Keywords

19.1 Introduction

Health monitoring, and particularly system identification, of civil structures has become an important field bridging the gap between numerical model-based response predictions and actual response of structures. The estimation of dynamic properties from measurement data recorded in situ on real [1, 2] structures or in the laboratory on large/full-scale structural specimens [3, 4] under realistic conditions allows to calibrate, validate and improve numerical (typically finite element) modeling techniques of structures. Experimental modal analysis and operational modal analysis are the main procedures to estimate the modal parameters (natural frequencies, damping and mode shapes) from recorded structural vibration data. Most of the studies done in this area were performed using data collected from completed structures; only a few have analyzed the variation of the dynamic properties of buildings during construction [5–8] and the effects of the nonstructural components (NCSs) on these dynamic properties [4, 9, 10]. However, these studies were carried out on real buildings and using output-only system identification methods; and it was not possible to closely follow the construction process in order to disaggregate the effects of different construction activities and NCSs, and the effect of the amplitude of the excitation on the dynamic characteristics of the buildings. This paper focuses on the investigation of a full-scale five-story reinforced concrete (RC) building fully equipped with a wide range of nonstructural components, which was tested on the NEES-UCSD shake table. Measured responses from ambient and forced vibration tests obtained during the construction and at different stages of installation of the NSCs are used to estimate the dynamic characteristics of the test specimen.

19.2 Description of Specimen and Construction Process

19.2.1 Specimen

The test structure is a full-scale 5-story cast-in-place RC frame building. It has two bays in the longitudinal direction (direction of shaking) and one bay in the transversal direction, with plan dimensions of 6.6 by 11.0 [m]. The building has a floor-to-floor height of 4.27 m, a total height of 21.34 [m] and an estimated total weight of 3010 [kN] for the bare structure and 4420 [kN] for the structure with all the nonstructural components (excluding the foundation, which has a weight of 1870 kN approximately), respectively. The seismic resisting system is provided by a pair of identical moment resisting frames in the North and South bays. The beams have different details at different floors. The specimen has six 66 × 46 [cm] columns reinforced with 6#6 and 4#9 longitudinal bars and a prefabricated transverse reinforcement grid (baugrid). The floor system consists of a 20.3 [cm] thick concrete slab for all levels. There are two main openings on each slab to accommodate the stair and elevator. Two transverse concrete walls 15.2 [cm] thick provide the support for the elevator guiderails. More details about the structural system and NCSs can be found in [11]. Figure 19.1aa shows a schematic view of the test building.

Fig. 19.1
figure 00191

(a) Schematic view of the test specimen, (b) bare structure (August 12, 2011) and (c) complete building (February 23, 2012)

Full size image

19.2.2 Construction Process and NCSs

The construction of the building began in May 2011 with the foundation system. The bare structure was completed on September 23, 2011. In parallel to the construction of the building, the installation of the stair began on August 13 and was completed on October 11, 2011 (see Fig. 19.1bb). The installation of the NCSs started immediately after completion of the structure. Each level of the structure was equipped with different nonstructural components and contents to support different occupancies. The first floor was designated as a utility floor, while the second floor was modeled as a home office and a laboratory environment area. The third floor had two computer servers representing important electronic equipments commonly damaged during earthquakes. Levels four and five were designated as hospital floors with an intensive care unit (ICU) and a surgery suit, respectively. A fully-operational passenger elevator, metal stairs, a ceiling subsystem and gypsum board partition walls were also installed in the building. The facade consisted of a light gauge metal stud balloon framing overlaid with synthetic stucco spanning over levels one to three and precast concrete panels on levels four and five. Anchored to the roof were a penthouse, an air handling unit and a large water-filled cooling tower. Table 19.1 summarizes the general properties and date of installation of the main NCSs. The building system was completed on February 22, 2012 (see Fig. 19.1cc).

Table 19.1 Properties of the main NCSs and dates of installation
Full size table

19.3 Instrumentation Array and Dynamic Tests

19.3.1 Instrumentation Array

Before pouring the concrete slab at the third level, on August 18, 2011, a temporary accelerometer array was deployed in the structure, consisting of four translational sensors per floor (two per translational direction). The data were sampled at 240 Hz and the raw acceleration time histories were detrended and filtered using a band-pass order 4 IIR Butterworth filter with cut-off frequencies at 0.25 and 25 Hz, frequency range which covers all the modes participating significantly in the response of the system.

