铝合金作为一种重要的轻量化材料，应用广泛，其性能主要取决于凝固过程中形成的枝晶、共晶组织和氢气孔等缺陷的形貌和分布。凝固过程是一个热质流交互作用的复杂相变过程。对凝固组织及氢气孔缺陷进行细致深入的研究，是掌握和控制铝合金凝固过程，获得理想材料性能的关键。通过计算机模拟仿真技术揭示凝固过程复杂的热质流交互作用，凝固组织演变规律以及与气孔间的相互作用，对于深入理解凝固机理，优化工艺方案和提升产品性能具有重要的理论价值，对于实现绿色制造、数字化制造及精确化制造具有重要的指导意义。针对多物理场作用下的组织演变数理模型不完善问题，本论文结合相场法（PFM）和格子玻尔兹曼方法（LBM），建立了相场-双分布函数格子玻尔兹曼模型（PF-LBM），PFM用于求解凝固组织演变，LBM用于求解温度场和流场演变。针对当前固液气三相场模型局限性大的问题，基于自由能泛函理论推导建立了新的三相场模型。该模型考虑了对流作用和气泡变形，能够精确刻画固液气三相相互作用，并建立了守恒相场模型精确追踪气液界面。针对相场模拟计算效率低的问题，从并行计算、多层网格和自适应调整等策略出发，基于空间曲线填充算法，开发了并行-自适应网格加密（Para-AMR）算法。该算法将PF-LBM和Para-AMR相结合，首次采用多层网格架构使时间步长增大2-3个数量级，采用Para-AMR进一步提高计算效率2-3个数量级，奠定了高效率仿真模拟的基础。本论文将PFM和LBM相结合，用于求解凝固过程温度场-溶质场-流场耦合作用下的枝晶和共晶组织演变，首次实现了多物理场耦合作用下、高固相率、复杂界面的三维组织模拟，研究了枝晶羽流造成的雀斑缺陷形成的微观机制，提出了流场作用下的共晶生长模型。为解决凝固过程固液气三相相互作用以及流场作用下的溶质输运和气液界面大变形，提出了固液气三相场模型和守恒相场模型，同时考虑晶体生长界面前沿的Gibbs-Thomson效应、合金溶质输运、气体组分输运、液体对流、高液气密度比、气泡运动和变形等，实现了对多物理场作用下凝固组织与氢气孔相互作用过程的仿真预测。
As one of the most important lightweight materials, aluminum alloys are widely applied in aerospace, automobile, and household appliances. The properties of the alloys are highly dependent on the solidified microstructures and could be deteriorated by defects such as the hydrogen porosity. Uncovering the complex thermo-solute-convection interaction during solidification is a prerequisite to control the solidification process and obtain materials with excellent properties. With the development of the computational materials science, numerical modeling is becoming an indispensable method to investigate the underlying physics during solidification. The complicated solid-liquid-gas interaction, together with the nonlinear thermo-solute-convection interaction, can be visualized and explored by numerical simulations. The deeper understanding of the solidification process is the key to optimize the process and improve the performance of products, which is of both scientific and technological significance for developing green, digital and precise manufacturing.In this work, a hybrid phase-field lattice-Boltzmann method (PF-LBM) is developed to simulate the microstructure evolution under multi-physical fields. The phase-field method (PFM) is used to simulate dendrite and eutectic growth involving Gibbs-Thomson effect and solute redistribution, while the lattice-Boltzmann method (LBM) is applied to simulate the thermo-convection evolution. To remove the limitations of the existing solid-liquid-gas models, a new multi-phase field is developed from the free energy functional in a thermodynamically consistent way. The liquid convection is incorporated in this model and a conservative phase-field model is embedded to characterize the liquid-gas interface with large deformation. To facilitate the large-scale numerical simulation, a high-performance computing algorithm named the parallel-adaptive mesh refinement (Para-AMR) algorithm is developed to improve computing efficiency. The Para-AMR algorithm is based on the space filling curves and composed of parallel computing, multi-block structure, and adaptive layout. The developed Para-AMR algorithm promotes the computing performance in two aspects. The first is the magnification of the time step by constructing the multi-block structure, which enlarges the time step by 2-3 orders of magnitude. The other is a further improvement of the computing efficiency by 2-3 orders of magnitude by combining the parallel computing and the adaptive layout. The development of the high-performance computing algorithm lays the foundation for the high-efficiency numerical simulation.The PF-LBM is developed to solve the thermo-solute-convection interaction during the growth of dendrite and eutectic. For the first time, the three-dimensional microstructure evolution with multi-physics, high solid fraction and complex interface is simulated and reproduced. The formation and evolution of the freckle defect due to the solute plume is revealed from the microstructure level by combining the phase-field simulations and in-situ experiments. The classical Jackson-Hunt theory for eutectic growth is improved by taking the convection effect into account. To solve the solid-liquid-gas interaction, the developed multi-phase field model can characterize the Gibbs-Thomson effect, transport of both alloy solute and gas component, convection effect, high liquid-gas density ratio, and movement and deformation of the pore in a unified frame. The difficulty in describing the multi-phase interaction and large-deformation liquid-gas interface is solved, which realizes both simulating and forecasting of the microstructure and hydrogen porosity evolution during solidification.
