Abstract

Wind-induced interference effects are caused by the presence of adjacent structures, resulting in an increase or a decrease in the wind loads on a building depending mainly on the geometry and arrangement of these buildings, their orientation with respect to the direction of wind and upstream exposure. A comprehensive analysis of literature underscores the seriousness of the problem whilst exposing the inadequacy of existing research efforts in explaining and solving the problems of wind interference. The overriding reason for this is the complex nature of the problem and the large number of variables involved, even for the basic two-interfering buildings situation. The main aim of this research is to tackle the problem of interference in a systematic manner in order to establish a generalized set of guidelines that will be of practical use to building designers and planners. An extensive experimental program is developed to quantify interference effects and a methodology employing neural network representations is developed to model these effects, to define the complex relationships among the various parameters involved and to provide knowledgeable assistance to building designers. Extensive wind tunnel experiments have been conducted to find the mean and fluctuating forces as well as their spectra on a building due to an adjacent building of small, medium and large sizes, for several wind directions and various upstream exposure conditions, including city-centre situations. Interference effects are presented in the form of non-dimensional Interference Factors that represent the aerodynamic forces on a building with interference from an adjacent building, relative to the forces on a single free-standing building. The detailed experimental results have been analyzed and simplified to yield simple Interference Influence Grids, generalized guidelines and regression equations. Finally, a database of experimental cases has been constituted and integrated with the qualitative knowledge-base developed during the course of this study, in order to provide valuable assistance to building designers at an early design stage. The database is used to model interference effects via a novel technique based on Neural Network representations. The methodology is capable of developing functional relationships among limited experimental data and documented cases. The ability of Neural Networks to generalize when presented with limited data makes them an attractive tool for knowledge acquisition on wind-induced interference effects for which there is no theory or empirical generalization at present. The developed knowledge-based design adviser is effective in analyzing wind interference situations and in suggesting suitable remedial measures to alleviate problematic interference effects.