پاسخ بلند مدت ساختمان به زلزله ها در منطقه خلیج سانفرانسیسکو Long-Period Building Response to Earthquakes in the San Francisco Bay Area

The northern San Andreas fault produced the devastating 1906 San Francisco earthquake. The same fault may produce a similar earthquake in the future, and the consequences of a similar earthquake in a modern, urban area are uncertain. The city of San Francisco and surrounding communities are significantly different than they were 100 years ago. Specifically, urban areas now include long-period buildings that have been built only in the last several decades. To better understand the possible performance of these long-period buildings in San Francisco’s next great earthquake, we study the response of two examples of long-period buildings: 20- story, steel, welded moment-resisting frame (MRF) and baseisolated buildings. Experience provides few examples of steel-frame responses in great (magnitude >7:5) earthquakes. Contemporary and modern reconnaissance reports of the 1906 San Francisco earthquake conclude that the steel frames existing in 1906 performed well in the severe ground motions (Soulé, ۱۹۰۷; Tobriner, 2006). However, the steel frames in 1906 were markedly different than modern frames. The tallest building in 1906 San Francisco was the 18-story braced-frame Call Building, and Soulé (۱۹۰۷) notes only nine steel-frame buildings between nine and 12 stories; modern steel-frame buildings are often much taller. Also, modern structural engineers can choose longer-period, MRFs rather than the shorter-period, braced frames, which were state-ofthe-art 100 yr ago. (See Hamburger and Nazir (2003) for a brief discussion of historic and modern steel frames.) Because modern steel frames may be taller and have modern designs, it is difficult to infer the performance of modern steel frames based on the reported response of 1906 buildings. Whereas one great earthquake tested steel frames designed a century ago, there are several examples of modern steel-frame responses in smaller earthquakes (Hamburger and Nazir 2003). In particular, tall steel-frame buildings showed either repairable or no damage following both the 1994 magnitude 6.7 Northridge and 1995 magnitude 6.9 Kobe earthquakes. However, in the Northridge earthquake, the largest recorded ground displacement near a building was 0.31 m near the Olive View Hospital. Somerville et al. (1995) provide recorded and simulated ground motions at seven building sites; none of these ground motions exceeds 0.3 m. Dynamic ground displacements beneath tall steel frames were less than 0.5 m in the Kobe earthquake (Building Research Institute, 1996). Furthermore, investigations of modern, steel MRF buildings after the Northridge earthquake demonstrated that many welds in existing moment-resisting joints are brittle (Gilani, 1997). The SAC Steel Project reports document this widespread problem (e.g., Krawinkler, 2000; Roeder, 2000). In this study, we explore the response of MRF buildings with either ductile or brittle welds to strong ground motions with large displacements. To predict the response of MRF buildings in large and great earthquakes, many research engineers employ numerical models. Luco and Cornell (2000) studied the effect of beam-to-column connection failure on the seismic response of 3-, 9-, and 20-story building models developed by the SAC Steel Project. The authors found that some of their largest ground motions induced interstory drifts in excess of 10% or caused collapse in the 20-story building models with the brittle connections they studied. However, the authors did not include internal gravity frames or shear connections in this study, but they found in a separate study that including these features typically reduced large drifts like these responses. Gupta and Krawinkler (2000) used the same SAC building models with no deterioration mechanisms to predict the seismic response at several seismic hazard levels. In some of the ground motions that represent the hazard level of 2% exceedance in 50 yr, the authors noted large drift demands in the 20-story building model designed for the Los Angeles area. They conclude that “the potential for unacceptable performance is not negligible.” Lee and Foutch (2006) designed several alternatives to the SAC building models with various strengths. From a nonlinear time history analysis, the authors found that some 20-story models exceeded their interstory drift capacities, including models with higher strengths. Krishnan et al. (2006) studied the response of a common type of building in the Los Angeles area to simulated ground motions from hypothetical magnitude 7.9 ruptures on the southern San Andreas fault. They used a fully three-dimensional building model to compare the responses of 18-story steel MRF buildings designed to the 1982 and 1997 Uniform Building Codes (UBCs). In the simulations, the MRF models showed large drifts, which would threaten life safety in many areas of Los Angeles. The present study augments previous work by applying thousands of simulated, large ground motions to different 20-story building models and by evaluating the building model responses on a regional level near San Francisco. Base-isloated buildings are a relatively new type of long-period structure, characterized by a purposely built flexible zone in the foundation that supports a superstructure. By design, the natural frequencies of the superstructure are high compared to the effective frequency of the isolation system. Base isolation can significantly reduce highfrequency vibrations of a building because the isolation level does not transfer high-frequency motions to the superstructure. The seismic forces in a building’s superstructure are significantly smaller for a base-isolated building compared to an identical building without isolators. However, all isolation systems have a limited range of motion. Depending on the individual system, base-isolated buildings may experience impacts between foundation walls and the superstructure (Heaton et al., 1995). Base-isolation systems performed well during the 1994 Northridge and 1995 Kobe earthquakes (Kelly, 2004) although no base-isolated buildings were in the near-source areas for these earthquakes. Furthermore, these earthquakes produced much smaller ground displacements than those produced in the 1906 magnitude 7.8 San Francisco earthquake. One goal of this study is to estimate isolator displacements that might occur in a large San Andreas fault event like the 1906 San Francisco earthquake. Detailed models of isolation systems would provide the best predictions of isolator behavior, but we do not presently have such models. Instead, we use an equivalentlinear approximation of the isolator system to estimate the isolator displacements in our considered earthquakes. The 2006 international building code (IBC) requires a dynamic analysis to design base-isolation systems in the San Francisco Bay Area. The designer must perform response spectrum and response history analyses to determine the design and maximum displacements of the isolators, among other design parameters (International Code Council, 2006). The code also specifies that these displacements must not fall below minimum values. Thus, the model response to ground motions may control the design, and the choice of motions may affect the design. If a design engineer uses ground motions larger than those required by the building code, then he or she will call for a system with a larger isolator displacement capacity than an engineer who uses smaller motions. There has been active discussion on the use of near-source ground motions—characterized by a large displacement pulse—for design (e.g., Hall, 1999; Kelly, 1999). Jangid and Kelly (2001) stated that base-isolation systems should be designed primarily to minimize damage to contents (measured by superstructure acceleration) in moderate earthquakes and secondarily to minimize isolator displacement in large pulse-type ground motions. The authors showed the existence of an optimum isolator damping that minimizes superstructure accelerations. This optimum damping does not minimize isolator displacement because isolator displacement monotonically decreases with increasing damping. Ryan and Chopra (2004b) compared the results of a nonlinear analysis of base-isolation systems to the equivalent-linear procedure of the 2000 IBC used to determine the design displacement. The authors found that the isolator displacements from the equivalent-linear procedure underestimated those from the nonlinear analysis by 20%–۵۰% on average. Thus, the minimum design displacements required by the code were not conservative for the strong ground motions, consistent with moderate earthquakes, used in their study. Since the ground motions in future earthquakes are uncertain, structural engineers cannot design a building for the specific ground motions that it will experience in its lifetime. Instead, structural engineers rely on building codes to define the types of ground motions that buildings must safely survive. The design response spectrum represents ground motions from events that are deemed likely to excite buildings in their assumed lifetime. From the design spectrum, structural engineers determine the minimum design forces that buildings must withstand. The building code also acknowledges that unusually large earthquakes occur in or near urban areas. The 2006 IBC describes the maximum considered earthquake (MCE) as “the most severe earthquake effects considered in this code” and defines the MCE as 1.5 times the design response spectrum (International Code Council, 2006). In an earthquake consistent with the MCE, structural engineers acknowledge that buildings will sustain significant damage to structural systems and building contents. However, the buildings may collapse—either partially or totally—only in ground motions that exceed the MCE (Hamburger, 2003). The design of MRFs like the ones in this study is consistent with this philosophy. The design of baseisolated buildings in the San Francisco Bay Area requires the use of ground motions consistent with the MCE, not the standard design spectrum (Structural Engneering Institute [SEI], 2006). In this way, structural engineers design most buildings for likely earthquakes while acknowledging that a large, unusual event will test the limits of the lateral force resisting systems. The purpose of the study described in this article is to predict the response of some long-period buildings to scenario and hypothetical earthquakes in the San Francisco Bay Area. We apply ground motions from simulations of the 1989 Loma Prieta and large, northern San Andreas fault events (including the 1906 San Francisco earthquake) to steel MRF and base-isolated building models. We evaluate the effect on overall building performance of rupture direction, MRF building strength and weld state, and base-isolation system period and damping. We use response spectra to compare the spectral accelerations predicted in the magnitude 7.8 earthquakes to the 1994 UBC, 1997 UBC, and 2006 IBC design response spectra. This study continues the work of Hall and Challa (1995), Heaton et al. (1995), and Hall (1998) by considering long-period building response to recent simulations of ground motions from great earthquakes on a large geographical region. Our analysis is limited in several ways. We only consider a few realizations of possible future great earthquakes. The next great northern San Andreas earthquake will almost certainly be different than our scenario and hypothesized events. Nonetheless, these events seem plausible because they are compatible with the current understanding of the 1906 San Francisco earthquake. Our study is also limited because we only consider several idealized building models. The buildings we consider are likely not the most vulnerable buildings in our considered earthquakes. We choose these buildings because their analysis is presently accessible to us. Furthermore, they are examples of very flexible structures. Increasing a building’s flexibility can help to limit the stress in a building subjected to high-frequency ground motions, which may result from moderately sized earthquakes. However, extremely flexible structures may experience very large deformations in the large long-period ground motions produced by unusual great earthquakes.