No abstract

DiscussionProfessor P. B. Morice (Professor of Civil Engineering, University of Southampton) said that the Author had clearly demonstrated that the general problem of structural design was at present beyond the scope of computers. He too had found this and recently, therefore, he had been considering a more limited aspect of the problem which was: given a general structural geometry, what were the dimensions that led to minimum cost? It was also of interest to know how cost varied with changes from the optimum dimensions. Such a solution made no attempt to provide complete designs but acted rather as an aide memoire to assist in the process of design.76. The first problem he had studied was that of single-storey roofing using twopinned steel portals and purlins for a building of inlinite extent in width and length. For given frame heights, loading, soil properties, and unit costs a simple programme led to the cheapest span and spacing, with a specification of member and footing details, giving costs per square foot very close to those currently accepted. In addition, plots were obtained of the variation of cost with stable designs of different span and spacing, producing what might be called "cost surfaces". A whole system of cost surfaces resulted from different loadings or storey heights.77. The fact that a few hours' programming and a few minutes' machine time could lead to the well-known answer was very satisfactory. It seemed possible to study new structural forms where there had not been many years of practical experience, and where there was no well-known answer to the problem of minimum cost. He believed that that was the case with plywood box beams which he was now investigating.78. The Author had said that new machines were on the whole easier to programme than the old ones. The next generation of machines, Professor Morice believed, would be designed specifically for autocodes, i.e. simplified methods of programming, which the engineer could use himself. A considerable number of civil engineers he knew had already made use of autocode programming with great success. That had avoided the disappointment some engineers had experienced in employing mathematicians as programmers. Confusion frequently seemed to arise owing to the inability of men of different disciplines to communicate sufficiently precisely, resulting in admirable arithmetical solutions to the wrong physical problem. In his experience, except for the preparation of the most general form of programme it was better for the engineer to do the work.79. The Author had asked what should be done for engineering undergraduates during their training. Professor Morice considered it very important that they should have some acquaintance with electronic computing machines and at least realize the scope of their application. In his own department a large proportion of the undergraduates had an opportunity to make use of computers in connexion with project work in their final year, and provided that the computers were regarded as standard apparatus, ...