Abstract
The modal parameters of the Robert A Memorial Millikan Library (referred to Millikan Library) building are estimated with forced vibration test techniques using various different instruments. The 10-story building is a reinforced concrete structure and permanently instrumented with a sensor array of 36-channel Force-Balanced accelerometers (referred to FBA) and two digital recorders primarily to monitor the building dynamic response during earthquakes. Three different tri-axial Micro-Electro Mechanical System (referred to MEMS) accelerometers were also deployed to measure the building response at the basement floor, at 5th floor, and at 9th floor during forced vibration testing. The amplitudes of the responses measured are compared in time domain. Whether the first two fundamental periods can be identified is mainly considered in frequency domain. SNR (abbreviation for Signal to Noise Ratio) is introduced to determine the data quality by fitting a sine function to recorded data. Amplitude characteristic analysis indicates that signal recorded from MEMS accelerometers shows occasional spikes. At low intensity shaking signal amplitudes tend to be equal and amplitude from MEMS is almost always higher than that from FBA. At higher amplitudes, amplitude mismatch becomes larger. The first two fundamental periods of the building in North-South and East-West directions can be clearly distinguished in frequency domain from the response signals recorded by FBAs. Responses recorded by MEMS accelerometers are in higher chaos at the basement floor and visual detection of fundamental frequencies of the building is infeasible The SNR computed for 9th floor response is larger than that of the basement floor response. That indicates that low amplitude response contains more noise than relatively higher response. For the same floor, the SNR computed for the response measured by FBA is always larger than those of responses measured by three MEMS accelerometers. Therefore, MEMS accelerometers should be applied for structural health monitoring with caution when ambient data is of concern.
Acknowledgments
We would like to thank USGS’s National Strong Motion Project technicians for setting up the instruments and retrieving the test data. Special thanks are extended to Erol Kalkan and Hasan Ulusoy providing us response data and the computing codes for data process. Shahneam Reza has kindly prepared the CAD drawings of the Building used in this manuscript. Last but not least, authors are grateful to Beijing earthquake agency for the financial support to Fei Wang’s staying and scientific research in US.