You are on page 1of 12

ACI STRUCTURAL JOURNAL TECHNICAL PAPER

MS No. S-2012-303.R3

Seismic Response of Reinforced Concrete Frames with


Haunch Retrofit Solution
by Akanshu Sharma, G. R. Reddy, R. Eligehausen, G. Genesio, and S. Pampanin
Haunch retrofit solution (HRS) has been investigated as a viable
alternative for retrofitting of beam-column joints of reinforced
concrete (RC) frame structures. For connecting the haunch
elements to the frame members, partially prestressed rods were
used. This solution, however, makes it difficult for implementation
in practice due to the requirement of drilling holes through the slab.
A more practical solution is to connect the haunch elements using
post-installed anchors, resulting in fully fastened HRS (FFHRS).
The solution has been tested at the subassembly level, and encour-
aging results were obtained. To verify the performance of the
FFHRS to retrofit the frame structures subjected to real-life seismic
loads, shake-table tests were performed on non-seismically detailed
two-dimensional (2-D) RC frame structures. This paper presents the Fig. 1—Schematic of haunch retrofit solutions tested at
results of the experiments with the detailed discussion of the perfor-
subassembly level.
mance of the FFHRS under earthquake ground motions.
To connect the haunch element, two types of connections
Keywords: haunch retrofit solution; non-seismic detailing; post-installed
anchors; reinforced concrete; seismic performance; shake-table tests. between diagonal axial element and plates—namely, hinged
and welded—were studied.3 In one specimen, a yielding-type
INTRODUCTION fuse element was used as an axial element. To connect the
The performance of reinforced concrete (RC) frame struc- haunch element with beams and columns, two partially
tures, especially designed and constructed with non-seismic prestressed external rods along with two anchors directly
provisions, under seismic loads is largely governed by the fastened to both beam and column were used. Figures 1(a) and
performance of beam-column joints. Several past earth- (b) show the schematic of the HRS used by Pampanin et al.3
quakes have exposed that one of the most severe deficien- It was reported1 that the retrofitted specimens displayed
cies in such RC frame structures that make it vulnerable a substantially enhanced response when compared with the
against earthquakes is the inadequate shear resistance of non-retrofitted specimens: damage to the joint was elim-
beam-column joints. Under the action of seismic forces, inated, and a flexural plastic hinge formed in the beam at
beam-column joints are subjected to large shear stresses in the location of the beam-haunch connection. One prohibitive
the core. The axial and joint shear stresses result in principal requirement of the connection proposed by them, however,
tension and compression that leads to diagonal cracking was to use external partially prestressed rods to connect the
and/or crushing of concrete in the joint core.1 These stresses haunch to the frame. Though, in laboratory tests, it could
in the joint core are resisted by the so-called strut and tie be easily achieved, to do the same in real structures would
mechanism.2 Due to the absence of shear reinforcement require drilling through holes in the slabs of the structure,
in the joints of old non-seismically detailed structures, the which renders the solution prohibitive.
joints suffer from insufficient shear strength, which leads to To eliminate this limitation, further investigations were
their failure in the event of an earthquake. To extract better carried out at the subassembly level to evaluate the perfor-
seismic performance from such structures, retrofitting of mance of haunch elements connected to the frame members
beam-column joints is essential. by using post-installed mechanical anchors.4-6 The exper-
Relatively recently, Pampanin et al.3 developed and imental program on beam-column joints retrofitted with a
validated the haunch retrofit solution (HRS) for RC beam- fully fastened haunch retrofit solution (FFHRS), as shown
column joints. The solution consists of a haunch element, in Fig. 1(c), produced encouraging results that proved the
which primarily consists of a diagonal axial element in the efficacy of the FFHRS to prevent the joint failure5 (Fig. 2).
form of a machined bar/plate connected to plates at both ends, The tests clearly demonstrated that FFHRS can successfully
connected to a beam and column (Fig. 1). The underlying modify the hierarchy of the joint subassembly, and lead to
principle of the HRS is to relocate the plastic hinge away desirable failure mechanisms such as beam flexure.
from the vulnerable joint panel/core while enhancing the ACI Structural Journal, V. 111, No. 1-6, January-December 2014.
global response of non-seismically designed RC joint subas- MS No. S-2012-303.R3 received March 8, 2013, and reviewed under Institute
publication policies. Copyright © 2014, American Concrete Institute. All rights
sembly by altering the hierarchy of strength suitably. This reserved, including the making of copies unless permission is obtained from the
copyright proprietors. Pertinent discussion including author’s closure, if any, will be
would reduce the shear force transferred to the joint core. published ten months from this journal’s date if the discussion is received within four
months of the paper’s print publication.

ACI Structural Journal 1


Fig. 2—Comparison of as-built and retrofitted joint5.

Fig. 3—Mechanics of conventional exterior joint.


