Substructures in computer aided structural design

• Main Author: Nasreldin, Hamdy Abdelaliem
• Published: University of Leicester 1976
• Subjects: 624
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.466848

Structural design is basically an iterative process with successive cycles of modification and re-analysis carried out until an acceptable design is obtained. Interactive graphics techniques permit the designer to rapidly view the results of his hypothesis and enable him to formulate the design changes quickly. Consequently, more design cycles can be carried out in a given time. Structural analysis is a major part of a design cycle and considerable effort has been spent in the development of computer based analysis methods. The finite element method, in particular, is now generally accepted and has a wide variety of applications. The rapid development of the finite element method has meant that increasingly large problems can be tackled in increasingly fine detail. This posed two problems. Firstly, the large size of input data makes the input process both tedious and error prone. Secondly, the interpretation of voluminous output can delay the designer's decision on the necessary modification for the next design cycle. The application of LUISA-1 to engineering problems has demonstrated the use of interactive graphics to overcome the above problems. The length of design cycle was reduced to a level which allowed the user to try several alternatives in few minutes while the computer response was maintained within conversational rates. However, the system was limited to small size idealizations, mainly because of large core storage requirements. This thesis investigates the use of substructuring to overcome core size limitations in an interactive graphics system, and the resulting effects on the system response and on the length of design cycle. An interactive graphics system, based on two-level substructuring, was developed as part of the investigation. The system was limited to two dimensional problems. It was implemented on a dedicated ICL 4130 computer and operated through an ELLIOT 4280 refresh terminal. Applying the system to industrial problems led to the following main conclusions: a. Employing two-level substructuring coupled with data paging techniques reduces considerably core size limitations. b. The division into substructures helps to maintain a conversational mode of man-computer communication during the data generation and results presentation phases of design. c. matrix analysis using a large substructuring system is not suited to a conversational mode of communication d. The modularity of substructuring helps to economize in both user time and computer time. e. Substructuring requires a rather complex data management scheme. f. The large size of program code involved is a major factor affecting the system response. The thesis suggests future work needed to increase the acceptability of an interactive substructuring system. This includes; extension to multi-level substructuring, application to three dimensional problems and implementation on a local computer linked to a large time sharing configuration.