Vol.5 No.2 2012

38/66

Research paper : Toward the integrated optimization of steel plate production process (K. Nishioka et al.)−112−Synthesiology - English edition Vol.5 No.2 (2012) steelmaking, which inevitably result in somewhat large fluctuation or variation in production. In the processes of intense heat and high temperatures, time constraints are substantial and no in-process stock is allowed, therefore, fundamentally its production structure is of the push-type. The “product” of the upstream processes of raw materials – ironmaking is mono-grade. During the early days of integrated steelworks, the final product types were limited and there were no such complicated requirements as today. Because of the above reasons, focus had been directed on the optimal control of high temperature processes, measures to deal with the production fluctuation/ variation of the upstream processes and the maximization of manufacturing lot sizes in the upstream processes by summarizing product order information. In contrast, very limited actions were taken for planning overall optimization of integrated manufacturing that examines, from the viewpoint of linking the quality control and delivery of ordered products, the manufacturing lead time and the in-process stock level in the processes of intermediate products such as hot rolling and plate mills, and for supporting scheduling.To search for optimization, the existence of various models is essential, but compared with the assembly industries represented by the automobile industry, in the steel industry, the fluctuation of daily production and operation is large and moreover plans are frequently changed. In addition, because the model is developed based on the past records that also contain external disturbances, it is difficult at present to make a highly accurate estimation using models, and therefore, it is understood that the optimal answer that depends on the models that contain many errors is naturally limited in its reliability.In such a situation, it is not realistic from the viewpoint of effectiveness, calculation load, and other work load, to conduct elaborate scheduling in daily routine work that rigorously searches for the answer to the overall optimization, and therefore, the approach to an optimal answer has to be continuous and asymptotic as shown by the example in this paper.The proposed model grasps actual data for every term and in addition to the result-based production control that performs control based on such data, it identifies the cause-based phenomena from which such data are born. It can be considered as a tool for conducting better control. As a practical problem, huge difficulties may arise if a comprehensive model is to be created that covers all the time strata extending from the micro events that occur in a short time called processing in individual processes to the macro events that occur over a long time called manufacturing lead time that extends from the rolling start to the completion of manufacturing. If models developed through trial and error are overviewed as a whole, the structure is such that respective models link the overall picture going over the time strata, and therefore, this model is called the multi-scale hierarchically structured model. In other words, the proposed model is cause/ result-based straddling of a certain level of time strata, and even though it contributes to the understanding of phenomena extending structurally from the micro to macro level, it is not a model that guarantees an optimal answer. However, it can be understood that it provides a means of approach to the optimal answer continually and asymptotically. The above viewpoint is added in 6.2 “Significance of the multi-scale hierarchically structured model.”5 Efficiency modelQuestion (Motoyuki Akamatsu)What is the difference between the efficiency model and the conventional methodology of conventional production control?Answer (Yasushi Mizutani)Processing efficiency is defined, in the rolling processes, as the weight of slabs processed per hour, and in the finishing equipment-heavy industries, or process industries where the renewal of existing equipment or changes of its installation layout are difficult, integrated optimization is feasible only within the time span of processing, and it is difficult to realize the integrated optimization for all. In other words, in order to ensure integrated optimization in many industries where the time structure of monozukuri extends from the micro to macro level, the understanding of phenomena that are beyond time strata is considered necessary, which means the development of a cause – result-based model is also necessary. There are three categories of these strata in the present case, but this naturally could be two or four depending on the process. What is important is to determine which strata are to be straddled and in what form the model should be developed, so that it contributes to the proper understanding of the phenomena of the process and to the integrated production control, and the present example presents one of such examples.In many industries, the shortening of time necessary for manufacturing is an essential element for developing a wide variety of value-added products and for manufacturing them as competitive products, and if the time structure in the production control is understood systematically by using the currently proposed model, we consider it will be of some help to it. For the future outlook, please see chapter 7.4 Application scope of the multi-scale hierarchically structured modelQuestion (Kanji Ueda)In the paper, the intent of overall optimization instead of partial optimization is stated. However, the methodology of this paper does not theoretically seek the overall optimization, and therefore, I think it is not guaranteed that an answer for overall optimization will be obtained. Therefore, I think you should refer to the effectiveness or limitation of the application of the present methodology.Question (Motoyuki Akamatsu, Human Technology Research Institute, AIST)The contention of this paper is the composition using three different time scale models, which enables quantitative evaluation extending from the micro to macro level and finds the optimal point. However, how do you integrate or interrelate multi-scale models and use them?Answer (Kiyoshi Nishioka)The model proposed at this time does not have the optimization evaluation function that realizes overall optimization, and does not guarantee the overall optimization theoretically and quantitatively, nor can it be used as it is by installing it in the production control system.The steel industry, the representative case of process industries, especially with integrated iron and steel manufacturers having blast furnaces, has physical constraints specific to its processes: its symbolic blast furnaces use natural resources as raw materials; its processes contain those of intense heat and high temperatures; and raw material yards, etc. are in the open air and are easily influenced by the weather/ climate. To overcome these problems, the steel industry has always pursued integrated continuous facilities, higher productivity and higher energy efficiency of high temperature processes by building larger-scaled works.For conducting smooth production activities in such integrated and continuous large-scale production facilities, an enormous amount of information on equipment control and production control has to be handled, and therefore, it introduced a production control system supported by large-scale computers earlier than other industries. The production control system of an integrated steelworks has constraints as described above, in the upstream processes of the raw materials, ironmaking and

※このページを正しく表示するにはFlashPlayer9以上が必要です