Search

An existing taxonomy of workload control rules is adapted to classify 25 workload control rules and their components examined in prior research to create a basis to model how three workload control rule components – order selection, work aggregation and buffer limits – affect shop performance in environments that differ in their bottleneck utilization and protective capacity levels. Two of the most frequently studied rules in the literature – CONWIP and DBR – were included in the simulation. Data analysis found that the workload control rules had a significant positive effect on the shop performance measures only in environments with the highest bottleneck utilization and the lowest protective capacity. However, the work aggregation method was found to be an important component of the workload control rule. Suggestions were made for future research that could increase our understanding of workload control rules and integrate this research with our understanding of lean production.

Processing times are difficult to accurately estimate in industries where custom-designed products are dominant. When schedulers are unable to make accurate estimates, what is the effect of estimation errors on the performance of scheduling policies? Processing time uncertainty (PTU) has become a focus of researchers in the production scheduling area. However, previous operational definition of PTU confounds the estimation error with the variation of actual processing time. In order to determine the effect of estimation errors, a thorough reexamination of prior studies becomes necessary. The results from our simulation experiments show that the estimation error has only a trivial effect on the performance of scheduling policies, and imply that the variation of actual processing time has a more important effect than the estimation error of processing time. The results from our simulation experiment have been verified by probabilistic analysis and Little's law.

Spare parts inventories are not intermediate or final products to be sold to a customer, and the policies that govern spare parts inventories are different from those which govern WIP and other inventories. This paper is an update of the discussion of maintenance inventories and a discussion of the future research needed. After a discussion of unique aspects of spare parts inventories, literature is reviewed which relates to management issues, age-based replacement, multi-echelon problems, problems involving obsolescence, repairable spare parts, and special applications.

Protective capacity is the `extra' capacity placed at non-bottleneck resources to absorb random disruptions in planned levels of performance so that the bottleneck resource is more effectively utilized. Although informative, the research related to protective capacity offers little insight into the preferred placement of protective capacity. By building on the existing research, our study develops hypotheses related to protective capacity positioning and then tested them by using a simulation model of a manufacturing cell. The results indicate that the placement of the protective capacity has no practical influence on mean flow time, but it strongly influences bottleneck shiftiness. Specifically, the study's results may indicate that under conditions of high work center utilization, the placement of protective capacity immediately before and after the bottleneck with a higher proportion distributed to the upstream work centers may significantly lower bottleneck shiftiness. Several future research needs are discussed.

A recent paper discusses a capacity rationing policy that allows make-to-order manufacturing firms encountering expected total demand in excess of available capacity to discriminate between two classes of products, one yielding a higher profit contribution per unit of capacity allocated to it than the other. The result is a selective rejection of orders for the class with lower unit contribution, yielding an increase in total profit when compared to a base case that implements no capacity rationing. Implementation of the policy requires forecasts of demand parameters for both product classes. In this paper we test the sensitivity of the capacity rationing policy to forecast errors in these parameters. The results indicate that, on average, the rationing policy is quite robust in improving profit even when actual demands are approximately twenty percent different from forecast values.

In this paper, we study the capacity allocation problem faced by make-to-order manufacturing firms encountering expected total demand in excess of available capacity. Specifically, we focus on firms' manufacturing short-life-cycle or seasonal products such as high fashion apparel. Using a decision-theory-based approach, we develop a capacity allocation policy that allows such firms to discriminate between two classes of products (one yielding a higher profit contribution per unit of capacity allocated to it than the other), resulting in selective rejection of orders for the class with the lower unit contribution. The effectiveness of capacity rationing is investigated under a wide array of conditions characterized by variations in factors such as the ratio of unit profit contributions of the two product classes, the ratio of total available capacity to expected total demand, the ratio of expected demands between the two classes, and the variability in demand for each product class. The results indicate that capacity rationing is very effective in increasing the total profit, and could therefore serve as a valuable decision tool for managers in such firms.

This article describes a new approach to learning curve estimation. Our approach is to formulate statistical procedures that conform to alternative learning curve theories. This leads to the development of nonlinear statistical models of the learning curves. For the three data sets analyzed, autocorrelation seems to be an important problem. Parameter estimates were derived using the maximum likelihood principle in the presence of first-order autocorrelation. Nonnested tests were used to select the appropriate formulation of the learning curve. Research conclusions are to use unit data when estimating a learning curve and to be prepared to treat autocorrelation if present.