استفاده از سیستم های کنترل تولید کامپیوتری در محیط تولید کارگاهی
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|18885||2000||19 صفحه PDF||سفارش دهید||12210 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Computers in Industry, Volume 42, Issues 2–3, June 2000, Pages 79–97
Job shops are the most perplexing and challenging of environments for computerized scheduling systems. Some vendors call them problematic and avoid them whenever possible. The development of a generic scheduling tool that can be widely installed has eluded the many vendors who have tried. Installations in job shops are still painfully slow and are heavily customized. The wide variety of types of industry and processes found in the job shop category explains part of the problem. However, there is an additional challenge facing a scheduling tool after it has been installed: will it be used and does it really help the scheduler do the task of scheduling? Is the tool used for rough planning in the next few days or is it really used for dispatching on the fly by the people sequencing work? Few studies have looked at these issues and vendors are not willing to share this information. Furthermore, few customers who have invested significantly in the purchase and installation of a scheduling tool perform an unbiased postmortem or will share the results. In this study, we will present two field studies where computer scheduling aids were considered and discuss the requirements that will support the scheduler in the daily dispatch task. We have found that while analytical and algorithmic aids have limited benefits to a typical job shop, the appropriate use of computer technology can address information overload, cue filtering, and assist the scheduler in problem solving. We describe seven steps of the job shop scheduling task and the implications of each.
“In recent years the mode of computer utilization has entered a new era where meaningful interaction has been achieved between the user and the computer. This approach has been particularly effective in solving ill-structured problems. In the interactive environment the person can replace some of the programming logic with his insight and experience. In this way the analyst becomes part of the solution-decision loop and he has a better control over unanticipated situations in arriving at a solution” . Written over two decades ago, the authors were describing the potential to be gained by supporting job shop schedulers via interactive scheduling tools. They showed that test subjects using an interactive system were able to create better schedules compared to a traditional dispatch heuristic when the problems had balanced loads and a wide variation in processing times. Unfortunately, the optimism voiced by the authors about the future of man-computer systems using OR and AI techniques has not translated into reality  and ). The task of detailed production control at a scheduler's desk remains an enigma. In this paper, we will focus on this enigma: releasing work into the factory, determining sequences, and figuring out what to do next — where. We will not address the larger problems of order acceptance and supply chain balancing nor the specialized areas of continuous flow manufacturing, just-in-time, process industries, or the ‘single large machine’ problem. We will look at the Achilles' heel of scheduling: the dynamic job shop where there are many machines, many products, and orders arrive as ‘jobs’ with quantities, requirements, and due dates attached. The most extreme job shop in terms of uncertainty builds new products constantly and basically sells its processing capability — “We can weld or fix anything”. The most stable job shop builds stock items but in an intermittent fashion. Between the two are a wide variety of manufacturing profiles which makes talking about the general ‘job shop problem’ very ambiguous. What happens in an intermittent job shop building a stable and mature product after several years of production is dramatically different from what happens in a ‘prototype’ job shop or a job shop facing rapid changes in materials, processes, and product design. Obviously, not all job shops have the same production control problem and not all job shops require the same solutions. Some shops may be easily managed via humans without the need of computerized systems while others are prime candidates for the use of technological aids. The questions addressed and discussed in this paper are: • Under what conditions are humans the best choice for performing the production control task? • Under what conditions is a human–computer solution preferred? • If a human–computer solution is warranted, then what are the tasks the human does and what can the computer do to help? To address these questions, the authors will rely upon two field studies involving intermittent job shops. In one field study, the human scheduler did an excellent job without the aid of technology. In the other, the human scheduler required a computer system to help manage the situation. The authors had several years of contact with each shop and developed an intimate understanding about the problem faced by the schedulers and how they tackled the problem facing them. The computer system used by the second site was written by one of the authors. Section 1 presents a brief overview of research involving the human scheduler. Section 2 introduces the two field studies and describes the production control situation in each. Section 3 compares the two sites and discusses the differences and similarities between the two. Section 4 presents a general discussion about the key issues relating to automated schedule generation. Section 5 presents the requirements for a hybrid approach to the problem. Section 6 concludes with a description of a system designed to address the key issues.
نتیجه گیری انگلیسی
We used brief descriptions of two extensive field studies to illustrate how different and how similar the scheduling task is in job shops. The system developed for Site B would never work at Site A and possibly only 30–40% of the system could be saved and reused in a site similar to Site B (same industry, same type of process). While not operationally similar, the spirit, intent, and purpose are the same. The schedulers would go through the same phases of scheduling and they would be looking for clues, opportunities for constraint relaxation, anchoring their schedule and then letting the remainder of the work flow into the plan according to normal, day-to-day heuristics. We recognize the unique nature of a context-sensitive, site-sensitive scheduling aid. The PPS tool at Site B has now been in use for over two and a half years and it has evolved from a simple scheduler's tool to supporting almost all of the operational processing requirements for planning — from the scheduler's desk to the plant manager's. The extent of its use was not anticipated, nor was the reliance of plant personnel on it. When SAP was installed, we expected the usage to decrease, but the reverse happened. We knew that the scheduler's position is often the hub and nerve center for manufacturing and this was again the case and PPS grew to fill the requirement. The next phase of research will investigate how the tool supports manufacturing tasks beyond the scheduler's desk. We will continue researching the design and use of such aids with a thrust towards identifying common building blocks and easier creation. The creation and wide spread use of computer tools to create job shop schedules looks deceptively simple. As we have shown, the requirements are complex and operational use of such tools is difficult to achieve. We hope that ongoing research into the scheduler's task and information requirements will reduce the effort in creating tools that will help yield better schedules.