Frictions Stir Welding (FSW) is a solid state joining process whereby a rotating tool is plunged into the join line between two clamped workpieces and traversed along the join line. The heat generated by friction at the join line reduces the localized yield stress of the workpiece material and hence can be readily stirred together. The temperature of the workpiece under the shoulder reaches 80%-95% of its melting temperature. The FSW process was analysed using ANSYS parametric models so that the operating parameters could be optimized in conjunction with modeFRONTIER. FSW joint performances are defined by tool rotation, plunge, dwell and feed. Three separate ANSYS FEM models were consequently developed to represent the tool rotation and reacting forces, plunge and dwell phases and the feed rate respectively.
In order to create a thermal model to represent the temperature distribution in the tool, using results from previous experiments found in literature, a modeFRONTIER workflow was developed in conjunction with ANSYS to minimize the difference between the results of the FEM and literature models. The resulting parameters which provided the minimum difference between the two models were subsequently used in the thermal models defined in ANSYS to represent the FSW process.
The desired process objectives were such that the rotational speed was minimized, the plunge rate maximized, the dwell time minimized and the feed rate maximized. Using modeFRONTIER, the process parameters were optimized by defining the relevant parameters as inputs to the respective ANSYS models and the resulting outputs were controlled using the required objectives and constraints. In the modeFRONTIER workflow logic the rotation and force model was executed first, the results of which were processed in Microsoft Excel and transferred to the plunge and dwell model. These results were then transferred to the remaining feed model. The order of the logic flow occurred in sequence similar to that of the actual FSW process. Using a Sobol design of experiments (DOE) and a MOGA-II genetic optimization algorithm, the completely parametric virtual welding model finally allowed an automatic and efficient selection of an optimum operating window for the selected aluminium alloy and thickness combination. The model drastically reduces the number of experimental trials necessary to define a robust “sweet spot” for the manufacturing process.
|