You are on page 1of 17

Composites: Part B 31 (2000) 481±497

www.elsevier.com/locate/compositesb

Strength and ductility of reinforced concrete moment frame connections


strengthened with quasi-isotropic laminates
A.S. Mosallam
Department of Civil and Environmental Engineering, California State University, Fullerton, CA 92834, USA

Abstract
This paper introduces an innovative technique using polymer composites for repair and retro®t of reinforced concrete (RC) moment frame
connections. In this pilot study, a total of six half-scale reversal cyclic tests were preformed. The tests were conducted on half-scale
specimens simulating interior beam-to-column sub-assemblages of a typical RC structure. Two connection specimens were used as control
specimens and were tested to failure. These two ªrepairableº damaged specimens were re-tested under a similar loading regime after being
repaired with both epoxy injection as well as carbon±epoxy and E-glass±epoxy quasi-isotropic laminates. To investigate the performance of
the composite systems as retro®t schemes, two other half-scale tests were conducted on undamaged specimens strengthened with both E-
glass±epoxy and carbon±epoxy quasi-isotropic laminates. Test results indicated that the use of composite overlays has led to an appreciable
increase in stiffness, strength, and ductility of these connections. The ductility and strength of the repaired specimens were increased up to 42
and 53% respectively, as compared to the control specimens. Discussion on the advantages and disadvantages of using E-glass±epoxy vs.
carbon±epoxy laminates is also presented. q 2000 Elsevier Science Ltd. All rights reserved.
Keywords: Seismic repair

1. Introduction government and the industry was allocated to verify,


support, and develop repair systems to upgrade the struc-
For the past few years, several major research projects tural performance of existing slabs, beams columns and
were launched to investigate the feasibility of using com- other structural members. In rehab applications, test results
posites in both seismic and corrosion repair of structural indicated that the use of composite laminates leads to an
systems made of reinforced concrete, steel, and wood appreciable increase of the loading capacity of the ¯oor
materials. The State and Federal Government, as a joint system (beam and slabs), columns and walls. The gain in
program with the industry, sponsored the majority of these the structural capacity is translated into the ability of adding
programs. The overwhelming experimental and analytical more loads to the existing structure. These additional loads
results have encouraged the practicing civil engineers and must follow a proper load path from the upgraded ¯oor
the construction industry to consider polymer composites as system to the upgraded columns and/or walls via some
an alternative construction material and system. One of the type of connection. If this connection is not properly
successful applications of polymer composites is the seis- upgraded to carry the additional loads, severe limitations
mic repair and retro®t of reinforced concrete columns. The could be imposed on the use of composite reinforcements
advantage of composites in this application is the ability of and on the allowable upgrade capacity. The main reason is
decoupling the stiffness and strength, by applying the major- that, by increasing the strength and stiffness of the
ity of ®bers in the hoop directions. This application has been connected members (e.g. beams and columns), the possi-
extended to cover not only the ductility enhancements, but bility of plastic hinge formation in the columns will increase
also in upgrading of both stiffness and strength of reinforced substantially. Structural engineers have avoided this
concrete columns. In addition, similar applications includ- concept when designing reinforced concrete Ductile
ing repair and retro®t of both reinforced and unreinforced Moment Resisting Frames (DMRFs). The design philoso-
bearing as well as shear walls, slabs, and beams have been phy of DMRF frames follows what is called ªweak girder±
adopted. A comprehensive state-of-the-art review of the strong columnº concept. According to this design concept,
different composite repair systems is presented by Mosal- the connection is designed in such a way that the joints and
lam [1] and by Marsh [2]. the column remain essentially elastic under the action of
As mentioned earlier, a heavy investment by both the earthquake forces, in order to ensure adequate energy
1359-8368/00/$ - see front matter q 2000 Elsevier Science Ltd. All rights reserved.
PII: S 1359-836 8(99)00068-2
482 A.S. Mosallam / Composites: Part B 31 (2000) 481±497

Fig. 1. Typical exterior (knee) frame connection failureÐNorthridge Earthquake, January 1994.

