With improved understanding of the dynamic behaviour and seismic performance of structures, numerous advancements in earthquake engineering have been developed in recent years. The structures can be designed, in the most cost effective and safe manner, by utilizing the specific ability to model the dynamic structural behaviour. For analytical modelling, the most widely used tool in structural engineering analyses and design, is finite element method (FEM). But "how accurately the established FEM is predicting the dynamic characteristics of the structure", is a question to ponder over. The reason for this issue is due to the errors present in the solution, resulting from the approximation involved in the formulation of FEM. However, our concern in engineering practice is to achieve a FEM which is capable of reproducing the real structural behaviour (Yang and Chen 2009). To overcome this, the FEM should be refined or updated by using the measured responses from the actual or prototype structure.
The importance of field measurements, from an actual structure response, stems from the simple fact that these represent the ground-truth about structural behaviour. However, most of the current testing methods, such as shake table or pseudo-dynamic experiments, are restricted to laboratory-scale structures and components. Moreover, such idealized laboratory experiments cannot account for the complexity of in situ structures, such as influence of environmental and operational conditions (e.g. temperature, humidity, loading etc.), non-structural components, soil-structure interaction, etc (Su et al. 2005; Trifunac and Todorovska 1999). While, measurements from in-situ structure even on few points, presents a great deal of information about the structural performance (Limongelli 2003).
New Zealand is situated in such a region where earthquakes are a common occurrence. It lies at the edge of the Australian and Pacific plates (Fig. 1). Past earthquakes in the region have caused major losses. The seismic response of structures therefore, is a major concern to the community. Like many other countries, several structures across New Zealand are instrumented under the banner of GeoNet project for measurement and understanding of the dynamic behaviour of structures during earthquakes. GeoNet is a New Zealand Government project to build and operate a modern geological hazard monitoring system. It comprises a network of geophysical instruments, automated software applications and skilled staff to detect, analyze and respond to earthquakes, volcanic activity, large landslides, tsunami, and the slow deformation that precedes large earthquakes. So far, GeoNet has placed an emphasis on recording ground motions and volcanic activity, but recent initiatives aim at extending GeoNet instrument arrays to monitoring of infrastructure such as buildings and bridges(David Baguley and Young 2008; Deam and Cousins 2002).
Figure 1. The distribution of New Zealand earthquakes and the boundary of the Pacific and Australian tectonic plates
(Courtesy: www.geonet.org.nz)Interaction with the GeoNet management and researchers enabled the University of Auckland an early access to strong motion data from building monitoring arrays in the Wellington and Canterbury regions. This PhD study will analyze these data. The present research will contribute towards better assessment of the behaviour of structures subjected to ground motion. This will also be a pilot study that will help to understand the nature of GeoNet data, and opportunities it presents and limitations it may have.http://www.geonet.org.nz/images/site/seis_map_large.jpg
2. OBJECTIVES
The primary objective of this research is "to investigate and improve understanding of the dynamic behaviour of instrumented buildings by developing analytical models and calibrating those using filed data so that these models can represent real behaviour during earthquakes". Following are the areas of focus to achieve this objective:
Investigating the effect of amplitude of response and the frequency content of many different earthquakes on the dynamic properties (frequency, damping) of the instrumented building using system identification
Exploring the effect of environmental and operational conditions i.e. Temperature variation, moisture condition and loading, on dynamic performance of buildings by using modal analysis and system identification
Investigating the effect of non-structural elements and soil-structure-interaction on structural response by incorporating them in FEM simulations and comparing to the real behaviour during earthquakes.
Exploring new ways of calibrating the Finite element models (FEM) using field data.
3. LITERATURE REVIEW
This section presents an overview of:
Seismic response of structures
Amplitude dependency of frequency and damping ratios of structures
Effect of soil-structure interaction
3.1. Seismic Response of Structures
Well-instrumented buildings present an excellent opportunity during moderate to intense earthquakes for studying their seismic response and to check the efficiency of seismic vulnerability assessment methods. In this regard system identification techniques can be used to extract dynamic characteristics of an instrumented building, which forms an important feature of recording the seismic motions (Miranda and Bertero 1996). These studies can also be useful for the improvement of methodologies involved in, design and analyses of structures, and earthquake hazard reduction programmes (Celebi 1997).
Laboratory tests can be useful but are not as complete as full-scale or in-situ experiments. From civil engineering point of view, the laboratory scale experiments are not able to incorporate many important features e.g. true effects of soil-structure interaction, environmental and operational conditions, and non-structural components. While the full-scale tests present the as built environment which includes all the physical properties of reality. However, the challenges of full-scale or in-situ experiments are to discover, record and interpret the reality. The popularity of in-situ experiments is increasing since the introduction of first strong motion accelerograph to record ground motion in early 1930's. Since then the instruments are installed in buildings, dams and bridges to capture their actual response during earthquakes and ambient vibrations (Trifunac and Todorovska 1999).
