Nonrobustness of the normal approximation of lead‐time demand in a (Q,R) system

For computing an optimal (Q, R) or kindred inventory policy, the current literature provides mixed signals on whether or when it is safe to approximate a nonnormal lead‐time‐demand (“LTD”) distribution by a normal distribution. The first part of this paper examines this literature critically to justify why the issue warrants further investigations, while the second part presents reliable evidence showing that the system‐cost penalty for using the normal approximation can be quite serious even when the LTD‐distribution's coefficient of variation is quite low—contrary to the prevalent view of the literature. We also identify situations that will most likely lead to large system‐cost penalty. Our results indicate that, given today's technology, it is worthwhile to estimate an LTD‐distribution's shape more accurately and to compute optimal inventory policies using statistical distributions that more accurately reflect the LTD‐distributions' actual shapes. © 2003 Wiley Periodicals, Inc. Naval Research Logistics, 2003     

Capacity expansion with lead times and autocorrelated random demand

The combination of uncertain demand and lead times for installing capacity creates the risk of shortage during the lead time, which may have serious consequences for a service provider. This paper analyzes a model of capacity expansion with autocorrelated random demand and a fixed lead time for adding capacity. To provide a specified level of service, a discrete time expansion timing policy uses a forecast error‐adjusted minimum threshold level of excess capacity position to trigger an expansion. Under this timing policy, the expansion cost can be minimized by solving a deterministic dynamic program. We study the effects of demand characteristics and the lead time length on the capacity threshold. Autocorrelation acts similarly to randomness in hastening expansions but has a smaller impact, especially when lead times are short. However, the failure either to recognize autocorrelation or to accurately estimate its extent can cause substantial policy errors. © 2003 Wiley Periodicals, Inc. Naval Research Logistics, 2003     

Determination of burn‐in parameters and residual life for highly reliable products

Today, many products are designed and manufactured to function for a long period of time before they fail. Determining product reliability is a great challenge to manufacturers of highly reliable products with only a relatively short period of time available for internal life testing. In particular, it may be difficult to determine optimal burn‐in parameters and characterize the residual life distribution. A promising alternative is to use data on a quality characteristic (QC) whose degradation over time can be related to product failure. Typically, product failure corresponds to the first passage time of the degradation path beyond a critical value. If degradation paths can be modeled properly, one can predict failure time and determine the life distribution without actually observing failures. In this paper, we first use a Wiener process to describe the continuous degradation path of the quality characteristic of the product. A Wiener process allows nonconstant variance and nonzero correlation among data collected at different time points. We propose a decision rule for classifying a unit as normal or weak, and give an economic model for determining the optimal termination time and other parameters of a burn‐in test. Next, we propose a method for assessing the product's lifetime distribution of the passed units. The proposed methodologies are all based only on the product's initial observed degradation data. Finally, an example of an electronic product, namely contact image scanner (CIS), is used to illustrate the proposed procedure. © 2002 Wiley Periodicals, Inc. Naval Research Logistics, 2003     

A general model of heterogeneous system lifetimes and conditions for system burn‐in

Burn‐in is a technique to enhance reliability by eliminating weak items from a population of items having heterogeneous lifetimes. System burn‐in can improve system reliability, but the conditions for system burn‐in to be performed after component burn‐in remain a little understood mathematical challenge. To derive such conditions, we first introduce a general model of heterogeneous system lifetimes, in which the component burn‐in information and assembly problems are related to the prediction of system burn‐in. Many existing system burn‐in models become special cases and two important results are identified. First, heterogeneous system lifetimes can be understood naturally as a consequence of heterogeneous component lifetimes and heterogeneous assembly quality. Second, system burn‐in is effective if assembly quality variation in the components and connections which are arranged in series is greater than a threshold, where the threshold depends on the system structure and component failure rates. © 2003 Wiley Periodicals, Inc. Naval Research Logistics 50: 364–380, 2003.     

An adaptive dynamic programming algorithm for a stochastic multiproduct batch dispatch problem

We address the problem of dispatching a vehicle with different product classes. There is a common dispatch cost, but holding costs that vary by product class. The problem exhibits multidimensional state, outcome and action spaces, and as a result is computationally intractable using either discrete dynamic programming methods, or even as a deterministic integer program. We prove a key structural property for the decision function, and exploit this property in the development of continuous value function approximations that form the basis of an approximate dispatch rule. Comparisons on single product‐class problems, where optimal solutions are available, demonstrate solutions that are within a few percent of optimal. The algorithm is then applied to a problem with 100 product classes, and comparisons against a carefully tuned myopic heuristic demonstrate significant improvements. © 2003 Wiley Periodicals, Inc. Naval Research Logistics 50: 742–769, 2003.     

A model for step‐stress accelerated life testing

Modern technology is producing high reliability products. Life testing for such products under normal use condition takes a lot of time to obtain a reasonable number of failures. In this situation a step‐stress procedure is preferred for accelerated life testing. In this paper we assume a Weibull and Lognormal model whose scale parameter depends upon the present level as well as the age at the entry in the present stress level. On the basis of that we propose a parametric model to the life distribution for step‐stress testing and suggest a suitable design to estimate the parameters involved in the model. A simulation study has been done by the proposed model based on maximum likelihood estimation. © 2003 Wiley Periodicals, Inc. Naval Research Logistics, 2003     

Optimal retail assortments for substitutable items purchased in sets

We consider the problem of optimizing assortments in a multi‐item retail inventory system. In addition to the usual holding and stockout costs, there is a fixed cost for including any item in the assortment. Customers' preferences for items include both probabilistic substitution patterns and the desire to purchase sets of complementary items. We develop a demand model to capture this behavior, and derive tractable approximations that allow us to formulate the optimization problem as a 0–1 mixed integer linear program. Numerical examples are solved to illustrate key insights into how both complementary and substitute items affect the optimal assortment and the expected profit. © 2003 Wiley Periodicals, Inc. Naval Research Logistics 50: 793–822, 2003.     

Exact algorithms for scheduling multiple families of jobs on parallel machines

In many practical manufacturing environments, jobs to be processed can be divided into different families such that a setup is required whenever there is a switch from processing a job of one family to another job of a different family. The time for setup could be sequence independent or sequence dependent. We consider two particular scheduling problems relevant to such situations. In both problems, we are given a set of jobs to be processed on a set of identical parallel machines. The objective of the first problem is to minimize total weighted completion time of jobs, and that of the second problem is to minimize weighted number of tardy jobs. We propose column generation based branch and bound exact solution algorithms for the problems. Computational experiments show that the algorithms are capable of solving both problems of medium size to optimality within reasonable computational time. © 2003 Wiley Periodicals, Inc. Naval Research Logistics 50: 823–840, 2003.