Contributions of the study of linkages in multi-echelon distribution systems

Most of the research on multiechelon distribution systems ignores the interdependence between echelons, and it assumes that echelon decomposition gives near optimal policies. The exceptions are the studies that consider (S-1.S) policies, some cases of (Q,r) systems and the recent work on "push Systems" This dissertation addresses some of issues related with the stochastic linkage between locations in the context of system operating under (s,S) policies. In the first part a framework for the study of the relationships between number of echelons, lead time and variability of demand is developped. The second part addresses the problem of determining optimal inventory policies when demand is serially correlated. Several structural results are derived. The third part studies the stabilizing effect of inventories in multiechelon manufacturing/distributions systems. Using the above framework we reexamine some of the recent results, presented in the literature, suggesting that inventories have a destabilizing effect. The fourth part examines the problem of estimating the distribution of demand at higher level stocking locations in a multiechelon system. A procedure for identifying the appropriate distribution family and for parameter estimation is presented. This procedure uses information based on demand and policy parameters for the lower level locations