19.3.2 Dynamic Tests

Vibration data from dynamic tests (more than 400) were recorded on the building during the period from August 2011 to February 2012, including ambient vibration, free vibration (shock) and forced vibration tests (low amplitude white noise base excitation) using the UCSD-NEES shake table. Ten minutes of ambient vibration data were recorded everyday and impact tests (using a pendulum consisting of a truck tire hanging at the end of a cable attached at the tip of the arm of a boom lift) were performed on average once a week between August 18 and December 22, 2011. White noise tests were carried out at key stages of the construction: bare structure condition on October 12, 2011 and complete building condition on February 23, 2012. Table 19.2 summarizes the recorded vibration data used in this study.

Table 19.2 Recorded vibration data used in this study
Full size table

19.4 System Identification Methods Used

In order to estimate the modal properties of the building specimen at different construction stages, two state-of-the-art output-only system identification methods, both assuming broad-band excitation, are used for the ambient vibration data: Data-Driven Stochastic Subspace Identification (SSI-DATA) and Natural Excitation Technique combined with Eigensystem Realization Algorithm (NExT-ERA). For the white noise base excitation data, in addition one input-output method is employed: Observer/Kalman Filter Identification combined with Eigensystem Realization Algorithm (OKID-ERA). Detailed explanations of these methods may be found in [12–14] for SSI-DATA, NExT-ERA and OKID-ERA, respectively.

19.5 Evolution of the Identified Modal Properties During Construction

The effects of six different main construction activities (Table 19.3) in the construction process are analyzed. The bottom row of Fig. 19.2 shows the time evolution of the construction in terms of the main activities, the left and right panels of the first four rows show the temporal evolution of the identified natural frequency and damping ratio, respectively, of each of the first four modes of the structure. Clearly, since the structure becomes taller and therefore more flexible as the construction progresses, the natural frequencies decrease with time.

Table 19.3 Main construction activities
Full size table
Fig. 19.2
figure 00192

Temporal evolution of the natural frequencies and damping ratios of the first four modes identified during construction

Full size image

It can be seen that the actions inducing the most significant abrupt changes in the natural frequencies are the pouring of the slab (A2) at the fifth and roof levels (September 6, 2011, and September 22, 2011 respectively) and pouring of the columns and shear walls (A4) at the fourth and fifth levels (August 24, 2011, and September 8, 2011 respectively). This can be explained by the fact that after the pouring, both activities increase significantly the mass of the building without considerably changing its lateral stiffness, which starts to increase as the concrete gains strength. The magnitude of the jumps in the variation of the natural frequencies due to concrete pouring reduces as construction progresses, since the added mass becomes progressively lower relative to the total mass of the building. On the other hand, as expected, the effect of formwork and shoring (A1 and A3) is negligible because their mass is low relative to the total mass of the structure and they do not contribute to the lateral stiffness of the building. Similarly, stairs installation (A6) and shoring removal (A5) do not induce any observable change in the dynamic properties of the building for the same reasons as for activities A1 and A3. Finally, it is observed that, from September 26 to October 10, 2011, the natural frequencies increased gradually (1.82–1.92, 1.84–1.94, 2.52–2.68 and 6.12–6.61 Hz for the first, second, third and fourth modes, respectively), which is due to the strength gain and corresponding stiffness gain of the concrete during the curing process over time. This last effect was also detected by [4].

Regarding the damping ratios, the identified values are mostly in the range 0.5–2.5% for all the identified modes when using ambient vibration data, which is consistent with previous studies on similar structures. However, as well known, the scatter of the damping ratio estimates is inherently larger than that of the identified natural frequencies. No clear trends can be observed between the variations of the modal damping ratios and the construction activities. Also, it can be seen that the agreement between the identified damping ratios obtained from different methods is better for the lower modes than for the higher modes.

Figure 19.3 shows the evolution of the mode shapes of the first four vibration modes identified from ambient vibration data when the slabs of the second, third, fourth and roof levels were poured. During the entire construction process, the first, second, third and fourth modes correspond to the first transversal + torsional (1-T+To), first longitudinal (1-L), first torsional (1-To) and second longitudinal (2-L) mode, respectively. It is observed that each of the identified modes keeps basically the same shape (or proportions) as the construction process evolves.

