Study of the behaviour of flexible production systems using neural nets

Owing to high development and acquisition costs, production systems require very detailed analysis for pre-design specifications, and an adjustment in operating philosophies. Detailed simulation models are one of the most important instruments to achieve this goal; normally the simulator must be custom-built, based on an accurate study of the specific industrial problem. The proposed management and economical analysis uses mathematical meta-models obtained from experimental designs produced by the simulator; today the detail level and precision required is very high, so these kinds of meta-model operate very close to their limit. Here we present a new methodology based on artificial neural networks that produces new types of meta-models which can correctly represent complex industrial systems. In this work we propose an application of this methodology together with a detailed analysis of the model development phase to highlight the potential of this instrument.     

Real-time part dispatching within manufacturing cells using fuzzy logic

A new fuzzy logic dispatching method is presented ff which o ers an improvement over certain existing common dispatching rules with respect to the performance measures, number of late parts, maximum part lateness, average producff tion times and machine bu er levels. The fuzzy logic dispatcher incorporates a wider range of scheduling issues than those considered by most dispatching rules, considers certain control ff issues that e ect scheduling, contains mechanisms for optimization and also allows for the addition of more control or scheduling rules.     

Rolling horizon master production schedule performance: A policy analysis

In this paper, a simulation experiment has been developed to examine the combined influence of the design, inventory and environmental factors on the cost performance of a rolling horizon master production schedule. Specifically, a 2 5 factorial design was used to examine the effects associated with three rolling schedule design policies, one inventory policy and one environmental condition of forecast error on MPS cost performance. The study was based on actual data from a paint company. Results suggest that the choice of appropriate lot-size and inventory policies have a significant influence on MPS costs and that there are indeed important interactions between these policies and other design factors of a rolling schedule.     

A decomposition technique for quantifying real process performance

A decomposition technique for quantifying the impacts of changes in product mix and process performance on aggregate process-related indicators is presented. Through application of the technique, the real performance of a process can be quantified. Changes in real performance over time can be monitored to provide useful information for process evaluation and production planning. Two case studies, one related to the aggregate defective rate for an assembly line of an integrated circuit fabrication plant and the other to the aggregate inventory turnover for a tyre distribution company, are presented to illustrate the application of the technique.     

TQM: Some issues for deliberation

In this note, the author puts forth some fundamental but generic issues arising from evolving theories and practices of TQM and suggests an agenda for furtherance of research.     

Resource smoothing: A cost minimization approach

From the available literature, there seems to be no defined approach to resource smoothing exercise except those attempted by Weist (1967, Management Science, 13, B359-B377) and Burgess and Killebrew (1962, Journal of Industrial Engineering, 13, 76-83). The aim of the smoothing exercise is to achieve optimal resource usage by avoiding high peaks and deep valleys in the project resource profile. The general approach has always been to move some activities with floats in the high peak regions to be started at a later date, and as this is done, the valleys will be filled to smooth the resource profile subject of course to time constraint. If this approach is followed as it is, it would be difficult to determine optimality especially when many resources are involved. A cost minimization approach is envisaged in the present study with no limitation on the number of resource inputs. In a situation where the resources are assumed to have the same value, the cost assigned to each of them should be similar. The method follows the general concept but with a difference; cost of the activity in question is considered. The exercise is continued until all the floats are exhausted. The optimum result would then be the one with the minimum cost profile. From examples used for the evaluation, the results obtained are comparable to those of the above two researchers, and some with better results in the majority of cases.     

The maximum loading and the optimum number of vehicles in a double-loop of an interbay material handling system

In this paper, the loading analysis of an automated double-loop interbay material handling system (AMHS) in a wafer fabrication was analysed, considering the effects of the vehicle numbers in the inner and outer loops. Discrete-event simulation models were developed in e-M Plant to study this AMHS system with a zone control scheme to avoid any vehicle collision. The layout of an interbay system is a combination configuration in which the hallway contains double loops and the vehicles have double capacity. The combination of the shortest distance with nearest vehicle (SD–NV) and the first-encounter-first-served (FEFS) dispatching rule was used in this study. The maximum loading of this interbay system can be determined by the simulation approach. We also found that the number of vehicles in the inner and outer loops can significantly affect the interbay performance. Furthermore, the optimum combination of the vehicle numbers in the inner and outer loops can be obtained by response surface methodology.     

Managing green productivity: Some strategic directions

The twentieth century has been the hallmark for managing total productivity, where the emphasis is towards maximization of outputs commensurate with the consumption of inputs. This is being challenged now to evolve the world class manufacturing strategy requiring a paradigm shift. This shift is towards green productivity, which looks forward to socially appropriate production and consumption aiming at value innovation and resource conservation. This paper is an attempt to outline the concepts and experiential learning about green productivity.     

Dimensions and design criteria for developing training reaction evaluations

Although there is research on training program evaluation, there is little systematic research on the design, development and use of training reaction evaluations. In order to obtain meaningful information from a program evaluation, evaluation professionals must be rigorous in the design and development of all aspects of an evaluation study, including instrumentation. The purposes of the study were to classify the dimensions of information sought using reaction evaluations and to establish design criteria for developing training reaction instruments. There were three major phases of the study: (1) classification of the dimensions and questionnaire design criteria used in reaction evaluations; (2) validation of the classified dimensions and the questionnaire design criteria by subject matter experts; and (3) assessment of a sample of training reaction instruments currently used in training programs in US corporations. The research findings were: eleven dimensions for reaction evaluation were identified and classified by purpose. Five overall design criteria, each consisting of several sub-criteria, were judged important in the design of reactionnaires. These include: introduction and directions; question format; question construction; questionnaire layout; and data analysis. It was concluded that a well-designed training reaction instrument integrates the proper application of design criteria with appropriate reaction dimensions. Most training reaction instruments used by US corporations consisted of questions representing only a few dimensions. The instruments varied in form and length. Few of them properly utilized the established questionnaire design criteria.