R. Nandan, T. DebRoy, The Pennsylvania State University, University Park, PA; B. Prabu, A. De, Indian Institute of Technology Bombay, Mumbai, India
Numerical simulation of heat transfer and materials flow in friction stir welding (FSW) process has led to a better understanding of both the welding process and the welds. Although the effects of variables such as the weld velocity, tool rotational speed and axial pressure on the temperature fields, cooling rates, weld geometry and torque can be computed, such models are essentially unidirectional. In other words, what is often desirable but cannot be obtained from the current models is the alternative set of welding conditions that can produce a given FSW attribute such as the cooling rate or torque. Here we show that a differential evolution based genetic algorithm can be combined with a transport phenomena based heat transfer and materials flow model to make the model bi-directional. As a result, multiple combinations of welding variables necessary to achieve a target cooling rate or torque can be systematically computed. The model capabilities were tested by extensive experimental work on the FSW of dissimilar welds involving 1000 and 6000 series aluminum alloys. Tailoring FSW weld attributes based on scientific principles provides an alternative to time-consuming and expensive trial and error approach.
Summary: Numerical models for heat transfer and plastic flow in friction stir welding are unidirectional in nature. Here our aim is to provide the ability to tailor weld attributes by combining the transport phenomena based model with genetic algorithms and using experimentally measured thermal cycles for different welding conditions to test and validate the model.
