Masayoshi Nakashima August 10,

Transcription

1 Masayoshi Nakashima Professor Disaster Prevention Research Institute, Kyoto University Gokasho, Uji, Kyoto JAPAN Director Director, Hyogo Earthquake Engineering Research Center (E-Defense), National Research Institute for Earth Science and Disaster Mitigation (NIED) , Nishikameya, Mitsuta, Shinjimicho, Miki, Hyogo, JAPAN Steel structures, Experimental techniques, Base-isolation My personal acquaintance with Luis began in 1987, when I visited Mexico City for the first time. I was a member of a Japan International Cooperation Agency (JICA) team, in Mexico to examine the possibilities of Mexican-Japanese collaboration on earthquake engineering research. One day, the team visited Luis office, and I shook hands with him. I am sure that he does not remember that occasion, but I recall vividly the striking modesty and kindness of this highly respected researcher in the earthquake engineering community. More than fifteen years later we met again, this time in Kobe, Japan. He was there as a distinguished guest at the inauguration ceremony of E-Defense, the world largest shaking table, on the outskirts of Kobe. It was a two-day ceremony, featuring an inauguration party held on January 15, 2005 and the inauguration s international symposium held on January 16, He was a keynote speaker at the symposium, delivering a very solid message about the past accomplishments, present problems, and future challenges of earthquake engineering and expressed great expectations for the contribution of E-Defense to earthquake disaster mitigation worldwide. We are very thankful to Luis for his strong support of E-Defense. On behalf of Luis s many Japanese friends, I express my sincere appreciation of his long-time leadership in the advancement of earthquake engineering, and wish him continuing health and leadership. Masayoshi Nakashima August 10,

2 TEST ON COLLAPSE BEHAVIOR OF STRUCTURAL SYSTEMS ABSTRACT Damage observed in the 1995 Hyogoken-Nanbu (Kobe) earthquake highlighted the importance of accumulating real data by experimentation regarding the earthquake response, damage, and collapse of structures. Full-scale tests to complete collapse in real time are indispensable. Four series of such tests conducted by the writer are presented, including complete failure tests applied to steel beams, columns, and connections, and full-scale tests applied to a steel moment frame. A newly built large shaking table owned by E-Defense is introduced, and its mission, strength, and status are outlined. Ongoing research projects including the NEES/E-Defense program are described. Lessons Learned from 1995 Hyogoken-Nanbu (Kobe) Earthquake The 1995 Hyogoken-Nanbu (Kobe) earthquake caused devastating damage to buildings and infrastructure in Kobe and its vicinity [Architectural 1995, Kinki 1995, Nakashima et al 1998a, Nakashima 2001]. The earthquake taught us lessons aboutstructural, economical, societal, cultural, and human factors. Since the earthquake, much research and development has been implemented for the mitigation of earthquake disasters. The 1995 Kobe earthquake, however, was not the sole motivation. Japan is destined to suffer from large earthquakes on a regular basis. Figure 1 shows a map of Japan, and the bold line indicates an ocean ridge called the Nankai trough, running deep along the Pacific Coast of Japan. The trough is divided into three regions, Tokai, Tonankai, and Nankai, reading from east to west. Slips and ruptures have occurred periodically in these regions. Figure 1. Nankai Trough and periodical large earthquakes. Table 1 shows the historical earthquakes related to the slips and ruptures of the three regions [Council 2004]. In some earthquakes, one or two of the regions ruptured; all three - 2 -

