Finite element modeling of the linear friction welding process

Linear Friction Welding (LFW) is a solid-state joining process used in high-added-value assemblies such as Blade Integrated Disk in the aeronautic industry. Four phases compose the LFW process: Initial, Transition, Equilibrium and Deceleration phases. It is a complex thermomechanical process with severe deformation that results in significant grain refinement. In the present research, a chained numerical model is proposed characterized by low computational calculation time for each phase along with specific outputs such as microstructural evolution. Ti-6Al-4V alloy was used as a case study to validate the reliability of the developed model. A thermo-mechanical analysis is carried out to define the transition between the Initial and the Transition phases. Two thresholds were defined depending on the approach used to simulate the Initial phase: Thermo-Mechanical Analysis (TMA) or Heat Transfer Analysis (HTA). Under a TMA approach, it has been shown that the Initial phase shall stop when the equivalent Von Mises stress at the edges reaches the yield stress to comply with the literature definition. It provides an accurate definition for numerical implementation and phase separation. Based on this analysis, a thermal threshold of 975 °C at the edges was proposed as an end phase criterion for the Initial phase simulated under a HTA approach. Three numerical models were built in this study. Two models are made to simulate the Initial phase with, on the one hand, a TMA approach and, on the other hand, a stationary TMA approach to identify the most efficient one. A subroutine was used in the stationary TMA approach to account for the heat generated by the oscillations. After comparison with experimental data from the literature, it was determined that the stationary TMA approach reduces the computational run time up to 99% with equivalent results to the TMA approach. The third model simulates the Transition and Equilibrium phases using a Single Body (SB) modeling approach to guarantee the axial shortening continuity. Furthermore, a subroutine was added to the SB to include microstructure transformation in the simulation. Considering the lack of information in numerical models about deformation parameters during the LFW, the JMAK equation was introduced as a microstructural transformation indicator. The latter counts the number of times that the recrystallized volume fraction reaches 0.99 , a value for which it is assumed that a grain is fully recrystallized. The material is then considered in an annealed state and is able to recrystallize again if further deformation occurs. The microstructure transformation indicator agrees with experimental metallography observations and helps to identify regions with high probability of strong transformation