dissipation and to provide proper lateral stability of the cracks and to provide adequate con®nement to the joint.
reinforced concrete structure. These stirrups are dif®cult to be placed in the very limited
joint zone and can generally result in improper compaction
of concrete at the connection area.
2. In¯uence of connection behavior of the overall seismic One of the determining factors in designing beam±
performance of DMRF structures column connections in DMRF structures is MR, the ratio
of column-to-beam ¯exural capacity given by
Stiffness or strength degradation of the connection in a
S MC
DMRF structure has a major impact on the lateral-load MR ˆ
resistance of such structures. For this reason, the joint has S MB
been identi®ed as the ªweak-linkº in DMRF structures. where MC, MB are the sum of the ¯exural capacities of the
During an earthquake, connection damages can lead to columns and beam, respectively. The experimental results
substantial drifts and can increase the possibility of building indicated that in order to avoid formation of the plastic
collapse due to what is called the ªP-deltaº effects. hinges in the joint, MR is 1.40. The reason behind that is,
The major in¯uence of beam±column connections on the if the frame structure is subjected to large seismic forces that
structural integrity and seismic performance of reinforced result in some deterioration of the column strength, there is a
concrete structures became more evident after the 1989 possibility that the plastic hinge may shift from the beams to
Loma Prieta, the 1994 Northridge earthquakes, the Kobe the column, which should be avoided as was mentioned
earthquake of Japan, and the August 1999 Earthquake in earlier.
Turkey. Post-earthquake reports of the Loma Prieta indi- The weak beam±strong column approach requires that in
cated that one of the main reasons behind the collapse of one side, the steel rebars will go from yielding to some or no
the Cypress Viaduct bridge, and the damage of the China compression at the other side of the joint. In this case, the
Basin and the I-80 Nemit Freeway is the failure of connec- beam steel rebars will be exposed to very high bond stresses.
tions. A site survey, conducted by the author, of several This will result in substantial degradation in the bond
parking structures in the Los Angeles in January 1994 strength, and possibility of slippage of the steel reinforce-
following the Northridge earthquake, indicated that collapse ment when exposed to seismic forces. In addition, during an
of several portal frame structures were mainly due to the earthquake, the joints will likely be exposed to biaxial
failure of beam±column and column±base connections (see forces that will result in an increase in the shear stresses
Fig. 1). at the joint area that can cause severe cracking and bond
Designing beam±column joints is considered to be a deterioration at the connection zone.
complex and challenging task for the structural engineer.
This is because although the sizing of the connection is
determined by the size of the framing members, these joints 3. Related research work
will be subjected to a different set of loads from those used
in designing both the column and the beams. For example, it Over the years, several seismic repair and retro®tting
is necessary to provide a relatively large number of trans- techniques, including pressure epoxy injection, epoxy
verse reinforcement (in form of steel stirrups) to control impregnation [3], external steel plating [4,5], and concrete
A.S. Mosallam / Composites: Part B 31 (2000) 481±497 483

jacketing [6] have been investigated. However, there are a of retro®tting older reinforced concrete bridge T-joints
number of drawbacks associated with these techniques. The using special dowels and post-tensioning rods. The use of
1995 Federal Emergency Management Agency (FEMA)-97 corrugated steel jacket attached to the joint region was
Report questions the effectiveness of the pressure injection studied by Biddah et al. [5] in 1997. The corrugated steel
technique in completely restoring bonds between the rein- jacket was found to be an effective system of upgrading the
forcement and concrete. The epoxy impregnation method, joint capacity by increasing the joint shear strength and the
while being able to overcome the partial ful®llment of air energy dissipation capacity. Bollo et al. [13] tested a three-
void problem, is extremely dif®cult to apply on large prac- span segment of the Cypress Street Viaduct bridge. In this
tical scales. The use of bonded steel plates has several disad- study, retro®tting techniques including external post-
vantages including high cost, possibility of steel corrosion at tensioning steel rods and the addition of external shear rein-
the interface resulting in a premature bond-line failure, as forcement were examined. The results indicated that these
well the requirements of heavy equipment at the job site. retro®tting techniques increased both strength and ductility
As compared to steel jacketing, the use of polymer with a minimal effect on the joint stiffness. Parvin and
composites repair and retro®t systems provide several Granata [14] presented a preliminary discussion on the use
unique advantages. The key advantage of composites in of FRP overlays at exterior beam±column connections.
these applications is its tailorability, which enables the They concluded, through ®nite element analysis, that the
engineer to decouple stiffness and strength (stiffer columns peeling stress at the junction of the tensile faces of the
would attract more forces during the earthquake which are beam and column is critical. The study further indicated
not included in the original design). This property is critical that the use ®ber wrap over the ®ber overlay would be
in the seismic repair of columns. Results of previous effective in preventing peeling of the overlay due to stress.
research studies demonstrated that ductility was enhanced Sexsmith et al. [15], investigated the cyclic behavior of a
without changing the stiffness of the members by applying 45% prototype of the Vancouver Oak Street Bridge retro-
the majority of ®bers in the hoop direction. In this con®g- ®tted with four different details. The details were a combi-
uration, the composite reinforcement was highly effective nation of cap beam longitudinal and vertical post-tensioning
because it con®ned the core concrete and prevented the steel and unidirectional E-glass±epoxy laminates. The composite
longitudinal rebars from buckling under reversal loading. laminates were applied on the cap beam and columns (and
The use of polymer composites in reinforced concrete repair not at the joint), while the cap-beam was longitudinally pre-
and rehabilitation is discussed in detail by Mosallam [1], stressed at 342 psi (2.4 MPa) using external Dywidag bars.
Haroun [7], and Duthinh [8]. Test results indicated that the composite retro®t system
Mosallam et al. [9] conducted a pilot study on the use of preformed very well up to a displacement ductility of 9,
3D braided graphite±epoxy composites and high-strength where the test had to be terminated due to the displacement
adhesives for repair and retro®t of damaged interior and limitation of the loading system. In 1996, a ®eld study was
exterior steel beam±column connections. Test results indi- conducted by researchers at University of Utah on the feasi-
cated that the proposed techniques satis®ed the ductility bility of using polymer composites for retro®t beam±
requirements recommended by the 1995 and 1997 FEMA column pier joints at Interstate 80 [16]. Preliminary results
guidelines for steel moment frame structures. The use of on the use of polymer composites for repairing concrete
polymer composites as a structural repair and rehabilitation joints were reported by Mosallam et al. [17].
for masonry walls was examined by Haroun and Ghoneam
[10].
Over the years, the American Concrete Institute (ACI
4. Experimental program
352) Committee has been working on developing state-of-
the-art reports to assist the structural engineers in designing
reliable reinforced concrete joints. Unfortunately, the 4.1. Materials and methods
majority of these valuable design recommendations are
only applicable to new construction. Based on the latest 4.1.1. Beam±column specimens
ACI 352R-91 [11], (the committee is currently working on A total of six half-scale beam±column connection tests
the ®nal draft of a new design document that will be avail- were conducted in this pilot study. The beam±column
able shortly this year), the majority of the older joints do not assemblages were designed using earlier design method-
meet the con®nement criteria set forth by this committee. ologies used in the 1960s. The specimen size and reinforce-
For this reason, there is an urgent need to develop reliable ments were limited by ^50 kips (222.40 kN) cyclic load
and cost-effective methodologies to repair and rehabilitate capacity of the hydraulic actuators and the test frame. The
older connections in existing reinforced concrete structures. yielding strength ( fy) of the steel rebars was 60,000 psi
As was mentioned earlier, several attempts using conven- (413.68 MPa). Number 5 (f 15.9 mm) rebars were used
tional materials and systems have been investigated. For for both column vertical reinforcement as well as for the
example, Lowes and Moehle [12], at the University of beam horizontal reinforcements. Number 2 (f 6.35 mm)
California at Berkeley, investigated the seismic behavior rebars were used for stirrups. The specimen ends were
484 A.S. Mosallam / Composites: Part B 31 (2000) 481±497