The seismic performance of any system or structure can mically detailed RC frame structures (especially joints),
be best assessed by performing a shake-table test. To verify including the seismic response and failure modes. It further
the suitability of the FFHRS for retrofitting of the joints of provides the results of test on the retrofitted structure to
structures when subjected to dynamic earthquake loading, understand the seismic behavior of RC structures with joints
tests were performed on two-dimensional (2-D) RC frame retrofitted with FFHRS under real seismic situations.
structures. Two RC frame structures exactly the same were
constructed and tested on the shake table: 1) without FFHRS MECHANICS OF JOINTS RETROFITTED WITH HRS
(as-built structure); and 2) with FFHRS (retrofitted struc- To understand the behavior of beam-column joints retro-
ture). The frames were designed such that the exterior joints fitted with HRS, it is important to understand the effect of
of the frames are a two-thirds scale replica of the joints the same in redistributing the forces and moments in a joint.
tested previously.4-6 This paper provides the details of the Consider an exterior beam-column joint subassembly formed
tests conducted on 2-D frame structures without and with by the assumed points of inflexion at midheight of columns
FFHRS. The experiments clearly proved the efficacy of the and midspan of beams, as shown in Fig. 3(a). The dashed lines
FFHRS for retrofitting and improving the seismic perfor- depict the physical dimensions of the members, while the solid
mance of non-seismically detailed RC frame structures. lines with rounded ends depict the frame element to model the
Further experimental data, however, is required to support members and their end nodes, respectively. The joint is applied
the findings of this study. by a beam end load Vb. The bending moment and shear force
diagrams for the joint subassembly are as depicted in Fig. 3(b)
RESEARCH SIGNIFICANCE and (c), respectively. Due to the applied load Vb, the joint panel
This paper presents an experimental study, in terms of is subjected to a horizontal joint shear force given by
shake-table tests, carried out to verify the suitability and effi-
cacy of the FFHRS to improve the seismic performance of Mb L L + 0.5hc 
the non-seismically detailed RC frame structures. The paper V jh = − Vc = Vb  b − b  (1)
Zb  Zb Lc
presents the results on the seismic behavior of non-seis-

2 ACI Structural Journal


Fig. 4—Mechanics of joint retrofitted with diagonal haunch elements.
where Zb is the lever arm of internal forces in the beam at fying the critical section for flexure and thereby reducing
column face. Other symbols are explained in Fig. 3. This the maximum bending moment in the beam for a given
shear force is the major cause of failure of the joint panel. applied load (effect of geometry); and 2) further reducing
Consider the same joint subassembly, retrofitted with two the bending moment at the face of the column due to redis-
diagonal haunch elements, as shown in Fig. 4(a). The phys- tribution of forces (effect of stiffness).
ical dimension of the joint is shown by a dotted line, and Therefore, by designing the haunch element suitably, the
the central solid line depicts the frame element to model the horizontal joint shear stress can be significantly reduced
same. To take into account for the physical dimensions of compared with the as-built joint. It should be considered,
beam and column, a rigid link is used to connect the end however, that at the point of connectivity of haunch with
nodes of the haunch element to the beam and column. The beam and column, the shear force suddenly rises. Therefore,
bending moment and shear force diagrams for the joint this system should not be used for the case where the members
subassembly are as depicted in Fig. 4(b) and (c), respec- are or may become shear critical due to the haunch element.
tively. As seen in Fig. 4(b), under the applied beam end load
Vb, the maximum bending moment in the beam occurs at the DESCRIPTION OF TEST STRUCTURE
point of connectivity of the haunch Mbh, given by The geometry of the structure considered for the experi-
ments is shown in Fig. 5. The structure is a plane frame RC
Mbh = Vb(Lb – Lh) (2) structure with two storys, and one bay configuration. The
sizes and reinforcement details of the structural elements
where Lh is the horizontal projected length of the were decided so that a typical exterior joint of the structure
haunch element. corresponds to two-thirds scale of the joint subassembly
Beyond the point of connectivity of the haunch element, tested by Genesio and Sharma.5 The scaling was essential to
the bending moment in the beam gradually reduces. The accommodate the limitations of the shake table.
rate of reduction on bending moment depends on the rela- Due to a linear scaling factor SL of 1.5, the column height,
tive stiffness of the beam and the haunch element.3,4 Thus, at which was 3230 mm (127.2 in.) in case of beam-column
the face of the column, the moment in the beam Mb′ is less joints,4-6 was reduced to 2150 mm (84.65 in.), while the
than the moment Mbh. Similarly, the bending moment was beam clear span, which was 3450 mm (135.83 in.) in case
reduced in the column at the face of the beam Mc′, and the of joints4-6 was reduced to 2300 mm (90.55 in.) in this struc-
reduced shear forces in the beam Vb′ and column Vc′, beyond ture. Similarly, the dimensions of the beam length, as well
the haunch element. The reduced joint horizontal shear force as beam and column sections, were linearly scaled. Due
for the haunch retrofitted joint Vjh′ can now be obtained as to limitations on material availability, the material scaling
factor was kept as unity for both steel and concrete. The
M b′ maximum size of aggregate (msa), however, was limited to
V jh = − Vc′ (3) 12 mm (0.47 in.), while the msa of 20 mm (0.79 in.) was
Z b′
used for casting of beam-column joints. For material scaling
factor of unity, according to similitude requirements,7 the
where Zb′ is the lever arm of the internal forces in the beam required model reinforcement area to provide the scaled
at the face of the column. A more detailed discussion on bar yielding force was calculated by dividing the area of
the mechanics of joints retrofitted with HRS can be obtained reinforcement in the prototype (beam-column joint) by the
from Reference 1. Based on the aforementioned mechanics, square of linear scaling factor. The evaluated area of rein-
it can be observed that the effect of haunch in reducing the forcement was provided within the constraints of avail-
joint horizontal shear force is essentially two-fold: 1) modi-

