PMMC cluster analysis
S. Yotte; J. Riss, D. Breysse, S. Ghosh

doi:10.3970/cmes.2004.005.171
 Source CMES: Computer Modeling in Engineering & Sciences, Vol. 5, No. 2, pp. 171-188, 2004 Download Full length paper in PDF format. Size = 234,721 bytes Keywords image analysis, distance, composite, simulation. Abstract Particle distribution influences the particulate reinforced metal matrix composites (PMMC). The knowledge of particle distribution is essential for material design. The study of particle distribution relies on analysis of material images. In this paper three methods are used on an image of an Al/SiC composite. The first method consists in applying successive dilations to the image. At each step the number of objects and the total object area are determined. The decrease of the number of objects as a function of the area is an indicator of characteristic distances. The second method is based on dilations of one particle among all the others. Then each time it touches a neighbor the number of the step i of the process is recorded and gives the distance to the n$^{\relax \fontsize {8}{9.5}\selectfont {\unhbox \voidb@x \hbox {th}}}$ neighbor. This is done for each particle of each image. Thus statistical parameters of the distribution of the distance to the six first neighbors are obtained and compared to the previous characteristics. The third method is the covariance method. These three methods are tested on synthetic images of known characteristics. Then the Al/SiC image is analyzed and once the characteristics are identified a statistically identical image could be created later.