AN EVALUATION OF INVENTORY MANAGEMENT ALTERNATIVES IN A MULTI-ECHELON, MULTI-LOCATION LOGISTICS SYSTEM (ALLOCATION, SIMULATION)

This dissertation addresses the question of whether an anticipatory, or push, system of inventory management might be more effective and efficient in the distribution operations of firms than a reactive, or pull, system. Improvements in the ability of firms to process and transmit large quantities of data rapidly and at low cost have made the push approach potentially more viable. A number of authors point out that a better understanding is needed of when push systems are superior to pull systems.; Three alternative inventory management systems are evaluated; the naive pull, the controlled pull, and the push. The pull systems allow replenishment orders to be placed by lower echelons on an autonomous basis in reaction to current demand. The upper echelon in the naive pull system has no information on the demand and inventory situation of facilities in the lower echelon. Orders are filled in the sequence received. The controlled pull system has inventory and demand information, which can be used to adjust the sequence in which orders are filled to improve system performance. The upper echelon in the push system determines which facilities to replenish based on the relative urgency of need. Unlike the pull systems, no inventory is held centrally, and all goods are distributed as produced.; The three alternatives are evaluated through the simulation of a hypothetical firm which produces and distributes a single case goods product from a central facility to five spatially dispersed distribution centers. The distribution centers attempt to satisfy exogenous demand. The objective of the firm is to achieve a specified level of customer service at the lowest distribution cost. The firm is affected by constraints on production and shipping, stochastic seasonal exogenous demand, and stochastic transit times. The simulation model replicates the firm's daily operations for three years with 16 combinations of independent factors. The final two years data are evaluated using the Friedman Rank Sums test, a nonparametric procedure.; The push system achieved the target service level with less inventory, smaller maximum storage requirements, and lower inventory costs than either of the pull systems. . . . (Author's abstract exceeds stipulated maximum length. Discontinued here with permission of author.) UMI...