Modeling of casting microstructures and defects

Abstract

Casting is an ancient art that has been a trial-and-error process for more than 4000 years. To predict the size, shape, and quality of a cast product, casting manufacturers typically cast full-size prototypes. If one part of the process is done incorrectly, the entire process is repeated until an acceptable product is achieved. One way to reduce the time, cost, and waste associated with casting is to use computer modeling to predict not only the quality of a product on the macro- scale, such as distortion and part shape, but also on the micro-scale such as grain defects. Modeling of solidification is becoming increasingly feasible with the advent of parallel computers. There are essentially two approaches to solidification modeling.The first is that of macro-modeling where heat transfer codes model latent heat release during solidification as a constant and based solely on the local temperature. This approach is useful in predicting large scale distortion and final part shape. The second approach, micro-modeling, is more fundamental. The micro-models estimate the latent heat release during solidification using nucleation and grain growth kinetics. Micro-models give insight into cast grain morphology and show promise in the future to predict engineering properties such as tensile strength. The micro-model solidification kinetics can be evaluated using first principles or they can be evaluated using experiments. This work describes an implementation of a micro-model for uranium which uses experimental results to estimate nucleation and growth kinetics.