This work focuses on the optimization of the returnable container management problem in a global and sustainable supply chain context. The problem consists in: (1) choosing the best way to reposition empty containers between several sites for next rounds of use and (2) purchasing the right quantity of new containers at the right period to meet system requirements for product transportation. For this problem a linear integer program model is proposed. Based on some partial network flow properties, a decomposition into two independent models, purchasing and transportation, has been developed to determine an optimal policy. The benefits of this decomposition, on the technical and industrial level, will be presented and discussed.