A Knowledge-based Design Aid for Superheaters Employing Pseudo-random Search

Abstract

This paper describes a knowledge-based design aid for creating low-cost superheaters in fossil-fuel boilers. Design problems in general are notably ill-structured. The superheater design problem, in particular, is made difficult by the presence of special pricing functions and discounts to the boiler-maker. The approach used in the design aid combines a frame-based representation structure for the knowledge used to create superheaters with a pseudo-random search technique for improving the cost of the design. The emphasis in this paper is on the use of the search technique. An application is presented which is typical of the results observed during testing of the design aid. The average savings in material costs by using this aid are 10% of total material costs per superheater.