accurately predict the effects of various process parameters such as scan pattern, laser power and scan speed. However, in order to increase the fidelity of the model and to fully capture the out of plane grain growth, a 3D extension of the CA method was implemented and an analysis was conducted on the effects of out of plane growth. Using a single model, the developed CA-FE simulation showed promising results in a range of experimental validations and could possibly be used to design and optimise AM processes in the future.