Date of Award
Doctor of Philosophy
Materials Science and Engineering
Materials Science and Engineering
Richard A. LeSar
Deposition of metals through additive manufacturing has garnered research interest as of late due to the large range of potential industry applications. In particular, direct metal deposition processes such as Laser Engineered Net Shaping (LENS) have the ability to construct near net shape parts, open cellular structures, compositionally graded parts, and parts with improved mechanical properties over those manufactured via traditional methods such as casting and forging. To utilize additive processes to their full potential, it is imperative that the relationships among process parameters, development of the molten pool, microstructure, and properties are understood. Our goal in applying computational modeling to this problem is to aid in our understanding of such relationships to guide future experiments towards sets of alloying additions and deposition conditions that produce preferred microstructures. Cellular Automata (CA) based modeling techniques provide a way to bridge the scales of the complex phenomena that occur during AM processes, reducing them to physics-based rules for the evolution of cell state variables; in particular, this makes these methods well-suited for large scale parallel computing problems and large ensembles of simulations. CA is applied at the scale of individual dendrites yielding quantitative agreement with analytical models for dendrite tip undercooling as a function of solidification velocity. For dendritic colonies, CA modeled microstructures yielded favorable quantitative and qualitative agreement with expected trends in primary arm spacing, side branching, solute segregation, and non-equilibrium growth phenomena such as solute trapping and banded growth morphology. CA is also applied at the scale of multiple grains to investigate the columnar to equiaxed transition in 2D and 3D with varied nucleation undercooling, alloying addition, and interfacial response function. The lattice Boltzmann (LB) method for fluid transport is combined with COMSOL Multiphysics simulations of melt pool dynamics and the dendrite-scale CA for coupled simulation of fluid flow, solute transport, and solidification, yielding good agreement on microsegregation and dendrite arm spacing with experimental results for LENS alloy deposition. A thermal lattice Boltzmann (TLB) model of the melt pool is also developed and combined with the grain-scale CA for parallel, concurrent multiscale simulation of fluid flow, heat transport, and grain growth for LENS-representative conditions, showcasing the model's ability to predict microstructure trends with changes in process conditions or alloying additions. The ability of CA to accurately predict many aspects of and trends regarding alloy solidification in additive processes show a promising future for using similar codes to augment experimental results for new alloy development, while the parallelizability and computational efficiency of CA show its potential for use in Exascale computing application codes.
Rolchigo, Matthew, "Use of cellular automata-based methods for understanding material-process-microstructure relations in alloy-based additive processes" (2018). Graduate Theses and Dissertations. 16870.