A comparison RSM and ANN surface roughness models in thin-wall machining of Ti6Al4V using vegetable oils under MQL-condition

Abstract

Thin-wall components as usually applied in the structural parts of aeronautical industry require significant challenges in machining. Unacceptable surface roughness can occur during machining of thin-wall. Titanium product such Ti6Al4V is mostly applied to get the appropriate surface texture in thin wall designed requirements. In this study, the comparison of the accuracy between Response Surface Methodology (RSM) and Artificial Neural Networks (ANN) in the prediction of surface roughness was conducted. Furthermore, the machining tests were carried out under Minimum Quantity Lubrication (MQL) using AlCrN-coated carbide tools. The use of Coconut oil as cutting fluids was also chosen in order to evaluate its performance when involved in end milling. This selection of cutting fluids is based on the better performance of oxidative stability than that of other vegetable based cutting fluids. The cutting speed, feed rate, radial and axial depth of cut were used as independent variables, while surface roughness is evaluated as the dependent variable or output. The results showed that the feed rate is the most significant factors in increasing the surface roughness value followed by the radial depth of cut and lastly the axial depth of cut. In contrary, the surface becomes smoother with increasing the cutting speed. From a comparison of both methods, the ANN model delivered a better accuracy than the RSM model.