EXPERIMENTAL STUDY ON THE SEISMIC RESPONSE OF BRIDGE COLUMNS USING E-DEFENSE. Abstract. Introduction

Transcription

1 EXPERIMENTAL STUDY ON THE SEISMIC RESPONSE OF BRIDGE COLUMNS USING E-DEFENSE Kazhiko Kawashima 1, Tomohiro Sasaki 2, and Koichi Kajiwara 3 Abstract This paper presents preliminary reslts of a large scale shake table experiment for stdying the failre mechanism of three reinforced concrete bridge colmns; a typical flexral dominant colmn in the 197s (C1-1), a typical shear failre dominant colmn in the 197s (C1-2) and a typical colmn designed in accordance with the crrent design code (C1-5). They were 7.5 m tall m diameter circlar reinforced concrete colmns. They were sbjected to a near-filed gronds motion recorded dring the 1995 Kobe, Japan earthqake. Preliminary reslts on the experiment and analytical correlation are presented. Introdction Bridges are a vital component of transportation facilities; however it is known that bridges are vlnerable to the seismic effect. Bridges sffered extensive damage in past earthqakes sch as 1989 Loma Prieta earthqake, 1994 Northridge earthqake, 1995 Kobe earthqake, 1999 Chi Chi earthqake, 1999 Bol earthqake and 28 Wenchan earthqake. A large scale bridge experimental program was initiated in 25 in the National Research Institte for Earth Science and Disaster Prevention (NIED), Japan as one of the three US-Japan cooperative research programs based on NEES and E-Defense collaboration. In the bridge program, it was originally proposed to condct experiments on two model types; 1) component models and 2) system models. They are called hereinafter as C1 experiment and C2 experiment, respectively (Nakashima 28). The objective of the C1 experiment is to clarify the failre mechanism of reinforced concrete colmns sing models with as large section as possible. On the other hand, C2 experiment was proposed to clarify the system failre mechanism of a bridge consisting of decks, colmns, abtments, bearings, expansion joints and nseating prevention devices. C1 experiment was condcted for two typical reinforced concrete colmns which failed dring the 1995 Kobe earthqake (C1-1 and C1-2 experiments) and a typical reinforced concrete colmn designed in accordance with the crrent design reqirements (C1-5 experiment). This paper shows preliminary reslts of the experiment and analysis on three C1 colmns. 1 Professor, Department of Civil Engineering, Tokyo Institte of Technology 2 Gradate Stdent, Department of Civil Engineering, Tokyo Institte of Technology 3 Senior Researcher, National Research Institte for Earth Science and Disaster Prevention

2 Photo 1 C1 on E-Defense 15@15=22513@3=39 15@15=225 28@3= (a) C1-1 D29@32bars D29@32bars D29@16bars D13@3mm @15=22513@3=39 15@15=225 16@3=48 1@3= D29@32bars D29@32bars D29@16bars 1 D13@3mm @15=84 27@3= (b) C1-2 (c) C1-5 Fig. 1 C1 colmn models D35@36bars D35@36bars D22@15mm Colmn Models Photo 1 shows the experimental setp of three colmns sing E-Defense (Kawashima et al. 29). Two simply spported decks were set on the colmn and on the two steel end spports. A catch frame was set nder the lateral beam of the colmn to prevent collapse of the colmn when it was excessively damaged. Tribtary mass to the

