Deterministic inventory models for two-level assembly systems

Abstract

The purpose of this study is to determine optimal or near optimal lotsizes for a two-level assembly system, in which a number of raw materials are required for the production of single final product. The criterion is to minimize the total variable system cost per unit time incurred at the final product and raw materials. This thesis deals with the following three areas of the two-level assembly system. Firstly, the decaying property of the raw material is studied. A model is developed under the integrality assumption in which the lot size of each raw material is such that it satisfies the demand requirements caused by an integer multiple of production lots of the final product. Based on the characteristics of the optimal policy, an iterative algorithm is developed to find an optimal inventory policy. Secondly, a new class of policies is proposed which facilitates both computation and implementation. Two kinds of lot-sizing policies, namely integer ratio lot requirements (IRLR) and powers of two policies, are studied. It is shown that the cost performance of the IRLR policy is a lways better than or at least equal to the existing policies, the integer multiple lot requirements and the integer split lot requirements policies. The computational experiences with the powers-of-two policy show that it takes about one fifth of the computation time of the existing algorithm only with minor cost increases. Thirdly, two types of joint replenishment policy of the raw materials, the basic-cycle and fixed-cycle policies, are considered. In the basic-cycle policy, the results of numberous test problems have revealed that the procedure proposed in this study achieves the results which is very close to the optimum in short steps. For the other policy, the raw materials are classified into groups and then optimum inventory schedules are determined for the product and each group.