PSEUDO-DYNAMIC TEST OF FULL-SCALE RCS FRAME:

PART 1 – DESIGN, CONSTRUCTION AND TESTING

Chui-Hsin Chen1, Wen-Chi Lai2, Paul Cordova3, Gregory G. Deierlein4 and Keh-Chyuan Tsai5

SUMMARY
This is the first in a two-part paper describing a pseudo-dynamic test of a full-scale three-story
three-bay RCS frame. This paper focuses on the design, construction, and experimental testing of
the frame; the second companion paper describes the validation simulation models and resulting
design implications. Based on design experience, composite RCS moment frames, consisting of
steel beams and reinforced concrete columns, can provide efficient and economical alternatives to
traditional steel or reinforced concrete construction. Apart from the economies achieved by
effective use of materials, research shows the viability of RCS beam-column connections to
provide strength and ductility exceeding that in conventional steel or reinforced concrete moment
frames. The full-scale RCS frame test was undertaken as a capstone project to validate the system
performance of composite frames designed according to current seismic code provisions. The test
offers additional opportunities to examine innovative pre-cast construction methods and to
validate performance simulation models. The frame was loaded pseudo-dynamically to simulate
the structural response under ground motions corresponding to earthquake hazards for a
high-seismicity site with 50%, 10%, and 2% chance of exceedence in 50 years. Following the
pseudo-dynamic tests, the frame was subjected to a monotonic push with inter-story drift ratios up
to 0.10. Ground motions scaled to the 10% in 50 year earthquake hazard caused peak interstory
drift ratios up to 0.025, accompanied by steel beam yielding and concrete spalling and plastic
hinging at the RC column bases. The 2% in 50 year earthquake hazard was more damaging with
peak interstory drift ratios of up to 0.055 accompanied by local flange/web buckling in the beam
hinge regions and significant spalling and cracking of the concrete columns. Throughout the
loading the composite beam-column connections exhibited only minor damage.

Keywords: pseudo-dynamic tests, reinforced concrete columns with steel beams (RCS), full-scale
tests, pre-cast construction methods

INTRODUCTION

The use of the RCS system connecting steel beams to RC columns in Taiwan is only limited to a few low-rise
warehouse buildings while the use of this system in the US has been limited to high-rise buildings in low seismic
regions. A number of tests have been conducted in the US-Japan Cooperative Research Program on Composite
and Hybrid Structures during the past few years. It has concluded that well designed and constructed steel
beam-to-RC column joints possess excellent strength and ductility characteristics for seismic applications.
However, information such as the system constructability and performance of a RCS frame building are rather
limited. Therefore, a large-scale test is launched in the National Center for Research on Earthquake Engineering
(NCREE) by the researchers from US, Japan and Taiwan. It is believed that the large-scale test will well provide
a proof of concept for this structural system.

The large-scale, 3-story RCS frame shown in Fig. 1 and Photo 1 is employed in this experimental research. The
typical bay width of 7000mm and typical story height of 4000mm have been found common in Taiwan and US

1
Research Assistant, National Center for Research on Earthquake Engineering, Taiwan, e-mail: [email protected]
2
Research Assistant, National Center for Research on Earthquake Engineering, Taiwan, e-mail: [email protected]
3
Research Assistant, J.A. Blume Earthquake Engrg. Center, Stanford University, e-mail: [email protected]
4
Professor, J.A. Blume Earthquake Engrg. Center, Stanford University, e-mail: [email protected]
5
Professor, National Center for Research on Earthquake Engineering, Taiwan, e-mail: [email protected]

107
building configuration. The 2150 mm wide concrete slab is adopted to develop the composite action of the
beams. Measuring 12 meters tall and 21 meters long, the frame is among the largest frame tests of its type ever
conducted. The three-story prototype structure is designed for a highly seismic location either in Taiwan or
California. Prior to the frame test, a series of subassembly tests have been completed in the structural laboratory
in NCREE and some recommendations have been concluded [Tsai and Chen 2002, Cheng and Cian 2002, Chen
and Lin 2002] for the design of frame specimen. Meanwhile, this experiment also provides the impetus to
explore international collaboration and data archiving envisioned for the Networked Earthquake Engineering
Simulation (NEES) initiatives and the Internet-based Simulations for Earthquake Engineering (ISEE) [Hsieh et
al. 2002, Yang et al. 2002] launched recently in USA and Taiwan, respectively.

