Kevin Solberg1, Luoman Li1, Brendon Bradley1, John B. Mander2, Rajesh P. Dhakal3
SUMMARY
This work will further investigate the validity of Damage Avoidance Design systems through an experimental
and computational study of a 80% scale sub-assembly. A coupler system is implemented whereby high
strength unbonded threaded rods running through the beams are coupled to rods within the columns. The
threaded rods are post-tensioned and additional energy dissipation devices are installed. Precast and cast
insitu solutions are considered. A multi-level seismic performance assessment is conducted considering
three performance objectives related to occupancy and collapse prevention. Bi-directional quasi-static cyclic
tests and quasi-earthquake displacement tests are performed. Critical earthquakes are selected using a
probabilistic computational method. Results indicate the DAD system satisfies all performance objectives.
Further design improvements are discussed.
INTRODUCTION
As part of a large research project in the United
Current seismic design accepts that damage will
States, the PRESSS program investigated the
occur in moderate to large seismic events,
behaviour of these systems through testing of
although attempts are made via special detailing
many sub-assemblages [3] and a five-storey 3D
to limit this damage to specific plastic hinge
frame and wall system [4]. The system performed
zones. These zones, designed to sustain severe
well with much less damage than would be
damage under multiple cyclic rotations, tend to act
expected with monolithic construction. Little
like a fuse, essentially protecting the structure
residual displacement was observed in both
from forming unfavourable mechanisms.
frames and walls. The joints , however, employed
Although this design philosophy ensures good
a concrete-concrete or high-strength grout
protection to occupants by preventing collapse,
interface, resulting in some damage at the joint
there is a strong likelihood a moderate to large
region.
earthquake will render a structure irreparable. As
a result, economic costs, both direct and indirect,
Mander and Cheng [5] proposed a new seismic
can be significant; this has been confirmed from
design and construction philosophy for bridges
recent earthquakes in the United States
called Damage Avoidance Design (DAD). In this
(Northridge, 1994) and Japan (Kobe, 1995). To
approach, joints are armoured with steel to protect
address this issue, alternative structural systems
them from damage incurred from rocking. This
have been proposed where precast concrete
concept was validated by uni-directional tests
elements are designed to remain essentially
performed on a scaled bridge pier, with bi-
elastic, with inelastic behaviour accommodated for
directional tests conducted more recently [6].
by rocking at specially detailed joints.
Results indicate little damage at the joint and
good bi-linear elastic behaviour.
The theoretical basis of rocking systems have
been investigated by many early researchers (e.g.
These concepts have been further developed in
[1],[2]). Although it was not until more recently [3]
that so called “hybrid” systems were introduced.
New Zealand and design guidelines for such
ductile jointed precast concrete systems have
These systems utilize full or partially unbonded
been introduced into the concrete code as an
post-tensioning to provide a restoring force and
appendix [7]. As part of an ongoing research
supplemental yielding devices to provide energy
program at the University of Canterbury, further
dissipation. By combining the hysteretic
experimental investigations have been conducted
behaviour of these two components, it is possible
[8-11] with the goal of refining detailing at the joint
for a joint to exhibit a combination of bi-linear
and providing cost-effective alternative solutions.
elastic (post-tensioning) and elasto-plastic
As a follow up to this previous work, this paper
(yielding devices) hysteresis behaviour. The
presents results from a combined experimental
result is a flag shaped hysteresis loop, displaying
and computational study on the bi-directional
good energy dissipation and re-centring
behaviour of DAD beam-column joints.
characteristics.
1
Research Assistant, Dept. of Civil Engineering, University of Canterbury, Christchurch
2
Professor, Dept. of Civil Engineering, University of Canterbury, Christchurch
3
Senior Lecturer, Dept. of Civil Engineering, University of Canterbury, Christchurch
Two research objectives will be addressed herein. EXPERIMENTAL INVESTIGATION
Firstly, previous research adopted quasi-static
testing, in which loading was composed of Prototype Structure
regulated displacement cycles. These cycles As shown in Figure 1, the prototype is a ten-storey
however, are not completely representative of the reinforced concrete frame building with three 10m
displacement demands due to seismic excitation. bays in each direction This generic structure,
Therefore, this study will adopt the Quasi- commonly known as the “red book” building [15],
Earthquake Displacement (QED) test [12], where was designed according to the New Zealand
the specimen will be subjected to displacement concrete standard [7] for intermediate soil in
profiles found analytically using real ground Christchurch, New Zealand. Keeping all other
motions. Using this approach, a multi-level variables constant, the same structure was
seismic performance assessment (MSPA) [13] will designed and detailed according to damage
be conducted, characterizing the performance of avoidance principles, thereby resulting in precast
the specimen at multiple levels of seismic beams and columns being connected via a post-
demand. tensioning system with other devices to provide
supplemental energy dissipation. The DAD
Secondly, further refinement of the beam-column building was designed with precast flooring units
joint details are needed to ensure a practical, running in the transverse direction and seated on
cost-effective solution. Li [10] investigated the the transverse beams, leaving the longitudinal
behaviour of a beam-column joint using a bent beams to resist predominately seismic forces.