Fig. 19.3
figure 00193

Evolution of the mode shapes of the first four identified vibration modes identified using NExT-ERA from ambient vibration data over different stages of construction

Full size image

19.6 System Identification of Bare Structure

A first set of dynamic tests making use of the shake table was performed on August 12, 2011, with the building specimen at the bare condition. The low-amplitude banded (0.25–25 Hz) white noise base acceleration excitation consisted of two 10 minutes runs, with RMS acceleration of 1.0 and 1.5%g, respectively. Table 19.4 reports the identified natural frequencies and damping ratios for ten modes of the structure using the white noise base excitation test data. The natural frequencies identified using different methods are in very good agreement for the same level of excitation. The identified damping ratios exhibit a higher method-to-method variability, consistent with previous studies [3, 15].It is observed that based on ambient vibration data the identified first and second modes correspond to 1-T+To and 1-L respectively (see Table 19.5), while these two modes shift when identified based on white noise test data. This mode-crossing is produced by a decrease of the stiffness of the structure (due to concrete cracking) in the longitudinal direction (direction of motion) during the white noise base excitation test.

Table 19.4 Natural frequencies and damping ratios of the bare structure identified from white noise base excitation test data
Full size table
Table 19.5 Natural frequencies of the bare structure identified from ambient vibration test data
Full size table

It is noticed that for a number of modes OKID-ERA provides an estimate of the damping ratio significantly lower than the output-only methods, especially for the longitudinal modes of vibration (direction of excitation). The identified natural frequencies decrease as the amplitude of the white noise base excitation increases from RMS = 1.0%g to RMS = 1.5%g. Although these differences are relatively small (less than 3.5% in the average), they clearly show the effect of cracking in the concrete, and the resulting loss of stiffness during the white noise tests. This fact is confirmed when the identified natural frequencies obtained from ambient vibration data before and after the white noise tests are compared (Table 19.5). They are practically the same, and are higher than their counterparts identified during the white noise tests. Also, the first two modes cross again, and their order before the white noise tests is recovered. This means that the concrete was cracked during the white noise tests, however, the amplitude of the ambient vibrations is too small to re-open a significant portion of the cracks, which remain closed due to gravity effects.

Since the mode shapes identified with the methods used in this study are complex-valued, the realized modes were obtained using the method proposed in [16]. Figure 19.4 shows the identified mode shapes obtained from NExT-ERA based on the first white noise base excitation test data (RMS = 1.0%g).

Fig. 19.4
figure 00194

Mode shapes for the bare structure obtained using NExT-ERA for white noise RMS = 1.0%g

Full size image

In order to compare the mode shapes identified from the three different methods, Figure 19.5 shows the MAC values between them. The high MAC values (close to one) indicate that the identified mode shapes using different methods are in very good agreement. Red lines are shown for the modes that could not be identified by one of the methods.

Fig. 19.5
figure 00195

MAC values between modes of the bare structure identified by using different methods (white noise RMS = 1.5%g)

Full size image

19.7 Effects of Nonstructural Components

During the installation of all the NSCs, daily ambient vibration data were collected and additionally low amplitude white noise base excitation tests were performed when the building, including all the NCSs, was completed on February 23, 2012. Table 19.6 shows the main NCSs considered in the analysis. Figure 19.6 presents the evolution of the natural frequencies and damping ratios of the first three modes of the structure identified using SSI-DATA and NExT-ERA during the time window of NSCs installation (October 13, 2011, to February 22, 2012). The bottom panels show the Gantt chart for the activities defined in Table 19.6.

Table 19.6 Main NCSs installed in the building
Full size table
Fig. 19.6
figure 00196

Evolution of the natural frequencies and damping ratios of the first three modes during installation of main NCSs

Full size image

Again, there is a good agreement between the natural frequencies identified using SSI-DATA and NExT-ERA, while the identified damping ratios, ranging between 0.4% and 3.0%, have a much higher method-to-method variability. The increase of the modal frequencies of the system due to the increase of the lateral stiffness of the building produced by the installation of partition walls can be clearly observed. Due to the installation of partition walls in the elevator shaft (C1) and South East corner (C2 and C3) of the building (from October 13 to October 21), the first, second and third natural frequencies increased from 1.88 to 1.94, 1.90 to 1.96, and 2.63 to 2.67 Hz, respectively. Later, between December 5 and December 19, the natural frequencies increased from 1.91 to 2.10, 1.96 to 2.15 and 2.68 to 3.01 Hz for the first, second and third modes respectively, as a result of the installation of the interior partitions on levels one to three (C6). Within the same time window, the roof-mounted (cooling tower and AHU unit) equipment (C7 and C8) were installed, yet despite these increases in mass at the top of the building, the added stiffness due to the partition walls more than compensated, and the natural frequencies increased. Finally, from January 10 to February 10, 2012, the natural frequencies gradually increased due to the installation of partition walls on the fourth and fifth levels (C11). During this time window, the effect of the extra mass added by the elevator counterweight and rails (C10), pipes (C14) and equipment (C15) is dominated by the stiffening effect of the added partition walls. Furthermore, the partition walls also slightly affect the values of the identified equivalent viscous damping ratios. This effect can be seen in the right column of Fig. 19.6 between the dates abovementioned. Similar results have been reported in [4, 10]. Later, from December 19 to December 20, 2011, the first three modal frequencies dropped from 2.15 to 2.00, 2.18 to 2.06 and 3.01 to 2.63 Hz, respectively, due to the placement of the precast concrete panels (C9) on the South and West faces of the building at the fourth and fifth levels. From December 20 to December 22, the same frequencies dropped from 2.00 to 1.85, 2.06 to 1.91 and 2.63 to 2.55 Hz, respectively, when the precast concrete panels (C9) on the North and East facades were installed. Results also show that the installation, between November 2 and November 23, 2011, of the balloon framing (C4), spanning the bottom three stories of the building, moderately increased the natural frequencies from 1.88 to 1.93, 1.93 to 2.00 and 2.62 to 2.70 Hz for the first three modes, respectively. Finally, installation of ceilings (C12), penthouse (C5) and elevator cabin (C16) did not induce any clear effects on the identified modal properties of the building or their effects were negligible compared to those induced by the installation of other NCSs at the same time.

