Abstract
The good performance of seismically isolated structures during the 1995 Kobe earthquake persuaded structural engineers, isolator manufacturers and housing construction companies to undertake a cooperative research project to investigate the possibility of introducing the seismic isolation technology in Japan’s private housing sector. As a part of this project, in this study, the seismic performance of a recently developed Friction Pendulum System (FPS) for houses is presented. In order to verify its behavior under recorded earthquake ground motions, 3-dimensional shaking table tests were conducted. A three-dimensional nonlinear model was used to simulate the experimentally recorded acceleration and displacement response. Experimental results showed that the FPS significantly reduced the acceleration response at both moderate and strong levels of input ground motion. The analytical model used in this study satisfactorily predicted the experimentally recorded acceleration and displacement response time histories. Keywords: seismic isolation; friction pendulum system; houses; 3-D shaking table tests; numerical simulation Introduction turers and housing construction companies to undertake Kobe earthquake of January 17, 1995 severely hit a cooperative research project to investigate the one of the most populated urban areas of Japan, possibility of introducing seismically isolated structures unfortunately located very close to the fault rupture area. in Japan’s private housing sector. In order to achieve the It was one of the worst natural disasters that Japan required performance for the isolators and base-isolated th experienced in the second half of the 20 century. The residences, isolators' manufacturers were focused on earthquake caused severe building damage and left the development of new types of devices. Based on the thousands of people without shelter. Scientists and fundamental concepts used for developing each of them, engineers involved in earthquake-related disciplines they can be grouped in three systems: (a) Rubber were shocked by the damage, deaths and injuries that bearing system, (b) Sliding system and (c) Rolling resulted from this earthquake. As a result, important system. Details of the fundamental characteristics for questions were raised about earthquake preparedness, each of the newly developed isolators and the outline of disaster response, seismic design, upgrading of the shaking table tests performed are given elsewhere earthquake-resistant structures and introduction of new (Iiba et al., 2000). In this study, the focus is put on one technologies, which can assure high safety levels of the isolators belonging to the group (b). Its against destructive earthquakes (Yamanouchi et al., fundamental characteristics are briefly discussed in the 1997; Iiba et al., 1998). following section. The high concentration of damage in conventionally built residences (Building Research Institute, 1996) and Characteristics of the Newly-developed Isolator the good performance of seismically isolated structures The recently developed friction pendulum system during this destructive earthquake (The Building Center (FPS) for houses, shown schematically in Figure 1, is a of Japan, 1995) persuaded structural engineers, compact frictional isolation system that combines a earthquake engineering researchers, isolator manufac- sliding action and a restoring force by geometry. Differently from the previously devised FPS, usually with an articulated slider moving on a spherical surface, Contact Author: Bujar Myslimaj, Research Fellow, the isolation device presented in this study has a sliding Department of Civil Engineering, McMaster University cylinder that moves on two coated spherical surfaces. 1280 Main Street West, Hamilton, Ontario, Canada Both sides of the sliding cylinder that are in contact L8S 4L7 with the spherical surfaces are coated with a Tel: +1-905-525-9140 Ext. 22044 Fax: +1-905-529-9688 low-friction composite material (Teflon). As the slider e-mail: bujar@mcmaster.ca moves over the spherical surfaces, it causes the (Received April 10, 2002; accepted September 5, 2002) Journal of Asian Architecture and Building Engineering/November 2002/24 17 supported mass (weight) to rise, providing thus the providing thus the experimental data base necessary for restoring force for the system. Friction between the the investigation of the effect of bi-directional and sliding cylinder and the spherical surfaces generates vertical earthquake motions on the characteristics of the damping in the isolator. The effective stiffness of the FPS, as well as the effect of mass eccentricity on the isolator and the isolation period of the system are response of the system. In Figure 3 is shown sche- controlled by the radius of curvature of the concave matically how the experimental setup presented in surfaces. Normally, the coefficient of friction depends Figure 2 was rearranged in order to create various on the vertical stress in the isolator and sliding velocity, conditions of mass/weight distribution in the system, i.e. but taking into account the relatively low vertical stress balanced weight (top), unbalanced weight Case 1 in case of houses and high sliding velocity, the (middle) and unbalanced weight Case 2 (bottom). influence of the above factors on could be considered The input-output for the experimental setup shown in negligible for practical purposes. A typical value for Figure 2 was monitored through 50 channels. As it is the FPS presented in this study could be 0.04. In Table 1 shown in the figure, accelerometers (strain-gauge type), are summarized the fundamental characteristics of the velocity transducers (servo type) and displacement FPS. transducers (servo type, reel type, and laser type) were installed on the specimen. Strong ground motions recorded during the El Centro low-friction material sliding cylinder Earthquake of 1940 and Kobe Earthquake of January 17, coated spherical surfaces 1995 (JMA Kobe Station record) were used as input motions at the shaking table. Based on the peak ground velocity value observed in each record, the input earthquake waves were proportionally adjusted to various intensity levels. The following correspondence between the horizontal components of the recorded motions and the geometrical axes of the specimen (see Fig.1. Schematic Drawing of the FPS Figure 2) was applied: EWX and NSY. In addition, in order to see the effect of vertical component of Table 1. Characteristics of the FPS earthquake ground motion on the restoring force b ( mm ) 450 characteristic of the FPS, a special case of h ( mm ) 97 three-directional earthquake excitation was also ( mm ) 80 considered, where only the vertical component of input Load carrying capacity ( tf ) 4.0 motion was doubled. Allowable displacement ( mm ) In order to investigate the effect of bi-directional and Natural period ( sec ) 2.8 vertical earthquake ground motion on dynamic Coefficient of friction 0.02-0.05 response characteristics of the FPS, one-directional (X,Y), bi-directional (XY,YZ) and three-directional The extremely simple mechanical modeling of the (XYZ, XYZ2) earthquake excitations (El Centro 1940, FPS makes it very attractive. However its simplicity is Kobe JMA 1995) were applied as input motions to the associated with the main disadvantage of the system. It shaking table. The above shown XYZ2 symbol is essentially a one-parameter system, and that corresponds to the special case of a three-directional parameter is controlled by the radius of the concave excitation, where the vertical component only was surface. In order for the various articulated surfaces to doubled. slide together all surfaces have to be spherical, resulting thus to a linear nature of the restoring force character- Results of the Tests and Discussion istic over the entire range of displacement. An experimental database covering all the series of tests mentioned above is already created and is being Outline of the Shaking Table Tests Performed analyzed, with the purpose of investigating: a) the In order to verify the FPS behavior under recorded effectiveness of the FPS, b) the effect of bi-directional earthquake ground motions, 3-dimensional shaking input ground motion, c) the effect of vertical ground table tests were conducted. Tests were performed on a motion and d) the effect of mass eccentricity. large-scale three-dimensional shaking table recently Effectiveness of the FPS installed at the Public Work Research Institute of In Figure 4 are shown the input and response Ministry of Construction (PWRI, 1997). This shaking acceleration time histories for El Centro (a) and Kobe table is specially designed for simulating earthquake JMA ground motion (b), adjusted to a peak ground ground motions of the same intensity as those ones velocity (PGV) of 50cm/s. Both the input and response recorded during the Northridge Earthquake of January correspond to the one-directional test in the Y direction 17, 1994 or Kobe Earthquake of January 17, 1995. and are observed at the center of the shaking table and The experimental setup is shown in Figure 2. It was the specimen, respectively. For both input ground used for testing the base-isolation system under one-, motions a significant reduction in the acceleration two- and three-directional earthquake excitations, 18 JAABE vol.1 no.2 November 2002 Bujar Myslimaj response can be noticed. The ratio of peak response considerable displacement response that should be acceleration Ar, to peak input acceleration Ai for the treated by the designers with special attention. In each two cases shown in Figure 4 (a) and (b) resulted to be FPS design, the designer should clearly specify the peak 0.26 and 0.21, respectively. It is important to note here displacement limits of the system in accordance with that the beneficial effect of the FPS in remarkably the performance and serviceability requirements for a reducing the acceleration response of the su- house. per-structure is basically achieved at the expense of a JAABE vol.1 no.2 November 2002 Bujar Myslimaj 19 The Ar/Ai ratio depends on the input motion intensity. Effect of bi-directional and vertical ground motion This can be clearly noticed from the plots shown in In order to investigate the effect of bi-directional and Figure 5, which represent the Ar/Ai - PGV relation for vertical earthquake ground motion on dynamic both X and Y reference axis of the system subjected to a response characteristics of newly developed three-directional input motion. As the input motion base-isolation system for houses, one-directional (X, Y), intensity increases from 10 to 25cm/s the Ar/Ai ratio bi-directional (XY, YZ) and three-directional (XYZ, drops remarkably from 0.75 level to less than 0.50, XYZ2) earthquake excitations (El Centro 1940, Kobe remaining after that almost unchanged even though the JMA 1995) were applied as input motions to the PGV increases up to 50cm/s. shaking table. Response acceleration Response acceleration -200 -200 -400 -400 0 5 10 15 20 25 30 0 5 10 15 20 25 30 Time ( s ) Time ( s ) Input acceleration Input acceleration 0 0 -200 -200 -400 -400 0 5 10 15 20 25 30 0 5 10 15 20 25 30 Time ( s ) Time ( s ) (a) El Centro (b) Kobe JMA Fig.4. The Input and Response Acceleration Time Histories for El Centro and Kobe JMA Ground Motions Adjusted to a Peak Ground Velocity of 50cm/s X - Direction Y - Direction 1.00 1.00 0.75 0.75 0.50 0.50 0.25 0.25 0.00 0.00 0 10 20 3040 5060 0 102030 405060 PG V (cm /s) PG V (cm /s) El Centro Kobe JMA El Centro Kobe JMA Fig.5. The Ar/Ai - PGV Relation for Both X and Y Reference Axis of the System Subjected to Three-directional Input Motions Adjusted to Various Intensity Levels In Figure 6 are shown displacement-shear force significant difference can be seen between them, coefficient relations in Y-direction, for one-directional although they correspond to one- and two-directional (Y) (a), bi-directional (XY) (b) and three-directional earthquake excitations. In case of three-directional (XYZ2) (c) Kobe JMA input motion, adjusted to a peak excitation (Fig. 6(c)), a bit higher response acceleration ground velocity of 50cm/s. Comparing first the compared to one- and two-directional excitation case hysteresis loops shown in Figures 6(a) and 6(b), no has produced a slightly “fatter” hysteresis loop, 20 JAABE vol.1 no.2 November 2002 Bujar Myslimaj Ar/Ai Acc. ( gal ) Acc. ( gal ) Ar/Ai Acc. ( gal ) Acc. ( gal ) demonstrating thus the way the vertical component of ity could affect the displacement response of the FPS, input motion is expected to affect the restoring force the ratio PDHS/PDLS, where PDHS and PDLS characteristics of newly-developed FPS for houses. represent respectively the Peak Displacement of the Effect of mass-eccentricity Heavier Side and the Peak Displacement of the Lighter Based on the experimental setup shown previously in Side, is plotted against mass eccentricity ratio (Figure Figure 3, series of tests were conducted with the 7). From both graphs shown in Figure 7 it can be purpose of investigating the effect of mass eccentricity noticed that, regardless of the input ground motion, the on the response of FPS. Three patterns of weight PDHS/PDLS increases as the mass eccentricity ratio distribution into the system (Figure 3) correspond to increases. On average, for the mass eccentricity ratio of mass eccentricity ratios of 0, 0.094 and 0.188, 0.188 the PDHS/PDLS results to be 1.15. It can be respectively (Inoue et al., 1999). It should be mentioned further noticed that the PDHS/PDLS ratio is almost here that these mass eccentricity ratios represent the unaffected by the input motion intensity level. Since the system’s mass eccentricity in reference to the center of FPS significantly reduces the acceleration response of geometry C, normalized by L/2 (see Figure 2). During the superstructure, as shown in previous sections, the these tests, El Centro 1940 and Kobe JMA input expected stress increase due to mass-eccentricity is motions were adjusted to the intensity levels of PGV expected to be negligible from the engineering practice 25cm/s and 50cm/s. point of view. In order to see at what extension the mass eccentric- 0.12 ( a ) 0.06 -0.06 -0.12 -100 -50 0 50 100 Displacement ( mm ) 0.12 ( b ) 0.06 -0.06 -0.12 -100 -50 0 50 100 Displacement ( mm ) 0.12 ( c ) 0.06 -0.06 -0.12 -100 -50 0 50 100 Displacement ( mm ) Fig.6. Displacement-Shear Force Coefficient Relation in Y-direction, for One- (a), Two- (b) and Three-directional (c) Kobe Input Motion Adjusted to a PGV of 50cm/s JAABE vol.1 no.2 November 2002 Bujar Myslimaj 21 Shear force coefficient Shear force coefficient Shear force coefficient Analytical Simulation of the Recorded Response response of the system. While 3-dimensional shaking table tests provided a In this study, a three dimensional nonlinear model is very necessary information for understanding the real used to simulate the experimentally recorded dynamic behavior and performance of the newly acceleration and displacement response time histories. developed base isolation system, in the design practice As shown in Figure 8, it consists of a rigid slab a reliable analytical tool is needed to predict the supported by six three-dimensional springs along the PGV = 50cm/s PGV = 25cm/s 1.50 1.50 1.00 1.00 0.50 0.50 0.00 0.00 0.00 0.05 0.10 0.15 0.20 0.00 0.05 0.10 0.15 0.20 Mass eccentricity ratio Mass eccentricity ratio E l Centro Kobe E l Centro Kobe Fig.7. PDHS/PDLS versus Mass Eccentricity Ratio CM 2775 2775 Fig.8. 3-dimensional Nonlinear Model Used in Analytical Simulation perimeter of the slab. Each of the 3-D springs models a W=16/62.7tf, one can easily estimate K as FPS isolator. As observed from cyclic loading tests on W/R=2.7/2, i.e. K =1.35tf/m. This stiffness produces 1/2 1/2 single FPS isolators, the restoring force characteristics an isolated system period T = 2(W/gK ) = 2(R/g) for horizontal springs k and k can be satisfactorily = 2.84sec. In order to assure sufficient enough rigidity x y modeled by a rigid-plastic type bilinear model. The in the system before sliding, the ratio K /K was set at 1 2 high initial stiffness (K ) before sliding occurs can be 1000, that is assuming K =1350tf/m. Finally, using 1 1 estimated as a multiple of post-yield stiffness K , which =0.04 as a typical value for the FPS presented in can be accurately estimated or predicted based on the this study, one can determine the yield shear force in vertical load acting on the isolator (W) and the radius of each isolator at sliding as Q = W=0.108tf. It should curvature R of the spherical surface. be noted here that Q = 0.108tf corresponds only to the Assuming that the average vertical load on each case of the balanced weight distribution in the system. isolator for the case of balanced weight (see Figure 3) is For the other cases of weight distribution shown in 22 JAABE vol.1 no.2 November 2002 Bujar Myslimaj PDHS / PDLS PDHS / PDLS Figure 3, Q is not any longer the same for each isolator. eccentric weight distribution arranged according to the First, the EW component of El Centro adjusted to a case 2 (see Figure 3). As it was mentioned above, for PGV of 50cm/s was used as input motion along the this case Q for the pair of isolators on the heavier side X-axis. The acceleration and displacement response at is larger than that one of the pairs on the lighter side. CM in the X direction were then analytically simulated This was taken into account in the modeling process by and compared with those recorded experimentally assigning different Q to each isolator based on its during the shake table tests. These results are shown in location, i.e. vertical load share. Here, the focus was put Figure 9, where a very good agreement between the on the acceleration and displacement response at the analysis and experiment can be noticed. heavier and the lighter side of the slab. These results are In order to demonstrate further the simulation shown in Figure 10. Here also, the satisfactory capabilities of the 3-D model, the NS component of El agreement between the analysis and experiment can be Centro adjusted to a PGV of 50cm/s was used as input clearly noticed. motion along the Y-axis of the system having an 100 100 50 50 0 0 0 5 10 15 05 10 15 -50 -50 -100 -100 Time ( s ) Time ( s ) Analysis Experiment Analysis Experiment Fig.9. Compared Acceleration and Displacement Time Histories at CM Heavier side 150 150 0 5 10 15 -75 -75 0 5 10 15 -150 -150 Time ( s ) Time ( s ) Analysis Experiment Analysis Experiment Lighter side 75 75 0 0 0 5 10 15 -75 05 10 15 -75 -150 -150 Time ( s ) Time ( s ) Analysis Experiment Analysis Experiment Fig.10. Compared Acceleration and Displacement Time Histories at Both Sides of a Mass-eccentric System JAABE vol.1 no.2 November 2002 Bujar Myslimaj 23 Acc ( gal ) Acc ( gal ) Acc ( gal ) Dis ( mm ) Dis ( mm ) Dis ( mm ) Summaries of Technical Papers of Annual Meeting of Architectural Conclusions Institute of Japan; Vol. B-2: 745-746 (in Japanese). Three-dimensional shaking table tests on a newly 5) Public Work Research Institute (1997) Large-scale developed Friction Pendulum System (FPS) for houses three-dimensional shaking table. Public Work Research Institute, were conducted. These tests were primarily designed to Ministry of Construction, Japan. investigate the effectiveness of the FPS; the effect of 6) Yamanouchi H., Midorikawa M. and Iiba M. (1997) Research and development on base-isolation technologies for houses - A survey on bi-directional input ground motion; the effect of vertical the performance based evaluation of base-isolated houses and ground motion and the effect of mass eccentricity. In shaking table tests on model houses. Kenchiku Kenkyu Shiryo, No. addition to the tests, a three-dimensional nonlinear 89, (in Japanese). model was conceived and elaborated in order to 7) The Building Center of Japan (1995) Sheet of outlines for simulate the experimentally recorded acceleration and base-isolated buildings. The Building Letter, June-1995, 57-58 (in displacement response. The most relevant observations Japanese). made in the previous sections are summarized as follows: 1. The FPS reduced significantly the acceleration response at both moderate and strong levels of input ground motion. Depending on the input motion intensity, the ratio of peak response acceleration to peak input acceleration calculated from the experi- mental data resulted to be in the range of 0.2-0.5. 2. No significant difference could be observed between the results obtained by one- and two-directional earthquake excitations. The effect of vertical component of ground motion on the characteristics of FPS appears to be negligible. 3. Tests results indicated that for a mass eccentricity ratio of nearly 20%, the ratio PDHS/PDLS could go up to 1.15. 4. The analytical model conceived and elaborated in this study satisfactorily predicted the experimentally recorded acceleration and displacement response time histories. Acknowledgments The authors would like to express their sincere thanks to all the members of the committee on “Development of seismically isolated houses”, the chairman of which is Mr. Shoichi Yamaguchi, president of Tokyo Kenchiku Structural Engineers, for providing data and valuable suggestions related to this study. The assistance of Dr. Keiichi Tamura and Mr. Takuo Azuma, respectively Head and Research Engineer of the Ground Vibration Division at the Earthquake Disaster Prevention Research Center of Public Work Research Institute in Tsukuba, in operating the large-scale three-dimensional shaking table is greatly appreciated. References 1) Building Research Institute (1996) A survey report for building damages due to the 1995 Hyogo-ken Nanbu Earthquake. Building Research Institute, Ministry of Construction, Japan. 2) Iiba, M., Yamanouchi, H., Midorikawa, M., et al. (1998) Research on performance of base isolated houses. Proceedings of the Second World Conference on Structural Control (2WCSC), Vol. 2, 1119-1126. 3) Iiba, M., Midorikawa, M., Yamanouchi, H., et al. (2000) Shaking table tests on performance of isolators for houses subjected to three dimensional earthquake motions. Proceedings of the 12-th World Conference on Earthquake Engineering. Paper No. 1765. 4) Inoue, K., Iiba, M., Myslimaj, B., et al. (1999) Three dimensional shaking table tests on seismic behavior of isolators for houses - Part 3: Effect of unbalanced weight on characteristics of isolators. In: 24 JAABE vol.1 no.2 November 2002 Bujar Myslimaj
Journal
Journal of Asian Architecture and Building Engineering
– Taylor & Francis
Published: Nov 1, 2002
Keywords: seismic isolation; friction pendulum system; houses; 3-D shaking table tests; numerical simulation