coupler system whereby high strength threaded
rods in a beam are coupled to diagonal rods
running through the column. The aim of such a
system is to allow for rapid on-site erection,
thereby reducing initial costs. From physical
testing, it was found that its performance was
satisfactory, however several design
improvements relating to the coupler system and
the armoured ends were suggested. This study
implements these design improvements and
ensures detailing satisfies performance objectives
relating to occupancy and collapse prevention.
Figure 1: The prototype structure, showing the
location of the subassembly.
THEORETICAL BEHAVIOUR
Two methods have been introduced for predicting To help ensure jointed precast systems are
the behaviour of rocking systems. Pampanin et adopted by the construction industry, the system
al. [14] proposed using a monolithic beam analogy must be relatively simple to erect. A major
approach. An iterative process is used to component of this is the post-tensioning system.
determine the neutral axis depth and strain in the This study investigates the use of a coupled high
compression concrete. Although it has been strength threaded rods to provide post-tensioning.
demonstrated this method agrees well with In this design, it is possible for complete beam
physical results, it was developed for precast and column sections to be cast off-site with rods
members without armouring. If armouring is already in place within the respective elements.
considered, it is reasonable to assume rigid body Once on-site, the beam’s rods are connected via
behaviour at the joint. As investigated by Mander a coupler system to a short rod in the column and
and Chang [5] and Li [10], the theoretical anchored to the column’s opposite face. A
behaviour of an armoured rocking system can detailed explanation of this, and other design
best be determined from coupling elastic elements follows.
deformation with rigid body kinematics. In this Specimen Sub-assemblage
method, the post-joint opening neutral axis is
An exterior joint on the second floor of the
assumed to be negligible, thus allowing one to
prototype structure was taken for the 3D beam-
presume the specimen rocks on an extreme edge.
column subassembly. Using constant stress and
Using this approach, it is possible to calculate the
strain similitude principles, the specimen was
moment capacity and stiffness at several key
scaled to 80%, and consisted of two beams in the
locations, namely gap-opening, yielding of the
longitudinal direction, and one beam in the
steel energy dissipaters, and yielding of the
transverse direction. Herein, the longitudinal and
tendons. This method has been confirmed by Li
transverse beams are dominated by seismic load
[10], where a more thorough explanation is given
and gravity loads (carrying the one-way precast
outside the scope of this study.
panels) and respectively are referred to as the
east-west seismic and north-south gravity beams.
Reinforcing details of the column are given in 700
3200
2709
precast beams were joined. The minimum
560
reinforcement ratio, ñt = 0.008 was provided using
TM
12 HD20 threaded bars (Reidbars ). This low DRAPED PT TENDON C
4550
310
A B 4@100 4@200 HR12 TYP @250 4@100
560
B
187
1015
NOTES:
EXTERNAL STEEL ENERGY DISSIPATERS NOT SHOWN
150
EAST BEAM STRUT LOAD CELL ANALYTICAL
LATERLA FORCE (KN)
-100
-150
2867 -200
STRUT -3% -1% 1% 3%
GRAVITY PIN LOAD
BEAM JOINT CELL DRIFT
UNIVERSAL JOINT
Figure 8: Experimental-analytical response.
EAST ELEVATION Earthquake Record Selection
Figure 6: Elevations of the testing apparatus. In the case of MSPA, it is necessary to select
earthquake records that represent the desired
level of ground excitation. Following current
trends, three performance levels were considered.
These levels correspond to an upper bound
design basis earthquake (DBE), which has a 10%
probability of occurrence in 50 years, and a
median and upper bound maximum considered
event (MCE), which has a 2% probability of
occurrence in 50 years.
Figure 7: Photograph of the specimen in the testing
apparatus.
