Knowledge that Transforms

We consider a two‐echelon inventory system with one warehouse and several stores. The warehouse as well as the stores are controlled by periodic review ( s, S) inventory policies. We study the interrelationship between the safety stocks at the warehouse and the stores. Stockouts at the warehouse will result in supply delays to the stores and cause the lead time to be stochastic. The stores may react by increasing their safety stock. However, there is a trade‐off between the safety stock at the warehouse and the safety stock at the stores. We use a service level at the warehouse to quantify the effect of warehouse stockouts on the lead time to the stores. The service level at the warehouse is considered a decision variable to find the best compromise between the various safety stocks by minimizing the overall costs. Using power approximations for the ( s, S) policies, we provide an iterative procedure for adjusting the lead time distribution to the stores; this can result in substantial savings, but it doesn't guarantee the overall optimality. Numerical studies are provided to test the accuracy of approximations. The effects of the different system parameters on the inventory policy give general guidelines for use of the policies.

The mantra of health care reform in hospitals in the United States and elsewhere, the spiralling costs of health care, and the development of diagnostic‐related groups (DRGS) that define health services as “products” are influencing the strategic role of operations planning and control in hospital delivery systems. We developed a new operations planning and control system that we call “Hospital Resource Planning” (HRP). This system is based on the concept of DRGS and the familiar concept of manufacturing resources planning (MRP‐II). TO determine the potential feasibility of HRP, we gathered longitudinal data from two hospitals, one 300‐bed community hospital and one 1,100‐bed teaching hospital. Our exploratory study indicated that while the concept of MRP‐II can be transferred to hospitals, the traditional MRP logic has shortcomings. HRP advances prior research in three ways: (1) consideration of DRGS as products with a bill of resources structure that simultaneously incorporates both capacity and materials resources, (2) implementation of a hospital‐wide (versus a functional) planning and control'system, and (3) gross‐to‐net requirements logic based on notions of treatment staging. Other feasibility issues that we addressed pertain to a stochastic bill of resources, trends toward hospital product standardization, and the coordination of a central planning system with decentralized decision making. The paper concludes with a description of areas for future research.

Total quality management (TQM) is a revolutionary approach to effective management. The research in TQM has emerged from practical needs of organizations embracing this philosophy, and the literature is mostly conceptual and practitioner‐oriented. There is a lack of sound theoretical framework classifying past efforts and guiding future research. To fill the void, a study of the published TQM literature is undertaken. A review, classification, and analysis of the research in TQM spanning the last two decades is presented. A total of 226 TQM‐related articles are identified from 44 refereed management journals published from 1970 to 1993. These articles are then classified and analyzed using the following two‐dimensional scheme: (1) article orientation (conceptual, case study, empirical, analytical, simulation, and overview) and (2) article focus using the Malcolm Baldrige National Quality Award criteria. The analysis of the literature presents pertinent developments in each of the seven criteria. In addition, it provides future research directions as well as a ready reference of the TQM literature. The suggestions for research should guide future developments in the TQM field and help transform it into a formal discipline.

A methodology for benchmarking manufacturing practices in the worldwide semiconductor industry is presented. Several metrics for measuring semiconductor manufacturing performance are defined and discussed. Interviews conducted during site visits are used to identify the managerial, technical, and organizational practices underlying superior metric performance. Multivariate statistical analyses of the performance metric data indicate that these performance metrics measure independent aspects of performance and expose significant performance differences among fabs.

We present a pavement management expert system developed by the University of Wisconsin‐Madison and implemented within a geographical information system for the Wisconsin Department of Transportation. The system uses pavement data regularly collected on the state's 12,000 miles of highway to assist engineers, planners, and budget analysts' management decisions about pavements to be included in 6‐year improvement and 3‐year maintenance programs. The system has a three‐layer architecture. The lowest level suggests treatments for each of a large number of small segments of highway. The middle layer aggregates segments, suggests alternative treatments, and estimates the cost of each. The top layer prioritizes the projects and incorporates them into intermediate‐range plans. The geographical information system environment enables integration of existing databases within the system using a topologically structured geographic database and specialized software.

Operational‐level scheduling decisions on the factory floor include the release policy and the dispatching rule. Recent work by researchers suggests that factory performance depends heavily on the release policy. Several heuristic factory release rules have been developed that generate release decisions based on the status of the bottleneck machine. The heuristic rules perform very well in reducing the work‐in‐process inventory. However, factory output schedules generated by these rules generally fail to match the demand pattern, creating excessive finished goods inventory and/or excessive backorders. To overcome these difficulties, a computationally efficient integer programming model is proposed for release scheduling.