A rule-based expert system approach to process selection for cast components

Abstract

A knowledge-based expert system at the discretion of casting product designers can be employed as a real-time expert advisor to assist product designers to achieve the correct casting design and select the most appropriate casting process for a given component. This paper proposes a rule-based expert system approach for casting process selection, and describes an ongoing rule prototype development. The system in its present development state consists of five interconnected levels each concerning a particular process selection parameter or group of parameters including alloy to be cast, casting geometric features, casting accuracy, production quantity and overall comparative costs. The system progressively evaluates the user's specifications against the capabilities of various casting processes and in each level selects the processes that satisfy the design parameters specified. The final comparative cost level compares the processes that have satisfied all the criteria in the previous levels and recommends the most economical option.

Introduction

Casting is one of the most widely used manufacturing processes for shaping metallic alloys. It involves pouring molten metal into a mould that contains the cavity of the desired shape where the metal solidifies to form a cast component. Casting processes are classified by the type of mould materials used (e.g. sand, metal, ceramic), moulding techniques and the methods by which the molten alloy is introduced into the mould cavity (e.g. gravity, low pressure, high pressure). Availability of many casting processes and a diverse range of casting alloys enable casting of almost any imaginable metal part, regardless of shape, size or intricacy for many different applications.

Casting design and manufacture is, however, a complex problem and involves the interactions of many interdependent casting process variables. Designing cast components and determining the correct casting process requires extensive knowledge of various casting processes and their practical capabilities and limitations. Quite an extensive experience curve is necessary in order for one person alone to be able to acquire all the knowledge and experience needed. It is, therefore, highly unlikely that a casting product designer will have all the knowledge needed to solve a whole range of casting design problems. On the other hand, casting experts, often using heuristics obtained through extensive experience, can assess a casting design and its suitability for a particular casting process, in a relatively short time.

In practice, casting product designers need to communicate with casting experts in order to ensure that the casting being designed is manufacturable and the most appropriate casting process is chosen. Lack of timely communication between these parties or lack of expertise support can lead to erroneous designs and extensive design lead times. The problems originated in such scenarios are considerably magnified when the design engineer is as yet inexperienced. A casting expert system at the discretion of casting product designers that has captured and incorporated the problem solving procedures applied by casting process experts for solving such problems would, therefore, be a viable alternative solution.

Since the late 1980s but more specifically during the 1990s there has been an active research interest in exploring possibilities of developing and applying expert systems in the fields of product design and manufacture with particular emphasis on process selection [1], [2], [3]. A number of authors have reported some developments in the specific areas of casting design and castability evaluation [4], [5], [6], [7], [8], and casting process and alloy selection [9], [10]. A few developments reported concern relatively narrow domains such as design of dies and die casting process selection [5], [11], [12] and design of rigging systems for investment casting [13]. Undoubtedly, these developments have made substantial contributions towards practical applications of expert systems in certain domains of casting design and manufacture. There is, however, still much scope for further research and development of expert systems for solving diverse casting problems.

A comprehensive tool that could assist product designers in selecting the most appropriate casting process for a given design situation is, therefore, the subject of this paper. It discusses the development and the features of an ongoing rule-based prototype expert system, and the features described here are confined to its present state of development.