Characterization of flee structure using cluster analysis

This work is based on a numerical classification method (Forgy's method or c-means clustering) to investigate flee structure, taking into account a ph ysically well-founded model. Two different kinds of morphologies, voluminou s, irregularly shaped, fluffy flocs and dense, regularly shaped, compact, p ellet-like aggregates, are analyzed. These structures are embedded in glyco l methacrylate and cut into thin sections by means of a motorized microtome . Scanning electron microscopic observations show that no serious damage to the aggregate structure due to the preparation technique is apparent. A se quence of image analysis steps reduces the total image information of the s ections, taken by a light microscope, to a few pertinent measurements, appl icable to cluster analysis. Thus a cluster-cluster aggregation mechanism is confirmed experimentally. The flee structure is separated into two cluster s. Cluster 1 consists of six microflocs and cluster 2 is made up of two sub -units, which are again subdivided into six and three microflocs. The pelle t-like aggregate is assembled from ten particles. Further interpretation of aggregate growth history remains unconsidered owing to the expected errors . In a second paper more aspects concerning cluster analysis of flocs will be published.