ACI Structural Journal 3


Fig. 5—Geometry of frame structure tested. (Note: Dimensions in mm.)
ability of reinforcement bar sizes. Because the density and ness of the haunch element, which depends on its dimensions
modulus of elasticity of concrete in the prototype and model and material, as well as by the stiffness of the anchors in
was practically equal, a mass simulation approach given in case of FFHRS. It is desirable to keep the size of the haunch
References 8 through 10 was followed to evaluate the addi- element in the range of approximately 0.1 to 0.2 times the
tional inertial mass to be placed on the beams. Basically, length of the beam, considering anchor layout and architec-
the amount of mass was defined to comply with a period of tural and structural requirements. A smaller size of haunch
vibration of the prototype. would result in a closer critical section to the joint, while a
Typical non-seismic design philosophy was followed for the bigger haunch size would make it closer to bracing, and may
structure. The details of the beam and column sections for the not be architecturally acceptable. If the haunch is designed
structure are given in Fig. 5. The clear cover to the reinforcing well and the anchors deliver good performance, the failure
bars of the beam was kept as 20 mm (0.79 in.), while that for the mode of the subassembly should essentially be beam flexure.
column was kept as 25 mm (0.98 in.). The beam reinforcement Pampanin et al.3 proposed an iterative procedure for the
bars were bent into the joint core for all the joints, while no design of the HRS, where a factor expressing the effectiveness
shear reinforcement was provided in the joint core. (capability of redistribution of shear) of the HRS is determined
The concrete mixture consisted of 371 kg/m3 (23.16 lb/ft3) of while giving due consideration to beam and column deform-
cement, 1005 kg/m3 (62.74 lb/ft3) of fine aggregate (sand), ability. Based on the tests at the beam-column joint sub-as-
and 872 kg/m3 (54.44 lb/ft3) of coarse aggregate. The semblies and numerical parametric studies, Genesio4 extended
water-cement ratio was maintained as 0.40. The average the guidelines to consider the stiffness of the anchors while
cubic concrete compressive strength at the day of testing was designing the FFHRS using post-installed anchors. The typical
obtained as 28.29 MPa (4103 psi). High-strength deformed input parameters for the design of FFHRS include the dimen-
bars were used as reinforcement. The average yield and ulti- sions, geometry, and material properties of members framing
mate strength of the reinforcing bars was 518 and 608 MPa into the joint and of the trial haunch element, along with
(75.13 and 88.18 ksi), respectively, for the 16 mm (0.63 in.) the stiffness of the anchors proposed to connect the haunch
bar; 534 and 620 MPa (77.45 and 89.92 ksi), respectively, element. The design of the anchors is performed based on
for the 12 mm (0.47 in.) bar; and 529 and 616 MPa (76.72 the guidelines given by Eligehausen et al.11 The details of the
and 89.34 ksi), respectively, for the 8 mm (0.31 in.) bar. design procedure can be obtained from References 3 and 4.
The design of the haunch elements in this work was performed
FULLY FASTENED HAUNCH RETROFIT SOLUTION following the design procedure recommended by Genesio.4
For joints retrofitted with FFHRS, the extent of redistribu- The haunch elements used in the retrofitting of the frame
tion of the shear around the joint core is decided by the stiff- structure consisted of three machined steel plates welded

4 ACI Structural Journal


Fig. 7—Haunch element installed using bonded anchors.

Fig. 6—Schematic of haunch elements used for retrofitting


of frame. (Note: Dimensions in mm.)
together. Figure 6 shows the schematic drawing of the
haunch elements used in this work. All plates used were
10 mm (0.39 in.) thick. The haunch elements were connected
to the frame using bonded anchors, which were comprised of
10 mm (0.39 in.) diameter high-strength threaded rods glued
using two-component epoxy-based mortar. The effective
embedment depth of anchors was 100 mm (3.9 in.). Figure 7
represents an installed haunch element using bonded anchors.

TEST SETUP Fig. 8—Layout on shake table.


The experiment was carried out on the triaxial shake-table
system. The system features a 100 kN (22,481 lb) payload and setup, and partly because it was not desirable to get too
capacity shake table of size 3 x 3 m (118.1 x 118.1 in.). In close to the haunch element to avoid any accidental pounding
this work, only unidirectional tests were performed due to of mass on the retrofit. The supporting structures were placed
the 2-D geometry of specimen. In such tests, certain out-of- on either side of the frame in such a way that the rollers at each
plane small vibrations are inevitable. To prevent any acci- floor level just barely touched the flange of the mass to prevent
dental out-of-plane movement of the 2-D frame structures, out-of-plane motion. Figures 9(a) and (b) show the arrange-
two steel frames were fabricated and placed on either side ment of mass on the first floor beam of the structure.
of the frame on shake table. The frames were provided with The instrumentation for the 2-D frame consisted of the
rollers, capable of rolling about the vertical axis at the first accelerometers to measure the response acceleration history
and top story level. Figures 8(a) and (b) show the layout of the at each floor as well as of the shake table (input), strain
as-built and retrofitted structure, respectively, along with the gauge on reinforcing bars to measure strains at critical loca-
supporting structure on the shake table. The shake table has a tions, and potentiometers to measure beam, column, and
grid of holes at 300 mm (11.81 in.) center-to-center spacing joint shear deformations. The locations of strain gauges on
to which the foundation of the test structure was bolted. reinforcing bars are included in Fig. 5.
To simulate the floor mass, 1.0 tonne (2205 lb) of mass The time history compatible to the shape of the site-spe-
concentrated in the middle 800 mm (31.5 in.) length of the cific response spectrum used for seismic design of safety-re-
beam was added at each floor level. Considering the clear lated structures in Mumbai was used in the tests. Figure 10(a)
dimension of the beam, the mass was concentrated in middle shows the 5% damped response spectrum for a peak ground
third length of the beam in the as-built structure, and middle acceleration (PGA) of 0.2g, which was used to generate the
half of the same in the case of the retrofitted structure. The compatible acceleration time history. The obtained time
center of the mass was approximately 200 mm (7.9 in.) from history was time scaled7,12 by a factor (1/SL)0.5. The typical
the top of the beam. The flanges of the mass were tightly bolted acceleration time history for a PGA of 0.2g, which was used
so that the mass had a tight grip on the beam, and slipping in the experiment, is shown in Fig. 10(b).
was prevented. It is true that distributed load is closer to real- After completing the test setup, a sine sweep test was
istically simulating the floor loads; however, in this case, the performed to evaluate the natural frequencies of the struc-
load was kept concentrated partly due to the ease of fabrication ture. The natural time periods were obtained by the half-