DESIGN OF SPECIMEN

Based on a perimeter frame concept, the frame is designed according to several standards, following the Seismic
Force Requirements for Building Structures in Taiwan, IBC 2000, the 1997 AISC Seismic Provisions, the ACI
318-99 and the 1994 ASCE Joint Recommendations. Several types of connection details in RCS system have
been tested in the subassembly tests conducted prior to the large-scale frame test, and those details adopted in the
3-story, 3-bay specimen are determined based on these test results. All steel beams, H396×199×7×11 for roof
level, H500×200×10×16 for second floor and H600×200×11×17mm, are of A572 GR50. All columns are 650
×650 mm square using 42Mpa (6000psi) concrete.

Beam-column joint

The joint panels of an RCS frame may have two kinds of major failure modes [Deierlein 1988]: 1) panel shear
failure and 2) vertical bearing failure. When the joint panel is subjected to unbalanced beam moment, the
deformation of the joint panel immediately adjacent to the steel beam is relatively severe. This region is defined
as the inner element shown in Fig. 2. The outer element, defined as the region outside the inner element, deforms
smaller than the inner element. The failure mechanisms in the inner element are usually the bearing failure of the
concrete and the panel shear yielding of the beam web. These two failure modes are shown in Fig.3. Unlike the
inner element, the failure modes of the outer element shown in Fig.4 are usually the bond failure of the
reinforcing bars and the shear cracks in the joint panel. The shear strength of the joint panel is dependent on both
the strength of the inner element and the outer element [Kanno and Deierlein 2000].

Fig.5 depicts the typical joint details adopted in the full-scale specimen, which include face bearing plates and
steel band plates. Face bearing plates can help develop the concrete strut within the beam top and bottom flanges.
Steel band plates can help confine the concrete above and below the beam and develop the outer concrete struts. In
addition, the stirrups are also required in the joint. Three transverse reinforcing patterns have been investigated in
the subassembly tests. One consists of 4L-shaped ties, second one includes double-U ties and the third one is
4-square ties (see Fig. 6). The experimental results of the subassembly tests illustrate that these patterns make
almost no difference in their seismic performance. Accordingly, the 4-square tie configuration turns out to be
more favorable than the L-shaped and the double-U ties because it is not necessary to drill holes in the beam web.
As for the beam-column joints on the top floor, Fig. 7 depicts the connecting details. The reinforcing bars are
plug-welded to a bearing plate on the top of the column using the details successfully applied in subassembly tests.
The bearing plate helps develop the force in the reinforcing bars. Further, it keeps the column from extending out
of the roof slab while providing sufficient anchorage of the reinforcing bars.

Column

1. Main bars

The amount of main bars in columns varies with the different demand for moment on each floor. Further, the
moment demand in the exterior columns is also different from that in the interior columns. All vertical
reinforcing bars are of Grade 60 (yield strength of 420Mpa). For exterior columns, the main bar requirements are
8#11 in the first story, 4#11 in the second story and the third story. Actually, based on the moment diagram form
the elastic design forces, the 8#11 bars are required within the first 1300mm of the column height in the first
story. The required development length for these bars per ACI-318-99 is 38.73db, which is 1387mm, and thus the
8#11 bars can be dropped off in the height of 2687mm. Before fabrication of the exterior columns, the 8#11 bars
are determined to continue through the entire height of the first story. For the interior columns, the main bar
requirement are 12#11 bars in the first and second stories and 8#11 in the third story. One of the #11 bars
continues through the first and second stories and stops at the height of 2700mm beyond the floor level on the

108
third floor to conform to the requirement of the development length stipulated in ACI-318-99.

2. Splice

Columns are spliced using mechanical sleeves as shown in Fig. 8. Columns are precast with embedded sleeves to
splice the other columns with reinforcing bars extending up. After the two columns are positioned, the sleeves
were fill up with grout.

A typical precast unit is shown in Photo 2. The columns are spliced on the top of slabs except for the column in
the first story. According to the findings in the subassembly tests in which experimental parameter is the location
of the column-to-footing splice, the specimen with column splice in the height of 1000mm above the footing
performs well in dissipating energy. Accordingly, in the large-scale test frame, the locations of the splice between
columns and footings are in the height of 1000mm from the top of the footings as depicted in Fig. 9.