From in-situ experiments, it has been revealed that there are many factors on which the response especially dynamic response of the structures depends. It is observed that response is sensitive to the intensity of the ground motion which governs the building deformations. Non-structural components (NSC) participation in dynamic response is a function of building deformations (Sashi et al. 2004). Many researchers reported the participation of NSC in the dynamic response during strong shaking (Lee et al. 2007; Liew et al. 2002; Torkamani and Ahmadi 1988). During the test of a tall building, it has been observed that NSC participation in total lateral stiffness of the building is as much as 87% (Su et al. 2005). The detailed discussion on the inclusion of NSC in modelling and their effect on response of structures will be discussed in the section 3.1.3.
Consideration of ground motion characteristics, local site conditions and soil-structure interaction (SSI) is reported to have considerable effect on the seismic response of the buildings. During different earthquakes the response of the building is varying because of the level of shaking, the site frequency which causes resonance, rocking and beating phenomenon which emphasize the need of soil-structure-interaction investigation (Celebi 1993; Celebi 1994; Celebi and Safak 1991; Celebi and Safak 1992). Therefore, in seismically active regions of the world, seismic monitoring of structures makes up an integral part of hazard reduction strategies. Countries like United States, Japan, Taiwan, Mexico, Chile, Italy, Turkey and Greece have established extensive programmes for monitoring of structures during earthquakes (Celebi 2007). The effect of response amplitude, frequency content of earthquake and soil-structure-interaction on dynamic response of structures are discussed in detail in section 3.1.1 and 3.1.2 respectively.
The response of structures to shaking over long term monitoring is extremely complex. Substantial research efforts have been dedicated to study the influence of environmental factors on dynamic characteristics through field measurements and dynamic tests (Farrar et al. 1997; Sohn et al. 1999; Xia et al. 2006). It has been concluded that temperature is the major environmental factor causing variation in modal frequencies and structural damping (Li et al. 2009; Ni et al. 2005; Yan et al. 2005). The influence of environmental conditions is detailed in the section 3.1.4.
3.1.1. Amplitude dependency of frequency and damping ratios of structures
Natural frequencies and damping ratios are very important parameters which affect the dynamic response of structures under dynamic actions such as wind or earthquake excitation. These dynamic characteristics of building structures are observed to depend on the vibration amplitude. The natural frequencies are usually determined from conventional methods with reasonable accuracy. The resonant acceleration of a vibrating structure is inversely proportional to the square root of the damping. Experience has shown that damping is one of the most difficult structural parameter to predict at the design stage. It is therefore, necessary to investigate damping using full-scale measurement on the existing structures. These full-scale experiments will help in (i) checking the dynamic characteristics and thus response calculations for the structure (ii) improving the understanding of damping mechanism, which will make possible, better theoretical models (Littler 1995).
Many researchers have investigated the amplitude dependency using Random Decrement Technique(RDT) (Tamura et al. 1993; Tamura et al. 1994a). In a study, RDT ranked by peak amplitudes is used to calculate effectively the wind-induced response of three towers (Tamura and Suganuma 1996). Two procedures were used by Tamura and Suganuma. In the first, RDT ranked by peak amplitudes is used. The second one calculates the dependence by arranging the dynamic characteristics, estimated using responses with various input levels, in the order of the average amplitude. They have observed for the investigated three towers that natural frequencies tend to decrease with increasing amplitude while damping ratio tend to increase (Tamura and Suganuma 1996).
Celebi in a study has compared the variation of damping and fundamental periods for 5 instrumented buildings, using low amplitude tests and strong motion records. He observed the percentages of critical damping and the corresponding fundamental periods from low-amplitude test data appreciably lower than those determined from strong-motion recordings. It has been concluded that soil-structure-interaction, to be discussed in the next section, and nonlinear behaviour during strong shaking is the main cause of this difference (Celebi 1996).
In another study, the apparent frequency changes form one earthquake to another has been investigated by taking 5 earthquake excitations, recorded on Van Nuys seven storey hotel and 2 ambient vibration tests. It was observed that the frequency changes depend on level of shaking. Ambient vibration tests produced the largest frequencies while the lowest frequencies were observed during the largest shaking (from 1994 Northridge earthquake). For EW and NS direction, the maximum change is by a factor of 2.2 (55%) and 3.5 (71%) respectively. It was concluded that these changes are due to the nonlinearity in the response of the foundation soil (Trifunac et al. 2001).
Recently many researchers have given serious thoughts to response of tall buildings under wind excitations. The reason behind the studies is the actual damping ratio being a non-linear parameter with amplitude-dependent property. To understand better this phenomenon, instrumented Di Wang tower in downtown Shenzhen, China is studied during the passage of several typhoons (Li et al. 2003; Wu et al. 2007). In 2001 Chicago Full-Scale Monitoring Program is established to evaluate the performance of high-rise buildings by comparing their measured and predicted response. Under this program 3 buildings in Chicago, USA, 1 in Seoul, Korea and 1 in Toronto, Canada are instrumented with GPS, accelerometers and anemometers. The results achieved from these studies show considerable amplitude dependency of frequency and damping ratios due to winds (Correa 2009). Wind-induced excitations are out of the scope of this research study so our focus will be on earthquake and ambient vibrations only.
