Article Abstract:
Bounds are provided on the value loss as influenced by optimal cost for restricting focus on a class of inventory stocking policies displaying myopic behavior for a given stopping time. The model used is a single-product inventory system with discrete time and linear procurement, holding and shortage costs. The myopic stocking policy is observed to be a default rule to adopt with immediate delivery systems with no economies of scale. Structural characteristics of the optimal value function that are unaffected by demand give bounds that strengthen the demand process subsequent to the stopping period. The bounding data derived facilitates consideration of possibly intractable demand processes in the analytical context. Consideration is given particularly to the additive and multiplicative shock models.
User Contributions:
Comment about this article or add new information about this topic:
Article Abstract:
The problem of accurately estimating service-level measures in multistage production inventory systems that are managed for high service levels is analyzed. High service levels result in rare stockouts, large backorders and unfilled demands which, in turn, make it extremely difficult to generate estimates using simple simulation. To resolve this problem, alternative estimators are developed which are based on altering the demand distribution to minimize the rarity of such events. It is found that the requirements of these importance sampling estimators continue to be bound for all parameter values. In contrast, straightforward simulation for a fixed relative error produce computational requirements that expand exponentially in some stock-level parameters.
User Contributions:
Comment about this article or add new information about this topic:
Article Abstract:
Two models of inventory control methods are developed using Bayesian formulation and solution techniques applied to problems of periodic inventory review, with one or more unknown factors affecting demand distributions. The models developed may be described as depletive inventory for consumable items and nondepletive inventory for reparable items. An analysis of these models indicates (1) that the Bayesian solution is reducible to the solution of a dynamic programming problem with one-dimensional state space, and (2) that an explicit form for optimizing the Bayesian ordering policy for the inventories described in each model can be achieved.
User Contributions:
Comment about this article or add new information about this topic: