Experimental Evaluation of Interfacial Surface Cracks in Friction Welded Dissimilar Metals through Image Segmentation Technique (IST)

Surface cracks on the friction welded interface of dissimilar metals are one of the earliest indications of degradation of the joint, which is a critical aspect for the welding strength. By manual inspection of the friction welded joint, observations of irregularities, porosity voids, crack lengths, cracked surfaces, and depth penetrations of two dissimilar metals can be made. Manual inspection purely depends on a quality expert’s experience of quantitative analysis and knowledge. In this research, an attempt has been made to effectively utilize the image segmentation technique (IST) in the estimation of the welded surface quality of a dissimilar joint by friction welding. The bonding strength between dissimilar metals in friction welding is more dependent on the coefficient of friction between the metallic surfaces. To demonstrate the capability of the image segmentation technique, experiments were conducted with various parameters, such as friction pressure, friction time, coefficient of friction, and torque speed of the rotating work piece. The effect of the coefficient of friction on friction welded surface quality by considering process parameters is estimated by using the proposed technique. Experiments were validated and the results claimed that the proposed image processing approach is efficient in fractured surface crack detection, reducing the computation cost, and providing a high-speed method with greater accuracy in the identification of welded surface defects. It was found that the friction coefficient is dependent mostly on the friction pressure and friction time. Its values range from 0.21 to 0.71, with the highest value of friction pressure at 120 MPa and 500 rpm. The present work deals with the detection of welding defects by means of segmentation based analysis of the welded interface. This method has a significant improvement in the fractured surface, crack detection, and non-welded areas’ detection in terms of pixels at the desired region, and is easy when compared to conventional detection techniques by using an operator’s decisions.