Beam

1. Bolted splice

During the subassembly tests of beam-to-column joints, the beam splice using only bolts evidently slipped. The
slippage of beam splice contributed to the difficulty of data processing because the digital readings of sensors are
affected by the slippage. Therefore, after the discovery of the splice slippage, the bolted beam splices were
strengthened using additional welds to prevent the abrupt variations of the collected data. As for the full-scale
test frame, the slippage of beam splice in local areas did not affect the digital readings so much. All the beams
were spliced using bearing-type bolts without additional weld. However, the large bangs were frequent and
evident during the test of the full scale frame specimen when the frame was undergone significant lateral
deformations.

2. Lateral support

According to the AISC Seismic provisions, the calculated maximum unbraced length is 2880 mm. The
H300x150x6.5x9 transverse beams were designed to separate one beam span into three segments equally in the
entire specimen. The center-to-center beam spacing is 2333mm which satisfies the requirement in AISC Seismic
provisions. However, to improve the bracing for beams in the first story, H396x199x7x11 transverse beams are
employed.

Lateral support system for the entire frame specimen

Two transverse lateral braced plane frames were designed to resist a lateral force of 3% of the peak actuator
forces. Assuming the ratio of the actuator forces is 3:2:1 in the roof, the second and the first floors, respectively.
The corresponding peak lateral forces are 90, 60 and 30 kN, simultaneously applied laterally on the lateral
support frame in the same direction at the third, the second and the first floors respectively at elevations where
the slab edge loading beams occur. The story drift limit of the lateral support is set at 0.001 radian in sizing the
column and struts of the braced frame. To accommodate a maximum story drift of 0.08 radian imposed during
the final push over of the full scale frame specimen, the lateral support system is not located exactly at the center
of the specimen, but closer to the side where actuators were mounted.

CONSTRUCTION OF THE FRAME SPCIMEN

Precast in plant

The test frame was assembled rapidly using 9 steel beams, 12 columns 4 isolated footings. All the precast
members include two categories: the column part (see Photo 2) and the footing part (see Photo 3). The footing
part includes a concrete footing and a column stub. Footings were first poured in the precast plant and on the
second day when the footing molds were removed, one-meter high columns were then poured with main steel
bars extending out of the top column end. As for the typical precast column member, a main girder segment and
two steel transverse beams were cast with the RC column as shown in Photo 2. These precast columns were cast
in the position as it laid down on the working platform. After concreting, some defects were formed near the top
steel band plates in the beam-column joint as depicted in Photo 4. These defects were filled with non-shrink
grout after the removal of the molds.

109
Construction in NCREE

In order to assure an uniform and close contact surface between the bottom of the footings and the strong floor in
the laboratory at NCREE, the footings were positioned with gypsum group underneath. This was also made
between the washer and the top surface of footings before tying down the footing of the specimen onto the strong
floor. The entire specimen was constructed in NCREE within two weeks from the positioning of the precast
footings to the final pouring of concrete slabs. The highly efficient construction performance of the test specimen
clearly demonstrated the excellent constructability of the RCS frame systems.

EXPERIMENTAL PROGRAM

Selection of earthquakes using PISA2D

Extensive nonlinear static and dynamic time-history analyses were performed using OpenSEES [Cordova et al.
2003] and PISA2D [Tsai and Chang 2001] prior to the tests. It was required to select suitable earthquake ground
accelerations and investigate the most probable ultimate lateral strength and the inelastic demands imposed on
the frame specimen under the simulated earthquake effects. The main focus of these studies was to predict the
possible peak responses of the frame during the test while verifying that both the force and stroke limitations of
the actuators were not exceeded.

Since this structure was based on a perimeter frame concept, the seismic mass used in the analytical models was
based on one-half of the dead load of the 3-D building. Using this same logic, a leaning column must also be
included to properly account for the P-Delta effect of the interior gravity columns. The load on this leaning
column was based on recommendations from FEMA356 considering the following provision, 1.1DL+0.275LL.
The fundamental period of the PISA2D frame model is 1.04 second. The material properties in the analytical
models were all based on those measured from material tests of the test frame. Nonlinear static pushover
analyses were also performed cyclically using PISA2D by increasing the story drift ratio gradually till 4% rad
followed by pushing the structure to a drift ratio of 8% rad (see Fig. 10). Note that the composite beams and
joints were not considered in the PISA2D model. The bi-linear beam-column model was adopted for the RC
column members while the two-surface plastic beam model was applied for the steel beam members. The
analytical results obtained from PISA2D show that the RCS frame performs adequately in dissipating energy,
demonstrating that illustrates that most of the plastic hinges are concentrated in the beams and at the column
bases (Fig. 11).

