High resolution operational modal analysis on a five-story smart building under wind and human induced excitation

• Published in: Engineering structures 2018-12, Vol.176, p.279-292
• Main Authors: Sarlo, Rodrigo, Tarazaga, Pablo A., Kasarda, Mary E.
• Format: Article
• Language: English
• Subjects: Accelerometers; Ambient vibration; Automation; Bridges; Buildings; Civil structures; Damping; Estimates; Excitation; Frequency dependence; High resolution; Human influences; Instrumentation; Modal analysis; Parameter identification; Parametric analysis; Resonant frequencies; Set theory; Smart infrastructure; Standard deviation; Stochastic models; Stochasticity; Structural health monitoring; Wind power
• ISSN: 0141-0296; 1873-7323
• DOI: 10.1016/j.engstruct.2018.08.060

•State of the art, high resolution operational modal analysis is performed on a five-story smart building.•Novel adaptation of automated OMA based on variance estimation.•Excitation type (wind and human-induced loading) dictates the observed vibration behavior.•Decreases in natural frequencies observed with increasing excitation.•Wind excites primarily in one direction while human activity excites all directions. The Goodwin Hall Smart Infrastructure facility at Virginia Tech is a five-story “smart building” with an integrated network of 225 wired accelerometers. This study utilizes a subset of 117 sensors to perform Operational Modal Analysis (OMA) of the structure under wind excitation and establish a high-resolution benchmark modal characterization. Frequency Spatial Domain Decomposition and Stochastic Subspace Identification results are compared to validate the extracted modal parameters. Twelve structural modes were identified, including five high frequency local modes. These local modes are crucial features for structures with complex geometries and can generally be identified only with high density instrumentation. Through a parametric analysis and the use of standard deviation estimates, we determine that 50–60 min time series were optimal for high confidence on frequency and damping estimates. Furthermore, we employ standard deviation estimates to improve existing OMA automation methods. This enables continuous modal parameter extraction over a four-day period to understand the characteristics of the two main forms of ambient excitation: wind and human-induced. Although similar continuous analyses have been conducted on bridges, few of this kind exist for buildings. In general, we observe that modal participation of the three fundamental modes is closely tied to wind and human activity and that the confidence in frequency and damping estimates of these modes improves as the excitation increases. Slight decreases in natural frequency with increasing participation occur for several modes, agreeing with behavior observed in bridge monitoring studies. Finally, wind is seen to excite primarily in one direction, whereas humans induce even excitation in all directions.