TY - JOUR
T1 - Wind effects on the world's longest spatial lattice structure
T2 - Loading characteristics and numerical prediction
AU - Fu, J. Y.
AU - Li, Q. S.
PY - 2007/10
Y1 - 2007/10
N2 - The 486-m long roof structure of Shenzhen Citizens' Centre is the world's longest spatial lattice structure. This paper presents some selected results from a combined wind tunnel and numerical simulation study of wind effects on the extra-long-span roof structure. In this study, simultaneous pressure measurements on its entire roof are made in a boundary layer wind tunnel, and the measured wind pressures, such as mean, root-mean-square (rms) and peak pressure coefficient distributions on the roof are presented and discussed. Based on the measured data from a number of pressure taps, a numerical simulation approach using backpropagation neural networks (BPNN) is developed for the predictions of wind-induced pressure time series at other roof locations which are not covered in the wind tunnel measurements. The BPNN is trained with the pressure data time series measured from adjacent pressure taps. The good performance of the developed neural network is demonstrated by comparing the predictions with the model test results, illustrating that the BPNN approach can serve as an effective tool for the design and analysis of wind effects on large roof structures in conjunction with wind tunnel tests. © 2006 Elsevier Ltd. All rights reserved.
AB - The 486-m long roof structure of Shenzhen Citizens' Centre is the world's longest spatial lattice structure. This paper presents some selected results from a combined wind tunnel and numerical simulation study of wind effects on the extra-long-span roof structure. In this study, simultaneous pressure measurements on its entire roof are made in a boundary layer wind tunnel, and the measured wind pressures, such as mean, root-mean-square (rms) and peak pressure coefficient distributions on the roof are presented and discussed. Based on the measured data from a number of pressure taps, a numerical simulation approach using backpropagation neural networks (BPNN) is developed for the predictions of wind-induced pressure time series at other roof locations which are not covered in the wind tunnel measurements. The BPNN is trained with the pressure data time series measured from adjacent pressure taps. The good performance of the developed neural network is demonstrated by comparing the predictions with the model test results, illustrating that the BPNN approach can serve as an effective tool for the design and analysis of wind effects on large roof structures in conjunction with wind tunnel tests. © 2006 Elsevier Ltd. All rights reserved.
KW - Long-span roof
KW - Neural networks
KW - Wind tunnel test
KW - Wind-induced pressure
UR - http://www.scopus.com/inward/record.url?scp=34547828022&partnerID=8YFLogxK
UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-34547828022&origin=recordpage
U2 - 10.1016/j.jcsr.2006.12.001
DO - 10.1016/j.jcsr.2006.12.001
M3 - RGC 21 - Publication in refereed journal
SN - 0143-974X
VL - 63
SP - 1341
EP - 1350
JO - Journal of Constructional Steel Research
JF - Journal of Constructional Steel Research
IS - 10
ER -