Toward reliable calculations of heat and plastic flow during friction stir welding of Ti-6Al-4V alloy

Heat transfer and visco-plastic flow during friction stir welding of Ti-6Al-4V alloy have been modeled in three dimensions by numerically solving the equations of conservation of mass, momentum and energy using temperature dependent thermo-physical properties and temperature and strain-rate dependent viscosity values. The computed results showed that five important model parameters, i.e., the spatially variable friction coefficient, the spatially variable slip between the tool and the workpiece, the extent of viscous dissipation, the mechanical efficiency and the spatially variable heat transfer rate from the bottom surface of the workpiece significantly affected both the temperature fields and the computed torque on the tool. An important problem in the modeling of friction stir welding is that the values of these five parameters cannot be specified from fundamental principles and, and as a result, computed results are not always accurate. Here we show that by combining the heat transfer and plastic flow model with a genetic algorithm based optimization scheme, the values of the five uncertain parameters can be determined from a limited volume of experimental data so that the model predictions of peak temperatures and cooling rates match well with the experimental results. The computed results show that for the welding conditions reported in this paper, close to sticking condition prevailed at the tool workpiece interface for all the experiments. The extent of viscous dissipation converted to heat was fairly low indicating lack of intimate atomic mixing in the stir zone. Computed three dimensional pressure distributions and streamlines were consistent with defect-free reliable welds for all conditions of welding studied.