3 colmn by two decks inclding for weights was 37 t and 215 t in the longitdinal and transverse directions, respectively. The tribtary mass was increased by 21 % from 37 t to 372 t in a part of C1-5 excitation. Three fll-size reinforced concrete colmns as shown in Fig. 1 were constrcted for the experiment. Colmns sed for C1-1, C1-2 and C1-5 experiments, which are called hereinafter as C1-1, C1-2 and C1-5, respectively, are 7.5 m tall reinforced concrete colmns with a diameter of 1.8 m in C1-1 and C1-2 and 2 m in C1-5. C1-1 and C1-2 are typical colmns which were bilt in the 197s based on a combination of the static lateral force method and the working stress design in accordance with the 1964 Design Specifications of Steel Road Bridges, Japan Road Association. Since it was a common practice prior to 198 to terminate longitdinal bars at mid-heights, the inner and center longitdinal bars were ct off at 1.86 m and 3.86 m from the colmn base, respectively. The ct-off heights were determined by extending a length eqivalent to a lap splicing length l ls (abot 3 times bar diameter) from the height where longitdinal bars became nnecessary based on the moment distribtion. On the other hand, longitdinal bars were not ct-off in C1-1. C1-1 and C1-2 had the same shape, heights, bar arrangement and properties except the ct-off. As a conseqence, C1-1 failed in flexre while C1-2 failed in shear, as will be described later. The shear failre de to ct-off was one of the major sorces of the extensive damage of bridges in the 1995 Kobe earthqake (Kawashima and Unjoh 1997). Combination of the lateral seismic coefficient of.23 and the vertical seismic coefficient of +/-.11 (pward and downward seismic force) was assmed in the design of C1-1 and C1-2. Deformed 13 mm diameter circlar ties were provided at 3 mm interval, except the oter ties at the top 1.15m zone and the base.95 m zone where they were provided at 15mm interval in C1-1. Ties were only lap spliced with 3 times the bar diameter. Lap splice was a common practice by the mid 198s. The longitdinal and tie bars had a nominal strength of 345 MPa (SD345), and the design concrete strength was 27 MPa. The longitdinal reinforcement ratio P l was 2.2 % and the volmetric tie reinforcement ratio ρ s was.32 % except the top 1.15 m and base.95m zones where ρ s was.42% in C1-1. P l and ρ s varied depending on the zones in C1-2; 2.2 % and.42 % at the base.95 m zone, 2.2 % and.32 % between.95 m and 1.86 m, 1.62 % and.21 % between 1.86 m and 3.86 m,.81 % and.11 % between 3.86 m and 4.85 m, and.81 % and.21 % at the top 1.15 m zone, respectively. On the other hand C1-5 was designed in accordance with the 22 JRA Design Specifications of Highway Bridges (JRA 22). Sixty for deformed 35mm diameter longitdinal bars were provided in two layers. Deformed 22 mm diameter circlar ties were set at 15 mm and 3 mm interval in the oter and inner longitdinal bars, respectively. The ties were developed in the core concrete sing 135 degree bent hooks after lap spliced with 4 times the bar diameter. The nominal strength of longitdinal and tie bars and the design concrete strength were the same with those in C1-1 and C1-2

4 colmns. The longitdinal reinforcement ratio P l was 2.19 % and the volmetric tie reinforcement ratio ρ was.92 % s Evalation of the seismic performance of C1-1 and C1-5 in the longitdinal direction based on the 22 JRA code is as follows: Becase the design response acceleration S A is m/s 2 for both C1-1 and C1-5, the yield displacement y and ltimate displacement are.46 m and.99 m in C1-1 and.45 m and.231 m in C1-5. The design displacement d is evalated from y and as d y = y + (1) α in which α depends on the type of grond motion (near-field or middle field grond motion) and the importance of the bridge. Assming α is 1.5 for a combination of the near-field grond motion category and the important bridges category, the design displacement d is.81 m in C1-1 and.169 m in C1-5. On the other hand, the displacement demand is.328 m in C1-1 and.168 m in C1-5 becase the force redction factor is 1.58 and 2.56 respectively. Conseqently, C1-1 and C1-5 were evalated to be nsafe and safe, respectively based on the crrent design code. Acceleration (m/sec 2 ) Acceleration (m/sec 2 ) Acceleration (m/sec 2 ) (a) Longitdinal (b) Transverse Table Motion Target Table Motion Target 5 Table Motion -5 Target (c) Vertical Fig. 2 1% E-Takatori grond motion (C1-5(1)-1 excitation)

