Abstract: The distribution network design of substitute products under the risk of supply failure was studied， and a mixed integer programming model including manufacturers’ production design， capacity allocation， distribution center location and emergency inventory decision was established. In the model， emergencies causing supply chain failure are represented by scenarios， and within the same scenario， more than one node of the distribution network may break down， which may result in a sharp increase of problematic scenarios. The sample average approximation method (SAA) is used for solution， whose validity is verified by experimentally comparing the gaps between the upper bound and the lower bond. Finally， the reliability of supply chain execution， and the relationship between product substitution degree and supply chain profit are analyzed.

Key words: supply failure risk, substitution, distribution network, optimal design