NEHRP Clearinghouse

Title
System Identification of Tall Vibrating Structures.
File
PB301064.pdf
Author(s)
Taoka, G. T.
Source
National Science Foundation, Washington, DC. Engineering and Applied Science., July 1979, 124 p.
Abstract
An investigation of the comparative accuracy of four different system identification methods for estimating frequency and damping parameters from identical ambient vibration records of tall structures is reported. Ambient vibration responses under natural wind conditions of five tall structures in Tokyo and Yokohama were recorded. The ambient data thus obtained were analyzed by four system identification methods: filtered correlation, spectral moments, spectral density, and two-stage least square. Factors that greatly affect the values of parameters estimates obtained from the ambient vibration record are record length, signal-noise ratio in the record, filter shape, and filter cutoff bandwidth. The filtered correlogram method was easy to program and generally produced reasonably accurate vibrational parameter estimates. The spectral moments methods was also easy to program and produced parameter estimates consistent with those of the filter correlogram method. Spectral density estimates are consonant with those from the other two methods. The two-stage least square method requires greater effort in programming and in analyzing data than the filtered correlogram or spectral moments methods. Building pictures, equations, data, graphs and references are included.
Keywords
Skyscrapers; Tall buildings; Earthquake resistant structures; Tokyo (Japan); Least squares method; Yokohama (Japan); Forced vibration; Dynamic response; Earthquake engineering; Power spectra; Japan; Earthquakes; Dynamic structural analysis