A Study on Pendulum Seismic Isolators for High-Rise Buildings

Transcription

1 ctbu.org/papers Title: Autors: Subjects: Keywords: A Study on Pendulum Seismic Isolators for Hig-Rise Buildings Ikuo Tatemici, Maeda Corp. Mamoru Kawaguci, Kawaguci & Engineers Masaru Abe, Hosei University Seismic Structural Engineering Seismic Structure Publication Date: 2004 Original Publication: Paper Type: CTBUH 2004 Seoul Conference 1. Book capter/part capter 2. Journal paper 3. Conference proceeding 4. Unpublised conference paper 5. Magazine article 6. Unpublised Council on Tall Buildings and Urban Habitat / Ikuo Tatemici; Mamoru Kawaguci; Masaru Abe

2 A Study on Pendulum Seismic Isolators for Hig-Rise Buildings Ikuo Tatemici 1, Mamoru Kawaguci 2, Masaru Abe 3 1 Maeda Corporation 2 KAWAGUCHI & ENGINEERS 3 Lecturer, Hosei University Abstract Seismic isolation systems ave recently been adopted for several ig-rise buildings in Japan. If ig-rises are to be isolated effectively from eartquake-induced motions, tey need to ave a longer period and iger reliability tan tose provided by conventional seismic isolation systems. Te autors, wo are developing seismic isolation systems based on te principle of pendulums, ave sown tat seismic isolation systems using translational pendulums are free from te influence of te weigt of te structure tey bear and are functionally stable. Te autors ave recently expanded te study and found tat non-parallel swing systems can provide longer periods effectively. In tis paper, a pendulum wit supports tat deviate from or come closer to eac oter symmetrically are referred to as te non-parallel swing system. Tis paper presents an experimental study on te effectiveness of te non-parallel swing system and on its applicability to ig-rise structures. Keywords: Pendulum; Isolator; Seismic; Experiment; Hig-Rise 1. Introduction In spite of recent developments in eartquake resistant engineering, eartquakes still inflict widespread damage in various countries. Seismic isolation is a very effective measure for protecting structures from eartquake damage. In Japan, te construction of seismically isolated structures progressed significantly following te Hyogoken- Nanbu Eartquake (wit a magnitude of 7.2) in 1995; te number of structures provided wit suc measures now exceeds 1,000 (excluding seismically isolated wooden residences). Tis figure is among te largest in te world. Te most commonly used seismic isolation device globally is te laminated rubber bearing, wic comprises layers of rubber and steel plates. However, te dynamic caracteristics of laminated rubber bearings vary depending on bearing stress, displacement, temperature and so on. Especially, te natural period of laminated rubber bearings is affected by te weigt of te structure, wic places considerable restrictions on te design of structures wit suc bearings. Laminated rubber isolators are difficult to apply to ligtweigt structures and te center of gravity of te upper structure must coincide wit te stiffness center of te isolation device. In Contact Autor: Ikuo Tatemici Maeda Corporation Fujimi Ciyoda-Ku, Tokyo, Japan, Tel: Fax: itatemit@jcity.maeda.co.jp addition, wen applied to ig-rise buildings or structures constructed on soft ground, te commonly used seismic isolation system as to be re-designed to ave a longer natural period. In order to address tese issues, various combinations of seismic isolation devices, including laminated rubber bearings, are being studied, and tose provided wit sliding mecanisms ave been developed. However, a device to replace te laminated rubber bearing as yet to be acieved. Te autors ave been studying a seismic isolation system wit iger performance and reliability tat uses bearings acting like pendulums 1)2)3). Te autors conducted analytical and experimental studies of pendulums symmetrically ung from two supports including translational pendulums as special examples, and presented te structural beavior of pendulums ung from two supports 4). Te same study also noted te possibility of developing a seismic isolation system using te caracteristics of pendulums ung by non-parallel angers from four supports tat enable te natural period of an isolated structure to be adjusted. 2. Principle of pendulum seismic isolators Te pendulum is one of te basic metods of seismic isolation, and is also used as te basic mecanism of seismograps. However, pendulums are rarely used for seismic isolation of structures. 5) As sown in Figure 1, pendulums used for engineering purposes include: (a) simple pendulums, (b) pysical pendulums, and (c) translational pendulums. 182 CTBUH 2004 October 10~13, Seoul, Korea

3 te floor swings te same way regardless of te eigt of te center of gravity of te substance above te floor (Figure 2). (a) Simple (b) Pysical (c) Translational Fig. 1. Various pendulum systems It is known tat te natural period T of a simple pendulum wit arc L is given as follows, were g is gravitational acceleration: T 2 (1) One advantage of te pendulum seismic isolator is tat te lengt of te anger L is te only parameter governing its natural period, and te mass of te object to be isolated or te tension of te anger as no effect at all. Tus, te desired period can be obtained merely by canging te anger lengt. Tis is te greatest advantage of te pendulum seismic isolator compared to laminated rubber bearing seismic isolators in wic te natural period is determined by te mass and rigidity of te isolation structure. As wit te amplitude of a pendulum, its natural period is elongated if te amplitude is made larger. Te elongation, owever, is minute, as indicated by te fact tat for an amplitude as large as 60 on one side, te increment in natural period is only about 8%. Tus, te above equation can be considered valid for practical purposes. Tis is anoter advantage of pendulum seismic isolators compared to laminated rubber bearings, of wic te deformability is limited. A wide selection of materials is available for te anger. For example, tecnology for fireproofing as already reaced a mature state if steel is to be used. Anoter point to note is tat wen tensile force is applied to a pendulum seismic isolator, compression stress is borne by te anger. A steel anger can witstand tis well, provided sufficient measures ave been provided to prevent buckling. As discussed above, seismic isolators using te pendulum principle possess considerable merits. Considering tat seismic isolators must also function as a part of te structural support, te simple pendulum sown in (a) of Figure 1 is obviously difficult to use. Te natural period of te pysical pendulum sown in (b), on te oter and, fluctuates wit te location of te center of mass of te object to be supported. Tus, te translational pendulum sown in (c), wose natural period is only affected by te anger lengt as in te simple pendulum, would be appropriate for use as a seismic isolation device. 3. Application for seismic isolation of floors One possible application of te translational pendulum seismic isolator is for individual floors. Tis isolation system as a more interesting aspect: L g Fig. 2. Sinple pendulum and translational pendulum systems A floor suspended from a girder of a building frame as sown in Figure 3 was adopted for te exibition rooms of an actual museum "Ceramics Park MINO" for pottery and porcelain (arcitectural design: Arata Isozaki & Associates, Structural design: Kawaguci & Engineers, Completion: May 2002). Potograp 1 sows te outside of te building and Figure 4 presents an isometric drawing of te isolated floor. Te area of te suspended floor is about 900 m 2, and its mass is about 1,000 tons. Hinges aving universal joints are used for te upper and lower ends of te anger as sown in Potograp 2. If te anger is made to be 4.5 m long, Equation (1) yields a natural period of more tan 4 seconds, wic is considered sufficiently long for seismic isolation. Te results of seismic isolation tests performed to confirm te design are described below. Fig. 3. Concept of te floor wit pendulum isolator Potograp 1. Outside of te building (courtesy of Arata Isozaki & Associates) CTBUH 2004 October 10~13, Seoul, Korea 183

4 sufficient isolation of te floor suspended as a translational pendulum. Fig. 4. Isometric drawing of seismically isolated floor Potograp 2. Installing te angers Te specimen comprises a test floor, suspended by four angers from a frame, as sown in Potograp 3. Te lengt of te angers was made equal to tat used in te actual structure. Te dimensions of te suspended floor were 1.5 m by 2.5 m. Fig. 5. Acceleration response of test floor Figure 6 sows a frequency response function under orizontal vibration by wite noise. As sown in te figure, te rate of amplification in response between te saking table and suspension frame was about 1, wic indicates tat te suspension frame is sufficiently rigid to convey te vibration of te table to te suspension ancor. Te isolated floor resonated significantly wit te input vibration in te vicinity of a period of 4 seconds. Wen te prevalent period of input seismic motion is less tan 2.7 seconds, owever, te isolated floor functioned well to reduce te response. Te input acceleration was reduced to about a alf wen te period was 2 seconds, and to less tan 10% wen it was 1 second or sorter. Furtermore, a free vibration test was conducted for te Ceramics Park MINO during construction to identify te period, friction and damping caracteristics of te suspended floor of te actual structure. Potograp 3. Specimen on te saking table Te specimen was placed on a saking table vibrating bot vertically and orizontally, and vibratory motions induced by wite noise and te actual seismic records were applied. Figure 5 sows te cange wit time in orizontal accelerations of te saking table and te suspended floor, under te vibration based on te El Centro 1940 N-S. It indicates Fig. 6. Frequency response function for te input wite noise 4. Application to seismic isolation of ig-rise buildings 4.1 Concept Seismic isolation is now being applied to structures wit slender elevations. Altoug stress due to seismic motion generated in structural members is not a major 184 CTBUH 2004 October 10~13, Seoul, Korea

5 issue for ig-rise buildings, seismic isolation is desirable for te comfort of occupants and safety against overturning. In te case of laminated rubber bearings, owever, te rigidity of te bearings is significantly reduced wen subjected to tensile force, and terefore its application to ig-rise buildings is deemed difficult if uplift due to overturning moment of te structure is considered. Tus, anoter possibility is application of a system tat uses a pendulum for base isolation of ig-rise buildings. Ground Building Column Pendulum device Damper elements Fig. 7. Conceptual drawing of pendulum for base isolation 4.2 Studies based on eartquake response analyses An eartquake response analysis was performed for a ig-rise structure provided wit a pendulum seismic isolation mecanism at its base, simulated by a single-mass system as sown in Figure 8 (a). Even wen uplift is generated at te base of te structure due to overturning moment, te period of te pendulum is determined only by te anger lengt L, provided te anger is a rod or oter elements tat are compression-resistant. Terefore, te pendulum can be simulated by a spring wit equivalent natural period, and modeled as sown in Figure 8 (b). Here, me, He, and Ke represent te equivalent mass and eigt of te building, and spring constant, respectively. Kp represents te equivalent spring constant of te pendulum. For simplicity, it was assumed tat me = m and He = 0.5H. Te equivalent spring constant of te pendulum was simulated by Equation (2), as it was (a) m B L H ig-rise building me Ke (b) Kp C Modeling He Fig. 8. Modeling of pendulum isolator for a ig-rise building evident from te analysis results tat te amplitude is sufficiently small. g Kp me (2) L A ig-rise building wit an aspect ratio H/B of 10 (te building 100-m tall wit a natural period of 2 seconds, and damping constant of te structure being 3%) was studied as an example. Te frequency of te pendulum seismic isolator was assumed to be 4 seconds, and te damping constant as 10%. Te total mass of te building was 2,000 tons. Te input seismic motion was tat of te Hacinoe (1968) N-S, and te maximum input acceleration was 500 gal. Te analysis metod was tat by direct integration. Table 1 sows te maximum response values. Also sown for comparison are te results for cases wit fixed foundation and tat witout a damper. Te response displacement between te ground and foundation represents te amplitude of te pendulum. Te angular amplitude of te pendulum was 5.7 for an input motion of 500 gal, wic is equivalent to a great eartquake. Hence, period sift does not need to be considered. Te magnitude of response displacement between te foundation and building structure represents te magnitude of base sear. Te obtained results were in a range between 0.25 and 0.3 times tose obtained for a case wit fixed foundation, wic is a significant reduction. Te maximum response uplift was insignificant, being about 66 tons even for cases witout a damper, wen te weigt of te building itself (1,000 tons) was subtracted. Tus, it was concluded tat pendulum seismic isolators aving a long period are effective for ig-rise buildings. Table 1. Maximum response values (Hacinoe N-S, maximum input acceleration of 500 gal) Fixed Pendulum isolator foundation =10% =0% Maximum response displacement (m) Between te ground and foundation Between te foundation and building Maximum response Foundation acceleration (gal) Building Maximum response uplift (ton) 3, , A study on te non-parallel swing system for seismic isolation 5.1 Concept of non-parallel swing system CTBUH 2004 October 10~13, Seoul, Korea 185

6 In te field of arcitecture, translational pendulums ave played a central role in seismic isolation because of teir ig applicability. From a geometric viewpoint, tere exists anoter type of pendulum system between (b) and (c) in Figure 1. Te type adopts pendulums tat are ung symmetrically. Te distance between te supports may be eiter longer or sorter tan tat for te translational pendulums in (c). Swinging bars in a park are an example of tis concept. In tis capter, a pendulum wit supports tat deviate from or come closer to eac oter symmetrically are referred to as te pendulum symmetrically ung from two supports, and its vibration properties are identified. A metod for applying te useful caracteristics of te pendulum symmetrically ung from four supports to seismically isolated structures is proposed and a basic model test is conducted. 5.2 Natural period of a pendulum symmetrically ung from two supports A solution tat provides te natural period of te pendulum symmetrically ung from two supports is a general solution applicable eiter to te pysical (Figure 1 (b)) or translational (c) pendulum. Figure 9 sows a vibration model of a pendulum symmetrically ung from two supports. As is obvious from Figure 9, te center of gravity G of te object supported by te pendulum rocks due to vibration. A special solution witout rocking is applicable to te case of a translational pendulum sown in Figure 1 (c) (a=b). Te ratio of te lengt of te suspended floor A-B to te distance between te supports C-D is expressed by =b/a. =0 and =1 represent te pysical and translational pendulums, respectively. Wen <1 (left figure), te angers (A-C and B-D) and te suspended floor rock in te same direction. Wen >1, te angers and te suspended floor rock in opposite directions. Wen =1 (in te case of a translational pendulum), te suspended floor does not rock. A rigid object was placed on te suspended floor of te model and free vibration was applied. Te period was obtained according to te mecanical energy conservation law (Equation (3)). Te suspended floor (A-B) was assumed to ave sufficient rigidity. 1 T 2 cos g 1 2 I M G G sin 1 cos 2 2 (3) 1 b a b were, tan, a were, G is te eigt of center of gravity, and I is te moment of inertia around te center of gravity of te rigid object. Te masses of te suspended floor and te angers were ignored. Figure 10 sows canges in natural period according to =b/a using G / as a parameter. Te figure sows tat te natural period fluctuated substantially wen te distance between te supports of te angers was varied. Te iger G /, te greater te fluctuation. Te natural period can be made longer tan tat of te translational pendulum by adjusting te distance between te supports of te angers. Tis is very effective for seismic isolation. 5.3 Metod of application to a seismic isolation system A system using te caracteristics of te pendulum symmetrically ung from two supports can provide a structure wit a longer natural period tan te seismic isolation system wit te translational pendulum. Te structure supported by te pendulum symmetrically ung from two supports is, owever, subject to rocking. Rocking is very detrimental to seismic isolation systems because amenity may be adversely affected or te seismic response of certain stories may not be reduced effectively. In tis study, a seismic isolation system tat can control rocking wile using te caracteristics of te pendulum symmetrically ung from two supports is discussed. Te seismic isolation system can be made by connecting te columns supported by te pendulums symmetrically ung from two supports to te superstructure by inges. A conceptual view of te system is sown in Figure 11. Te columns Cs supported by te pendulums rock due to vibration, but te column tops E and F remain at te same elevation. Fig. 9. Vibration model of a pendulum symmetrically ung from two supports 186 CTBUH 2004 October 10~13, Seoul, Korea

7 Tus, translational movement is predominant for te superstructure. Sligt vertical displacement naturally occurs in te superstructure as te restoring force of te pendulums is generated. 2 T g G Fig. 10. Natural period of a pendulum symmetrically ung from from two supports expressed as Equation (3) using te sign sown in Figure 12. Te system can deal wit tree-dimensional movement by canging te distance of te top ends of te angers spatially. Te effectiveness of a seismic isolation system wit te pendulums symmetrically ung from four supports was examined by a vibration test using a scale model. In te test, a tower structure wit an aspect ratio of approximately ten tat simulated a ig-rise building was adopted to facilitate testing. Te test models are composed of an upper structure and four base isolation devices. Potograp 4 sows four base isolation devices and Potograp 5 sows te specimen on te saking table. Te superstructure was 100 cm ig and 10 cm wide and was made of acrylic resin boards. Te distance between te supports of te angers was adjustable. Te upper model simulates ig-rise buildings of 100 m, 200 m and 300 m eigt. Tose simulated models produce te first mode natural periods of 2 sec, 4 sec and 6 sec, respectively. To perform tese simulations, te time scales for te eartquakes are compressed by scaling laws. Simultaneously wit te experiment, we also performed numerical analysis. Te upper model is replaced by a two-dimensional dynamic analysis E F C C Fig. 11. Conceptual view of application to a seismic isolation system Potograp 4. Specimen of four base isolation devices 5.4 Experimental study on application of pendulum seismic isolators to ig-rise buildings However, a pendulum symmetrically ung from two supports considered only orizontal one-directional movement. Terefore, te seismic isolation system of te pendulum wit four symmetrical angers is suggested as an advanced system as sown in Figure 12. Te natural period T of tis system can be similarly 2b 2a Fig. 12. Conceptual view of te pendulum wit four symmetrical angers Potograp 5. Upper model on four base isolation devices CTBUH 2004 October 10~13, Seoul, Korea 187

