A Study of Vegetable oil Trim E950 During Machining of SAE AMS4413B Alloy using DOE and Machine Learning Model
Main Article Content
Abstract
The lightweight Aluminum alloys are the optimal material choice for aerospace industries compared to other metals. The Aluminum Lithium (Al Li) alloys as a light weight alloy have lower density and improves weight carrying capacity of the aircrafts. An attempt is made to understand the efficient machining of Aluminum Lithium aerospace alloy (SAE AMS4413B) for achieving high quality machining such as Surface Roughness (Ra) and Heat Affected Zone (HAZ) by varying input parameters such as cutting speed, feed, depth of cut, coolant concentration, and tool nose radius. SAE AMS4413B alloy using Vegetable oil Trim E 950 which has been chosen as an environmentally friendly cutting fluid that fulfills the important aspects of biodegradability and sustainability. The machinability study is conducted by a robust DOE based technique. The response variations of the output parameters studied using Central composite design. The individual and interaction effect of the input parameters on Ra and HAZ are studied and Analysis of Variance (ANOVA) applied for checking the model adequacy. The results obtained for R2 values of 95.01% for Ra and 96.86% for HAZ. The results revealed that the vegetable oil gives the lower roughness values of 0.177µm and 0.211µm at both 5000 rpm and 17000 rpm respectively with 12% coolant concentration, 0.5mm depth of cut and 0.4 tool nose radius. The HAZ value of 20.82 IACS were obtained at the 5000 rpm and 11000 rpm respectively for 2.5 mm depth of cut and 8 & 10% coolant concentration. Thus, there exists a possibility for the use of vegetable oil for machining of Aluminum Lithium SAE AMS4413B alloys. The integration of Machine Learning (ML) in mechanical engineering research has led to improved predictive modeling and validation techniques. This study employs a Random Forest Regressor model to validate the experimental results. The dataset, obtained from controlled experiments, was used to train the ML model, achieving a high degree of accuracy in validating the experimental results. The study highlights that ML-based predictive models can significantly reduce the time and cost associated with traditional experimental methods.