Studying the behaviour of shear wall-floor slab connections using artificial neural network (ANN)

Abstract

The behaviour of shear wall-floor slab connections has been studied by using artificial neural network (ANN). An artificial neural network is an information-processing system based on the observed behaviour of biological nervous systems. A neural net consists of a large number of simple processing elements called neurones. Each neurone is connected to another neurone by means of direct communication links, each with an associated weight. The strength of the connections is dictated by the weights, which connect different neurones. . Each neurone accepts a set of inputs from other neurones and also from external sources and generates an output. Structural design requires engineering judgement, intuition, experience and creative abilities in addition to number of options available to the designer. The ANN approach has the capabilities to incorporate some of the above-mentioned requirements for development of computer programs in structural design. ANN has been used in this thesis to (i) to determine the effective width of slabs coupling walls of different shapes and (ii) to determine the critical perimeter and ultimate punching shear capacity of the shear wall-floor slab junction, A computer program is also developed for determining the effective width and the ultimate punching shear capacity of the shear wall-floor slab junction. The effective width and ultimate failure load of slabs coupling walls of different shapes as predicted by ANN is compared with the experimental and theoretical results available. The agreement is found very good. The effect of shear wall thickness, opening between shear walls, effect of flange wall etc. on effective width and ultimate failure load of slabs are also analyzed. Once the neural network is developed, it will be able to predict the effective width and ultimate failure load for different new parameters within fraction of minute. It has been revealed that application of ACI and British Code in case of shear wall structures for calculating punching shear strength may lead to an overestimation of strength. In most practical cases, the formula proposed by Bari M. S. gives a conservative estimation of punching shear strength of shear wall structures.