In the selection of the ground accelerations, more than 10 records of Chi-Chi earthquake were considered as
input ground motions. Based on the uniformity of the inter-story drift demands as well as the minimal residual
drift, only one out of these records was chosen as the most suitable for the experiment. Similarly, the third author
of this paper considered ground accelerations recorded during the 1994 Northridge and 1989 Loma Prieta
earthquakes [Cordova et al. 2003]. The test specimen were to be subjected to increasingly larger seismic
demands that represent incipient damage level earthquakes (50% chance of exceedance in 50 years), design level
earthquakes (10% chance of exceedance in 50 years) and the maximum considered level earthquake (2% chance
of exceedance in 50 years). TCU082 EW component record of 1999 Chi-Chi Taiwan earthquake record was
chosen because the drift ratios in all the three floors (shown in Fig. 12) are more uniform than those obtained
from using other records. Furthermore, the story energy time history as shown in Fig. 13 shows it is increasing
evenly across the during of the earthquake effects. In the PISA2D analyses, it was also indicated that the top of
the second floor columns had the potential of forming plastic hinges. Thus, it was hoped that by the strategy of
changing the mass distribution, this phenomena could be reduced. Three different mass distributions were then
investigated. From the first floor to the third floor, the variations of the mass distribution included: 1:1:0.8,
1:0.8:1 and 0.93:0.93:0.93. Based on these analytical results, the mass distribution of 0.93:0.93:0.93 was found
most suitable for the test. The estimated distribution and the extent of the plastic hinges are shown in Fig.14. The
maximum lateral frame displacement less than 0.3 m fits quite well within the stroke limitation of the actuators.
The results of the analyses demonstrate that the story shear of each floor shown in Fig. 15 is less than 2510 kN.
It is noted that the peak story shear in each story usually do not occur simultaneously. Therefore, it does not
directly associate with the peak story force. Fig. 16 demonstrates that the maximum story forces of 2621 kN
might be developed at the 1st floor, 1852 kN at the 2nd floor and 1629 kN at the 3rd floor. In selecting the
earthquake records, the distribution of the peak story forces has been the most important factor to ensure the
force capacity of the actuators can satisfy the force demand. The final record chosen from the Northridge and
Loma Prieta events is the LP89g04 NS component. More analytical results are discussed in detail in the
companion paper [Cordova et al. 2003]. The acceleration time history and the response spectrum of the selected

110
ground accelerations (Chi-Chi_TCU082EW and LP89g04NS) are shown in Figs. 17 and 18, respectively.

Testing techniques—ISEE

Internet-based Simulations for Earthquake Engineering (ISEE) has been developed by the researchers in NCREE
as a prototype of Internet-based cooperative structural experimental environment [Yang et al. 2002]. During the
full-scale RCS frame test, ISEE was activated to allow remote participants witness the real time video images of
the specimen in the laboratory as well as the digital response data through the Internet. The schematic of the
ISEE framework configured for this study is shown in Fig. 19. It consists of three major parts: the Data Center,
the Analysis Engine and the Facility Controller. The Data Center is a database server, which processes all the
prescribed data sent from the Analysis Engine and the Facility Controller. The Analysis Engine, considering both
the analytical and experimental responses of the specimen, deals with the numerical integration of the dynamic
responses of the entire system using Newark explicit scheme with a time step size of 0.02 second. The Facility
Controller is the software bridging the experimental facilities and the Data Center. When the target displacement
was satisfactorily imposed, the load cells measured the force responses of the test specimen and sent all the
response data (about 300 channels) to the data logger while sending the actuator force and other prescribed
response data (some specific displacement transducer or strain gauge data) back to and the Data Center for
real-time Internet distribution. The related information was sent to the Analysis Engine to compute the target
displacement for the next time step. Because the Data Center was too important to be interrupted, the
web-casting of the digital response data have been done through an independent Web Server. Meanwhile, the
Real Video Server broadcasted the video images to allow data viewers witness the experimental responses in the
laboratory.

