Modeling and Response Time Analysis of Process Computer System in Steel-making Industry
- Modeling and Response Time Analysis of Process Computer System in Steel-making Industry
- Date Issued
- Diverse computer systems for factory automation in the steel-making industry have been configured in a hierarchical manner for various reasons. A process computer system (PCS) is in charge to supervise and control the factory. The PCS collects field data from programmable logic controllers (PLCs), calculates reference values, and sends them to PLCs which control field equipment directly. Also the PCS receives production orders from a Manufacturing Executive System (MES) and manages production data. The PCS must respond with time constraints because the delayed response may result in defective products. So, when we design a new PCS or improve already deployed one, it is very desirable to check in advance whether the anticipated system meets the requirements because correcting a system after implementation requires more time and effort. This dissertation proposes a method for modeling and response time analysis of a PCS. To the best of our knowledge, there is no tool to evaluate the response time of a PCS in design stage.
Performance analysis method has been developed for middleware itself, an asynchronous Web server and a distributed component-based E-business system. For the Web server and the E-business system, overall average response time was evaluated. Notably the PCS of a steel-making process handles the burst messages from PLCs. The arrival rate of the messages is higher than the service rate in the short burst period, so that some messages may experience longer response time than others. Since the response times to messages are different from one another, it is necessary to evaluate the response time to each individual message. Especially, our attention is focused on evaluating the response time for the burst messages. There are many approaches to the response time analysis, but we hardly find a good one for a PCS.
This dissertation presents a modeling method to set up an analytic model of a PCS for the response time analysis using Colored Generalized Stochastic Petri Net (CGSPN). The model is analyzed by means of transient queueing analysis and simulation. To validate our approach, a test bed system is built that is identical to a part of a PCS in a continuous casting plant which plays the crucial role in time critical operation of the PCS, and measure response times. Analytic results that are obtained by simulating our analytic model match the measured values well. Then our approach is applied to the whole PCS in the same plant to estimate the response times which are presented and discussed in the dissertation. The results show that the proposed approach can be applied to the PCS of any other plant in steel-making industry to effectively predict the response time prior to implementation.
- Article Type
- Files in This Item:
- There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.