Design and Development of Aluminium Matrix Composites with Improved Performance
Co Supervisor: Dr. Shubhabarata Datta (SRM Institute of Science and Technology, Tamil Nadu)
The worldwide demand for weight reduction in transport industries urges the optimization in the design of aluminium matrix composites (AMCs) that could perform in a high frictional environment. The present work explores hard alumina ceramic reinforced AMCs for the optimum tribo-mechanical properties computationally, followed by a few experimental trials. In the case of computationally designing, the AMCs ‘Materials Informatics’ approach based on different machine learning tools was employed for modelling and data analysis along with genetic algorithm for design optimization.
The purpose, motivation, and objectives of the present work on designing alumina-reinforced AMCs with improved tribo-mechanical properties were outlined in Chapter 1. A detailed literature review was presented in Chapter 2. It highlighted an understanding of controlling parameters that influence the properties of alumina-reinforced AMCs. A general overview of the material informatics-based design of composites using artificial intelligence and machine learning was summarized in Chapter 3. In Chapter 4, composition- morphology-process-property correlations of alumina reinforced AMCs were developed using suitable data-driven predictive modelling techniques, viz. multiple linear regression, artificial neural network, and neuro-fuzzy hybrids. For mechanical properties, hardness, yield strength, tensile strength, and total elongation of composites were modeled as output responses considering the composition of the matrix alloy (e.g., Si, Fe, Cu, Ni, Mn, Zn, Mg, and Ti), amount and size of alumina reinforcement and heat treatment state as input variables. In the case of tribological properties, wear rate and coefficient were modelled considering parameters related to tribo-system, i.e., load, speed, and hardness of the counter body apart from parameters associated with alloy matrix and reinforcement. Dimensionality reduction and features selection techniques were used to simplify the computational process and identify influential inputs in Chapter 5. The other data-driven modelling techniques, viz. principal component analysis, Rough set analysis, and genetic programming were employed for the comparison on and cross-verifications in Chapter 5. It was found that amongst the element elements, Cu, Mg, Mn, and Ni are the most important for all six tribo-mechanical properties. Characteristics of alumina reinforcement strongly influenced the tribological properties. Optimum values of output variables were obtained using modeling based on multi-objective optimization technique viz. genetic algorithm as detailed in Chapter 6. It was found that aluminium alloy with 4-5 wt.% Cu and 1 wt.% each of Mg and Mn is best as the matrix for AMCs. However, the content and size of alumina reinforced varied with the objectives of the design. Alumina content in the range of 10-15 wt.% provided high strength with excellent tribological properties. The details of the experimental trials with the results of the mechanical, tribological and microstructural characterizations were reported in Chapter 7. The findings of the Chapters 4, 5 and 6 were utilized to design the composites. The composites were developed and characterized experimentally. The designed and developed composites showed quite encouraging mechanical and tribological properties. Savings in terms of cost and time, the present study highlighted an effective scheme for design optimization of advanced composite for a specific purpose.