5 Three colmns were excited sing a near-field grond motion as shown in Fig. 2 which was recorded at the JR Takatori Station dring the 1995 Kobe earthqake. It was one of the most inflential grond motions to strctres. However dration was short. Taking accont of the soil strctre interaction, a grond motion with 8% the original intensity of JR Takatori record was imposed as a command to the table in the experiment. This grond motion is called hereinafter as the 1 % E-Takatori grond motion. Excitation was repeated to clarify the seismic performance of the colmns when they were sbjected to near-field grond motions with longer dration and/or stronger intensity. Only C1-5 was excited sing 125 % E-Takatori grond motion with 21 % increased deck mass to stdy the seismic performance nder a stronger grond motion than the JR-Takatori Station grond motion. Progress of failre Seismic Performance of C1-1 and C1-5 C1-1 was sbjected to the 1 % E-Takatori grond motion twice. Photo 2 shows the progress of failre at the plastic hinge on the SW srface where damage was most extensive. NS and EW direction correspond to the transverse and longitdinal directions, respectively, of the model. Dring the first excitation (C1-1-1 excitation), at least two oter longitdinal bars from S to W locally bckled between the ties at 2 mm and 5 mm from the base. Dring the second excitation (C1-1-2 excitation), both the covering and core concrete sffered extensive damage between the base and.7 m from the base on the SW srface. Three ties from the base completely separated at the lap splices. Eleven oter and three center longitdinal bars locally bckled between ties at 5 mm and 5 mm from the base. (a) C1-1-1 excitation (8.35s) Photo 2 Progress of damage of C1-1 (b) C1-1-2 excitation (7.71s) On the other hand, C1-5 was sbjected to the 1% E-Takatori grond motion twice (C1-5(1)-1 and C1-5(1)-2 excitations). After the mass was increased by 21 % from 37 t to 372 t, C1-5 was sbjected to the 1% E-Takatori grond motion once (C1-5(2)

6 excitation). Then C1-5 was sbjected to the 125% E-Takatori grond motion twice (C1-5(3)-1 and C1-5(3)-2 excitations). Photo 3 shows the progress of failre of C1-5 at the plastic hinge dring C1-5(1)-1, C1-5(2) and C1-5(3)-2 excitations. Dring C1-5(1)-1 excitation, only a few flexral cracks with the maximm width of 1mm occrred arond the colmn at the plastic hinge. Therefore it is noted that the seismic performance is enhanced in C1-5 than C1-1 nder the first 1% E-Takatori excitation. The damage progressed dring C1-5(2) excitation sch that the covering concrete spalled off at the 5 mm base zone from WSW to SSW. Dring C1-5(3)-2 excitation, the failre extensively progressed. The core concrete crashed de to repeated compression, and blocks of crashed core concrete spilled ot from the steel cages like explosion. Sch a failre was never seen in the past qasi-static cyclic or hybrid loading experiments. Becase the maximm aggregate size was 2 mm, the concrete blocks after crashed can be as small as 2-4 mm. Becase the gaps of longitdinal bars and circlar ties were 132 mm and 128 mm, respectively, it was possible for the blocks of crashed core concrete to move ot from the steel cages. Frthermore twelve oter longitdinal bars and nineteen inner longitdinal bars locally bckled on SW and NE-E srfaces. The 135 degree bent hooks developed in the core concrete still existed in the original position althogh the core concrete arond the hooks sffered extensive damage. (a) C1-5(2) excitation (8.8s) Photo 3 Progress of damage of C1-5 Response displacement and moment capacity (b) C1-5(3)-2 excitation (7.17s) Figs. 3 and 4 show the response displacement at the top of C1-1 and C1-5, respectively, in the principal response direction (nearly SW-NE direction). The peak displacement of C1-1 was.179 m (2.4 % drift) dring C1-1-1 excitation while the peak displacement of C1-5 was.84 m (1.1 % drift) dring C1-5(1)-1 excitation. Becase the ltimate displacement in accordance with JRA 22 code was.1 m and.235 m in C1-1 and C1-5, respectively, the above peak response displacements corresponded to 179 % and 36 % the ltimate displacement in C1-1 and C1-5, respectively.

