Chapter 16Probabilistic Inventory Models.ppt

Chapter 16Probabilistic Inventory Models to accompany Operations Research: Applications and Algorithms 4th edition by Wayne L. Winston Copyright (c) 2004 Brooks/Cole, a division of Thomson Learning, Inc. Description No we consider inventory models in which demand over a given time period is uncertain, or random; single period inventory models, where a problem is ended once a single ordering decision has been made; single-period bidding models; version of the EOQ model for uncertain demand that incorporate the important concepts of safety stock and service level; the periodic review (R,S) model, and the ABC inventory classification system. 16.1 Single Period Decision Models In many situations, a decision maker is faced with the problem of determining the value q for a variable. After q has been determined, the value d assumed by a random variable D is observed. Depending on the values of d and q, the decision maker incurs a cost c(d,q). We assume that the person is risk-neutral and wants to choose q to minimize her expected cost. Since the decision is made only once, we call a model of this type a single-period decision model. 16.2 The Concept of Marginal Analysis For the single-period model, we now assume that D is an integer-valued discrete random variable with P(D=d)=p(d). Let E(q) be the decision maker抯 expected cost if q is chosen. Then In most practical applications, E(q) is a convex function of q. Let q* be the value of q that minimizes E(q). From the graph we see that q* is the smallest value of q for which E(q*+1) ?E(q*) ∶ 0. Thus if E(q) is a convex function of q, we can find the value of q minimizing expected cost by finding the smallest value of q that satisfies inequality. If E(q) is not a convex function, this argument may not work. Our approach determines q* by repeatedly computing the effect of adding a marginal unit to the value of q. It is called marginal analysis. 16.3 The News Vendor Problem: Discrete Demand Organizations of