Blind identification of the Millikan Library from earthquake data considering soil-structure interaction

Abstract

SUMMARYThe Robert A. Millikan Library is a reinforced concrete building with a basement level and nine stories above the ground. Located on the campus of California Institute of Technology (Caltech) in Pasadena California, it is among the most densely instrumented buildings in the U.S. From the early dates of its construction, it has been the subject of many investigations, especially regarding soil-structure interaction effects. It is well accepted that the structure is significantly interacting with the surrounding soil, which implies that the true foundation input motions cannot be directly recorded during earthquakes because of inertial effects. Based on this limitation, input-output modal identification methods are not applicable to this soil-structure system. On the other hand, conventional outputonly methods are typically based on the unknown input signals to be stationary whitenoise, which is not the case for earthquake excitations. Through the use of recently developed blind identification (i.e. output-only) methods, it has become possible to extract such information from only the response signals because of earthquake excitations. In the present study, we employ such a blind identification method to extract the modal properties of the Millikan Library. We present some modes that have not been identified from force vibration tests in several studies to date. Then, to quantify the contribution of soil-structure interaction effects, we first create a detailed Finite Element (FE) model using available information about the superstructure; and subsequently update the soil-foundation system's dynamic stiffnesses at each mode such that the modal properties of the entire soil-structure system agree well with those obtained via output-only modal identification.

Bayesian identification of soil-foundation stiffness of building structures

“…The identification is conducted using a single observation of =10.585 rad/s for the fundamental natural frequency and ={0.0256, 0.0064, 0.4474, 1} T for the fundamental mode shape of the building. Ghahari et al presented this information by applying a blind identification technique on the recorded response of the Millikan Library during the 2002 Yorba Linda, California, earthquake. The height of the building is 43.28 m, and the dimensions of the plan are 23 m by 21 m. The total mass of the building is estimated at 11.95 × 10 6 kg .…”

“…Only a few of the past studies have employed real‐life seismic data to quantify the dynamic soil‐foundation stiffness . Most pertinent to the present effort, Ghahari et al proposed an identification methodology based on a deterministic model updating technique for estimating the dynamic stiffness from strong motion records. In that approach, an error function is minimized to match the response of a soil‐structure system to real‐world observations of accelerometer records by optimizing the underlying unknown parameters of the system, including the soil‐foundation representative springs.…”

“…In that approach, an error function is minimized to match the response of a soil‐structure system to real‐world observations of accelerometer records by optimizing the underlying unknown parameters of the system, including the soil‐foundation representative springs. To this end, Ghahari et al modeled the superstructure using detailed finite element models. However, as suggested by several past studies, soil‐structure interaction (SSI) primarily affects the fundamental natural mode of vibration.…”

Blind modal identification of non-classically damped structures under non-stationary excitations

“…Time‐frequency‐based methods —which we have previously adopted for civil engineering applications —are among the different available solution techniques for BSS problems and are capable of handling unknown input excitations that are non‐stationary. We have successfully verified and applied these newly developed methods through the use of data sets from various experiments and earthquake events .…”

Application of Timoshenko beam theory to the estimation of structural response

“…The differences in the beam theory used in the derivation of each of these beam models is discussed later. Timoshenko beam has been extensively used in literature to simulate building motions [2], to predict the propagation of waves in buildings [3], and in the performance of system identification [6].…”