Non-stationary ordering policies for multi-item inventory systems subject to a single resource constraint

This paper studies the multi-item inventory problem with a single constraint. Two well-known approaches to this problem are the Lagrangian method and the fixed-cycle approach. Both of these methods generate stationary ordering policies. A competing approach is presented here that generates non-stationary ordering policies, with order quantities that vary over time. These non-stationary policies are, in all cases, preferred to those generated by the Lagrangian approach, and in some cases preferred to those generated by the fixed-cycle approach. Computational results comparing the three methods are also given.