The Community for Technology Leaders
Green Image
<p>Numerical interval simulation combines qualitative and quantitative knowledge to predict when a complex process is violating acceptable margins of productivity and risk. A Brazilian steel plant plans to use the system as an adviser to its workers. </p> <p>Complex industrial processes pose special problems for modeling and simulation. On the one hand, precise information for modeling is generally not available or is hard to determine, because such processes are typically complex and experimental. On the other hand, safety, reliability, and productivity are often prime concerns, requiring precise predictions of system behavior. Thus, any effort to develop a simulation system must first create a sufficiently accurate and precise model that reflects the information actually available. It must then determine how to successfully simulate that model within a relatively narrow acceptable operating range. </p> <p>In this article, we describe how we created a semiquantitative model of the sintering process at Companhia Sider?rgica de Tubarao, a Brazilian-Japanese steel company in Vit?ria, Brazil. A semiquantitative model is one that combines quantitative knowledge, such as a known, observable value, with qualitative knowledge, such as an interval of possible values for "high" pressure. We used this model as input to numerical interval simulation, a method to produce accurate and tight predictions of semiquantitative models. Our goal was to improve productivity while ensuring that the process did not become overly risky. The semiquantitative simulations have been very successful, and work is underway to incorporate the simulator (model and NIS) into operations at the plant. The resulting advisory system will extrapolate process behavior and advise operators when the process might enter either a low-productivity or high-risk region. CST also plans to use the system as a component of its operator-training program.</p> <p>We believe this approach is viable for modeling and simulating a range of complex industrial processes, including chemical, nuclear, and thermal. We hope that our approach will help fill in the information gaps in these processes and enable better monitoring </p>

M. M. Lamego, M. R. Vescovi and A. Farquhar, "Modeling and Simulation of a Complex Industrial Process," in IEEE Intelligent Systems, vol. 12, no. , pp. 42-46, 1997.
86 ms
(Ver 3.3 (11022016))