Robotics and Computer Integrated Manufacturing 18 (2002) 267–274 A rule based expert system for rapid prototyping system selection S.H. Masood*, A. Soo Industrial Research Institute Swinburne (IRIS), Swinburne University of Technology, Hawthorn, Melbourne 3122, Australia Received in revised form 25 July 2001 Abstract The availability of a growing number of rapid prototyping (RP) systems with wide ranging capabilities has created a problem of selecting an appropriate RP system for industry as well as educational sector to suit their requirements. This paper presents an expert system based RP system selection program incorporating 39 RP systems commercially available from 21 RP manufacturers worldwide. The program allows the user to choose one of the four options, namely, quick selection, detailed selection, build technology or machine style for system selection, with each option considering a systematic selection criteria. The program is a rule based expert system, and recommends the RP system along with its full specifications on the basis of interactive question–answer session with the user. The system is believed to be the first expert system based RP selection program and has the potential for future expansion into a full-fledged RP selector system. r 2002 Elsevier Science Ltd. All rights reserved. Keywords: Rapid prototyping; Selector program; Expert system; Benchmarking; Intelligent system 1. Introduction With the growing market and proliferation of rapid prototyping (RP) technology, the selection of an appropriate RP system is becoming an increasing difficult task for organisations intending to adopt the technology to serve their specific needs. The choice of a RP system depends on a number of factors or selection criteria such as price, accuracy, build envelope, build material, build speed, surface finish, and end applica- tions. At present there are over 20 RP vendors around the world, marketing a large range of RP systems, with each system having its own strengths, limitations and applications [1]. The selection process is further compli- cated by a general shortage of benchmark standards and industry experience with most of these systems. More- over, there are regular introduction of newer RP systems in the market. Early attempts in selecting a RP system were based on studies that compared different RP systems. These mainly included benchmarking studies carried out either by the user companies or by independent researchers [2–6]. In benchmarking studies, a benchmark part is created on a CAD system with features that will test the capabilities of the RP machines. The stereolithography (STL) file of the benchmark part is then used to create the prototype part on the RP machine. An extensive measurement and inspection of the part is then conducted to establish the accuracy or discrepancy of the desired features. Benchmarking studies are used to confirm the strength and weakness of the RP systems being tested. Some past benchmarking studies have confirmed that the accuracy and surface finish quoted in sales brochures can only be achieved under ‘‘ideal laboratory condi- tions’’. Benchmarking studies can be time consuming and quite expensive. Often, multiple models of the same part are required to be built to check the repeatability. Unless there is a compelling need, such as to justify the purchase of a very expensive RP system, the bench- marking studies are often hard to justify. Moreover, previous attempts to compare RP systems through benchmarking trials have met with limited success due to the inability of the RP vendors to agree on a common benchmark part. More recent efforts in the direction of selecting an RP system has been directed to the development of computer based selector programs. These are aimed at assisting the potential purchaser to select the right RP *Corresponding author. E-mail address: smasood@swin.edu.au (S.H. Masood). 0736-5845/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved. PII:S0736-5845(02)00017-0