Some Implications of Uncertainty in Periodic-review Inventory Theory

Abstract

The first part of this thesis is concerned with finding optimal ordering policies for multistage inventory systems in which stocks are reviewed periodically. A new exact method of calculating critical stock levels is developed in order to avoid difficulties in the classical dynamic programming approach. The equations developed for the new method also provide a clearer picture of the way in which optimal policies for periodic-review processes of finite duration converge to those for processes of infinite duration. Models with lead time are then considered in away which leads to a new generalisation of earlier work on stochastic lead times, Consideration of adaptive inventory systems leads on to the second part of the thesis, where the emphasis is on Bayesian methods of treating imperfectly known demand distribution. Important original results are obtained when the adaptive behaviour of several well known distributions is considered.