Laser powder bed fusion represents the future for metal additive manufacturing. Advance of this emerging technology is bottlenecked by the unavailability of high-fidelity prediction tools for cost-effective optimization on printing design. Simulations of selective laser melting of metals must tackle a complex granular solids and multiphase fluids system that undergoes intra- and inter-phase interactions and thermal-induced phase changes, including melting, vaporization, and solidification, which are challenging to model. We develop a high-fidelity computational tool to provide high-resolution simulations of the multiphase, multiphysics processes of selective laser melting (SLM). Key to this tool is a multi-phase, semi-coupled resolved Computational Fluid Dynamics (CFD) and Discrete Element Method (DEM). It contains innovative features including (1) a fully resolved immersed boundary CFD with fictitious particle domain coupling with DEM for resolving mechanical interactions and heat transfers between solid particles and surrounding fluid; (2) An evaporation model in consideration of the Knudsen layer implemented in the volume of fluid (VOF) method which is enriched by two sharp interface capture schemes, isoAdvector and MULES, for accurate identification of the vaporization process and phase boundaries of fluids with different Courant numbers; and (3) a ray tracing model compatible with the VOF method for high-resolution of absorbed laser energy. We demonstrate the proposed method can quantitatively reproduce key observations from synchrotron experiments and captures critical interdependent physics involving melt pool morphology evolution, vapor-driven keyhole dynamics and powder motions. This new computational tool opens a new avenue for quantitative design and systematic optimization of laser powder bed fusion and may find wider engineering applications where thermal induced phase changes in a multi-phase system are important.

激光粉末床融合代表了金属增材制造的未来。由于无法使用高保真预测工具来对印刷设计进行经济高效的优化，这一新兴技术的发展受到了瓶颈。选择性激光熔化的模拟金属必须处理复杂的颗粒状固体和多相流体系统，该系统会经历相内和相间的相互作用以及热诱导的相变，包括熔化、汽化和凝固，这对建模具有挑战性。我们开发了一种高保真计算工具，以提供对选择性激光熔化 (SLM) 的多相、多物理场过程的高分辨率模拟。该工具的关键是多相、半耦合解析计算流体动力学 (CFD) 和离散元法 (DEM)。它包含创新功能，包括 (1) 完全解析的浸入式边界 CFD，具有与 DEM 耦合的虚拟粒子域，用于解析固体颗粒与周围流体之间的机械相互作用和热传递；(2) 蒸发模型考虑Knudsen 层在流体体积 (VOF) 方法中实施，该方法通过 isoAdvector 和 MULES 两种尖锐界面捕获方案进行了丰富，用于准确识别具有不同流体的汽化过程和相界库朗数字；(3)与VOF方法兼容的射线追踪模型，用于高分辨率吸收激光能量。我们证明了所提出的方法可以从同步加速器实验中定量再现关键观察结果，并捕获涉及熔池形态演变、蒸汽驱动的锁孔动力学和粉末运动的关键相互依赖的物理。这种新的计算工具为激光粉末床熔合的定量设计和系统优化开辟了一条新途径，并可能找到更广泛的工程应用，其中多相系统中的热诱导相变很重要。