ACI Structural Journal 5


Fig. 9—Arrangement of mass placed on structure.

Fig. 11—Response spectrums from roof response time


history corresponding to 0.1g PGA.
obtained as 0.076 and 0.062 s (13.16 and 16.13Hz) for the
as-built structure and retrofitted structure, respectively. The
periods obtained are also shown in the 5% damped accel-
eration and displacement response spectrums for the two
structures in Fig. 11(a) and (b), respectively, obtained from
the roof response time history corresponding to 0.1g PGA
seismic loading. Due to this difference in the fundamental
time period of the two structures, the zero period acceler-
ation (ZPA) for the as-built structure (0.48g) is higher than
Fig. 10—Response spectrum and time history corresponding the ZPA for the retrofitted structure (0.41g).
to 0.2g PGA used for experiment.
Results of simulated seismic tests on
power bandwidth method.13 The structure was then subjected as-built structure
to simulated seismic loading by providing the acceleration The structure was subjected to uni-axial seismic loading
time history corresponding to the spectral shape, as shown stages for increasing levels of PGA. During the ground motion
in Fig. 10. The tests started with the time histories corre- corresponding to a PGA of 0.1g, no cracks were observed in
sponding to the PGA of 0.1g, and it was gradually increased the structure, suggesting that the structure behaved essentially
in steps of 0.1g until 1.0g, which was the limit of the shake- in linear elastic fashion. The very first cracks, as expected,
table facility. After each loading wave, the test was paused appeared at the base of the column and on the column just
to mark the crack patterns on the structure. below the first floor beam during the loading wave with a
PGA of 0.2g. After the loading wave corresponding to a PGA
EXPERIMENTAL RESULTS of 0.3g, several new cracks at the base of the column were
Results of sine-sweep tests observed (Fig. 12(a)), as well as at the ends of the beams. It
The structures were first subjected to a sine sweep test to is well known from the fundamental structural mechanics that
evaluate the natural time periods of the structures. The time for such frame structures subjected to lateral loads as in the
period corresponding to the first mode was obtained as 0.31 s case of earthquakes, the maximum bending moments occur
(3.23 Hz) for the as-built structure, and 0.23 s (4.35 Hz) for at the base of the structure, which is the reason for typical
the retrofitted structure, while the same for second mode was flexural cracking at the base of the columns. In the case of

6 ACI Structural Journal


Fig. 12—Crack patterns corresponding to different input acceleration levels.
0.4g PGA, several new cracks were observed at the first floor roof level, respectively, obtained for the as-built structure
beam level on both sides (Fig. 12(a)). This is attributed to corresponding to an input loading wave with a PGA of 0.2g,
the fact that once the cracks form at the base of the column, as shown in Fig. 10. The peak floor acceleration (PFA) was
the local stiffness reduces, and the increase in moment is obtained from the measured acceleration-time histories for
not observed. Consequently, the beam is subjected to higher each floor. The plot of PFA versus PGA thus obtained for the
bending moments, resulting in significant cracking. During as-built structure is depicted in Fig. 15. It can be observed
the loading wave of 0.5g PGA, the first diagonal shear cracks that during the initial loading waves, the PFA rises linearly
appeared at the joint (Fig. 12(a)). This highlights the vulnera- (though not proportionally) with the increasing PGA. When
bility of the beam-column joints of non-seismically designed the PGA reaches a value of 0.5g, the rise in PFA is blocked,
structures subjected to earthquakes. Further, this confirms that which is attributed to the joint cracking at this stage. The
to develop a suitable retrofit for this structure, the joints are PFA again rises with increasing PGA before it gets saturated
required to be strengthened or safeguarded, which is essen- for PGA values of 0.8g, 0.9g, and 1.0g. This is attributed
tially one of the major focuses of this work. During the loading to the corresponding increase in hysteretic damping of the
wave of 0.6g PGA, the diagonal shear cracks appeared on the system due to induced damage. This also explains why no
right-side joint as well (Fig. 12(a)). A few hairline diagonal further damage was observed in the tests during last three
cracks also appeared on the joint. On the left-side joint, the loading waves. A summary of experimental results on the
existing diagonal cracks propagated further and the existing as-built structure is given in Table 1.
cracks opened up. At the base of column, a few more flexural To further investigate the system characteristics, the time
tensile cracks appeared. Corresponding to the loading wave history obtained from the experiment corresponding to each
of 0.7g PGA, on the joint, the existing cracks opened, and a PGA level was used to generate a Fourier spectrum from
small amount of spalling was observed from the point of inter- which the natural time period of the structure corresponding
section of the diagonal cracks. From this point onward, the to fundamental mode was obtained. Figure 16 shows the
existing cracks opened, and more spalling was observed from plot of the variation of vibration period corresponding to
the joints. Figure 13(a) represents the final crack patterns and first mode as a function of the PGA. The plot clearly shows
failure modes observed in the test for all joints of the as-built that the fundamental time period of the structure elongates
structure. As expected, the damage suffered by the roof corner with increasing PGA due to the damage inflicted in the
joints was much less compared with the exterior joints of structure. The fundamental time period was increased from
first floor. It can be concluded that joint shear failure was the 0.31 s (for undamaged structure) to 0.45 s after completion
fundamental mode of failure for the structure. of the whole test series. This implies that the structural stiff-
For each loading wave, the response at the first floor and ness for the damaged structure is approximately 50% of the
roof level were measured using accelerometers. Figures structural stiffness of the undamaged structure. As per the
14(a) and (b) show the typical response at the first floor and response spectrum corresponding to input ground motion,

