Selecting Distribution Centers in Disaster Management by Network Analysis and Composition of Probabilistic Preferences

Abstract

AbstractDisaster management operations require adequate and rapid planning to efficiently serve the affected populations. In this context, an Area of Operations must be dimensioned with the main locations capable of receiving logistic facilities. These locations need to be connected by networks capable of ensuring the flow of critical supplies. The objective of this article is to identify the most influential nodes in a disaster management supply chain, by network analysis and a multicriteria decision aid method. Network analysis provides measures that indicate the most influential nodes in the network. These measures, called centralities, are selected through Principal Component Analysis (PCA). The nodes and their centrality evaluations configure a decision matrix to be modelled by the Composition of Probabilistic Preferences (CPP). CPP is an alternative to other MCDA methods used in this type of problem, such as TOPSIS, by exploring its non-linearity to highlight the best nodes in the network. The R software and specific packages were used to compute results. An undirected and weighted network of 12 nodes was used to test the model. Two nodes identified the best access to the affected area and suggested the best location to establish the distribution center or other logistics facilities.KeywordsDisaster managementNetwork analysisComposition of probabilistic preferences