3 regions ruptured simultaneously in other cases. The frequency of occurrence was between one hundred and one hundred and fifty years. Observing the pattern of these previous earthquakes, Japan is to be hit by large earthquakes in the middle of the twenty-first century. Contemporary science is not able to predict the date and time of the next rupture precisely, but the public throughout Japan is fully aware that within several decadesjapan will be hit by a very large earthquake. Table 1. Historical records of Nankai, Tonankai, and Nankai ruptures. Year Earthquake Toka Tonankai Nankai 1605 Keicho 1701 Hoei 1854 Ansei 1944 Tonankai 1946 Nankai 20xx NEXT??? In 2005, the Council of National Disaster Mitigation, chaired by the Prime Minister of Japan, disclosed an estimate of the damage that Japan will sustain if the Nankai trough were ruptured again [Council 2004]. Table 2 shows the damage statistics in terms of the number of houses and buildings collapsed, the death toll, and direct capital loss for various combinations of ruptures. Should the three regions rupture all together together, about forty million people, equivalent to one-third of the entire population of Japan, would be affected, about one million houses and buildings would collapse, about twenty-five thousand people might lose their lives, and the economic loss might amount to close to one trillion US dollars. As Tables 1 and 2 clearly indicate, the earthquake disaster was, is, and will remain the most critical national problem in Japan. Table 2. Damage estimates for possible Nankai, Tonankai, and Nankai ruptures. Tokai Tonankai Tokai 1995 & Nankai & Tonankai Kobe & Nankai Collapse (x 1000) Death 9,200 17,800 24,700 6,400 Loss (billion $) Needs of Structural Test Regarding the 1995 Kobe earthquake, the writer is convinced that the following two lessons are most notable in the structural aspect. (1) Cities and towns throughout Japan have large stocks of old buildings and infrastructural systems whose seismic capacity is insufficient. To prepare for future large earthquakes, it is crucial to accurately evaluate their existing seismic capacities and then to retrofit and rehabilitate them accordingly. (2) Much larger shaking than that contemplated in current seismic design is known to be possible. Evaluation of the reserve seismic capacity of existing buildings and infrastructural systems, development of design and construction technologies to enhance the seismic capacity, and implementation of these technologies for real design and - 3 -

4 construction are critical. As evidence of (1), Fig. 2 shows a photo taken in downtown Kobe immediately after the 1995 Kobe earthquake. Two RC buildings, standing side by side, disclosed a clear contrast in damage; the one on the right side lost the third story completely, while the one on the left side looked nearly intact from the exterior. The ages of the structures were significantly different. The severely damaged building was nearly forty years old and had been constructed according to obsolete design and construction practices, whereas the undamaged building was relatively new. This distinctive contrast demonstratesthat earthquake-resisting capacity can differ significantly among structures. Figure 2. A contrast of damage observed in 1995 Kobe earthquake. As evidence of (2), Fig. 3 shows the pseudo-acceleration spectra of eleven strong motions recorded in the 1995 Kobe earthquake [6], together with the design spectrum stipulated for large earthquakes in the current Japanese seismic design code. It is evident that quite a few records possess significantly larger pseudo-accelerations than the code acceleration. Ground motions that would exceed those considered in the seismic code were also obtained in other recent earthquakes, e.g. the 1999 Tottoriken-Seibu, 2000 Geiyo, and 2004 Chuetsu earthquakes. 25 S a (m/s/s) Code spectrum Figure Period (s) Pseudo accelerations of strong motions recorded in 1995 Kobe earthquake. In light of lessons (1) and (2) cited above, it is crucial to identify the state of complete - 4 -

5 collapse in which the structure no longer can sustain gravity and as a result will kill people in the structure. This need is relevant to the characterization of the collapse margin, defined as the reserve capacity that the structure possesses for loads greater than that specified in seismic code up to the collapse. The collapse margin is difficult to characterize because of the scarcity of real data. Severe earthquake ground motions that would cause structure collapses occur very rarely, which makes it difficult to monitor or measure the real behavior of structures subjected to such events. The interactions between members and system behavior are known to be complex; hence tests on a structural system that involve force redistribution due to member yielding and plastification are indispensable. Building structures, however, are massive, and it is difficult to fabricate and load them in the laboratory, while miniature models are known to fail to duplicate real building behavior because of lack of similitude. Advances in numerical analysis methods, particularly those using the finite element method, are notable, but the analyses insufficiently duplicate the behavior of structures to collapse, which involves significant material nonlinearity, strength and stiffness degradations, and topology changes such as fracture, separation, and detachment. For the past ten years, much research has been conducted in the name of performance-based seismic design on the development of innovative systems by which to enhance the functionality, operability, and safety of structures. Base-isolation and passive dampers are typical examples; indeed, numerous inventions have been proposed toward this end. Ll such research and development must, however, be checked for expected actual performance before being transferred with confidence to real design and construction practices. Here, experimentation again plays a very important role to provide real data for performance checking. The accumulation of such data is not sufficient, because tests in the full-scale are rare due to limitations of loading devices. Furthermore, not a few materials used for new inventions are affected significantly by the rate of loading, which aggravates the situation because of the scarcity of facilities that are capable of conducting large dynamic loading tests on a realistic scale. The need for real data obtained by experimentation is deemed extremely urgent for the advancement of earthquake engineering, particularly for issues pertinent to collapse (relative to mildly inelastic, rather stable action), rate-of-loading (dynamic loading relative to quasi-static loading), realistic-scale (relative to miniature models), and structural systems (relative to components) Example Tests Conducted to Reproduce Complete Failure and Collapse Over the past few years, the writer and his group have conducted tests that focused directly on the issues of collapse in realistically scales structural systems. They are summarized below. Behavior to Complete Failure of Steel Beams Subjected to Cyclic Loading [Liu et al 2003] An experimental study was conducted on steel beams subjected to cyclic loading to extremely large deformations (Fig. 4)

6 Figure 4. Test setup for reproduction of complete failure of steel beams. The study aimed to collect information on beam hysteretic behavior up to complete failure, in the belief that such information is needed for the establishment of performance-based design. Test beams were about 1/10-scale models, and the effects of RBS details and lateral braces arranged at beam top flanges were examined. Behavior up to the cyclic loading amplitude of 0.06 rad was commensurate with behavior observed in many previous studies [Fig. 5(a)]. Behavior in extremely large deformations from 0.1 to 0.5 rad amplitudes was significantly different from the behavior in large deformations (to 0.06 rad amplitude) (Fig. 6). The RBS beam failed earlier in the reduced cross-section, primarily due to strain concentrations at the section [Fig. 5(b)]. Lateral braces also caused strain concentrations, leading to earlier fractures. Significant increase in the maximum resistance was observed in extremely large deformations for beams not braced laterally. Tensile axial forces induced in the beam according to the geometry change were responsible for the increase. Figure 5. (a) (b) Behavior of beams in large deformations: (a) Lateral torsional buckling; (b) fracture

7 2 1 M/M M/Mp p 0-1 θ (rad) θ (rad) 0.5 x Figure 6. End moment end rotation relationships to complete failure. Instability and Complete Failure of Steel Columns Subjected to Cyclic Loading [Nakashima and Liu 2005] In a testing system designed for large deformations, structural columns were loaded to complete failure, defined as either complete separation of the column or inability to sustain the prescribed axial load. The test system consisted of very large stroke quasi-static jacks, digital displacement transducers that can ensure accurate measurement of large deformations, hydraulic pump units capable of controlling oil flow, controllers that control the jack motion, and separate personal computers for operating the jack controllers and for supervising and measuring data (Fig. 7). These components were connected on-line for data and signal operations, which enables automatic and accurate load control for tests that lead specimens to complete failure. Six columns having a square tube cross-section were tested in cyclic loading conditions, with axial load and column length as major parameters. The load-deformation relationships obtained from the tests were presented in detail, and relationships among the deformation capacity, failure mode, slenderness, and axial load were discussed (Fig. 8). An intermediate axial load of 30% of the yield axial load was effective in retarding the occurrence and growth of cracks, resulting in larger deformation capacity to complete failure. Finite element analysis accurately duplicated the experimental behavior up to a large inelastic range including material yielding, strain hardening, and local buckling. It failed to simulate the experimental behavior in a very large deformation range where the column surfaces crashed and contacted each other (Fig. 19). More experimental data is strongly needed on the behavior of structural systems and elements at and near complete failure

8 Figure 7. Test setup for reproduction of complete failure of steel columns. Figure 8. Test specimens at end of loading: (a) No axial load; (b) medium axial load; (c) large axial load. Figure 9. Comparison with test and finite element analyses: (a) Slip option; (b) glue option

9 Tests of Welded Beam-Column Subassemblies I: Global Behavior and II: Detailed Behavior [Nakashima et al 1998b, Suita et al 1998] Cyclic loading tests were applied to fourteen full-scale beam-column subassemblages (Fig. 10). Efforts undertaken in the Japanese steel community in response to damage observed at welded beam-to-column connections in the 1995 Hyogoken-Nanbu (Kobe) Earthquake were introduced. The major test parameters chosen in this study were: type of steel, type of connection, type of weld access holes, type of weld tabs, and type of loading. The test results were presented in terms of the ductility capacity of the test specimens. Major findings were as follows. All specimens developed plastic rotations and cumulative plastic rotations of 0.03 rad and 0.3 rad. respectively, suggesting that the ductility capacity of the specimens was sufficient in light of present Japanese seismic design. Dynamic loading had no detrimental effect on ductility capacity (Fig. 11). A significant rise in temperature observed in the dynamic loading tests was the likely cause of the larger ductility capacity and more ductile fracture. Fracture surfaces were examined from fractography analysis. Changes in material hardness before and after the test are also investigated, and the correlation between the hardness increase and cumulative plastic strain was quantified (Fig. 12). Modified details for the weld access hole had the effect of preventing cracks initiating from the toe of the weld access hole. Figure 10. Test setup for fracture of steel beam-to-column connections. Figure 11. (a) (b) End moment end rotation relationships: (a) Quasi-static loading; (b) dynamic loading

10 Figure 12. (a) (b) Fracture surfaces: (a) Brittle fracture in quasi-static loading; (b) ductile fracture in dynamic loading. Test on Full-Scale Three-Story Steel Moment Frames and Assessment of Numerical Analysis to Trace Inelastic Cyclic Behavior [Nakashima et al 2006] A test on a full-scale model of a three-story steel moment frame (Fig. 13) was conducted, with the objectives of acquiring real information about the damage and serious strength deterioration of a steel moment frame under cyclic loading, studying the interaction between the structural frame and nonstructural elements, and examining the capacity of numerical analyses commonly used in seismic design to trace the real cyclic behavior. The outline of the test structure and test program was presented, results on the overall behavior were given, and correlation between the experimental results and the results of pre-test and post-test numerical analyses was discussed. Pushover analyses conducted prior to the test predicted the elastic stiffness and yield strength very reasonably. With proper adjustment of strain hardening after yielding and composite action, numerical analyses were able to duplicate the cyclic behavior of the test structure with great accuracy up to a drift angle of 1/25 (Fig. 14). The analyses could not trace the cyclic behavior for larger drifts, in which serious strength deterioration occurred due to the fractures of beams and anchor bolts and the progress of column local buckling (Fig. 15). 9, ,700 3,5001,300 South CG 3 1 2C G CG 1 3C 1 1C 1 2G 2 North 3C 1 2C 1 1C 1 3C 1 2 C 1 1 C 1 3 G 1 2G 1 3C 2 2 C 2 1 C 2 3 G 1 2G 1 3C 1 2 C 1 1 C 1 2,850 1,500 8,250 3,975 6,000 6,000 12,000 North C G C G C South Figure G 2 B 1 2 G 2 B 1 a a a a 2C 1 G G 2 1 2C C 1 2CG 1 B 3 B 2 2 CG 1 B 3 6,000 6,000 12,000 A three-story full-scale steel moment frame B 2 2 G 2 1,500 8,250 2 CG

11 Figure 14. Comparison of test and analysis for behavior with mild plasticity. Figure 15. Comparison of test and analysis for behavior with serious strength deterioration. Development and Completion of E-Defense Over the last ten years, the National Research Institute for Earth Science and Disaster prevention (NIED) had been constructing a shaking table facility, known as E-Defense [Hyogo 2005]. E-Defense was completed in March 2005, and its operation started in April The Hyogo Earthquake Engineering Research Center was established on October 1, 2004, to manage research projects using E-Defense and to operate and maintain the facility. E-Defense has the unique capacity to experiment with life-size buildings and infrastructural systems in real earthquake conditions, and stands as a tool of ultimate verification. With this feature, E-Defense should help expedite the transfer of various research outputs into the practice of earthquake disaster mitigation. Figure 16 is a bird s eye view of E-Defense, located in a city called Miki on the north of Kobe City. The heart of the facility is a jumbo shaking table in the center of the site. The table is attached to five actuators in each horizontal direction and supported by fourteen actuators installed vertically underneath the table (Fig.17). Table 3 shows the major specifications of the table. The table is 20 meters by 15 meters in the plan dimension. It can accommodate a specimen up to a weight of 12 MN (1,200 metric ton). The unique feature of the table is that can produce shaking of a velocity of two meters per second and a displacement of one meter in the two horizontal directions simultaneously. As far as the

12 capacity is concerned, the table owned by E-Defense appears to be the largest shaking table in the world. Figure 16. A bird eye view of E-Defense. Figure 17. Shaking table of E-Defense

13 Talbe 3. Major specifications of shaking table. 3D Full-Scale Earthquake Testing Facility Payload Size Driving Type 12 MN (1,200 tonf) 20 m x 15 m Accumulator Charge Electro-Hydraulic Servo Control Shaking Direction Max. Acceleration (at Max. Loading) Max. Velocity Max. Displacement Max. Allowable Moment X & Y Horizontal >9 m/s/s 2 m/s 1 m Overturning Moment 150 MN x m Z Vertical >1.5 m/s/s 0.7 m/s 0.5 m Yawing Moment 40 MN x m Construction of E-Defense was in near completion in the fall of 2004, and since that time a series of tests on performance calibration have been conducted, without specimens in the first phase and with real-size specimens in the second phase. Figures 18 shows examples for the table performance check. One shows how the table motion was duplicated with the basic control, while the other shows how it was improved by the application of a specially designed advanced control. In both cases, the JMA Kobe record (6.17 m/s/s, 8.18 m/s/s, and 3.32 m/s/s in the maximum acceleration of the EW, NE, and vertical components), a strong motion recorded at the 1995 Kobe earthquake, was applied to the table. The basic control was found to be satisfactory, and the advanced control augumented the accuracy. Figure 18. Reproduction of ground motion record (JMA Kobe record). Regarding the performance test with specimens, the very first application was made for a two-story wood house tested on January 15 and 16, 2005 at the inauguration ceremony of E-Defense. The house had two stories, with ten meters by eight meters in plan and ten meters in height (Fig.19). The JMA Kobe record shook the house in all three dimensions. The total weight of the specimen was about 800 kn, far lighter than the maximum weight accommodated by the table; hence the control was easy, and the test ran successfully. The test was carried out for demonstration to the public, and the specimen was designed to be very earthquake-resistant (to avoid any inconvenience). Accordingly, the specimen revealed

14 only minor damage even for the unscaled JMA Kobe record. Figure19. A two-story wood house tested at E-Defense. The ongoing test featured a five-story steel braced frame (Fig. 20) on the table. The frame is twenty meters tall, fifteen meters by ten meters in plan, 6MN in weight, and built very strongly so that the frame would remain elastic even under the strongest shaking. The natural frequency of the frame is 5.0 Hz when all braces are installed, and 3.0 Hz when they are removed. According to the performance test of the table without any specimen, the resonant frequency of the table system (reflecting the table dynamics) is about 4.4 Hz. The two natural periods assigned to the specimen sandwich the table resonance period; hence the overall performance can be checked even in the most difficult circumstances of control. Furthermore, an overturning moment approximately equal to 150 MN x meter is to be imposed onto the table, which is another challenge for the table control. The test is ongoing at the time of this writing, and more details will be published soon. Figure 20. A five-story steel frame for tests of table performance checking. Ongoing Projects at E-Defense E-Defense participates in a comprehensive research project named Special Project for Mitigation of Earthquake Disaster in Urban Areas sponsored by the Japanese Ministry of Education, Culture, Sports, Science, and Technology, nicknamed MEXT. The project started in 2002 and will extend for a period of five years. Four major thrust areas have been

15 established; first is the evaluation of earthquakes and strong motions; second is the evaluation and enhancement of the earthquake resistance of structures; third is the simulation for disaster responses; and fourth is the integration of the first three projects for the enhancement of better countermeasures to be taken by our society. Among the four areas, E-Defense deals primarily with the second category. In the project, three targets were chosen. One is wood houses, another is reinforced concrete buildings, and the third comprises soils and foundations. There are sensible reasons for the choice of the three targets. Wood is by far the most popular material for Japanese houses. People are always very keen about the safety of their shelters against earthquakes. Reinforced concrete is used most commonly for apartment buildings and schools. This building type is also very involved in the daily life of the Japanese public. Soils in particular liquefaction and lateral spreading are of serious concern throughout Japan. At the time of writing, the project has completed its first three years. E-Defense was not available until 2005; hence a variety of tests in this project were conducted using other facilities. Full-scale tests are scheduled in 2005 and 2007 for the three types of structures (Fig. 21). Figure 21. (a) (b) (c) Three tests scheduled at E-Defense for 2005 to 2007: (a) Wood houses; (b) RC frame; (c) soil-structure interaction. Figure 22(a) shows a 1/3-scale wall-frame, fabricated as a replica of the first full-scale RC test to be tested in the winter of 2005 to 2006 at E-Defense. The 1/3 scale frame was sixstoreys tall, with two by three spans in plan and a total weight of 1.5 MN. The test was conducted last winter on the shaking table owned by NIED at Tsukuba, Japan. The frame was shaken to a complete first-story collapse, as shown in Fig. 22(b). (a) (b) Figure 22. One-third scale six-story RC frame: (a) Before test; (b) after test exhibiting

16 first-story collapse. Collaboration Between E-Defense and NEES Stimulated in 1995 by the Kobe earthquake, discussions and plans for the construction of E-Defense bore fruit after ten years, and the facility was completed in March The United States of America also implemented a national project on the upgrade of experimental facilities used for earthquake engineering, named the George E. Brown, Jr. Network for Earthquake Engineering Simulation (NEES), completed in the i fall of 2004 [George 2005]. Earthquake disaster and its mitigation is a very critical problem in both countries; NEES and E-Defense facilities have similar missions and functions in research on the mitigation of earthquake disasters, and the two countries have a very long history of collaboration on earthquake engineering research and practices. In consideration of this, a very natural outcome is research collaboration through complementary usage of the two facilities. Since the spring of 2004, the research communities in the United States and Japan have conducted an extensive discussion regarding visible and close research collaboration. The two communities met a few times including three planning meetings held in April 2004 in Kobe, July 2004 in Washington DC, and January 2005 in E-Defense, respectively. To strengthen and formalize the collaboration, a memorandum of understanding (MOU) between NSF and Japanese Ministry of Education (MEXT), and another MOU between NEES and NIED are being prepared. As a result of the series of meetings, the parties reached an agreement that steel buildings and bridges would be the immediate targets of research collaboration between the two countries (Fig. 23). In addition, NEES and E-Defense have formalized collaboration on the advancement of cyberinfrastructure in both countries. Details of the collaboration can be found in [13]. Figure 23. (a) (b) Structures considered in NEES/E-Defense joint project: (a) steel frame; (b) bridge. E-Defense and International Collaboration As indicated in the previous sections, E-Defense is a very large shaking table, probably the largest in the world, but NIED is in no manner boasting about the size of E-defense. NIED fully understands that large is not synonymous with good. After all, good and useful research is achieved only through intellect and enthusiasm of the participants in the test. To this end, E-Defense tries its best to recruit as many experts available in Japan as possible for research projects conducted at E-Defense, and wishes to implement

17 community-based research that involves all layers of researchers and professionals engaged in earthquake engineering. E-Defense also has a goal of positive and effective collaboration within the international community of earthquake engineering to collect and make the best use of the intellect and enthusiasm throughout the world and to collaborate on the mitigation of earthquake disasters in all the regions that are prone to earthquake disasters. References Architectural Institute of Japan (1995). Preliminary Reconnaissance Report of the 1995 Hyogoken-Nanbu Earthquake, 216pp. Council of National Disaster Migitation (2004). George E. Brown, Jr. Network for Earthquake Engineering Simulation (NEES) (2005). Hyogo Earthquake Engineering Research Center, National Research Institute for Earth Science and Disaster Prevention (NIED) (2005). Kinki Branch of the Architectural Institute of Japan (1995). Reconnaissance report on damage to steel building structures observed from the 1995 Hyogoken-Nanbu Earthquake, Steel Committee, Osaka (in Japanese with attached abridged English version). Liu, D., Nakashima, M., and Kanao, I. (2003). Behavior to complete failure of steel beams subjected to cyclic loading, Journal of Engineering Structures, 25(3), Nakashima, M., Inoue, K., and Tada, M. (1998a). Classification of Damage to Steel Buildings Observed in the 1995 Hyogoken-Nanbu Earthquake. Engineering Structures, 20,(4-6), Nakashima, M., et al. (1998b). Tests of welded beam-column subassemblies I: Global behavior," Journal of Structural Engineering, 124(11), Nakashima, M., Matsumiya, T., and Asano, K. (2000). Comparison in earthquake responses of steel moment frames subjected to near-fault strong motions recorded in Japan, Taiwan, and the U.S., International Workshop on Annual Commemoration of Chi-Chi Earthquake, Taiepi, Taiwan, Nakashima, M. (2001). APPENDIX C. Overview of Damage to Steel Building Structures Observed in the 1995 Kobe Earthquake, Past Performance of Steel Moment-Frame Buildings in Earthquakes, Federal Emergency Management Agency, Report FEMA-355E, C-1-C24. Nakashima, M. and Liu, D. (2005). Instability and complete failure of steel columns subjected to cyclic loading, Journal of Engineering Mechanics, ASCE, 131(6), Nakashima, M., Matsumiya, T., Suita, K., and Liu. D. (2006). Test on full-scale three-story steel moment frames and assessment of numerical analysis to trace inelastic cyclic behavior, Journal of Earthquake Engineering and Structural Dynamics (accepted for publication). Suita, K., Nakashima, M, and Morisako, K. (1998). Tests of welded beam-column subassemblies II: Detailed behavior, Journal of Structural Engineering, ASCE, 124(11),