Identification of Vortex-Induced Vibration of Tall Building Pinnacle Using Cluster Analysis for Fatigue Evaluation: Application to Burj Khalifa

Abstract

Pinnacles on top of tall buildings are vulnerable to vortex-induced vibrations (VIVs). These structures may undergo large-amplitude vibrations that can lead to fatigue damage accumulation. To assess the performance of buildings and its appendages, numerous structural health monitoring (SHM) programs have been installed on tall buildings. This continuous monitoring generates more than 1 trillion data points per year per building. Also, on many occasions, the data generated by SHM programs contain missing observations. The evaluation of fatigue life using conventional methods becomes an impossible task in this case. This paper introduces the use of machine-learning techniques as a potential solution to deal with the burgeoning data generated by tall building monitoring systems. In particular, the present study involves the evaluation of the crosswind fatigue life of the pinnacle of Burj Khalifa subject to VIVs using cluster analysis. This unsupervised machine-learning technique is used to develop a generalized framework robust to missing data to effectively identify and extract VIVs from a large pool of other responses recorded by the monitoring system. The data generated from 2010 to 2014 by the SmartSync monitoring system installed on Burj Khalifa are utilized for this study. The proposed framework is validated using a wind tunnel dataset of a bridge sectional model undergoing VIVs. The VIVs extracted from the SmartSync system through cluster analysis are used to evaluate the crosswind fatigue damage of the pinnacle of Burj Khalifa using conventional closed-form approximations. The proposed cluster analysis framework uses a step-by-step data-driven decision-making approach, thus widening the applicability of the method to other SHM programs.