Author: Chen, Sixin
Title: Field monitoring and prediction of temperature-induced responses of a supertall structure using neural networks
Advisors: Xia, Yong (CEE)
Degree: M.Sc.
Year: 2017
Subject: Hong Kong Polytechnic University -- Dissertations
Tall buildings -- Thermal properties
Buildings -- Environmental engineering
Department: Faculty of Construction and Environment
Pages: ix, 84 pages : color illustrations
Language: English
Abstract: The non-uniform distribution of temperature of tall structures often causes displacement and thermal stress. As these changes mix up with those resulting from structural damage or unusual events, and cause false-positive or false-negative damage diagnosis, it is important to eliminate them. However, studies on temperature actions on tall structures are limited due to insufficient real measurement data. Apart from that, few studies on modelling the thermal response have been released, opposite to those on temperature-frequency relationship, With a comprehensive monitoring system installed, the Canton Tower, a tube-in-tube supertall structure, provides abundant data for the data analysis, data visualisation and model training. In this study, temperature and strain measurements of Canton Tower in two summers in 2011 and 2012 were used for investigation. Daily and seasonal variations of temperatures and strain responses at different heights and facades were first presented and analysed. Systematic methodologies for processing the raw data, separating datasets, training models, selecting the proper architectures, and evaluating model's performance were then applied. Consequently, 20 neural networks were generated to capture the relationship between the temperature distribution and 20 strains of various components. After that, both the indices to evaluate models' generalisation capability, and comparisons between predicted strains and measured strains in 2 summers, were analysed. Finally, the models were applied for separating the wind-induced responses from temperature-induced responses on a typhoon day. It is revealed that segments with functional floors enclosed with curtain walls have lower and less fluctuant temperatures, and higher sections tend to be cooler due to elevation. For sections of the inner tube that are exposed to the environment, the east part is the hottest and the time when different facades get warm follows a specific sequence. A 6-hour temperature delay between concrete and steel surface of CFT columns, and an obvious temperature difference between the inner and outer tube are observed. Compared with temperature, thermal responses show a more complicated pattern. In general, when the sun begins to rise, the concrete in both the inner tube and outer CFT columns is under tension, while the steel tube of CFT columns is under compression. It is found that PCA can significantly compress the size of temperature data and slightly improve the prediction performance. The performance evaluation of the generalised neural networks shows that, generally, the strains in the summer of 2011 can be predicted well with high correlation coefficient and low residuals. Although most daily variation patterns of strains in 2012 can be captured, for some strains, the bias between predictions and measurements are observed. After separating the temperature-induced and typhoon-induced responses, it is found that the decreasing of temperature leads to compression of the concrete in CFT columns which shares the same magnitude as wind-induced responses.
Rights: All rights reserved
Access: restricted access

Files in This Item:
File Description SizeFormat 
991022142757603411.pdfFor All Users (off-campus access for PolyU Staff & Students only)7.91 MBAdobe PDFView/Open

Copyright Undertaking

As a bona fide Library user, I declare that:

  1. I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
  2. I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
  3. I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.

By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.

Show full item record

Please use this identifier to cite or link to this item: