Anda di halaman 1dari 15

PERFORMANCE OF SEISMICALLY ISOLATED STRUCTURES IN THE JANUARY 17, 1994 NORTHRIDGE EARTHQUAKE Peter W.

Clark', Masahiko Higashino2, James M. Kelly' 1) University of California at Berkeley Earthquake Engineering Research Center 2) Takenaka Corporation, Japan INTRODUCTION Although seismic isolation is a relatively new technology with a limited number of completed applications in the United States, several seismically isolated structures in the Los Angeles area were affected by the MW 6.8 Northridge earthquake which struck January 17, 1994. Four steel-frame buildings within 40 km of the epicenter were strongly shaken, and strong motion recordings at three of these buildings which were occupied at the time of the earthquake indicated peak ground accelerations in excess of 0.20 g. (The fourth building was under construction at the time of the earthquake and did not yet have strong motion instruments installed.) While it is apparent that the isolation systems in these structures were activated to some extent, reconnaissance visits to the building sites immediately after the earthquake combined with analysis of available strong motion data has shown that not all of the buildings responded as intended. At three other isolated buildings 66 km and greater from the epicenter, the intensity of the ground motions was significantly lower, but strong motion recorders were still triggered at two of the sites. However, the recordings suggest that the isolation systems in these structures were not fully activated with the result that there was some amplification of accelerations over the height of the buildings. Table 1 summarizes all of the isolated buildings affected by the Northridge earthquake. Name Private Residences University of Southern California Teaching Hospital LA Emergency Operations Center (under construction)
LA Fire Command and Control Center

Epicentral Distance (km) 21 36 -38 38 66 90 -100

Isolation System GERB LRB HDR HDR LRB HDR LRB

Lateral System Steel braced frame Steel braced frame Steel braced frame Steel braced frame Concrete moment frame Steel braced frame Steel braced frame

(g)

PGA

Roof Accn. (g) 0.63 0.21 NA 0.32 0.15 0.10 NA

0.44 0.37 NA 0.22 0.08 0.05 NA

Rockwell Computer Center - Seal Beach Foothill Communities Law and Justice Center Kaiser Computer Center Corona (not instrumented)

Table 1. Seismically Isolated Buildings Affected by the Northridge Earthquake

The Northridge earthquake has provided the most significant test of modern isolated buildings to date, and a valuable data set of the response of various types of structures incorporating various types of isolation systems has been obtained. This paper summarizes the response of the three most strongly shaken instrumented buildings including physical evidence gathered during a reconnaissance of the sites. Implications for the design, construction, and maintenance of isolated buildings are also outlined. UNIVERSITY OF SOUTHERN CALIFORNIA TEACHING HOSPITAL The isolated building most strongly shaken in terms of peak ground acceleration during the Northridge earthquake was the University of Southern California Teaching Hospital in eastern Los Angeles, located approximately 23 miles (36 km) from the epicenter. The USC Hospital was completed in 1991 and is an 8-story concentrically-braced steel frame supported on 68 lead-rubber isolators and 81 elastomeric isolators. Design and construction aspects of the building are provided in [1], [2], and [3]. The foundation system consists of spread footings and grade beams on rock. Because of functional requirements, both the building plan and elevation are highly irregular with numerous setbacks over the height (Figure 1). Two wings at either side of the building are connected through what is referred to as the "necked-down" portion of the building, and in the original fixed-base design the irregular configuration led to both coupling between the lateral and torsional vibration modes and very large shear force demands in the slender region between the two wings. (Even in the isolated design steel trusses are required to carry the shears in the necked-down region.) These were two of the main reasons that seismic isolation was eventually chosen for this structure. The design base shear of the hospital is 0.15 g and the design comer bearing displacement including torsion is 10.25 inches (26.0 cm). The effective period of the isolation system at this displacement is approximately 2.3 seconds. To minimize torsional response in the isolated structure, the lead-rubber bearings are distributed on the perimeter of the building under the braced frames while the elastomeric bearings are used in the interior. To minimize local vertical uplift at the bases of the braced frames, the base level diaphragm directly above the isolators is a 10-inch reinforced concrete slab with integral deep beams along the frame lines. The fundamental period of the fixed-base superstructure is approximately 0.7 seconds. The specific details of the structural system, including the isolator dimensions and force-displacement relationships, are not available at this time. The USC Hospital was instrumented by the California Strong Motion Instrumentation Program (CSMIP) soon after its, completion, and digitized acceleration, velocity, and displacement recordings from the Northridge earthquake have recently been made available [4]. The instrumentation layout and overall plan and elevation views of the building are given in Figure 1. An interesting aspect of both the free-field and foundation acceleration time histories recorded at this site and other sites south of the epicenter is the presence of two distinct S-wave arrivals. The first P-wave arrives at approximately 5 seconds into the record, and the first Swave is seen just over 10 seconds into the record. However, a second, stronger S wave arrives at approximately 15.5 seconds. These two arrival times are consistent with reports that there were two fault ruptures associated with the Northridge event [5]. The strongest motions recorded at the site were in the north-south direction. In the free-field a peak acceleration of 0.493 g and a peak velocity of 12.2 in/sec (31.1 cm/sec) were measured, while at the center of the foundation the peak acceleration was 0.366 g and the peak velocity was 10.2 in/sec (25.9 cm/sec). The pseudo-acceleration response spectrum of the north-south foundation motion in Figure 2 shows significant shortperiod energy in the range of 0.2 to 0.6 seconds but relatively little energy above a period of approximately 1.3 seconds. At periods greater than 2 seconds, the corresponding 10-percent damped spectral displacement is less than 1.2 inches (3 cm). In the east-west direction the peak acceleration is only 0.163 g, and the peak velocity is 3.5 in/sec (9.0 cm/sec). However, the corresponding 10-percent damped

spectral displacement at a period of 4 seconds is approximately 2.8 inches (7 cm). This is due to a longperiod displacement component in the east-west motion that starts about 19 seconds into the record and that is absent from the north-south motion (Figure 3). Although this long-period motion is similar to what might be seen in an alluvial basin, the soil underlying the USC Hospital has been described as rock [1]. Additional information about the site geology and analysis of other motions recorded in the area are required to fully understand this effect. It should also be noted that a similar response was seen in the east-west free-field displacement, and although there is a slight reduction in the high frequency content of the foundation northsouth acceleration response spectrum compared to that in the free-field, in general there are no significant differences between any components of the free-field and foundation motions Figures 4 and 5 show the relative displacement time histories of the isolation system in the northsouth and east-west direction, respectively. The relative displacement is calculated by subtracting the absolute displacement of the instrument at the foundation from that of the instrument just above the isolators at the base level. Both components exhibit high-frequency response in the time interval between the arrival of the first Swave at 10 seconds and that of the stronger, impulsive S-wave at approximately 16 seconds. The initial S-wave arrival had a peak ground acceleration of approximately 0.2 g in the north-south direction, and even though the frequency of the response appears to be relatively high - roughly 2.0 Hz - this input is effectively filtered out by the isolation system as structure accelerations up to the 6th level are below 0.08 g. The two components of the displacement response after 16 seconds are significantly different. The peak north-south displacement is 1.38 inches (3.5 cm) to the south, and apparently consists of a single spike at a frequency considerably higher than that of the subsequent cycles. However, this is likely a result of energy from a higher mode superposed onto the primary response at a period of approximately 1.4 seconds. After this spike there is no clear interval of freevibration decay as the amplitude of the five cycles between 19 and 26 seconds is approximately constant. It is interesting that the peak reversed displacement (to the north) is 0.71 inch (1.8 cm) and does not occur within the same cycle as the peak to the south. In contrast, the east-west response from 16 to 22 seconds appears more typical of what would be expected of a highlydamped system in free-vibration response to an impulsive load. The peak displacements both occur within the same cycle, and the vibration period of the system during the decay is approximately 1.5 seconds. The north-south pseudo-acceleration response spectra at the first floor and at the roof are shown in Figures 6 and 7, respectively. The peaks in the spectra at approximately 1.3 seconds reflect the fundamental period of the response of the isolation system after 15 seconds into the record. Somewhat more energy is contained in the response at the roof than at the first level, which indicates that the fundamental mode shape involves some superstructure deformation. Superstructure drift time histories between the roof and the first level also show motion at approximately 1.3 seconds, but the maximum drift in the north-south direction is only 0.83 inch (2.1 cm), and approximately 30% of this corresponds to drift in the more flexible top two floors. This localized flexibility also contributes to the relatively large peak acceleration recorded at the roof (0.205 g) as compared to those in the first through sixth floors of the structure (less than 0.130 g). It appears that this peak is due to higher-mode response as evidenced by the substantial energy in the roof spectra at approximately 0.6 seconds. The pseudo-acceleration spectra from the instrument at the sixth floor contain substantially less energy at this period. Finally, it should be noted that the peak in the first floor spectra at approximately 0.3 seconds likely represents the higher-frequency response of the building in the initial portion of the record before the isolation system was fully activated. The transfer function of first level N-S acceleration to foundation N-S acceleration is shown in Figure 8. In this figure, the isolation frequency is approximately 0.7 Hz. Also, the height of the first peak is almost 3.5, ignoring sharp peaks and dips. This translates into a damping ratio of approximately 14%. The first peak in the E-W direction was at 0.75 Hz and has a height of 4, from which the damping ratio of 13% was derived. These results indicate that the system exhibited desired isolation frequency as well as good damping capacity.

It is apparent that a great deal can be learned about the response of base-isolated buildings from relatively straightforward analyses of the records obtained at the USC Hospital. In this earthquake, the structure was effectively isolated from ground motions strong enough to cause significant damage to other buildings in the medical center. However, more detailed study is recommended, which should include evaluation of the site conditions, validation of structure and isolation system models, and extension to more severe ground motions. LOS ANGELES COUNTY FIRE COMMAND AND CONTROL FACILITY The Los Angeles County Fire Command and Control Facility (FCCF) is a 2-story braced steel frame supported on 32 high-damping rubber isolators and serves as the headquarters from which fire equipment is dispatched throughout Los Angeles county. It is located east of downtown, approximately 39 km from the epicenter of the Northridge earthquake, and was completed in 1990. The total mass of the building is 4230 kips (18.89 kN), and the isolation system is designed for a peak comer displacement of 9.6 inches (24.4 cm). One unique feature of the bearings on the perimeter of the building is a chain installed in the center of the bearings intended to serve as a displacement restraint in the event of a beyond-design level earthquake. The chains are initially slack but are designed to be activated at a displacement of 12.5 inches (31.8 cm). Details of the analysis, design, and construction of the FCCF are given in [6], [7], and [8]. Analog acceleration time histories of the response of the FCCF were made available by CSMIP within several weeks of the Northridge earthquake (9]. Figure 9 shows the layout of the structure and instrumentation and Figure 10 shows the recorded acceleration time histories. Digitized data are not yet available for this building. The recorded response of the FCCF in this event was unusual for a base-isolated structure because several high-frequency spikes were apparent in the east-west acceleration records. Although the peak foundation accelerations in this direction were between 0.19 and 0.22 g, the first floor accelerations were between 0.21 and 0.35 g, and the roof accelerations were between 0.24 and 0.32 g. The corresponding amplification ratios are therefore substantially greater than 1.0. In the north-south direction the building performed as expected, with amplification ratios of approximately 0.4 and 0.5 at the 1st floor and the roof, respectively. The peak foundation acceleration in this direction was approximately 0.18 g. The FCCF remained fully functional during and after the earthquake. Although the FCCF remained fully functional during and after the earthquake, the presence of highfrequency spikes in the accelerograms was cause for concern. An inspection of the site four days after the mail shock revealed that architectural details at an east-facing tile entryway near the north wall of the building may have compromised the isolation gap in the east-west direction. The tiles are not part of the isolated portion of the building and were designed as sacrificial elements that would be dislodged by the steel grillwork that overhangs the isolation gap from the main structure. Figure 11 shows a view of the tile entryway before the earthquake, and Figure 13 shows the intended design schematic. This joint as originally designed previously had been damaged under moderate shaking in both the June 28, 1991 Sierra Madre earthquake and the June 28, 1992 Landers earthquake. The superstructure acceleration time histories recorded during these two events indicated some high-frequency response, perhaps associated with pounding at this joint, but the relatively low levels of shaking in these events make the evidence inconclusive. However, after the Landers earthquake, it appears that an outside contractor strengthened the tile detail to resist damage in future earthquake, as shown in Figure 14, with the result that the newly installed tiles provided more lateral restraint than expected to the overhanging grill, imparting an impulsive force at the first floor of the structure as the grill pounded on the tiles. Figure 12 shows the damage to the tiles of the entryway four days after the earthquake.

An alternate explanation for the high-frequency spikes is that the chains in some of the perimeter bearings were prematurely activated. However, the mechanism of pounding at the north entryway is consistent with the observation that the high-frequency acceleration spikes are larger at the north side of the building near the entryway than at the south wall, and the fact that the spikes indicate amplified accelerations only toward the west. No high-frequency response is seen in the north-south acceleration time histories. One other interesting observation that can be made from the analog acceleration time histories is the lengthening of the period of the structure as the input increases. The foundation acceleration recorded at the FCCF is similar to that recorded at the USC Hospital in that there are two distinct S-wave arrivals with the second being of greater intensity than the first. The east-west response of the south wall at the second floor does not show significant high-frequency components, and well-defined peaks in the response allow the period of vibration to be evaluated directly. If the four cycles of response prior to the arrival of the second S-wave at 16 seconds into the record are evaluated, the average response period is approximately 0.9 seconds and the peak-to-peak acceleration (double-amplitude) is approximately 0.15 g. This translate into a displacement of Sa 2 T
2

(0.075) (386) 2 = 0.59 inch 0.9

A similar analysis of the two cycles of vibration after 16 seconds give a period of approximately 1.3 seconds and a double-amplitude acceleration of about 0.2 g, which translates into a displacement of about 1.8 inches (4.4 cm). The total height of rubber in the FCCF bearings is 10 inches (2.5 cm), so that the peak shear strain before 16 seconds was approximately 6%, while the peak shear strain after this time was approximately 18%. The high-damping elastomer compound used in the FCCF bearings has a similar modulus-shear strain relationship to the 243-62 compound used for the Foothill Communities Law and Justice Center [10]. This compound has a shear modulus of approximately 700 psi at 6% strain that drops to approximately 350 psi at 18% strain. The stiffness ratio is thus 0.5, and the squared frequency ratio observed during the two phases of the response of the FCCF is (0.9 ) 2 = 0.48 (1.3 ) Thus, a simple analysis of the analog data can lead to a useful confirmation of the observed period shift based on the material characteristics of the rubber compound. One final set of observations taken during the site reconnaissance was the relative displacements at several points around the perimeter of the structure. Estimates of the peak displacements across the isolation interface were made based on deposits of scraped concrete and dirt particles surrounding grills at simple entryways (without the tile overlay) of the building. It appears that the center of mass of the building moved approximately 22 mm south, 18 mm north, 24 mm east, and 16 mm west. Because of the pounding at the-north entryway tiles, the south wall of the building moved more in the east-west direction than did the north wall. However, it should be noted that these observations were taken four days after the mainshock and contradict to some extent the displacements calculated in the analysis of the analog data above. The main point to be made here is that in the absence of digital data, it can be very difficult to evaluate one of the most significant parameters of the response of a base-isolated building, that is, the peak bearing displacement. Even if displacement information from doubly-integrated acceleration time histories is avail-

able, even a simple displacement transducer would prove valuable in providing a check on the peak calculated quantities. GERB RESIDENCES The final isolated structures evaluated were two identical 3-story braced steel frame residences in Santa Monica, each supported at its comers by GERB helical springs and viscous dashpots [11]. Additional springs are distributed around the building perimeters. Figure 15 shows an overall view of the residences. The site is 22 km from the epicenter, and one of the structures is instrumented by the United States Geological Survey (USGS). Figure 17 shows the recorded acceleration time histories with peaks at the foundation of 0.44 g and 0.16 g, respectively [ 12]. Accelerations above the isolation system were generally larger with increased highfrequency content; at the roof the peak horizontal acceleration was 0.63 g while the peak vertical acceleration was 0.46 g. It appears from both the records and from a survey of the buildings performed after the earthquake that the isolation system is not as effective as it could be because several architectural and construction details limit movement of the superstructure. For example, slight damage was observed at locations where steel girders from the isolated portion of the structure framed into both a concrete footing and a masonry block wall attached to the ground Figure 16). Also, a series of square glass blocks distributed around the perimeter of the structures at the foundation level may have contributed to the large amplifications of the vertical motion. The owner indicated that these blocks were adjusted so that they would be damaged at increments of vertical displacement, and the rows of blocks adjusted to 1/2 inch and 3/4 inch were broken while the row of blocks adjusted to 1" remained intact. However, pounding between these blocks and the foundation of the structure may have led to the high-frequency vertical acceleration spikes. Unfortunately, the poor quality of the accelerograms combined with the peculiar details of the building makes a thorough interpretation of the response of this building difficult. It is hoped that a higher-quality acceleration trace will eventually be made available. CONCLUSIONS In summary, a set of very interesting records have been obtained from a variety of base-isolated structures as a result of the Northridge earthquake. The records from the USC hospital are particularly encouraging in that they represent the most severe test of an isolated building structure to date. These records deserve significant study to refine modeling techniques and extrapolate the response of the building to more severe input intensities. The results from the FCCF and GERB structures. are also worthy of further study, and illustrate the importance of careful maintenance of the seismic gap around isolated structures. Unfortunately, neither of these buildings performed as designed, but it is hoped that the lessons learned from this earthquake can be used to improve their performance in future events. It is suggested that all instrumented isolation buildings be equipped with simple displacement transducers which could measure, at a minimum, peak isolator displacements in an earthquake. This would provide a direct measure of this important response parameter as well as a check on relative displacements obtained from doubly-integrated acceleration records. ACKNOWLEDGEMENTS The efforts of the Northridge earthquake reconnaissance team from the University of California at Berkeley Earthquake Engineering Research Center were funded in part by a grant from the National Science Foundation. The authors express their appreciation to Mr. Robert Bachman of Fluor Daniel, Inc. for his information regarding the Los Angeles County FCCR The assistance of Robert Darragh at CSMIP is also very much appreciated.

REFERENCES [1] Asher, J.W, and VanVolkinburg, D.R., "Seismic Isolation of the USC University Hospital," Proceedings of Structures Congress '89, San Francisco, CA, May 2-5, 1989. [2] Asher, J.W, et. al., "Seismic Isolation Design of the USC University Hospital," Proceedings of the Fourth U.S. National Conference on Earthquake Engineering, Palm Springs, CA, May 20-24,1990. [3] KPFF Consulting Engineers, "USC University Hospital, Project Summary", Santa Monica, CA, 1991. [4] "Processed Data for Los Angeles 7-Story University Hospital from the Northridge Earthquake of 17 January 1994", Report No. OSMS-94-11E, Office of Strong Motion Studies, California Division of Mines and Geology, August 1994. [5] Hall, John F. ed., "Preliminary Reconnaissance Report of the January 17, 1994 Northridge Earthquake," Earthquake Engineering Research Institute, Oakland, CA, May 1994. [6] Bachman, R.E., Gomez, M.J., and Chang, K.C., "Verification Analysis of the Base-Isolated Los Angeles Fire Command and Control Facility," Proceedings of the Fourth U.S. National Conference on Earthquake Engineering, Palm Springs, CA, May 20-24, 1990. [7] Anderson, TL., "Seismic Isolation Design and Construction Practice," Proceedings of the Fourth U.S. National Conference on Earthquake Engineering, Palm Springs, CA, May 20-24, 1990. [8] Anderson, TL., Bachman, R.E., and Grant, R.R., "Base Isolation Response to Extreme Ground Motions," Proceedings of the Tenth World Conference on Earthquake Engineering, Madrid, Spain, July 19-24, 1992. [9] Shakal, A., Huang, M., et al., "Strong Ground Motions from the Northridge, California Earthquake of

January 17, 1994", CSMIP Report OSMS 94-07, Department of Conservation, Division of Mines and Geology, State of California, February 1994. [10] Tarics, A., Way, D., and Kelly, J.M., "The Implementation of Base Isolation for the Foothill Communities Law and Justice Center, Rept. No. RTA-84, Reid and Tarics Assoc., San Francisco, CA., 1984. [11] Hueffman, G.K., "Base Isolation of Two Residential Buildings in Los Angeles with Helical Springs and Viscodampers," Proceedings of the 11th SMiRT Post-Conference Seminar on Seismic Isolation of Nuclear and Non-Nuclear Structures," Nara, Japan, August 26-27, 1991 [12] Porcella, R.L., Etheredge, E.C., et al., "Accelerograms Record at USGS National Strong-Motion Network Stations During the Ms=6.6 Northridge, California Earthquake of January 17, 1994", Open File Report 94141, U.S. Geological Survey, February 1994.

Anda mungkin juga menyukai