Computational modeling of heat transfer and visco-plastic flow in Friction Stir Welding
Open Access
- Author:
- Nandan, Rituraj
- Graduate Program:
- Materials Science and Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- August 22, 2008
- Committee Members:
- Tarasankar Debroy, Committee Chair/Co-Chair
Long Qing Chen, Committee Member
Christopher L Muhlstein, Committee Member
Panagiotis Michaleris, Committee Member - Keywords:
- torque
reliability
differential algorithm
viscoplastic flow
friction coeffiecient
finite volume method
parametric study
friction stir welding
aluminum alloys
heat generation
heat transfer
variable slip - Abstract:
- Friction stir welding (FSW) is a relatively new welding technique and a review of pertinent literature reveals that a quantitative understanding of the process is just beginning. FSW is characterized by complex physical processes like non-Newtonian viscoplastic flow, frictional and deformational heat generation and stick--slip flow boundary condition at the tool workpiece interface. To add to the complexity, large convective heat transport aided by viscoplastic material flow makes the process three dimensional in nature. A review of literature reveals the following gaps in numerical modeling of FSW: (a) no three dimensional model exists which considers spatially variable heat generation, variable slip condition at the tool workpiece interface and viscoplastic flow, (b) robustness of existing models has not been tested by modeling FSW of different alloys, (c) the transport of alloying elements across weld interface in dissimilar welds has not been studied numerically and (d) the existing models do not have mechanism for improving reliability and they cannot work backwards, i.e. provide a set of welding process variables that will result in the desired weld characteristics. The goal of this thesis is to address these important issues. With a focus to develop a quantitative understanding of the FSW process, a comprehensive three dimensional heat transfer and plastic flow model is developed. The model can predict variables such as temperature and velocity fields and torque based on the given welding parameters like weld velocity, tool rotational speed and axial pressure. It considers tool design dependent spatially variable heat generation rates, deformational work, non-Newtonian viscosity as a function of local strain rate, temperature and the nature of the material and temperature dependent thermal conductivity, specific heat capacity and yield stress. It is shown that the temperature fields, cooling rates, the plastic flow fields and the geometry of the thermo-mechanically affected zone (TMAZ) can be adequately described by solving the equations of conservation of mass, momentum and energy in three dimensions with appropriate boundary conditions and constitutive equations for viscosity. The model is tested for four different alloys: 1) AA 6061-T6, 2) 1018 Mn steel, 3) 304L stainless steel and 4) Ti--6Al--4V which have widely different thermophysical and rheological properties. Numerically computed temperature fields, variations of peak temperatures with FSW variables and TMAZ geometry were compared with the experimental results. Currently, due to unknown parameters in existing transport phenomena based models, the computed temperature and velocity fields and torque may not always agree with the corresponding experimentally determined values and may not show the same trend as experimental results for a range of welding variables. Here, it is shown that this problem can be solved by combining the rigorous phenomenological process sub-model with a multivariable optimization scheme called Differential Evolution. The values of the uncertain model input parameters from a limited volume of independent experimental data which includes temperature measurements obtained using thermocouples and torque measured using dynamometers. This approach resulted in agreement between the phenomenological model and the experimental results with a greater degree of certainty. It is tested for FSW of: 1) dissimilar AA 6061-T6 to AA 1200, 2) 1018 Mn steel and 3) Ti--6Al--4V. Independent thermocouple and dynamometer measurements are also used for validation and verification of results. Improvement in the reliability of the numerical model is an important first step towards increasing its practical usefulness. Also, one of the reasons why current models do not find extensive applications is because they cannot be used to tailor weld attributes. The aim of the present research is to develop a reliable bi--directional model which can find wide use in manufacturing and process control. It is shown that by coupling a reliable model with an evolutionary search algorithm, we can find multiple sets of welding parameters to achieve a target peak temperature and cooling rate in welds. The model is tested for dissimilar welds of AA 6351 and AA 1200. FSW is being increasingly used for dissimilar metal joining. Models are needed to calculate the redistribution of alloying elements when two alloys with dissimilar alloying element contents are joined. The transport and mixing of magnesium from Mg--rich AA 6061 alloy into a commercially pure aluminum AA 1200 was examined experimentally and numerically at various locations in the welded workpiece. The concentration of the solute is measured in transverse cross-sections across the weld-center line at various depths from the top surface of the workpiece. The measurement was done using electron probe micro-analysis (EPMA) of polished transverse-cut friction-stir welded samples. The comparison of the experimental and computed concentration profiles of magnesium shows imperfect mixing of the plasticized alloys during FSW. The plasticized material seem to move in layers without significant diffusive interlayer mixing. A comprehensive model for FSW is developed with capability of calculating temperature fields, material flow patterns and concentration fields in both similar and dissimilar welds in three dimensions. The model is tested for the FSW of alloys with widely different thermophysical properties. A mechanism for improving reliability and ability to provide guidance to tailor weld attributes is incorporated into the model to increase its practical usefulness. This is done by by combining the transport phenomena based model with Differential Evolution algorithm to minimize the objective function based on limited volume of experimental thermal cycles and torque measurements.