EXPERIMENTAL OBSERVATION

Four earthquake load effects were imposed on the specimen on 11, 14, 15 and 16 of October 2002, respectively.
During the tests, the actuators were paused to allow researchers to examine damages, mark the cracks and take
photographs. These specific pauses in terms of time step are marked by vertical lines in Figs. 20 to 23. The
observations are described following the sequence of the earthquakes applied:

50% chance of exceeding in 50 years (TCU082EW, CC50/50, PGA=0.276g)

At the time step of 5.76 second in the loading time history, an abrupt loud noise was heard from the specimen. It
was speculated, and later confirmed, to be the slippage of the beam splice. At the point of 7.14 second, hairline
cracks were found in the columns in the first story. At the point of 18.08 second, cracks were found in all the
columns, and all the slabs. Meanwhile, the beam-column joints on the first and the third floors began to form
hairline cracks. At the point of 23.50 second, the white wash were found flaking in some of the beam flanges on
the first and second floors as well as beam flanges and beam webs on the third floor (see Photo 5). All the
conditions kept becoming more and more severe, but no drastic damage occurred through out the entire time
history.

10% chance of exceeding in 50 years (LP89g04NS, LP10/50x0.8, PGA=0.341g)

At the 5.74 second time step, some of the beam flanges on the first and third floors buckled locally as dipicted in
Photo 6. None of the beams on the second floor were found buckled. In the end of this time history, severe
concrete spalling shown in Photo 7 was found in the column bottom ends in the first story.

2% chance of exceeding in 50 years (TCU082EW, CC2/50, PGA=0.622g)

At the step of 12.82 second, concrete spalled under the steel band plate of column on the second floor.
Meanwhile, on the third floor, severe local buckling was found in the beams and the concrete slab was crashed
near columns. On the first floor, the steel beams were also found severely buckled. At the point of 19.36 second,
all the concrete under the steel band plates on the tops of the columns in the first story were crashed due to
bearing failure. Columns in the second story were found crashed in the bottom ends along with the crash of the
concrete slab around them. Some columns in the second story were found damaged on the top and plastic hinge
appeared there. At the step of 27.96 second, the test was stopped because the roof displacement has reached the
preset limit (about the stroke capacity) of the actuators. In the end, the column bottom ends in the first story were
crashed so drastically that the stirrups of the columns were exposed. At the same time, all the beam-to-column
panel joints in the first story were found damaged with shear failure and bearing failure as illustrated in Photo 8.
On the second floor, bearing failure was found under the steel band plate, and the beam splice plates yielded
slightly.

111
10% chance of exceeding in 50 years (LP89g04NS, LP10/50x0.8, PGA=0.341g)
Since the test frame had experienced the most severe earthquake in this series of tests, new damage was not as
easy to find. In the end of this time history, only the plastic hinges in the bottom end of the columns in the first
story were found more pronounced.
Final pushover test

After the final pushover test to reach a roof lateral drift of about 1.0 meter, main bars in column bottom ends in
the first story were severely buckled (see Photo 9), and the plastic hinges were evidently formed in the range of
about a column-depth high from the top of the footings. Moreover, a beam splice plate at one first floor beam
bottom flange also fractured as shown in Photo 10. The damage conditions on the second and the third floors
were evidently more severe than those found after all the pseudo-dynamic tests.

PISA3D VERSUS EXPERIMENTAL RESPONSES

Based on the experimental responses of the test frame and those of the subassembly test specimen, PISA3D [Tsai
and Lin 2003] has been used to simulate the responses of the full-scale frame specimen. Analytical parameters
adopted for the full-scale frame were calibrated using the experimental responses obtained from the subassembly
tests noted previously. Material properties are all based on the actual material test results, including concrete
cylinder tests conducted on the same day of frame test. The analytical element for RC columns uses the
three-parameter degrading beam-column element considering stiffness degrading, strength deterioration and
pinching effects of RC elements. The analytical beams are represented by using the two-surface plastic
hardening model. To take into account of the composite effect of the steel beams and the RC slabs, the positive
bending strength of the beam is assumed higher than the negative bending strength of the beam by 10 percent.
Furthermore, a panel joint element, considering 3-parameter degrading effects noted above, was adopted in all
the beam-to-column panel zones in the PISA3D frame model. Thus, offset of rigid end zones has been activated
in formulating the flexural stiffness of all the beam and column elements. The fundamental period of the
PISA3D model is 1.08 second.

