Application of expert systems in heavy capital plant

• Main Author: Bate, Nick
• Published: Sheffield Hallam University 1996
• Subjects: 620.0042029; Steel rolling
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.307971

This thesis addresses the design of large bespoke pieces of equipment for the oil, steel and chemical process industries, the Heavy Capital Plant Industry (H.C.P.I.). The production of bespoke equipment relies heavily on experience, which makes it a prime candidate for the application of expert systems. A company operating in this area, producing equipment for the steel industry, is Davy International (Sheffield), a company with a world-wide reputation. The company had instigated a programme to capture selected areas of their expertise, using design manuals. Each manual focused on the design of one mill component. During this programme, they identified that there might be potential for the use of expert systems. This resulted in a collaborative research programme, with the aim of identifying the benefits of using expert systems in this application. Two distinct design activities, namely that of the individual components and that of the mill layout, were identified and prototype expert systems were built in these areas. As a source of component design knowledge the roller table design manual was selected, and this was used to build a rule based expert system prototype. Part of the purpose of this prototype was to gain experience using expert systems and to gain an appreciation of the company's approach to design. These lessons were used to help in the production of the second prototype. Knowledge of the design of steel mill layouts was built into the second prototype expert system. This used both object oriented and rule based representations. This thesis, through the use of prototype expert systems, demonstrates the effectiveness of expert systems in the H.C.P.I. A major advantage of these systems is their ability to cope with the many different layout situations encountered, which are often exacerbated by the size of the plant. The suitability of a combined object oriented and rule based toolkit is highlighted for this area, with the benefits of explanation for layout design systems being elucidated.