Open AccessPREDICTION OF REGRESSION BASED WEAR BEHAVIOUR MODELS OF ALUMINIUM ALLOY 356 – ZrSiO4 COMPOSITES
Author(s) -
J.Althaf Hasan Khan,
SA Jahan,
Alok Kumar,
V. Murali,
A. Arul Marcel Moshi
Publication year - 2022
Publication title -
journal of manufacturing engineering
Language(s) - English
Resource type - Journals
ISSN - 0973-6867
DOI - 10.37255/jme.v16i4pp124-126
Subject(s) - materials science , taguchi methods , alloy , composite material , aluminium , toughness , aluminium alloy , orthogonal array , casting , ultimate tensile strength , composite number , stiffness , metal matrix composite , metallurgy
The term composite is a combination of two materials with different physical and chemical properties. When combined, they create a specialised material to do a certain job, for instance, to become stronger, lighter or resistant to electricity. They can also improve strength and stiffness. Metal matrix composites have much improved properties, including high tensile strength, toughness, hardness, low density and good wear resistance compared to alloys or any other metal. Aluminium alloys are becoming important today, especially in the automobile, space and electrical industries. Unfortunately, due to poor wear resistance, aluminium alloy can deteriorate quickly. So the present investigation aims at developing Aluminium 356 alloy (AA356) composites reinforced with 5 wt.% Zirconium Silicate (ZrSiO4) with better wear resistance. The composites have been fabricated using the ‘stir-casting’ method in which the particles were added to molten metal during the stirring process at a rotating speed of 700 rpm. A wear test has been performed on a pin on the disc apparatus. Three process parameters have been considered: normal load, sliding velocity, and sliding distance at three different levels. An experimental plan has been made using Taguchi’s L9 orthogonal array table. The output responses such as wear rate and coefficient of friction have been considered for the investigation. Regression models have been generated for each output response. Using the generated regression models, one can predict the value of the output parameters even without actually performing the experimentation within the range of input factor combinations.