7 Figs. 5 and 6 show the moment at the colmn base vs. lateral displacement at the colmn top hystereses of C1-1 and C1-5, respectively, in the principal response direction. The compted moment vs. lateral displacement relations based on the 22 JRA code are also shown here for comparison. The moment capacity of C1-1 dring C1-1-2 excitation was MNm which deteriorated by 19 % from the moment capacity dring C1-1-1 excitation of MNm. On the other hand, the moment capacity of C1-5 colmn progressed from MNm dring C1-5(1)-1 excitation to 2.14 MNm and MNm dring the C1-5(2) and C1-5(3)-2 excitations, respectively. However since the moment capacity of C1-5 dring the C1-5(3)-1 excitation was MNm, the moment capacity of C1-5 deteriorated by 3% dring C1-5(3)-2 excitation. The compted moment capacities are close to the experimental vales in both C1-1 and C1-5, however the compted ltimate displacement are very conservative compared to the experiment. Displacement (m).3 C C Fig. 3 Response displacement at the top of C1-1 in the principle response direction Displacement (m).6 C1-5(1)-1.5 C1-5(2) C1-5(3) Fig. 4 Response displacement at the top of C1-5 in the principle response direction Moment at the base (MNm) y y d C1-1-1 C1-1-2 JRA Lateral Displacement (m) Moment at the base (MNm) 2 1 y y d C1-5(1)-1-1 C1-5(2) C1-5(3)-2-2 JRA Lateral Displacement (m) Fig. 5 Moment at the base vs. lateral Fig. 6 Moment at the base vs. lateral displacement displacement of the top of the top hysteresis of C1-5 in the principle hysteresis of C1-1 in the response direction principle response direction

8 (a) NW (1) 6.5s (b) SE (a) NW (b) SE (2) 6.87s Photo 4 Progress of damage of C1-2 Seismic Performance of C1-2 Progress of frailre Photo 4 shows the progress of failre of C1-2 on NW and SE srfaces. A horizontal crack first developed at 4.1s along NW to E srface, and it progressed to a shear crack at 4.33 s. Another horizontal crack developed at 4.6s along W to SE srface, and it extended to at least two diagonal cracks at 4.87s. Among two diagonal cracks developed at 4.33 s, a crack on NW srface extended to W srface, and the other crack on SE srface extended to S at 5.37s. The core concrete started to crash de to shear, and the blocks of crashed core concrete started to move ot from the inside of the colmn near the pper ct-off on N and NW srfaces at 6.4 s. The same bt more extensive failre occrred on S and SW srfaces at 6.54 s. The blocks of crashed core concrete progressively moved ot from steel cages associated with the colmn response in the SW direction. At 6.87 s, the bottom of lateral beam hit with the pper srface of catch frame de

9 to excessive response displacement. Three circlar tie bars completely separated at their lap splice and the longitdinal bars deformed in the otward direction. Extensive failre of core concrete and deformation of longitdinal bars progressed on W, NW, N, NE and E srfaces. It shold be noted in the above process that the failre of core concrete was extensive and a large nmbers of blocks of crashed core concrete as well as deformed longitdinal bars moved ot from inside of the colmn dring very short time (less than 3 s). It was like an explosion. Response and shear capacity Fig. 7 shows response displacement of C1-2 in the principal response direction. As described above, since bottom of the lateral beam hit with the pper srface of catch frame at 6.87 s, the colmn response after 6.87 s was affected by this contact. Withot the catch frame, the colmn possibly overtrned. Therefore the response displacement after this contact is plotted by dotted line in Fig. 7. At s, right after the contact, the colmn response displacement reached its peak of mm and 253. mm in the longitdinal and transverse directions, respectively. Residal drifts of 24.5 mm and mm were developed after the excitation. Displacement (mm).5.4 f.3.2 c.1 a e -.1 b d Fig. 7 Response displacement at the top C1-2 in the principle response direction Fig. 8 shows the lateral force at the pper ct-off vs. lateral displacement at the colmn top hysteresis in the principal response direction. The hysteresis after the contact of the colmn with the catch frame is plotted by dotted line. The shear capacity of the colmn F was evalated based on the trss theory (Priestly 1996). s The shear stress at the pper ct-off vs. the lateral displacement at the colmn top relation was evalated as shown in Fig. 9, in which τ c is normalized in terms of α c and α defined as pl