8 model of 10-mass structure, and te isolation devices are substituted by equivalent linear axial spring elements. In addition, te viscous damping of te test model is replaced by a stiffness-proportional damping element using te first-mode damping ratio obtained by te experiments as a criterion. First, a free vibration test was conducted. Figure 13 sows te rate of increase in natural period in relation to te natural period of te translational pendulum (=1). Two kinds of experiments, te case were te upper structure simulates ig-rise buildings, and te case were te upper structure is rigid, were conducted. In Figure 13, te square mark and te round mark sow eac experiment result respectively. Te calculated value from Equation (3) and te two-dimensional eigenvalue analysis result are given in Figure 13 for comparison. Figure 13 indicates good agreement between te analytical value and te measurement. In te case were te upper structure is rigid te effect of te coupled vibration of te superstructure and te isolation layer on te natural period was eliminated. Figure 13 sows a decreasing rate of increase in period of te specimen because of te coupled effect of te vibration of te superstructure. It was, owever, evident tat te effect of te increase in period owing to te seismic isolation system of te pendulum wit four symmetrical angers was maintained. Nnatural period (sec) Experiments, wit upper structure Experiments, wit rigid upper structure Eigenvalue analysis result Equation (3) g/=1.65 Fig. 13. Natural period of isolated superstructure Figure 14 sows te rates of te observed acceleration responses wen te seismic motion based on te records of actual eartquakes were applied to te vibration table. As sown, te response in te upper structure on te isolating layer was reduced sufficiently. It was also confirmed tat te effect depended on te distance of te top ends of angers. Figure 15 sows one of te results of saking experiments and eartquake response analyses. Te figure indicates good agreement between te analytical value and te measurement. Top 10t floor 9t floor 8t floor 7t floor 6t floor 5t floor 4t floor 3rd floor 2nd floor 1st floor Input wave acceleration ( gal ) Simulated 2 sec building and 3 sec isolator Input wave is Taft(EW) 350 cm/sec 2 Witout isolator Fig. 14. Isolation effects of te maximum response acceleration based on actual eartquakes time ( sec ) Fig. 15. Results of experiment and analysis input wave experimet analy sis 6. Conclusions Te autors ave been developing igly reliable ig-performance seismic isolation systems based on te simple principle of te pendulum. For applying te pendulum principle to seismic isolation, translational pendulums are easier to use tan any oter type of pendulum. In tis study, we developed a seismic isolation system wit pendulums symmetrically ung from four supports, wic could use te benefits of translational pendulums and provide enanced capacity. 1) Tis system can lengten te first mode natural period by increasing te lengt of angers and by extending te distance of te top ends of angers. 2) Extending te distance of te top ends of angers can reduce te seismic beavior of upper buildings wen lengtening te natural period by a translational pendulum does not provide sufficient effect. Te system may be able to isolate ultra-long-period motions and used to isolate structures wit predominant long-period components suc as ig-rise buildings. References 1) Kawaguci, M., Tatemici, I., Seismic isolation systems and teir application in space structures, Proceedings of te IASS-MSU International Symposium, Istanbul, Turkey, , ) I. Tatemici, M. Kawaguci (2000) A new approac to seismic isolation: possible application in space structures. International Journal of Space Structures, Vol. 15, No. 2, ) M. Kawaguci, I. Tatemici, M. Abe, T. Ide (2003) Development and testing of a seismically isolated floor system using translational pendulum principle. IASS-APCS 2003 Symposium, Taipei, Taiwan, 188 CTBUH 2004 October 10~13, Seoul, Korea

9 On CD-ROM. 4) Z. H. Cen, M. Kawaguci, I. Tatemici, M. Abe (2003) A study on te non-parallel swing system for seismic isolation. IASS-APCS 2003 Symposium, Taipei, Taiwan, On CD-ROM. 5) Garza Tamez et al. (1994). Test results and implementation of seismic base isolation system based on pendular action. Proceedings, Vol. 1, Second International Conference on Motion and Vibration Control, 1-6. CTBUH 2004 October 10~13, Seoul, Korea 189