Figs. 20 to 23 depict the roof displacement and the base shear versus time relationships of both the experimental
and analytical responses. In the CC50/50 test, the roof displacements are accurately captured from the 15th
second to the 28th second, while the PISA3D model performs satisfactorily in simulating the base shear effects
through the first 28 seconds. In the first LP10/50x0.8 test, the PISA3D analysis is satisfactory in the first 7
seconds in capturing the roof displacement; and the base shear can be captured well through out the first 33
seconds. In the CC2/50 test, although the simulation is not entirely satisfactory, the response trends of the test
frame can be represented especially in the magnitude of the base shear. Finally, in the second LP10/50x0.8 test,
the analysis can no longer be as good as those in the previous tests but the permanent roof displacement can be
well captured as observed in the test frame. Comparing the four analytical simulations, it is found that the
analytical model seems less and less capable of simulating the experimental responses with satisfactory precision.
This variation might come from the differences between the damage conditions of the real test frame and those
considered in the analytical model. After the devastating CC2/50 excitation, more or less, the RC columns, the
steel beams and the RC slabs are evidently damaged. The complex damage conditions may contribute to the
changes in the strength, the stiffness, the period and the damping ratio of the specimen and therefore increase the
difficulties of capturing the behavior of the specimen precisely. Through adjusting the modeling parameters,
perhaps, the simulations could be more satisfacroty.

CONCLUSION

This experimental program has allowed the international researchers jointly explore the system performance of
RCS moment resisting frame structure of realistic size subjected to various levels of earthquake load effects. The
4-square stirrups adopted in the joint panels provide excellent constructability while perform well with
satisfactory ductility and strength. Moreover, the experience of constructing the full-scale frame proves that the
construction of RCS system is highly efficient. Test results confirm that properly designed and constructed RCS
system is a viable alternative to all-steel or all-concrete frames in high-seismic regions. However, designers
should note that bolted beam splices using bearing type design might result in slippage and large bangs when the
structure is subjected to significant story drift. Analytical studies suggest that numerical model constructed using
PISA3D programs are effective in capturing overall seismic responses of the specimen especially when the joint
element is employed in the analytical model. It incorporates the contributions of the deformations of the

112
beam-to-column panel zone joints to the global responses of the RCS frame. The software and hardware
implemented in this program for the internet-based simulations for earthquake engineering (ISEE) has allowed
NCREE effectively disseminate key information before, during and after the actual tests. It clear demonstrates
the potential benefits of many networked features promised in the NEES program initiated in the US. Additional
information can be found in the web site at: http://rcs.ncree.gov.tw. This program has allowed the mutual
exchanges of experiences not just on technical issues, but also on educational perspectives among the
participants across the continents. A similar test program focuses on concrete filled steel tube column and
buckling restrained braced frame has also been launched in the year of 2003 with a similar test schedule
(http://cft-brbf.ncree.gov.tw).