10 3 24 1/ 1/ 3.12 α c = ; α pl = f (2) c pl In Fig. 9, shear stress evalated for two 1.68m tall 4mm diameter scaled model colmns with different shear vs. flexre strength ratio is inclded for comparison (Sasaki et al. 28). τ c / α cα pl of C1-2 is.68 MPa which is 15% larger than the vale evalated based on the shear eqation (.59 MPa) by Kono et al (Kono et al. 1996). Lateral Force (kn) Lateral Displacement (m) Fig. 8 Lateral force at pper ct-off vs. lateral displacement at the colmn top hysteresis in the principle response direction τ c / α c α pl (MPa) C1-2 Scale model (γ=.84) Scale model (γ=.74) Lateral Displacement (m) Fig. 9 Shear stress of concrete Analytical Correlation for C1-5 Analytical correlation for C1-5 dring C1-5(2) and C1-5(3)-2 excitations is shown here. The colmn was idealized by a 3D discrete analytical model inclding P - effect as shown in Fig. 1. The colmn was idealized by fiber elements. A section was divided into 4 fibers. Fig. 1 Analytical model The stress vs. strain constittive model of confined concrete is assmed based on

11 Hoshikma et al (1997) and Sakai and Kawashima (26). The Modified Menegotto-Pinto model was sed to idealize the stress vs. strain relation of longitdinal bars (Menegotto and Pinto 1973, Sakai and Kawashima 23). Fig. 11 shows the analytical correlation on the response displacements at the top of the colmn in the principal direction dring C1-5(1)-1 and C1-5(2) excitations. Fig. 12 compares the measred and compted moment at the base vs. lateral displacement at the colmn top hystereses dring the two excitations. Becase nonlinear hysteretic response was still limited dring C1-5(1)-1 excitation, the compted response displacement and moment vs. lateral displacement hysteresis are qite in good agreement with the experimental reslts, however as C1-5 sffered more damage, the accracy of analytical prediction decreases. Displacement (m) Displacement (m) Analysis Experiment (a) Displacement Acceleration (m/sec 2 ) (1) C1-5(1)-1 excitation Analysis Experiment (a) Displacement Acceleration (m/sec 2 ) (2) C1-5(1)-2 excitation Analysis Experiment (b) Acceleration Analysis Experiment (b) Acceleration Fig. 11 Analytical correlation for the response displacement and acceleration at the colmn top in the principle response direction Conseqently, it is reqired to develop an analytical model that can predict the response of the colmns ntil collapse for realizing reliable performance based seismic design.

12 Moment at the base (MNm) 2 1 y y d -1 Analysis -2 Experiment JRA Lateral Displacement (m) (a) C1-5(1)-1 excitation Moment at the base (MNm) 2 1 y y d -1 Analysis -2 Experiment JRA Lateral Displacement (m) (b) C1-5(2) excitation Fig. 12 Moment vs. lateral displacement at the colmn top in the principle response direction Conclsions A preliminary reslt on a series of shake table experiment and analysis to three fll-size reinforced concrete colmns was presented. Based on the reslts presented herein, the following tentative conclsions may be dedced; 1) C1-1 which is a typical colmn in the 197s sffered extensive damage nder C1-1-1 excitation. The progress of damage dring C1-1-2 excitation was extensive even thogh it was anticipated before the experiment that damage wold not progress nless the intensity of second excitation was mch larger than that of the first excitation. This reslted from the extensive deterioration of the lateral confinement de to separation of ties at the lap splices. It is highly possible that colmns withot sfficient lateral confinement have a similar progress of damage dring a long-dration near-field grond motion or strong aftershocks. 2) C1-5 which is a typical colmn in accordance with the crrent design criteria sffered only a few nmbers of horizontal cracks with the maximm width of 1 mm nder C1-5 (1)-1 excitation. The ltimate drift was 2.9 % which was 2.2 times larger than that of C1-1. Conseqently, enhancement of the seismic performance of C1-5 compared to C1-1 is obvios. However the progress of failre of C1-5 was extensive when it was sbjected to 25 % stronger excitation nder 21% added mass (C1-5(3) excitations). Blocks of crashed core concrete spilled ot like explosion from the steel cages. The seismic performance of C1-5 sbjected to longer dration near-field grond motion has to be careflly evalated. 3) C1-2 failed in shear at the pper ct-off. As soon as circlar ties at the pper ct-off yielded, a small diagonal cracks developed. As they extended to several major diagonal cracks, C1-2 completely failed in shear within less than 2.5 s since the initiation of a cople of small diagonal cracks. Concrete blocks crashed by shear and deformed

