Laser and metal powder based additive manufacturing (AM), a key category of advanced Direct Digital Manufacturing (DDM), produces metallic components directly from a digital representation of the part such as a CAD file. It is well suited for the production of high-value, customizable components with complex geometry and the repair of damaged components.
Currently, the main challenges for laser and metal powder based AM include the formation of defects (e.g., porosity), low surface finish quality, and spatially non-uniform properties of material. Such challenges stem largely from the limited knowledge of complex physical processes in AM especially the molten pool physics such as melting, molten metal flow, heat conduction, vaporization of alloying elements, and solidification. Direct experimental measurement of melt pool phenomena is highly difficult since the process is localized (on the order of 0.1 mm to 1 mm melt pool size) and transient (on the order of 1 m/s scanning speed). Furthermore, current optical and infrared cameras are limited to observe the melt pool surface. As a result, fluid flows in the melt pool, melt pool shape and formation of sub-surface defects are difficult to be visualized by experiment. On the other hand, numerical simulation, based on rigorous solution of mass, momentum and energy transport equations, can provide important quantitative knowledge of complex transport phenomena taking place in AM.
The overarching goal of this dissertation research is to develop an analytical foundation for fundamental understanding of heat transfer, molten metal flow and free surface evolution. Two key types of laser AM processes are studied: a) powder injection, commonly used for repairing of turbine blades, and b) powder bed, commonly used for manufacturing of new parts with complex geometry.
In the powder injection simulation, fluid convection, temperature gradient (G), solidification rate (R) and melt pool shape are calculated using a heat transfer and fluid flow model, which solves the mass, momentum and energy transport equations using the volume of fluid (VOF) method. These results provide quantitative understanding of underlying mechanisms of solidification morphology, solidification scale and deposit side bulging. In particular, it is shown that convective mixing alters solidification conditions (G and R), cooling trend and resultant size of primary dendrite arm spacing. Melt pool convexity in multiple layer LAM is associated not only with the convex shape of prior deposit but also with Marangoni flow. Lastly, it is shown that the lateral width of bulge is possibly controlled by the type of surface tension gradient.
It is noted that laser beam spot size in the powder injection AM is about 2 mm and it melts hundreds of powder particles. Hence, the injection of individual particles is approximated by a lumped mass flux into the molten pool. On the other hand, for laser powder bed AM, the laser beam spot size is about 100 µm and thus it only melts a few tens of particles. Therefore, resolution of individual powder particles is essential for the accurate simulation of laser powder bed AM.
To obtain the powder packing information in the powder bed, dynamic discrete element simulation (DEM) is used. It considers particle-particle interactions during packing to provide the quantitative structural powder bed properties such as particle arrangement, size and packing density, which is then an inputted as initial geometry for heat transfer and fluid flow simulation. This coupled 3D transient transport model provides a high spatial resolution while requiring less demanding computation. The results show that negatively skewed particle size distribution, faster scanning speed, low power and low packing density worsen the surface finish quality and promote the formation of balling defects.
Taken together, both powder injection and powder bed models have resulted in an improved quantitative understanding of heat transfer, molten metal flow and free surface evolution. Furthermore, the analytical foundation that is developed in this dissertation provides the temperature history in AM, a prerequisite for predicting the solid-state phase transformation kinetics, residual stresses and distortion using other models. Moreover, it can be integrated with experimental monitoring and sensing tools to provide the capability of controlling melt pool shape, solidification microstructure, defect formation and surface finish.