Loading...

System Identification of Soil-Structure Systems through Seismic Signal Processing

Ghahhari, Farid | 2013

1195 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 44931 (09)
  4. University: Sharif University of Technology
  5. Department: Civil Engineering
  6. Advisor(s): Ghannad, Mohammad Ali
  7. Abstract:
  8. Dynamic characteristics of structures—viz., natural frequencies, damping ratios, and mode shapes—are central to earthquake resistant design (model validation) and health monitoring. For civil structures like buildings and bridges, these characteristics can be highly influenced by surrounding soil in addition to the super-structure’s properties and cannot be correctly estimated using numerical modeling. For example, soil’s flexibility decreases the natural frequencies of the system; and in most cases, soil provides additional damping due to material hysteresis and radiation. On account of this fact, system identification techniques must be applied to extract such information from field data, i.e., ambient vibration or earthquake-induced responses. Traditionally, both input and output signals must be available for employing system identification techniques; however, measuring input motions for civil structures is hard and in most cases impossible. For ambient data, it is not easy to measure input excitations such as environmental traffic loads. For earthquake-induced vibrations, there are two main reasons which prevent measuring input motions: 1- soil-structure interaction effects, and 2- noise effects. Dynamic response of the structure and the foundation, produced by the inertial soil-structure interaction, makes the foundation response to be different from actual input, while vibrating structure as a new wave propagating source causes differences between Free-Field Motions (FFMs) recorded on the nearby ground surface and actual input excitations. That is, neither the FFMs nor the recorded responses of the foundations may be assumed as input. Additionally, kinematic soil-structure interaction, produced by the contrast between the stiffness of a (nearly rigid) foundation and the surrounding soil, causes the motion experienced by the foundation, i.e., Foundation Input Motions (FIMs), to differ from the FFMs. Even in cases in which both kinematic and inertial soil-structure interactions are negligible, in many instances, the foundation responses are not recorded or are recorded with a low signal-to-noise ratio.
    Due to limitations listed above, several output-only system identification methods have been already developed, in which only response signals are used for identification. Unfortunately, most of the existing output-only methods are limited to free vibration data, or weak stationary ambient, wherein the identified modal properties are related typically to small-amplitude vibrations. However, it is well known that the dynamic characteristics of most civil structures are amplitude dependent; and thus, parameters identified from low-amplitude responses do not match well with those from strong excitations—which arguably are more pertinent to seismic design. Moreover, in techniques devised for ambient vibrations, it is usually assumed that the input excitation is a broadband stochastic process, which is modeled as stationary Gaussian white noise. On account of these facts, most of available output-only identification techniques cannot be employed for non-stationary response signals recorded during strong ground motions.
    In this study, first, an existing blind source separation technique is adapted for output-only identification of soil-structure systems from free/ambient vibrations. As mentioned earlier, existing output-only identification methods are able to extract modal information for such data; however, this technique is an extension of the Second Order Blind Identification (SOBI) method, which is fairly well established in a number of other areas including sound separation, image processing, and mechanical system identification. The relative ease of implementation of this output-only identification technique has been the primary source of its appeal. Here, the SOBI is adapted so that it can be applied on the data from systems with complex mode shapes, which is commonly occurs in soil-structure problems.
    Due to the non-stationary nature of earthquake excitations, the above mentioned SOBI method cannot be employed for output-only modal identification of structures using seismic response (output) signals. Herein, a new identification method is presented, which is applicable for such problems, i.e., blind identification under earthquake excitations. Although this method can be also used for ambient vibration data, it is not recommended, because the SOBI technique is computationally much cheaper. In this method, first, the response signals’ Spatial Time-Frequency Distributions are used for blindly identifying the mode shapes and the modal coordinate signals (or their time-frequency distributions). Second, the modal coordinates are analyzed to determine the system’s natural frequencies and damping ratios on the premise of linear behavior for the system. The proposed method is even applicable for systems with sparse instrumentations, in which the number of active modes is greater than the number of instrumented floors. Both simulated and real-life data set are used to verify the method. Results show that this novel output-only modal identification technique can successfully extract modal parameters of civil structures without having input motion signals.
    Employing above mentioned blind identification techniques, new solutions are suggested in this study through which frequency-dependent soil-foundation impedance functions can be extracted from both free/ambient vibrations and earthquake-induced response signals.
  9. Keywords:
  10. Blind Source Identificatin ; Modal Identification ; Output-Only Techniques ; Strong Ground Motion ; Earthquake Record ; Soil-Structure Interaction ; Time-Frequencey Distribution ; Dynamic Stiffness ; Natural Frequency ; Free Vibration

 Digital Object List

 Bookmark