Adhesion area estimation using backscatter image gray level masking of uncoated tungsten carbide tools

Abstract

Machining tribology research reveals that adhesion of the workpiece material to the cutting tool is an important aspect that governs a number of physical parameters within the tool-chip interface. Adhesion is especially a concern during machining difficult-to-cut materials such as nickel-based alloys. Adhesion may cause tool wear that leads to premature tool failure, or it may trigger other wear modes, hindering machinability and reducing product quality. Many researchers are investigating adhesion fundamentals, adhesion quantification being an indispensable tool. The research suggests strategies to reduce adhesion's undesirable effects. However, when many experimental trials are required, time-consuming adhesion quantification may cause a bottleneck. This study proposes an efficient adhesion area quantification method using image processing of discrete gray intensities on backscatter images of uncoated tungsten carbide inserts, revealed by scanning inserts through an electron microscope. The obtained images are analyzed statistically from their gray-level intensity distribution. A recognizable gray-level range, located around a peak in the histogram, corresponds to adhered material and can easily be distinguished from the background tool material. Hence, adhesion pixels are masked and counted in a statistically controlled manner. The total adhesion area is subsequently quantified by summing the total number of adhesion pixel areas. This method is applied to quantify the resulting adhesion for a variation of the cutting speed, based on a number of orthogonal machining trials carried out on nickel-based superalloy NiCr19Fe19Nb5Mo3 (2.4668), using uncoated tungsten carbide inserts. K-means clustering algorithm is applied for image segmentation, and elemental mapping is obtained by energy dispersive x-ray spectroscopy; both are used to evaluate the effectiveness of the proposed method.