ACI Structural Journal 7


Fig. 13—Final failure modes observed in structures.
with an elongated time period, the structure would attract they make an angle αj with horizontal. The joint deforms in
less seismic forces compared with the undamaged structure. such a way that one potentiometer is elongated by an amount
Further, more damage implies more energy dissipation, and δj, and the corresponding compression in the other diagonal
hence, higher damping. These two reasons are responsible potentiometer is δj′. The joint shear deformation γj can then
for the saturation of peak floor accelerations and damage in be obtained as4
the structure. With further increase in PGA or with a higher
duration of loading, however, more damage could possibly δ j − δ ′j  1 
be inflicted; it was not, however, the objective of the test. γ j = γ1 + γ 2 =  tan α j + 
2l j  tan α j 
Figure 17 displays the plot of the peak displacement (4)
obtained from the displacement time histories for the first
floor and roof as a function of PGA. For the as-built structure, Figure 19 displays the variation of maximum joint shear
it can be seen that with the increasing PGA, the story displace- deformation for exterior joints of the frame as a function of
ments keep rising, which is attributed to increasing PFA (Fig. increasing PGA. It can be observed that there is a significant
15) as well as an elongated time period (Fig. 16). From the increase in the joint shear deformation corresponding to 0.5g
theory of structural dynamics, it is known that the flexible PGA, which is attributed to the development of joint shear
structure (elongated time period) has a higher displacement cracks. A further increase in PGA results in a high increase of
response for the same acceleration. Thus, in the case of the joint shear deformation, and the maximum shear deformation
as-built structure, the displacement demand is as high as 121 corresponding to 1.0g PGA was obtained as 0.0088 radians.
mm (4.76 in.) at the roof (approximately 2.8% global drift). The experiment clearly displayed the vulnerability of the
Joint shear deformations were measured by using diago- beam-column joints of non-seismically designed frames
nally installed potentiometers on the joint panel (Fig. 18). subjected to earthquake loading, and the structure success-
The undeformed length of diagonal potentiometers is lj, and fully served its purpose as a benchmark specimen.

8 ACI Structural Journal


Table 1—Summary of experimental results on as-built and retrofitted structure
As-built Retrofitted
Peak response acceleration, g Peak response acceleration, g
PGA, g Roof First floor Observations Roof First floor Observations
0.1 0.48 0.30 No cracks 0.39 0.29 No cracks
Flexure cracks at base and top of Single flexural crack at base of
0.2 0.81 0.56 0.69 0.52
bottom column bottom column
New flexural cracks at base of bottom New cracks at base of bottom
0.3 1.20 0.83 0.96 0.76
column and at ends of first-floor beam column
Single crack on first floor beam at
0.4 1.53 1.12 New cracks at ends of first-floor beam 1.29 0.98
farthest anchor of haunch
Diagonal shear cracks on left-side Cracks on top of bottom column at
0.5 1.68 1.32 1.59 1.17
exterior joint (first-floor level) farthest anchor
Diagonal shear cracks on right-side
0.6 1.79 1.38 1.78 1.33 Extension of earlier cracks in beam
exterior joint. New cracks in column
Diagonal shear cracks opened and
0.7 2.02 1.66 2.08 1.54 Extension of earlier cracks
propagated. Little spalling observed
0.8 2.23 1.89 Existing cracks opened and propagated 2.43 1.60 Extension of earlier cracks
0.9 2.34 1.89 Existing cracks opened and propagated 2.76 1.86 Extension of earlier cracks
1.0 2.41 1.88 Existing cracks opened and propagated 2.90 2.01 Extension of earlier cracks

Fig. 14—Acceleration time histories measured corresponding to 0.2g PGA input excitation.
Results of simulated seismic tests on floor level as well. During the test corresponding to a PGA of
retrofitted structure 0.4g, a single crack appeared also on the beam at the farthest
The retrofitted structure was also subjected to same test anchor of the haunch. The anchor farthest from the face of the
schedule as that of the as-built structure. For 0.1g PGA column serves as the critical section for flexure in the beam
input ground motion, no cracks were observed in the struc- for the retrofitted structure, and therefore, the first flexural
ture. At 0.2g PGA, a single crack at the column-foundation crack appears there. No flexural cracks, however, appeared
interface was observed. Several new cracks at the base of the on the column in the vicinity of the first floor level. This
column were observed during 0.3g excitation (Fig. 12(b)). shows that the desirable hierarchy of strength is maintained
No cracks, however, were observed at the beam of the first in the joint region with beam cracking before joint or column.
floor level, which was in contrast to the test on the as-built During the loading wave of 0.5g PGA, on the left-side
structure where the cracks appeared at the beam of the first column, the cracks appeared on the column just below the

ACI Structural Journal 9


Fig. 17—Peak floor displacements as function of PGA for
two structures.
Fig. 15—PFA as function of PGA for two structures.

