Engineering Computational Technology
Chapter 3 T.J. McCarthy
Manchester Centre for Civil and Construction Engineering, UMIST, Manchester, United Kingdom Keywords: knowledge based systems, expert systems, KBS, KBES, ES, design, structural steelwork.
Computer based design of structural steelwork has been around since the 1970's. Knowledge based systems have been developed in structural engineering applications since the 1980's. This paper takes a retrospective view of the developments in KBS research in structural steelwork design and in doing so exemplifies the broader shifts in engineering application software development. The paper examines how engineers have moved from developing software to do calculations rapidly and accurately to adding a higher level of decision support. The nature of structural design has evolved beyond simply providing structures that are strong enough to ones that meet other requirements such as buildability and cost reduction. The paper concludes with the author's personal view on the future developments in this field. Structural steelwork accounts for approximately 50% of the market for building frames in the UK. A key factor in structural steelwork maintaining its share of the market and its dominance in certain sectors is the availability of high quality guidance in codes of practice, design guides, software and information and advisory services. As the quality and quantity of these codes and other publications has improved so too has computer hardware and software technology. It is now easier than ever to create electronic books with hypertext explanations and user friendly navigation features. Researchers have worked for decades to find the most appropriate computational mechanism to deliver this expertise to users in a timely and efficient manner. This paper looks at the evolution of the use of Expert Systems, also known as Knowledge Based Systems, in this search for an efficient paradigm. The paper argues that, although the combined effort in developing numerous expert system prototypes has not produced the so called ``killer application" that solves all our steelwork problems, it has resulted in a necessary restructuring of the way we think about design. It has paved the way for today's buzz words: knowledge economy, knowledge management, organisational learning, etc. This paper charts the early progress of Expert system and Knowledge Based System development from Carnegie Mellon University, Waterloo University and Professor Gero's group in Sydney [1]. This became a global effort during the 1990's with many researchers from Europe and the rest of the world joining the activity. Expert systems have been developed in many structural steelwork domains: representation of codes of practice; conceptual design of bridges; high rise building frames; detailed design and fabrication of steel bridges; design and costing of structural steelwork systems; detailed design of connections and many more. Following a catalogue of many structural steelwork expert systems, the paper examines the role of information modelling and product data modelling in the future of knowledge systems. The internet has given us XML and XSD, while international standards such as STEP [2] are becoming widely used. Expert systems are more likely to be called decision support systems and are more likely to be embedded in more conventional software packages. It is acknowledged that the Expert system cannot function alone but must be part of an overall package to satisfy the structural engineers' needs. These needs include, good analysis tools, good advice and good information. The advent of good information models and an improved understanding of how to make them will facilitate the creation of integrated knowledge based systems. These will link with case based systems and engineering databases across the INTERNET to deliver the required information in a timely manner to enable the structural designer to make good decisions. With more and more knowledge readily available and in highly usable forms, the age of the knowledge based system is only just beginning.
References
| ||||||||||