13 longitdinal bars extensively moved ot from the inside of colmn. 4) The lateral confinement in the flexre dominant colmns is not niform arond the ties as it is crrently assmed in design. More importantly, the lateral confinement of mlti layered ties is very complex. Strains of ties are not similar among the mlti-layered ties, and they are related to the degree of constraint exerted for preventing local bckling of longitdinal bars. Strains are generally larger in the oter ties than the inner ties. This implies that the lateral confinement by Eq. (2) can be overestimated. 5) Compted response for the flexre dominant colmns is satisfactory while response ndergoes the moderate nonlinear range, however accracy of the analytical prediction deteriorates once the colmns ndergo the strong nonlinear range. An analytical model which can predict response of the colmns ntil failre shold be developed for enhancing the reliability of the performance based seismic design. Acknowledgments The C1 experiment was condcted based on the extensive spport of over 7 personnel in the Overview Committee (Chair, Professor Emerits Hirokaz Iemra, Kyoto University), Execting Committee of Large-scale Bridge Experimental Program (Chair, Professor Kazhiko Kawashima, Tokyo Institte of Technology), Analytical Correlation WG (Chair, Dr. Shigeki Unjoh, Pblic Works Research Institte), Measrements WG (Chair, Professor Yoshikaz Takahashi, Kyoto University), Dampers & Bearings WG (Chair, Dr. Masaaki Yabe, Chodai) and Blind Analysis WG (Chair, Professor Hiroshi Mtsyoshi, Saitama University). Their strong spport is greatly appreciated. Invalable spport and encoragement of Professor Stephen Mahin, University of California, Berkeley and Professor Ian Bckle, University of Nevada, Reno are greatly appreciated. References Nakashima, M., Kawashima, K., Ukon, H. and Kajiwara, K. (28), Shake table experimental project on the seismic performance of bridges sing E-Defense, S (CD-ROM), 14 WCEE, Beijing, China. Hoshikma, J., Kawashima, K., Nagaya, K. and Taylor, A.W. (1997), Stress-strain model for confined reinforced concrete in bridge piers, Jornal of Strctral Engineering, ASCE, 123(5), Kawashima, K. Sasaki, T., Kajiwara, K., Ukon, H., Unjoh, S., Sakai, J., Takahashi, Y., Kosa, K., and Yabe, M. (29), Seismic performance of a flexral failre type reinforced concrete bridge colmn based on E-Defense excitation, Proc. JSCE, A, Invited paper, 1-19, JSCE. Kawashima, K. and Unjoh, S. (1997), The damage of highway bridges in the 1995

14 Hyogo-ken nanb earthqake and its impact on Japanese seismic design, Jornal of Earthqake Engineering, 1(3), Kono, H., Watanabe, H. and Kikmori, Y. (1996), Shear strength of large RC beams, Technical Note, 3426, Pblic Works Research Institte, Tskba, Japan. Priestley, M.N.J., Seible, F. and Calvi, G.M. (1996), Seismic design and retrofit of bridges, John Wiley & Sons, New York, USA. Menegotto, M. and Pinto, P.E. (1973), Method of analysis for cyclically loaded R.C. plane frames inclding changes in geometry and non-elastic behavior of elements nder combined normalized force and bending, Proc. IABSE Symposim on Resistance and Ultimate Deformability of Strctres Acted on by Well Defined Repeated Loadings, Sakai, J. and Kawashima, K. (23), Modification of the Giffre, Menegotto and Pinto model for nloading and reloading paths with small strain variations, Jornal of Strctral Mechanics and Earthqake Engineering, JSCE, 738/I-64, Sakai, J. and Kawashima, K. (26), Unloading and reloading stress-strain model for confined concrete, Jornal of Strctral Engineering, ASCE, 132 (1), Sasaki, T., Krita, H., and Kawashima,K. (28), Seismic performance of RC bridge colmns with termination of main reinforcement with inadaqate development, S (CD-ROM), 14WCEE, Beijing, China.