This dissertation explores macrosegregation in directionally solidified aluminum castings. Two methods of interpolating thermocouple data are presented. A method using Lagrangian polynomials to interpolate thermocouple profiles is described and gives the best results for steady state furnace conditions. Using cubic splines to interpolate temperatures works best under transient conditions. A simple model, neglecting convection, is presented for predicting macrosegregation during melting, holding, and solidification of a sample and is compared with existing models. The model is able to accurately capture macrosegregation in microgravity experiments and is verified by experimental results. A two dimensional model of solidification, including convection, is presented and used to simulate samples grown in microgravity and terrestrially. The terrestrial samples exhibit steepling convection, while the microgravity samples do not. Causes of the steepling convection are explored and quantitative comparisons are made against experimental samples, with good agreement. The role of the furnace temperature profile is discussed and it is shown how it can be used to manipulate the steepling convection. Simulations of directional solidification through changes in cross section are presented for four experiments in graphite molds and one hypothetical experiment in an alumina mold. When solidifying through a contraction in cross section, the mold material is shown to have a strong influence on the convection and resulting macrosegregation. When solidifying out of an expansion, there is less of a difference between the two mold materials. Qualitative comparisons are made against experimentally obtained microstructures and good agreement is found. Stray grains were found, at the expansion, in some of the experimental samples and an explanation based on the results of the simulations is given.