The issue of supply chain safety has received broad attention which has led to a wide range of methodologically different approaches; for a survey see (Pfohl, Köhler & Thomas, 2010). The present paper introduces a novel quantitative algorithm that provides a multiple covering of the commodity graph via constrained clustering. In fact, we construct supply chain components in the overall supply network of a company, each being able to account for some percentage of the company’s overall production. They are all isomorphic to and can hence be viewed as different realizations of the commodity graph which are most independent with respect to known hazards. Consequently, suppliers (of each level) are assigned to supply chain components so as to minimize the probability for a total (or severe enough) breakdown. The basic new model is given in detail, complemented by an outline of more involved ramifications that are able to deal with realistic scenarios. Also, we give computer simulations that indicate the favorable behavior already of our basic model in terms of risk reduction.