REFERENCES

ASCE Task Committee on Design Criteria for Composite Structures in Steel and Concrete (1994), Guidelines
for Design of Joints between Steel Beams and Reinforced Concrete Columns, Journal of Structural
Engineering, ASCE, Vol. 120, No. 8, pp.2330-2351.
Chen, C.C. and Lin, N.J. (2002), Seismic Behavior of Steel Beam-to-RC Column Joint, Technical Report,
National Center for Research on Earthquake Engineering. (in Chinese)
Cheng, C.T. and Cian, P.H. (2002), Composite Behavior of Slab and Steel Beam-to-RC Column, Technical
Report, National Center for Research on Earthquake Engineering. (in Chinese)
Cordova, P., Chen, C.H., Lai1, W.C., Deierlein, G.G. and Tsai, K.C. (2003), “Pseudo-Dynamic Test of
Full-Scale RCS Frame: PART 2 – Analyses and Design Implications,” International Workshop on Steel and
Concrete Composite Construction, Taipei, Oct.
Deierlein, G.G. (1988), Design of Moment Connections for Composite Frames, Ph.D. Dissertation, University of
Texas, Austin, Texas.
FEMA, 2000, Prestandard and Commentary for the Seismic Rehabilitation of Buildings, Report FEMA 356,
Federal Emergency Management Agency, Washington D.C.
Hsieh, S.H., Tsai, K.C., Yang, Y.S., Wang, K.J., Hsu, J.W., Wang, S.J., Loh, C.H. (2002), “An Internet-based
Environment for Collaborative Networked Pseudo Dynamic Tests,” Proceedings, the 4th Seminar on
Earthquake Engineering for Building Structures, Seoul, Oct.
Kanno, R., and Deierlein, G.G. (2000), Design Model of Joints for RCS Frames, Composite Construction in
Steel and Concrete IV, ASCE.
Tsai, K.C. and Chen, P.C. (2002), A Study of RC Column-to-Foundation and Steel Beam-to-RC Column Joints
for an RCS Frame Specimen, Technical Report, National Center for Research on Earthquake Engineering.
(in Chinese)
Tsai, K.C. and Lin, B.C. (2003), “Development of an Object-Oriented Nonlinear Static and Dynamic 3D
Structural Analysis Program,” Center for Earthquake Engineering Research, National Taiwan University.
(in Chinese)
Tsai, K.C., Chang, L.C. (2001), “The Platform and Visualization of Inelastic Structural Analysis of 2D Systems
PISA2D and VISA2D,” Report No. CEER/R90-08, Center for Earthquake Engineering Research, National
Taiwan University. (in Chinese)
Yang, Y. S., Hsieh, S. H., Wang, K. J., Wang, S. J., Hsu, C. W., and Tsai, K. C. (2002), “Numerical Analysis
Framework for Distributed Pseudo-Dynamic Tests,” Proceedings of the 2nd International Conference on
Structural Stability and Dynamics, Singapore.
2.15m

b
bj Mc
Inner Vc
element
dj
[email protected] = 12m

Vb
Mb
d
Outer element
Effective Column
joint region
[email protected] = 21m
Fig. 2 Inner element and outer element (after
Fig. 1 Plan and elevation of test frame Kanno and Deierlein 2000)

113
 Concrete M Mbo
bi Panel shear
crushing yielding
Vbi Vbo
Vth Vtv Vth
Tie
Vtv yielding
Bond
failure
Gap Shear
Concrete cracks
crushing
Fig. 3 Failure modes of inner element Fig. 4 Failure modes of outer element
(after Kanno and Deierlein 2000) (after Kanno and Deierlein 2000)

Steel Band

#4 tie FBP(PL15x566x94)

4-square tie
Face Bearing Plate

Fig. 5 Beam-column joint of the test frame Fig. 6 4-square ties used in test frame

Rein. Bar

High Strength
Mortar

Column
Top Sleeve

Flow Out Flow Out

Flow Out Flow In

Column

Fig. 7 Beam-column connection on the roof Fig. 8 Details of column splice

3500

2500 3-Story
Grouted
sleeve RCS Frame
Base Shear (kN)

1500
500
-500
1000

-1500

-2500 PISA2 D
Op e n Se e s
-3500
-10 -5 0 5 10
Drift Ratio (% radian)

Fig. 9 Column-to-footing connection Fig. 10 Simulated nonlinear static response of the test frame

114
 Peak Story Drift Ratio
3
PISA2D
-2.22 1.55

Floor Level
TCU082EW
-2.44 2%c in 50 yr(PGA=0.622g) 2.61

-1.84 1.86

0
-3 -2 -1 0 1 2 3
IDR (% rad)

Fig. 11 The distribution of plastic hinges based on Fig. 12 The predicted peak interstory drift ratio of the
analytical cyclic loading (PISA2D) test frame

4000
TCU082EW 2%c in 50 yr(PGA=0.622g)
Steel,M1:M2:M3 = 0.93:0.93:0.93
3000
3rd Story
Energy (kN-m)

2nd Story
2000
1st Story

1000

0
0 9 18 27 36 45
Time (sec)

Fig. 13 The predicted energy distribution time history in Fig. 14 The distribution of plastic hinges based on
each level (PISA2D) dynamic analysis (PISA2D)

Peak Story Shear Peak Story Force

3 3
TCU082EW,Steel -1629 1360
2%c in 50 yr
-1629 1360 (PGA=0.622g)

2 2
-1362 1852
Floor Level
Floor Level

-1872 2119
1
1 -1648 2621

-2510 PISA2D 2453 PISA2D 2%c in 50 yr

TCU082EW,Steel (PGA=0.622g)
0
0
-4000 -3000 -2000 -1000 0 1000 2000 3000 4000 -3000 -2000 -1000 0 1000 2000 3000
Shear (kN) Force (kN)

