Full-field simulations of dynamic and meta-dynamic recrystallization
Shah, Vitesh; Roters, Franz (Thesis advisor); Korte-Kerzel, Sandra (Thesis advisor); Bos, Cornelis (Thesis advisor); Diehl, Martin (Thesis advisor)
Aachen : RWTH Aachen University (2022, 2023)
Dissertation / PhD Thesis
Dissertation, Rheinisch-Westfälische Technische Hochschule Aachen, 2022
The challenges related to sustainability and reduction of carbon dioxide emissions require development of new alloys. These newer sustainable alloys should be able to achieve material properties that are at least similar to the conventional alloys albeit with higher recycled scrap content or produced using alternative processing routes. Development of new alloys involves development of new processing routes optimized for such alloys. However, this requires extensive laboratory scale experiments and plant trials. This makes alloy development a time-consuming and expensive process involving significant material waste and loss of potential revenue. Simulations which capture the physics and metallurgy of diﬀerent processes along the production chain have potential to speed up new alloy development. This could be achieved by utilizing the knowledge gained from such simulations to design a limited number but eﬀective plant/laboratory trials. Such simulations can be a part of a ‘virtual laboratory’, where diﬀerent parameters can be tested to provide in-depth data necessary for improved understanding. It is in the interest of industry if laboratory scale experiments and plant trials can be reduced by such simulations as it can reduce the costs and speed up alloy development at the same time. One of the more challenging processes to understand, control and get in-depth data for better understanding is hot-rolling. The high temperatures, high deformation levels and high strain rates involved make this process diﬃcult to replicate in a laboratory. Therefore, simulations that can capture the microstructure evolution during hot-rolling are of great interest. Dynamic Recrystallization (DRX) and Meta-Dynamic Recrystallization (MDRX) are important processes that occur during hot-rolling. These metallurgical processes inﬂuence the microstructure evolution during hot-rolling. However, these metallurgical processes are diﬃcult to study due to concurrent plastic deformation and microstructure evolution. In this thesis, a simulation framework that can simulate DRX and MDRX has been developed. This has been achieved by sequentially coupling a full-field dislocation density based crystal plasticity model and a cellular automaton model. A regridding procedure is used to transfer data between the deformed mesh of the large-strain crystal plasticity model and the regular grid of the cellular automaton. The modified microstructure from cellular automaton is introduced back into the crystal plasticity model for further deformation. This approach is able to model the concurrent plastic deformation and microstructure evolution. A physics based nucleation criterion has been developed based on dislocation density diﬀerence and changes in orientation due to deformation. A coupling criterion based on maximum possible growth of a random interface is used to determine the time at which the simulation framework switches from crystal plasticity model to cellular automaton. The criterion ensures that the switch to Cellular Automata (CA) occurs only when this maximum possible growth of a random interface becomes significant. This helps in reducing the number of computationally expensive coupling steps necessary to simulate DRX. The Crystal Plasticity Fast Fourier Transform (CPFFT) method combined with the cellular automaton method along with the coupling algorithm enables DRX/MDRX simulations of virtual microstructures undergoing large deformations with a high number of orientations/grains and a reasonable spatial resolution. The developed simulation framework has been used to study MDRX during double-hit compression tests. The characteristic features of the mechanical response in case of MDRX, such as reduction in yield strength in the subsequent deformation step due to recrystallization and higher work-hardening rate of partially recrystalllized microstructures are reproduced by the model. Moreover, analysis of fully recrystallized microstructures showed that the recrystallized grain sizes are in similar range of magnitude as observed experimentally. This suggests that the number of nuclei detected by the nucleation criterion are in the correct range of magnitude. Both the nuclei orientations and the final recrystallization texture are correctly captured by the model. The analysis of the MDRX kinetics showed non-constant Avrami exponents with low values, mainly driven by impingement of recrystallized grains. Therefore, the model is shown capable in capturing the physics relevant to DRX and MDRX. The coupling frequency between the crystal plasticity model and the cellular automaton model, strongly controls the DRX behaviour. However, it is also shown that for the current simulation framework, the highest coupling frequencies are not ideal to capture DRX reasonably. This makes the simulation framework quite attractive as reasonable DRX simulations can be performed with lower coupling frequencies and thus, it provides a pathway for computationally less expensive DRX simulations. The simulation framework allows investigation of microstructure evolution during DRX in detail. It is observed that once DRX has initiated, the most favourable interface for nucleation is the interface between a recrystallized grain and an original grain. This leads to the phenomenon of necklacing which is commonly observed in DRX. The simulations also show that the mechanism behind nucleation at an interface between a recrystallized grain and an original grain is due to progressive grain rotation due to deformation accumulation in recrystallized grains. It is shown that the DRX grains stop growing not only due to deformation accumulation but also due to impingement from the new nuclei. As the simulation framework allows large deformation, it has been possible to apply it to investigate multi-stand hot-rolling. The eﬀects of strain distribution across the rolling stands and alloy chemistry on microstructure homogeneity have been studied. In all considered cases, it is observed that DRX and MDRX lead to grain refinement across hot-rolling. It is observed that DRX promotes MDRX during interpass annealing. Moreover, it is seen that a uniform distribution of strain over all stands results in a more homogeneous microstructure with finer grain size, because DRX occurs even in case of the final stands. The case of higher alloying content results in inhomogeneous microstructures with larger grain sizes due to lower grain boundary migration rates. To summarize, a model to simulate DRX and MDRX during hot-rolling is presented in this thesis. These simulations can be applied to multi-stand hot-rolling of realistic virtual microstructures. This enables the possibility to gain detailed insights into these processes. This framework can be a part of a ‘virtual laboratory’, which can be used at the design stage of new alloys to investigate the eﬀects of alloying chemistry or processing parameters on microstructure evolution during hot-rolling.
- Division of Materials Science and Engineering 
- Chair of Materials Physics and Institute for Physical Metallurgy and Materials Physics