Abstract
An efficient Monte Carlo procedure is applied to the study of dynamic recrystallization in polycrystals. The approach is essentially the same as in part I of this series of papers and consists of representing the microstructure by a discrete two-dimensional lattice. The crystallographic orientation of each lattice element and its neighbours determines the position of grain boundaries and their energies. Hot deformation is modelled by adding recrystallization nuclei and stored energy continuously with time. Nucleation of recrystallization is modelled by introducing a three-site nucleus and is represented by two generic types of model: either random nucleation or preferential nucleation in regions of high stored energy. An energy-storage-rate schedule was used to model both work hardening and dynamic recovery of metals by way of a stage III work-hardening law. The novel approach taken in this paper was to model a non-uniform deformation substructure by using a mean cell size L and a mean distance R between cell-wall dislocations. Experimental evidence suggests that both L and R are inversely proportional to the deformation stress. These observations were used to develop a phenomenological expression for heterogeneous distribution of the stored energy used in the simulations. The observed dependencies of the evolution of stress and mean grain size with strain were similar to those obtained with the use of a homogeneous stored-energy distribution. However, the relations obtained between transient and steady-state parameters show distinct differences and demonstrate that any realistic model of dynamic recrystallization must account for the presence of the dislocation substructure. Finally, the kinetics of dynamic recrystallization in the simulations were shown to be sensitive to the rate of deformation and to be consistent with experimental observations.