Fig. 18—Calculation of joint shear deformation.


ting the joints using FFHRS. Figures 14(c) and (d) show the
typical response at the first floor and roof level, respectively,
obtained for the retrofitted structure corresponding to an input
Fig. 16—Variation of fundamental time period of structure loading wave with PGA of 0.2g.
with increasing input acceleration. In Fig. 15, the PFA recorded in the experiment on the retro-
fitted structure for the first floor level and roof level are plotted
first floor beam at the farthest anchor location of the haunch
as a function of the PGA. In the case of the retrofitted structure,
(Fig. 12(b)). During the tests on the as-built structure, at the
this plot is practically linear until the end of the test series, unlike
base excitation level of 0.5g, clear shear cracks appeared in
the as-built structure. This is expected because the damage in
the joint. The same is not the case for the retrofitted struc-
the structure is minimal; therefore, large hysteretic damping
ture, however, proving the efficacy of the HRS. For 0.6g
is not expected in the system. Figure 16 also shows the plot
PGA input excitation, the crack patterns were essentially
of the variation of time period corresponding to the first mode
the extension of the previously formed cracks on the beams
as a function of the PGA as obtained from the Fourier spec-
(Fig. 12(b)), while no new cracks were formed at the base
trum of the response corresponding to each PGA level. In this
of the columns. On further loading, the earlier-formed cracks
case, the change in fundamental time period of the structure
propagated further in the beams, and new cracks were formed
is nominal with increasing PGA, and it increased from 0.23 s
at the anchor locations and beyond the haunch element in the
(for the undamaged structure) to 0.265 s after completion of
beam. Figure 14(b) presents the final crack pattern observed
the whole test series. The only significant variation in the time
in the test. Unlike the test on the as-built structure, no diagonal
period occurs for a PGA of 0.3g, which corresponds to various
shear cracks were observed in the joint panels for the retro-
cracks formed at the column base. A summary of experimental
fitted structure, clearly showing the efficacy of the FFHRS in
results on an as-built structure is given in Table 1.
safeguarding the beam-column joints. The crack patterns were
Figure 17 also displays the plot of the peak displacement
limited to the crack formation in beams and columns, while
for the first floor and roof as a function of PGA for retro-
the joint region was essentially crack-free. Also, the cracks
fitted structure. In the case of the retrofitted structure, the
formed in beams and columns were only hairline cracks. No
maximum displacement demand is approximately 50.54 mm
spalling was observed in the test. No diagonal cracks were
(1.99 in.) at the roof (approximately 1.18% global drift).
observed in the corner joints, and only a few minor cracks
Thus, due to the lower time period of the retrofitted structure
were observed at the bottom of the beam. As expected, the
until the end of the test, the displacement demands are rather
damage at the roof members was much less compared with
limited compared with that of the as-built structure.
that at the first floor. The test was stopped due to the limitation
Figure 19 also displays the variation of the maximum joint
of the shake table to provide maximum input acceleration as
shear deformation for exterior joints of the frame as a func-
1.0g. Nevertheless, the experiment clearly showed a signifi-
tion of increasing PGA for the retrofitted structure. It can
cantly improved seismic behavior of the structure by retrofit-

