Author:

Abstract:

Additive manufacturing (AM) has revolutionized the production of metal parts, offering unparalleled flexibility in design and efficiency in material use, thereby facilitating lightweight structural applications and minimizing waste. This innovation, however, diverges significantly from traditional manufacturing methods such as casting, primarily due to the extreme cooling rates that affect the microstructure and, consequently, the material properties of the produced parts. Despite identical alloy compositions, parts manufactured through AM and traditional methods exhibit marked differences in microstructure and properties. Furthermore, alloys optimized for conventional manufacturing often prove unsuitable for AM, resulting in diminished ductility, decreased corrosion resistance, and susceptibility to hot cracking. While extensive experimental work has advanced AM processes, computational modeling has lagged, with many simulations lacking direct experimental comparison or transparency in methodology, thus limiting their predictive capacity. Simulation frameworks can provide an effective tool for understanding the physical mechanisms behind these high-cooling-rate microstructures and allow for the optimization of alloy design and heat treatments. The phase-field framework, a popular tool for simulating microstructural evolution based on thermodynamic principles, is computationally intensive and struggles with instability due to high cooling rates and associated thermodynamic driving forces. The first part of this thesis addresses the limitations of the classical phase-field model by developing a new high-driving-force phase-field (HDPF) formulation. This model, validated through numerical benchmarks, can accommodate any thermodynamic driving force while maintaining quantitative predictive accuracy. It enables simulations on a larger physical scale, significantly reducing computational time. Subsequently, the model was applied to simulate nucleation and growth during the Beta to Alpha-phase transition in pure titanium, a material critical to medical applications and frequently produced via AM. The simulations were directly compared with experimental results obtained during this research and existing state-of-the-art data. A notable challenge was the scarcity of thermodynamic and kinetic input parameters for phase-field simulations, which are difficult to source from literature and measure experimentally. Although density functional theory (DFT) calculations and molecular dynamics (MD) simulations can provide these parameters, their high computational demand renders them impractical in acquiring a comprehensive set of parameters. Thus, the second part of the thesis introduces a novel computational framework, Molecular Statistics (MST), designed to calculate the necessary thermodynamic and kinetic parameters at a lower computational cost. This framework underwent a rigorous development and validation process, starting with the thermodynamics of pure materials and extending to the study of vacancies and their kinetics, culminating in a new analytical description of metal self-diffusion based on statistical mechanics. The validation of the MST framework against existing databases confirmed its accuracy and efficacy. The MST and HDPF frameworks developed in this thesis are complementary, bridging microscale and mesoscale simulations and enabling a multiscale approach to material science research. The HDPF framework allows for quantitative simulations of microstructural evolution at high cooling rates, and the MST framework can be used to obtain the required thermodynamic and kinetic input parameters for the AM process. While inspired by the challenges of additive manufacturing, these frameworks offer broad applicability across computational materials science, promising new avenues for future research.