Current seismological studies predict peak ground damage from frequent earthquakes. The second
acceleration (PGA) at various return periods. and third levels of response relate to rare
However, it is not correct to simply apply any earthquakes. At this level, one should have
earthquake record that conforms to this definition, moderate confidence the structure will be
th
as structural response is dependent on a repairable (50 MCE), and high confidence the
th
multitude of factors. Therefore, it is necessary to structure will not collapse (90 MCE). Given
extract earthquake records from a suite of likely these objectives, the DAD specimen will be
candidates that will result in the most severe monitored to ensure these objectives are met, if
structural behaviour. Such a method has been not exceeded.
proposed by Dhakal et al. [13] whereby
Incremental Dynamic Analysis (IDA) [18] is used RESULTS
to probabilistically determine earthquake records
Results are presented only for a bi-directional
representing multiple performance objectives.
quasi-static test to 2% drift and the QED tests
This method has been adopted herein and is
using the earthquakes selected for MSPA. In all
illustrated in Figure 9. Once an IDA has been
tests, each post-tensioned rod was stressed to
conducted earthquake records representing
50% of its yield limit (i.e. 200kN). This provided a
different percentile response at a given intensity
total of 400kN of post-tensioning force at each
measure (IM) can be extracted. In this study,
joint. The energy dissipaters were replaced after
records were chosen to yield responses that have
each test.
non-exceedance probabilities of 90% at the DBE,
50% at the MCE, and 90% at the MCE, from a Quasi-static Test Results
suite of 40 records consisting of medium and Figure 10 presents results of bi-directional testing
near-source ground motions. The selected to the design level drift of 2%. The results shown
records are given in Table 1. are for a bi-directional “clover leaf” test, where
total drift is calculated considering both X and Y
1.2
10th 50th
components. Note that the individual plots are
projected to one another, allowing an easy
1 comparison to be made between the NS and EW
90th direction.
0.8
i
0.6
at the bottom edge of each beam, running
between the edge of each flange. This crack can
0.4 90% MCE
be attributed to the vertical component of the
0.2 90% DBE
diagonal tendons, approximately a 120kN upward
force at the joint. This force in effect pulled the
0
beam up the face of the column. The bottom steel
0% 2% 4% 6% 8% 10% flange however, resisted this due to high friction
DRIFT forces, causing tearing just above the angle, as
Figure 9: Earthquake record selection for the evidenced by this crack.
MSPA, using Incremental Dynamic Analysis.
Opening of the gap was observed at
Table 1: Earthquake records selected for QED approximately 0.5% drift, at which point the steel
testing, N indicates a near-source record and M
dissipaters yielded in tension almost immediately
indicates a medium source record.
(as evidenced from strain gauges). In the east
1
Event Year Level Max Drift (%) beam, two hairline cracks formed just before
2 reaching the target drift of 2%, propagating
N-Palos Verdes 1992 90% DBE 2.1 100mm out from the dissipater anchor plate. The
N-Tabos 1974 50% MCE 2.8 west beam (high-strength concrete) did not suffer
additional cracking. Due to the bi-directional
M-Loma Prieta 1989 90% MCE 4.7 rocking, localised crushing was observed behind
1. Absolute peak interstory drift, considering X and Y radial drift.
2. Simulated ground motion. the top angle of the east beam over a 10mm
source: SAC Strong Motion Database square area at the top concrete face. At
approximately 1% drift, slight buckling of the steel
Performance Objectives dissipaters occurred as the gap began to close.
This was more severe for the gravity beams than
Performance objectives must be defined for the
th the seismic beams, attributed to their longer
MSPA. At the first level (90 DBE), there needs
length. Throughout testing, no damage was
to be a high level or reliability that no damage
observed on the column. A photograph of the
needing repair will occur (i.e. immediate
east beam after testing is given in Figure 11.
occupancy). This relates to the general
philosophy that a structure should incur no
I it
100 Nevertheless, these tests will give a more
accurate assessment of response from ‘real’
1% 1% 1%
of the energy dissipaters occurred at around the
DRIFT it
0% 0% 0%
same drift as in previous testing (~0.5%). No new
-1% -1% -1% cracks or additional crushing was observed on the
-2% -2% -2% seismic beams. A flag shaped hysteresis loop
200 100 0 -100
EW LATERAL LOAD (kN)
-200 -2%
0
-1% 0%
DRIFT NS
1% 2% 0 200 400
STEP
600 800 1000
was observed during the initial pulse, however for
200 the remainder of the test response was mostly
elastic. Some post-gap opening stiffness
STEP it
400
600
degradation was observed, likely due to yielding
800
and buckling of the energy dissipaters. The
1000
maximum gap opening, recorded from the
-2% -1% 0% 1% 2% potentiometers was approximately 5mm.
DRIFT NS
iit
200
0 0 0
DRIFT NS i
1% 1% 2%
DRIFT NS
DRIFT NS
0% 0% 0%
400 600
STEP it
STEP it
STEP it
400
600
600
800 1200
800
1000
1000 1200 1800
-3% -2% -1% 0% 1% 2% 3% -3% -2% -1% 0% 1% 2% 3% -6% -4% -2% 0% 2% 4% 6%
DRIFT EW DRIFT EW DRIFT EW
(a) (b) (c)
Figure 12: QED test results for the seismic beams (EW direction) for: (a) the 90% DBE; (b) the 50% MCE; and (c)
the 90% MCE.