"Optimal" operating rules for automated manufacturing systems

In modeling automated manufacturing systems such as transfer lines and flexible manufacturing systems it is necessary to make assumptions about how such systems will be operated. Hence, it is desirable to determine the "optimal" operating rules. In this paper a number of examples are given of how the "optimal" rules are determined so that consideration is given to the control options that can be used, the information available to the operator and the typical multilevel nature of the system's control.     

An Algorithm for Symbolic Reliability Computation with Path-sets or Cut-sets

An algorithm obtains a symbolic expression for system reliability, given a set of events (path-sets or cut-sets) as input. The algorithm is simple in structure and uses few probability laws. The resultant symbolic expression is compact and requires relatively few arithmetic operations in its evaluation.     

Models for first-pass FMS investment analysis

Consider a firm that could choose either an inexpensive product-specific (dedicated) facility or a costly flexible facility, or a combination of the two, in order to satisfy demand for two product groups. Flexible technology offers benefits of scale and scope economies. However, it may have added operational costs due to the need to have excess capacity (to permit changeovers) and to maintain cycle stocks. As a consequence, the economic viability of flexible technology is significantly affected by the choice of the operating doctrine. This article presents first-pass decision models to help the firm choose the optimum sizes of facilities and the degree of flexibility for the flexible facility such that the contingent operational costs are simultaneously optimized. Two variations of the problem are considered. Whereas, the first applies to a situation in which the demand rates are constant and predictable, the second considers random demands. In each case, this study provides a formulation of the problem, structural results, and the sensitivity of the results to cost parameters. These results have modest data and computational requirements, making them suitable for first-cut attempts at narrowing available choices.This research has been supported by the Natural Sciences and Engineering Research Council of Canada through operating grants to the authors.     

Reliability Evaluation of Systems with after Fault Switching

A new method is presented for reliability evaluation of power systems where components may be switched-out after faults. The method utilizes the concept of boundary conditions for the computation of the probability and frequency of system failure. The truncated state enumeration method applied traditionally to the analysis of such systems is now replaced by a network approach which does not, in principle, require truncation rules and can be applied to the components having arbitrary continuous probability distribution of time to failure or repair time. Examples are described in the text to demonstrate the application of the proposed method and to compare the results with the approximate calculations.     

The Future of Industrial Engineering As An Academic Discipline

Industrial engineering as an academic discipline appears to have reached a crisis. While the typical industrial engineering curriculum based on the Roy report results in graduate engineers who are valued by many employers, it has proved to have some adverse implications. Industrial engineering departments have not been able to maintain cohesiveness and a common view of their discipline and the engineering content has been at too low a level to put industrial engineers at the forefront of technological development. The problems and some possible solutions are described; however, this paper is primarily intended to identify some of the key issues that need to be discussed.     

Hedging against uncertainty in new technology development: the caseof direct steelmaking

Companies considering a major facility investment face a dilemma if a new and superior process technology may be available in the near future. Should the new technology be successfully developed, existing technology may be rendered obsolete, making the investment proposed for today a poor decision in retrospect. On the other hand, waiting for the new technology to become available may entail lost profits today as a result of inadequate capacity or poor product quality. In addition, the company could be vulnerable if the availability of the new technology is delayed or, even worse, if the development program is a failure. In this paper we show how a multiperiod stochastic programming model may be used to develop possible hedging strategies for dealing with such a situation, illustrating our ideas with a representative real-world example-direct steelmaking, For this case study, we find that planning on the basis of either scenario, success or failure, is a poor strategy compared to the available compromise solutions     

The Structure of Manufacturing Systems: Insights on the Impact of Variability

A classic issue in manufacturing strategy is positioning, that is, the appropriate structure for the manufacturing system. At its simplest, the choice is between a job shop and a flow line; but if the system has to produce a variety of different products,then it also is necessary to decide on focus, the degree of specialization. While effective use of resources is essential, it also is necessary to consider the ability of the system to cope with variability and disturbances. Using a variety of queueing models of manufacturing systems, it is possibleto get some useful insights into positioning and focus.     

Cut-Set Intersections and Node Partitions

With any cut set of a graph can be associated a partition of the nodes into two cells. We show that each cut-set intersection term in the cut-set inclusion and exclusion formula can be associated with a k-cell partition of the nodes. Thus, the number of distinct terms in the cut-set inclusion and exclusion formula cannot exceed the number of partitions of the set of nodes. This leads to a simplified formula for graph reliability: the node partition formula, When finding the overall reliability in complete graphs no terms cancel, thus the number of terms is equal to the number of partitions. In other graphs we show that the number of non-cancelling terms in the cut-set inclusion and exclusion formula is equal to the number of minimal partition sets of the graph. It follows that cut-set inclusion and exclusion is inherently an inefficient method for the exact calculation of network reliability measures.