HEURISTICS FOR COMPLEX INVENTORY SYSTEMS (Tinbergen Institute Research, No 54)

Efficient algorithms to find cost minimizing strategies for complex stochastic inventory systems fail to exist. A practical approach is to apply a rolling horizon procedure. Periodically a deterministic problem is solved and the decision for the first period is implemented in the stochastic system.

In the first part of this book deterministic single-product, multi-product and distribution problems are examined. Heuristic and optimal solutions for problems with constant and time-varying demand are discussed. Special attention is paid to the choice of the planning horizon. In the second part rolling horizon procedures are applied in stochastic single-product, multi-product and multi-retailer systems.

Average costs are compared with costs of alternative strategies or with a lower bound on the optimal costs. Furthermore, the usefulness of the generated replenishment schedules is considered, that is, the system nervousness. Several ways are investigated to reduce the average costs and the system nervousness. System dependent procedures are developed to adjust the solution of the deterministic problem before implementation in the stochastic system.

Extensive simulation experiments show the usefulness of rolling horizon procedures for practical problems.