Computational modelling of friction stir processing
thesisposted on 01.03.2017, 05:42 by Fagan, Timothy
Friction Stir Processing (FSP) is a new and exciting processing technique to locally modify the grain structure and improve mechanical properties of metals. Numerical modelling of the process will allow for improved understanding of the deeply complex thermo-mechanical processes that occur within this seemingly simple technique. An accurate numerical model will increase understanding of the process and reduce the number of experimental trials required to achieve the desired result. In FSP, a cylindrical non-consumable tool, generally consisting of a pin and shoulder, is rotated and plunged into the surface of a metal workpiece. The larger diameter shoulder prevents the surface of the metal workpiece from flowing outwards, while the specially designed pin induces a stirring action, producing a combination of frictional and adiabatic heating allowing the metal to flow around the pin from the advancing side to the retreating side. Localised severe plastic deformation occurs generally resulting in grain refinement. It is the grain refinement that has attracted the attention of many researchers and prompted development of numerical models. The typical numerical methods applied in literature rely on Eulerian or Lagrangian meshes which struggle to overcome large mesh deformation and track material history respectively. In addition, phenomenological models are generally applied which do not have a physical basis. To this end, this work will implement a particle based numerical method with a physically based constitutive law. The proposed three-dimensional fully-coupled thermo-mechanical model is able to concurrently determine the temperature field, material flow and microstructure evolution dependent on the processing conditions of FSP. The results of the complete model agree well with experimental thermocouple measurements, material flow and microstructure development.