A framework for controlling the grain size in friction stir processing -

Abstract

Introduced in this dissertation is a framework for controlling the grain size of Mg AZ31B via friction stir processing (FSP), a solid state processing technique widely recognized for its microstructure refinement capabilities. While other research have attempted to accomplish such a goal via multi-pass FSP, this dissertation proposes to ‘dial-in’ a desired average grain size target to be accomplished in a single pass. At the heart of the framework are a number of mathematical equations, referred to here as ‘framework relations’ that relate the output state variables (Zener-Hollomon parameter and grain size) to the input process parameters (tool rotational speed and traverse feed). Relations between intermediate state variables (temperature and strain rate) and the input process parameters were also developed at select points within the stir zone. Results from 3D thermo-mechanically coupled FEM simulations of a test matrix of the process parameters were used to construct the framework relations. The FE model was also used to optimize the process parameters as well as to investigate effect of in-process cooling on the process. In order to have high fidelity simulations, this work introduces a methodology for selecting suitable constitutive relations of mechanical behavior to be used in FEM friction stir processes. This selection methodology was successfully implemented to evaluate the performance of different constitutive relations in modeling friction stir processing of twin roll cast wrought (TRC) AZ31B. It was concluded that an HCP-specific Zerilli–Armstrong constitutive equation had an advantage over all other considered models and was, therefore, selected as most suitable for numerical modeling of FSP of twin roll cast wrought AZ31B. For this Zerilli-Armstrong constitutive equation, the material constants were determined for TRC wrought AZ31B from experimental tensile tests of this material over a range of temperatures and strain rates. The results of the FEM model were experi