Optimal Policies and Approximations
For the inventory systems investigated in the previous chapters, the stationary optimal policy can be determined in principle by finding the global minimum of the corresponding cost rate function. (For previous work in this area, see, for example, Veinott and Wagner, 1965; Archibald and Silver, 1978; and Federgruen and Zipkin, 1984). However, computational difficulties make the exact models unattractive in practice. This has motivated several approaches for the determination of approximately optimal (s, S) policies that require less computational effort and demand information. For continuous review systems, most of the earlier work was confined to the unit demands case under an order-quantity, reorder-point policy. This literature is based on approximations derived by Hadley and Whitin (1963) and Wagner (1969) which require the simultaneous iterative solution of two equations (cf. Gross and Ince, 1975; Gross, Harris and Roberts, 1972; Nahmias, 1976; and others). For periodic review systems, approximately optimal policies that require little computational effort were proposed previously by Naddor (1975, 1980), Ehrhardt (1979), Ehrhardt and Kastner (1980), Ehrhardt and Mosier (1984), Freeland and Porteus (1980), Federgruen and Zipkin (1983), and others. These and other approaches have been reviewed by Porteus (1985).
KeywordsLead Time Accuracy Condition Optimal Policy Asymptotic Approximation Batch Size
Unable to display preview. Download preview PDF.