On the polyhedral structure of two-level lot-sizing problems with supplier selection

In this article, we study a two-level lot-sizing problem with supplier selection (LSS), which is an NP-hard problem arising in different production planning and supply chain management applications. After presenting various formulations for LSS, and computationally comparing their strengths, we explore the polyhedral structure of one of these formulations. For this formulation, we derive several families of strong valid inequalities, and provide conditions under which they are facet-defining. We show numerically that incorporating these valid inequalities within a branch-and-cut framework leads to significant improvements in computation.