A greedy heuristic for a three-level multi-period single-sourcing problem

In this paper we consider a model for integrating transportation and inventory decisions in a three-level logistics network consisting of plants, warehouses, and retailers (or customers). Our model includes production and throughout capacity constraints, and minimizes production, holding, and tansportation costs in a dynamic environment. We show that the problem can be reformulated as a certain type of assignment problem with convex objective function. Based on this observation, we propose a greedy heuristic for the problem, and illustrate its behaviour on a class of randomly generated problem instances. These experiments suggest that the heuristic may be asymptotically feasible and optimal with probability one in the number of customers.