Provide a Modeling Algorithm for Mechanical Properties of Friction Stir Welding of 5 Series Aluminum and Pure-Copper Based on Fuzzy Logic

• Published in: Iranica journal of energy & environment Vol. 13; no. 2; pp. 169 - 175
• Main Authors: Maraki, M. R., Tagimalek, H., Pasoodeh, B.
• Format: Journal Article
• Language: English
• Published: Babol Noshirvani University of Technology 01.03.2022
• Subjects: 5 series aluminum; friction stir welding; fuzzy logic; microhardness; pure copper
• ISSN: 2079-2115; 2079-2123; 2079-2123
• DOI: 10.5829/IJEE.2022.13.02.08

The copper/aluminum composite is very important and practical due to its light weight, optimal thermal and electrical conductivity. The high weight resistance ratio, along with its inherent properties, makes it attractive for new applications. In this regard, the use of composites with high mechanical properties has significantly increased. In this research, 5000 series aluminum and pure copper samples in 1st, 2nd, 3rd, and 4th passes have been subjected to friction stir welding (FSW) and then the mechanical and metallurgical properties of the welded samples have been compared with the original samples. In order to further study the results of tensile tests, metallography and microhardness tests have been performed. Microstructural evaluation of welded samples showed that the mixing zone of the samples was determined by combining aluminum and copper layers. The results showed an increase in yield strength in the welding zone and ultimately an improvement in hardness and ultimate strength in the weld zone compared to the prototype. Compared to stretched samples, the greater the distance from the nugget weld, the less the improvement in mechanical properties and microhardness. By changing the parameters and increasing the inlet temperature, the mixing and uniform dispersion of the particles is performed more appropriately and ultimately increases the tensile strength. Finally, in the research, experimental data were modeled using fuzzy logic method and considering that the presented model was obtained in two indices R-Sq (pred) and R-Sq (adj), 96 and 99%, respectively. The comparison between the experimental data and the model data indicated an acceptable error in the experimental data.