Numerical and experimental study of constrained solidification
Abstract
Infiltration and solidification processing has been studied to produce selectively-reinforced metal-matrix composites. It has been observed that when the spacing between reinforcement phases in the preforms is small enough, the microstructure of the metal is modified and microsegregation reduced due to the constraint of the reinforcement phase. Different aspects of the constrained solidification have been studied experimentally and numerically, but the mechanisms for the change of microstructure and microsegregation are not fully understood. In order to study the effect of the reinforcement on dendritic solidification, a 2-D cellular automaton for solid-liquid phase change coupled with finite volume calculation of heat and mass diffusion is developed. The behavior of the model with various numerical parameters, such as time step and grid size, is studied and an operation window for producing reliable results is found. Constrained solidification through preforms of parallel planar channels and stacked bars with various spacings is simulated with the CA-FV model for Al-4.5 wt% Cu. It has been found that as the spacing decreases below the primary dendrite spacing, lateral solute diffusion in the liquid is first constrained, resulting in shorter secondary arms which may disappear at a later stage due to coarsening, but the tip velocity is not affected. When the spacing decreases further, the liquid diffusion layer at the growth front is also constrained, and the growth velocity is decreased, leading to higher undercooling at the growth front. As a result, the copper concentration in the primary phase is higher. Because the primary phase is cellular and tends to fill the small spacings, solute in the liquid is pushed ahead of the growth front rather than trapped behind, resulting in lower eutectic fraction. Infiltration and solidification experiments are also conducted for the study of solidification through parallel planar channels of different widths. The numerical prediction matches well with the experimental results.
Degree
Ph.D.
Advisors
Trumble, Purdue University.
Subject Area
Materials science
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.