Predictive System for Wind Induced response on tall structures using Machine Learning

Abstract

Tall buildings serve as a solution for the lack of land in most of the metropolitan cities. As the demand for taller buildings increases, the structural concerns related to tall buildings have become more prominent in the areas of research. For buildings with lesser height, it is easy to approximate the effect of aerodynamic forces. For tall structures, it is imperative to find accurate values of aerodynamic parameters instead of approximating them. With the use of wind tunnel experiments for studying the responses of tall structures, there has been a radical change in the method of evaluating the response towards wind effects. These developments recommended mandatory revisions in the existing codes and practices. Preliminary design plays a pivotal role in the design of tall structures. The data required for the preliminary design is acquired through wind tunnel tests or the CFD approach which provides the wind response of the structures for the desired wind speed and terrain conditions. Though wind tunnel provides valuable data, the procedure itself is time consuming and tedious which requires careful modelling of the building and its surroundings. On the other hand, the CFD approach is time consuming and computationally expensive since a simple simulation can take several hours to produce results and requires a powerful computer for the execution. These facts stand as a hindrance for a detailed study of wind effects on tall structures. A solution to bypass the complications of wind tunnel tests and time consuming CFD analysis is provided in the study. Analysis of aerodynamic and structural responses, the humungous process, is made easy with the help of a predictive system that can aid in the preliminary design of tall structures. This study investigates the aerodynamic and structural characteristics of tall buildings using wind tunnel, virtual wind tunnel experiments, and software analysis.