Fig. 15 The predicted peak story shear in each level Fig. 16 The predicted peak story force on each floor

115
 Original Ground Acceleration Time History
0.22
0.18 TCU082EW , PGA = 0.217g
Ground Acceleration (g)

0.14
0.1 Normalized Elastic Acceleration
0.06
0.02 Response Spectrum
-0.02
4
ξ=5% , PGA=1g
-0.06 TCU082EW
-0.1 (Sa1=1.475g)
-0.14 3 LP89go4NS
-0.18 (Sa1=1.690g)
-0.22

Sa (g)
0 5 10 15 20 25 30 35 40 45
Time (sec) 2

Original Ground Acceleration Time History

1
0.22
0.18 LP89g04NS , PGA = 0.212g
Ground Acceleration (g)

0.14
0.1 0
0.06 0 1 2 3
0.02 Period (sec)
-0.02
-0.06
-0.1
-0.14
-0.18
-0.22
0 5 10 15 20 25 30 35 40 45
Time (sec)
Fig. 18 Normalized response

Fig. 17 The original ground accelerations used in test (before scaling) spectrum of the two records

Internet-based Simulations for Earthquake Engineering

Data Center

Facility controller Data Viewer

Web server
SD

PRO

SD
SD P rofessional Workstat o
i n 6000

1A-001
1A-001

Analysis Real video Image capture

engine Dynamic Data server system
Static Data logger
logger
Video

Sensor

Fig. 19 ISEE framework employed in the pseudo-dynamic test

20
Roof Disp (cm)

CC(50/50) 0.276g
10
0
-10 TEST
PISA3D
-20
Time (sec)
Story Shear

4
CC(50/50) 0.276g
(10 kN)

2
3

0
-2 TEST 1F
PISA3D 1F
-4
0 5 10 15 20 25 30 35 40 45
Time (sec)
Fig. 20 Comparing the PISA3D analytical responses with experimental results under
CC50/50 excitation (damping ratio=1％)

116
 30
Roof Disp (cm)
20 LP(10/50)R1 0.341g
10
0
-10
-20 TEST PISA3D
-30
Time (sec)
Story Shear

4
LP(10/50)R1 0.341g
(10 kN)

2
3

0
-2 TEST 1F
PISA3D 1F
-4
0 5 10 15 20 25 30 35 40
Time (sec)
Fig. 21 Comparing the PISA3D analytical responses with experimental results under the first
LP10/50x0.8 excitation (damping ratio=1％)

40
Roof Disp (cm)

20 CC(02/50) 0.622g
0
-20 TEST
-40 PISA3D
-60

Time (sec)
Story Shear

4
(10 kN)

CC(02/50) 0.622g
2
3

0
-2 TEST 1F
PISA3D 1F
-4
0 5 10 15 20 25 30
Time (sec)

Fig. 22 Comparing the PISA3D analytical responses with experimental results under
CC2/50 excitation (damping ratio=1％)

20
Roof Disp (cm)

LP(10/50)R2 0.341g
0
-20 TEST
PISA3D
-40
Time (sec)
Story Shear

3
LP(10/50)R2 0.341g
(10 kN)

2
1
3

0
-1 TEST 1F
-2 PISA3D 1F
-3
0 5 10 15 20 25 30 35 40
Time (sec)
Fig. 23 Comparing PISA3D analytical responses with experimental results under the second
LP10/50x0.8 excitation (damping ratio=1％)

117
 Photo 3 A typical precast footing

Photo 1 Large-scale, three-story, Photo 2 A typical precast

three-bay RCS moment frame column
Photo 4 Defect under steel plate

Photo 5 White wash flaking at Photo 7 Concrete spalling in the

Photo 6 Local buckling of beam
the beam bottom flange column bottom ends in the first story
flange (1st LP10/50x0.8 5.74 sec)
(CC50/50 45 sec) (1st LP10/50x0.8 39.96 sec)

Photo 8 Shear failure and bearing Photo 9 Buckling of main bars in Photo 10 Fracture of beam splice
failure of beam-column connection in the column bottom end in the steel plate (Pushover)
the first story (CC2/50 27.96 sec) first story (Pushover)

118