10 ACI Structural Journal


Analysis of response time history showed a significant elon-
gation in fundamental time period for the as-built structure,
while the same remained almost unchanged for the retrofitted
structure. Significant joint distortion for joints was obtained
in the case of the as-built structure, while in the case of the
retrofitted structure, the joint distortion was rather limited.
To have an estimation of the effectiveness of the FFHRS,
strain data obtained from the strain gauge located on the first
floor beam closest to the column face (critical location for
as-built structure) was evaluated. The highest strain obtained
at that location, corresponding to the PGA of 1.0g, was
obtained as 1321 microns for the as-built structure, and as
827 microns for the retrofitted structure. This clearly indi-
Fig. 19—Joint deformation as function of increasing PGA.
cates a strong influence of the presence of haunch elements in
be observed that the maximum joint shear deformation in reducing the bending moments in the beam, and hence, joint
this case (0.0032 radians) is much less compared with that shear forces. Based on the crack patterns, joint shear deforma-
obtained in the case of the as-built structure (0.0088 radians). tions, acceleration and displacement responses, time period
The measured shear deformation suggests no significant variation, and strain data, it can be said that the experimental
distortion of the joint, as also seen from the crack patterns. program clearly displayed and successfully proved the effi-
The experiment clearly displayed that the structure retro- cacy of the retrofit solution in improving the seismic perfor-
fitted with the HRS behaved in an almost linear fashion until mance of non-seismically designed RC frame structures.
the end of the test, and successfully proved the efficacy of The static tests at joint level and the dynamic tests at struc-
the retrofit solution. tural level highlight certain interesting aspects. The experi-
ments at the beam-column joint sub-assemblies were carried
CONCLUSIONS out with different post-installed anchor systems, namely
In this paper, the experimental investigations carried bonded anchors, concrete screws, and expansion anchors.4-6
out on 2-D RC frame structures without and with FFHRS It was observed that in the case when anchors served their
are presented. Previously, HRS was investigated at the purpose well, for example, bonded anchors, a clear beam hinge
subassembly level using various configurations to connect was formed instead of the joint shear failure, as shown in Fig.
the haunch elements with the frame members. The solu- 20(a). In the case when the anchor system did not work well,
tion tested by Pampanin et al.3 where partially prestressed however, (for example, in case of concrete screws) the anchor
external rods were used to connect the haunch element with failure was immediately followed by the joint shear failure,
the frame members, proved the efficiency of the system, but which resulted in a non-ductile undesirable failure mode
has practical limitations for implementation in real life. The (Fig. 20(b)). This aspect cannot be commented upon based
FFHRS system using post-installed anchors to connect the on this work because only one type of anchor has been used.
haunch element with frame members as tested by Genesio It is important to note, however, that in the case of the tests at
and Sharma5 promises to remove this limitation. The tests the subassembly level, once a crack is initiated at the anchor
at the subassembly level proved the efficacy of the FFHRS level (Fig. 2(b)), it grew severely and resulted in concrete
system in improving the joint behavior provided that the cone failure. In the case of the dynamic tests at the structural
anchor system behaves in desirable fashion. level, the crack initiated at the farthest anchor location of the
No tests, however, have been carried out to test the retrofit haunch did not grow very severely, and therefore, the perfor-
solution at structural level under real dynamic loads. The mance of the FFHRS was better. This may be attributed to the
same has been attempted in this work. It should be noted that fact that in the case of seismic loading, the crack growth is
the conclusions provided herein are based on the measured only instantaneous due to the dynamic reversing nature of the
response of one retrofitted specimen only. Further experi- loading. Based on these findings, it appears that the FFHRS
mental data is needed to augment the conclusions. may offer a practical and efficient solution to retrofit struc-
In the case of the as-built structure, severe diagonal shear tures against seismic loading, but these conclusions must be
cracks were observed in the exterior joints, which initiated at supported by further tests. Though, the FFHRS seems to be
input ground motion corresponding to PGA of 0.5g. Most of a practical and efficient retrofit solution, the following points
the damage was concentrated in the beam-column connec- must be kept in mind while suggesting the system for seismic
tion region. It should be noted that beams had high-strength retrofitting of existing structures:
reinforcement, which may also be the reason for lower 1. As discussed previously, based on the tests at joint level,
damage levels in the beams. In case the beams were cast the FFHRS is clearly sensitive to the performance of the anchors
with plain round bars, the damage in the beams may have used to connect the haunch element to frame members (Fig. 20);
arguably been more. The retrofitted structure behaved in an 2. It has been shown from the tests that there is a high prob-
almost linear fashion until the end of the test program at 1.0g ability that fasteners installed in non-cracked concrete will be
PGA. Only hairline cracks were observed in the beams and intercepted by a crack when cracks form.11,14 Therefore, to
columns, while the joint region remained essentially elastic. connect the FFHRS, it is recommended that only anchors suit-
able for performing in cracked concrete should be used; and

ACI Structural Journal 11


successfully conducting the tests, and to K. K. Vaze for providing necessary
support and encouragement to carry out the work.