Since many researchers (Celebi 1996; Celebi et al. 1991; Trifunac and Todorovska 1999) concluded that one of the major factors, in variation of dynamic properties like frequency and damping ratio under earthquake excitation, is the non-linearity in foundation soil and possible soil-structure-interaction, therefore in the following section effect of soil-structure-interaction will be discussed in detail.
3.1.2. Effect of soil-structure-interaction (SSI)
A structure, its foundation and the surrounding soil constitute a system. Due to the flexibility of soil, the system period can be longer than the period of the fixed base building. The period can further prolong if under large amplitude excitations, the structure or the foundation soil or both experience nonlinearity in their response. Longer periods have been observed of the buildings during strong earthquake excitations (Trifunac et al. 2001; Udwadia and Trifunac 1974). Since building period constitutes an important part in the design and analysis of earthquake resistant structures so it should be determined with utmost care. Soil-structure-interaction investigations therefore, are necessary to see the actual response of structures during earthquakes.
Numerous research studies have reported the significance of soil-structure-interaction investigations. Papageorgiou and Lin has studied the response of a 14 storey reinforced concrete Hollywood Storage building during Whittier Narrows earthquake and observed clear evidence of SSI in the longitudinal direction and weak SSI in the transverse direction. From the recorded displacements, it was observed that stiffness in longitudinal direction is more as compared to transverse direction. This is due to the stiffer exterior longitudinal panels than the transverse frames. The soil at the site is relatively stiff. The flexible transverse frames did not develop large base shear to deform it while on longitudinal direction the stiffer frames and panels did so causing significant SSI effects (Papageorgiou and Lin 1991).
Evaluation of SSI effects during strong motion events was extensively studied by Celebi and Safak. The data from instrumented buildings was analyzed using Fourier amplitude spectra and rocking and beating phenomenon were observed. It was concluded that SSI strongly affects the structural response of the structures during the event of strong shaking and can change the fundamental frequency significantly (Celebi and Safak 1991; Celebi and Safak 1992; Safak 1993).
In another study, SSI effects were explored by measuring the response of the instrumented building during ambient vibrations and Loma Prieta earthquake (LPE). The measured frequency ratio of LPE over ambient vibrations ( fLPE/fAMB) was 0.70 for the first mode response (E-W translation) and was 0.68 for the second mode response (N-S translation). While damping estimates observed were smaller in ambient vibrations than LPE using system identification techniques, to be discussed in section 3.3. A computer model of the building was developed assuming (i) fixed base and (ii) soil springs. The frequency ratios between these two assumptions were matching the measured frequency ratios. It was concluded that the frequency difference between low ambient vibrations and large earthquake excitations was mainly due to SSI (Phan et al. 1994).
Bhattacharya, Dutta et al. assessed the SSI effect by considering number of scenarios in low-rise buildings on isolated, grid footings and raft foundations using computer modelling. They idealized the structure using springs to incorporate the soil flexibility and in-filled brick walls were idealized as equivalent compression only diagonal struts. The studies showed that SSI effects can considerably influence the response of low-rise structural systems (Bhattacharya and Dutta 2004; Bhattacharya et al. 2004; Dutta et al. 2004). To see the effect of dynamic soil-structure-interaction on response of asymmetric buildings, Shakib and Fuladgar formulated a time domain approach. Three-dimensional system for asymmetric building was studied on different soil conditions. Soil is idealized as linear elastic solid element and the contact surface between foundation and soil is modelled as linear plane interface elements with zero thickness. It was deduced that SSI effects reduce the lateral and torsional displacements of asymmetric buildings causing a decrease in time period of the structure (Shakib 2004; Shakib and Fuladgar 2004).
A parametric system identification technique was used by Stewart and Fenves to evaluate SSI effects in buildings from strong motion records. For fixed and flexible base modal parameters, recordings of base rocking, lateral roof, foundation and free-field motions are required. This technique was implemented at 11 instrumented sites and the results achieved were compared well with known parameters (Stewart and Fenves 1998). Another parameter identification technique was developed by Lin et al. to study the SSI with torsional coupling (TC). The foundation rocking as well as translational and torsional motions of the foundation floor are required as input for the identification. The developed technique was applied to 2 instrumented buildings in Taiwan, namely Civil and Environmental Engineering Building and Seismology and Applied Geo-physics Department Building at National Chung Hsing University. The strong motion record measured at these two buildings during 1999 Chi-Chi Taiwan earthquake was used for the study. There was 21% difference in fundamental period of the superstructure on a stiff soil site. It was concluded that all of the foundation motions should be included in system input to avoid longer than actual periods of structures (Lin et al. 2008).
