Artificial Neural Networks to Aid Conceptual Design

Mezzanine Floors Two more contributors have joined this discussion. Richard Harris writes from Bournemouth: I have read with interest the recent correspondence on mezzanine floor structures and the effective lengths of their columns. For some years, I have been refusing to accept a factor of 1.5 for unbraced frames. The designers are always unable to justify the figure. Occasionally, when an impasse is reached, I visit the site. Surprisingly, even when there appears to be no adjacent supporting structure according to the plans, there is often actually something robust enough to provide lateral restraint, bringing the effective length down below 1.0. I suspect that this, and vigilance by building control bodies, is an important factor in preventing collapse.

Mr W. Parlor (Fire Safety Development Group) We read the ‘preliminary conclusions’ of the paper with avid interest, and must again express our alarm at the lack of definition of these conclusions.

Every structural design should be produced in accordance with, and checked for conformance against, the appropriate standards. The term ‘standards processing’ is commonly used to describe this process. This paper presents a hybrid environment for automated standards-processing, named ‘SADA’: standards automated design assistant. One objective of the SADA environment is to overcome the drawbacks commonly associated with standards processing and currently available, commercial computer packages presently employed to facilitate standards processing. The functionality of existing packages is discussed, the focus of this discussion being the application of these tools to standards conformance checking. Illustrative examples of the SADA environment are then presented, demonstrating its ability to overcome drawbacks associated with standards documentation and existing packages. A.I. Thompson, B. Kumar and Professor I.A. MacLeod