NOTATION
Hh = vertical projected height of haunch element
hb = overall depth of beam
hc = overall depth of column
Lb = length of beam from loading point to face of column
Lc = length of column between point of contraflexure above and
below beam
Lh = horizontal projected length of haunch element
Mb = bending moment in beam at column face in conventional joint
Mb′ = bending moment in beam at column face in retrofitted joint
Mbh = bending moment in beam at face of haunch element
Fig. 20—Influence of anchor performance on failure mode Mc = bending moment in column at beam face
of retrofitted joints4-6. Mc′ = bending moment in column at beam face in retrofitted joint
Mch = bending moment in column at face of haunch element
3. The HRS leads to a reduction in shear forces transferred SL = linear scaling factor
Vb = beam end load
to the joint, but increases shear forces in the members, as Vc = shear force in column
shown in Fig. 2(c). In the case of the retrofitted joint, the beam Vjh = horizontal joint shear force
member experienced a peak shear force of approximately Zb = lever arm of internal forces in beam at column face
120 kN (27 kip) against 80 kN (18 kip) experienced by the
beam of the as-built joint. Therefore, one should verify while REFERENCES
1. Sharma, A.; Eligehausen, R.; and Reddy, G. R., “A New Model to
designing the HRS that it does not make a structure shear crit- Simulate Joint Shear Behavior of Poorly Detailed Beam—Column Connec-
ical, resulting in an undesirable brittle failure mode. tions in RC Structures under Seismic Loads, Part I: Exterior Joints,” Engi-
neering Structures, V. 33, 2011, pp. 1034-1051.
2. Paulay, T., and Priestley, M. J. N., Seismic Design of Reinforced
AUTHOR BIOS Concrete and Masonry Buildings, John Wiley & Sons, Inc., New York,
Akanshu Sharma is a Postdoctoral Research Engineer at the Institute for 1992, 768 pp.
Construction Materials, University of Stuttgart, Stuttgart, Germany, before 3. Pampanin, S.; Christopoulos, C.; and Chen, T. H., “Development
which he served as a Scientific Officer at Bhabha Atomic Research Centre and Validation of a Metallic Haunch Seismic Retrofit System for Existing
in Mumbai, India. He received his PhD from the University of Stuttgart Under-Designed RC Frame Buildings,” Earthquake Engineering & Struc-
in 2013. He completed his bachelor’s and master’s degrees in civil and tural Dynamics, V. 35, 2006, pp. 1739-1766.
structural engineering from the Regional Engineering College, Silchar, 4. Genesio, G., “Seismic Assessment of RC Exterior Beam-Column
India, and the Indian Institute of Technology (IIT), Delhi, India, in 2001 and Joints and Retrofit with Haunches using Post-Installed Anchors,” PhD
2004, respectively. His research interests include inelastic seismic analysis thesis, University of Stuttgart, Stuttgart, Germany, 2012, 311 pp.
and retrofitting of reinforced concrete structures and behavior of concrete 5. Genesio, G., and Sharma, A., “Seismic Retrofit Solution for Rein-
under high loading rates. forced Concrete Exterior Beam-Column Joints Using a Fully Fastened
Haunch, Part 2-2: Retrofitted Joints,” Test Report No. WS 221/08-10/02,
G. R. Reddy is a Senior Scientific Officer and Head of the structural and IWB, University of Stuttgart, Stuttgart, Germany, 2010, 136 pp.
seismic engineering section of the Reactor Safety Division at Bhabha 6. Genesio, G.; Sharma, A.; Eligehausen, R.; Pampanin, S.; and Reddy,
Atomic Research Center. He received his master’s degree from IIT Bombay, G. R., “Development of Seismic Retrofit Technique of RC Frame Using
Mumbai, and PhD in earthquake engineering from Tokyo Metropolitan Fully Fastened Haunch Elements: Static to Dynamic Testing,” 14th Sympo-
University, Tokyo, Japan, in 1998. His research interests include seismic sium on Earthquake Engineering, Paper No. A 0017, Indian Institute of
behavior and retrofitting of structures. Technology Roorkee, Roorkee, India, 2010, 10 pp.
7. Harris, H. G., and Sabnis, G. M., Structural Modelling and Experi-
R. Eligehausen, FACI, is Professor Emeritus at the Institute for Construc- mental Techniques, CRC Press LLC, Boca Raton, FL, 1999, 808 pp.
tion Materials at the University of Stuttgart, Germany. He received his 8. Morcarz, P., and Krawinkler, H., “Theory and Application of Exper-
Doctor of Engineering from the University of Stuttgart in 1979. He is a imental Model Analysis in Earthquake Engineering,” Report No. 50, John
member of ACI Committees 349, Concrete Nuclear Structures, and 355, Blume Earthquake Engineering Center, Department of Civil and Environ-
Anchorage to Concrete; and Joint ACI-ASCE Committee 408, Development mental Engineering, Stanford University, Stanford, CA, 1981, 263 pp.
and Splicing of Deformed Bars. His research interests include fasteners for 9. Quintana-Gallo, P.; Pampanin, S.; Carr, A. J.; and Bonelli, P., “Shake
concrete structures, bond behavior of reinforcement, and seismic behavior Table Tests of Under-Designed RC Frames for the Seismic Retrofit of
of beam-column joints. Buildings—Design and Similitude Requirements of the Benchmark Spec-
imen,” Proceedings of New Zealand Society of Earthquake Engineering
G. Genesio works at Eligehausen and Asmus, Stuttgart, Germany. He (NZSEE) Conference, Wellington, New Zealand, Paper No. 39, 2010, 12 pp.
received his PhD from the University of Stuttgart. His research inter- 10. Sharma, A.; Reddy, G. R.; and Vaze, K. K., “Shake Table Tests on a
ests include beam-column joints, seismic retrofitting of structures, and Non-Seismically Detailed RC Frame Structure,” Structural Engineering &
anchorage in concrete construction. Mechanics, V. 41, No. 1, 2012, pp. 1-24.
11. Eligehausen, R.; Mallée, R.; and Silva, J., Anchorage in Concrete
S. Pampanin is an Associate Professor of structural design and earthquake Construction, Wilhelm Ernst & Sohn, Berlin, Germany, 2006, 391 pp.
Engineering in the Department of Civil and Natural Resources Engineering 12. Bracci, J. M.; Reinhorn, A. M.; and Mander, J. B., “Seismic Resis-
at the University of Canterbury, Christchurch, New Zealand. He studied tance of Reinforced Concrete Frame Structures Designed for Gravity
civil engineering at the University of Pavia, Pavia, Italy; received his Loads: Performance of Structural System,” ACI Structural Journal, V. 92,
master’s degree in structural engineering at the University of California, No. 5, Sept.-Oct. 1995, pp. 597-609.
San Diego, San Diego, CA; and his PhD in earthquake engineering at the 13. Chopra, A. K., Dynamics of Structures: Theory and Applications to
Politecnico di Milano, Milano, Italy. Earthquake Engineering, Prentice Hall Inc., Upper Saddle River, NJ, 2007,
912 pp.
14. Eligehausen, R., and Sharma, A., “Seismic Safety of Anchorages in
ACKNOWLEDGMENTS Concrete Structures of Nuclear Power Plants,” Proceedings of Post-SMiRT
The tests reported herein were carried out at the Central Power Research Conference Seminar on Advances in Seismic Design of Structures, Systems
Institute (CPRI), Bangalore, under the funded project by BARC. The and Components of Nuclear Facilities, Mumbai, India, 2011, pp. 209-228.
authors are extremely thankful to R. Ramesh Babu and his team at CPRI for

12 ACI Structural Journal

You might also like