Application of Finite Element Model and Artificial Neural Network in Characterization of Al Matrix Nanocomposites Using
- PDF / 1,051,471 Bytes
- 8 Pages / 593.972 x 792 pts Page_size
- 28 Downloads / 211 Views
TRODUCTION
ALUMINUM alloys are promising materials in high technology fields owing to their excellent specific mechanical properties.[1–4] Aluminum matrix composites (AMCs), in which hard ceramic particles are dispersed in a relatively ductile Al matrix, have widespread applications in the areas of ground transportation (auto and rail), thermal management, aerospace, industrial, recreational, and infrastructure industries owing to functional properties that include high structural efficiency, excellent wear resistance, and attractive thermal and electrical characteristics.[5–7] The size of particulate reinforcements in AMCs generally ranges from a few micrometers to several hundred micrometers.[8,9] A decrease of the reinforcement particle size from micrometric to nanometric scale brings a superior increase in the mechanical strength of the composite, but the tendency of particle clustering and agglomeration also increases.[10] It is important to note that a homogeneous distribution of the reinforcing particles is essential for achieving the improved properties.[11,12] In comparison with Al matrix microcomposites, the research on Al matrix nanocomposites is still limited. The key reason is perhaps related to the difficulty in synthesizing Al matrix nanocomposites due to the higher agglomeration and clustering of particles. At present, particulate-reinforced composites are being produced by different methods, such as stir casting, powder metallurgy, and spray deposition technique.[9] Among these methods, stir casting is considered to be easily adaptable and economically viable due to its low MOHSEN OSTAD SHABANI and ALI MAZAHERY, Researchers, are with Karaj Branch, Islamic Azad University, Karaj, Iran. Contact e-mail: [email protected] Manuscript submitted March 14, 2011. Article published online February 11, 2012 2158—VOLUME 43A, JUNE 2012
processing cost and high production rate. An additional benefit of this process is the near-net-shape formation of the composites.[6,7] A number of reports are available on Al alloys + Al2O3 composites. However, the effects of nano-Al2O3 particles on the microstructure and mechanical properties are meager. Therefore, the objective of the present study is to develop a stir casting process to produce nano-Al2O3 reinforced cast Al alloy matrix composites, and to investigate their microstructure and mechanical properties. The microstructure and mechanical properties of the as-cast composite materials are compared with unreinforced materials. The modeling studies were also carried out on the mechanical properties of nano-Al2O3 reinforced A356 matrix composites. Neural networks are powerful tools for approximation of unknown nonlinear functions and have gained wide applications in a variety of fields. The neural network is able to learn the basic relationship from a collection of training samples.[13–16] Five types of training algorithms have been used in the literature in order to predict the experimental results. (1) Random order incremental training with learning functions trains a
Data Loading...
