A recent article in the Journal of Manufacturing and Materials Processing explored the friction stir processing (FSP) of Cu over Al1050 aluminum alloy. The study combined machine learning (ML) techniques to optimize the surface alloying process with experimental measurements to analyze the properties of the resulting surface alloy.
![Friction Stir Surface Alloying of Al1050-Cu Alloy](https://d12oja0ew7x0i8.cloudfront.net/images/news/ImageForNews_63482_17234526370029422.jpg)
Image Credit: Evannovostro/Shutterstock.com
Background
FSP has gained popularity for processing the surfaces of aluminum-based alloys, composites, and copper alloys. High-strength aluminum alloys are particularly valued for their excellent strength-to-weight ratio, cost-effectiveness, and superior wear resistance, making them ideal for lightweight structures in aviation and automotive industries.
Reinforcing copper alloys enables easy modification of aluminum using low-cost casting and mixing methods. FSP can enhance the microstructure and mechanical properties of both aluminum and copper alloys, but research on the process parameters and fatigue behavior of dissimilar friction stir welds between different alloys remains limited.
ML provides a robust framework for analyzing complex datasets and revealing intricate relationships between process parameters and material characteristics, facilitating more precise predictions in FSP applications.
This study focused on the FSP-based processing of Al1050 aluminum alloy, introducing Cu powder to its surface using experiments and the ML method of Genetic Programming (GP).
Methods
An Al1050 sheet, 5 mm thick, served as the base metal, while Cu was used as the alloying powder for FSP. The groove method of FSP was applied using a universal milling machine, creating a groove 2.5 mm deep and 1 mm wide along the rolling direction of the aluminum surface. The cu powder was then filled into the groove, which was sealed with a pin-less tool.
Various FSP parameters, including tool rotation speed (1250 and 630 rpm), feed rate (20, 50, and 80 mm/min), and the number of passes (1, 3, and 6), were varied to produce different alloy specimens.
The mechanical properties of the resulting surface alloys were analyzed through tensile and microhardness tests. Microhardness was measured using a Vickers machine with a 100 g load, while an optical microscope was employed to assess the microstructural changes in the alloys.
The ML method employed in this research was Genetic Programming (GP), a symbolic optimization technique. Specifically, arithmetic operations and mathematical functions in the standard GP algorithm were used for model representation. The proposed ML method was implemented using MATLAB software under the freely provided Toolbox.
The relevant parameters for the GP model were chosen according to complexity, computational viability, and appropriate solution exploration. Finally, the GP-predicted hardness and ultimate tensile stress values were compared with those experimentally measured.
Results and Discussion
FSP enhanced the mechanical properties of Al1050-Cu alloy, including microhardness and tensile strength. The cross-section of the FSP alloyed specimens exhibited various zones: thermomechanical zone (TMAZ), stir zone (SZ), heat zone (HAZ), and base metal (BS). The presence of Cu particles in the cross-sections indicated successful alloying, though some large agglomerated Cu particles were also observed.
Different processing parameters influenced the alloy hardness. Generally, higher tool rotation speed resulted in a higher hardness, attributed to Cu traces and powder agglomeration. However, the hardness decreased with the increasing number of passes due to increased heat input, which promoted grain growth and reduced hardness.
A higher feed rate improved alloy hardness by reducing heat input to the alloyed zone, limiting grain growth. This trend, however, varied in some cases due to powder agglomeration at high feed rates. Increasing the number of passes enhanced alloy strength, resulting in better powder distribution and a more homogeneous microstructure.
The coefficient of determination (R2) between the GP-predicted and observed hardness were 0.1633 and 0.5277, respectively. The R2 between the predicted and observed tensile stress values were 0.7288 and 0.7621, respectively.
The mean absolute error (MAE), mean square error (MSE), and root mean square error (RMSE) between the predicted and experimentally recorded hardness of FSA products were 10.65, 1.88, and 13.70, respectively. The corresponding values for tensile stress were 12.95, 2.39, and 15.47, respectively.
Conclusion
This study demonstrated the effectiveness of FSP in fabricating an Al1050 surface alloy with Cu powder, significantly enhancing its mechanical properties. The base metal strength increased from 115 MPa to 192 MPa due to FSP.
The initially non-uniform stir zone was homogenized with increased passes, resulting in better integration of compact copper particles with aluminum. Consequently, the base metal's strength improved, attributed to copper reinforcement and grain refinement. Parameters such as tool rotation speed, feed rate, and the number of passes significantly influenced the alloy’s microstructural evolution and mechanical behavior.
ML provided valuable insights into the mechanisms governing alloy formation during FSP. The findings from this study can inform the development of high-performance surface alloys for various industrial applications.
Journal Reference
Pedrammehr, S., et al. (2024). Experimental and Machine Learning Study on Friction Stir Surface Alloying in Al1050-Cu Alloy. Journal of Manufacturing and Materials Processing. DOI: 10.3390/jmmp8040163, https://www.mdpi.com/2504-4494/8/4/16
Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.