Abstract

This dissertation addresses production/inventory systems for families of related items as well as distribution systems that service a network of geographically dispersed retailers or sales outlets. The systems are driven by stochastic demand processes. In the former, finished goods are produced in two stages, each with its own lead-time. In the first stage a common intermediate product is manufactured; differentiation into the final products occurs in a second stage. In the latter, the distribution process may be arranged in two stages: in the first stage, goods are ordered from a supplier and shipped to a regional distribution center, and from there to the individual retailers.

In particular, we consider: (1) capacitated production/inventory systems in which the amount produced/ordered in a given period is limited by a capacity constraint. (2) The efficient matching of capacities and demands when their patterns fail to be synchronized; e.g. when demands vary greatly due to seasonal and promotional factors. We analyze two mechanisms to alleviate the problems and costs incurred by this synchronization problem, namely: smoothing of the demand pattern, and the adoption of flexible capacity. (3) Models in which some of the parameters of the demand distributions fail to be known at the beginning of the planning process. These models are analyzed in a Bayesian framework assuming that initial information about the parameters is captured via prior distributions. (4) Models in which consecutive demands are correlated. We show how our planning models need to be adjusted to incorporate intertemporal correlations, and how these impact on the structure of optimal strategies and various performance measures.

Finally, we show how our models can be used to characterize the benefits of information sharing and Vendor Managed Inventory (VMI) partnerships in decentralized supply chains. VMI, also referred to as direct replenishment, is a growing agile logistics partnership agreement where the supplier takes on the responsibility of managing the inventory at the retailer's site for the products it supplies, i.e., monitoring, planning, and directly replenishing the inventory in the retailer's distribution network.