As a general trend during the entire period of installation of the NCSs, it can be noticed that the identified damping ratios tend to slightly increase in time, but again their estimation variability is large compared to that of the natural frequencies. Most of this additional damping is probably due to source of energy dissipation generated by the friction at the interface between the NCSs and the structure.

19.8 System Identification of the Complete Building

After all the NCSs were installed in the building (complete building), a second set of dynamic tests using the shake table were performed on February 23, 2012. The same white noise base excitation sequence described for the bare structure was used. This white noise sequence was repeated three times considering different positions of the elevator. The modal parameters were estimated from the recorded data using the output-only and input-output methods. Additionally, output-only methods were applied to ambient vibration data recorded before and after the white noise base excitation tests. Table 19.7 reports the natural frequencies and damping ratios for ten modes of the building identified using the white noise base excitation data for the configuration with the elevator counterweight located at the top of the building. It is important to note that the results obtained for the other two configurations of the elevator system do not differ significantly, with differences less than 3% for the natural frequencies and 15% for the damping ratios.

Table 19.7 Natural frequencies and damping ratios of the complete building identified from white noise base excitation test data
Full size table

Similar observations to those made for the bare structure apply to the complete building. First, it is noticed that the identified natural frequencies decrease as the amplitude of the excitation increases from RMS = 1.0%g to RMS = 1.5%g. The reductions are relatively higher than for the bare structure. This is due to both cracking in the concrete and interaction between the structure and the NCSs. Furthermore, the same mode shapes as those for the bare structure are identified: the first five longitudinal (1-L, 2-L, 3-L, 4-L and 5-L), the first two coupled transversal-torsional (1-T+To, 2-L+T) and the first three torsional (1-T, 2-T and 3-T) modes. The MAC values between corresponding modes of the bare and complete building are practically equal to unity, indicating that the NCSs do not have a significant effect on the vibration mode shapes for this level of excitation. Additionally, the MAC values between the corresponding mode shapes identified from the different white noise excitations (RMS = 1.0%g and 1.5%g) were computed as very close to unity, implying no changes in identified mode shapes with increasing amplitude of the base excitation.

19.9 Conclusions

The effects of the construction process and NCSs on the identified modal properties of a 5-story full scale RC building were investigated. Ambient vibration data were recorded daily during the construction of the specimen and during the installation of different NCSs. Additionally, white noise base excitation tests were performed for both the bare structure and the complete building. Modal parameters were identified from the recorded acceleration response using both output-only and input-output system identification methods.

It was observed that during the construction, the pouring of the structural elements induced abrupt changes in the natural frequencies due to the extra mass added to the system. Also, the natural frequencies gradually increased due to the strength and corresponding stiffness gain of the concrete, during the curing process over time. The mode shapes of the four lowest identified modes remain practically unchanged during the construction process.

The results show that the NCSs affect significantly the modal properties of the system. The natural frequencies increase due to the contribution of the partition walls to the lateral stiffness of the building, and these partition walls also slightly increase the damping ratios. The precast cladding adds significant mass to the building, inducing abrupt decreases in the natural frequencies.

From the modal properties identified using white noise base excitation data, it was observed that the natural frequencies decrease as the amplitude of the excitation increases, but the natural frequencies identified using ambient vibration data before and after the white noise tests are unchanged, suggesting that the cracks formed in the concrete during the white noise tests do not open during ambient vibrations, due to the very low amplitude of the building vibration and the gravity effects.