Fig. 2. Beam-to-column connection test setup.

supported using a special hinged ®xture using steel sleeves the ®bers, and excess epoxy was squeezed off to eliminate
embedded in the reinforced concrete. the chances of creating air voids and epoxy-rich weak links.
All the laminates were bonded to concrete surfaces, and no
4.1.2. Composites and epoxy system wrapping was used. Although wrapping would have been
Both E-glass±epoxy and carbon±epoxy composites were effective in providing con®nement near the joint area, it
used in these experiments. The ®ber architecture of all was decided not to use it. The reason behind this decision
laminates was quasi-isotropic (08/908/^458). This laminate is that these pilot tests were designed to match the site
design was based on the cyclic loading demand and the conditions for a typical DMRF building, where it would be
anticipated directions of both ¯exural and shear cracks. impractical to wrap around the beam section because of the
The E-glass laminates were preformed using stitching existing ¯oor slab. However, the second phase of this
techniques. Due to the unavailability of the preformed program is underway, where both overlays and wrapping
stitched Quadra-axial carbon ®ber laminates at the time of are used.
the experiment, the quasi-isotropic laminates were
constructed using two layers of 08/908 laminates with an 4.1.3. Test ®xtures and instrumentation
offset of ^458. In addition, two unidirectional diagonal The specimens were tested in a 30 ft (9.15 m), 2D test
straps placed at the top surface of the laminate (at the frame. This test frame is equipped with dual hydraulic
joint). The objective of using these straps is to minimize actuators and each actuator has the capacity of
the possibility of premature delamination failure at the ^50,000 pounds (^222.40 kN). Load, de¯ection, and
corners due to the development of high radial tensile stres- strain were automatically recorded using a computerized
ses at the curved corners. E-glass±epoxy laminates were data acquisition system. The relative rotation between the
used for repair specimen SP-1, and for retro®tted specimen beam and the column was captured using four LVDTs as
SP-3. For connection specimens SP-2 and SP-4, carbon± shown in Fig. 2.
epoxy laminates were used.
In these experiments, three types of epoxies were used, 4.1.4. Loading protocol
namely, a high-modulus±high-strength epoxy paste, a high- The reversal loads were applied to the top of column
modulus±low-viscosity±high-strength epoxy to ®ll cracks centerline using ^50 kips (^222.40 kN) hydraulic actuator.
and voids on the damaged SP-1 and SP-2, and a two-part During the load-control regime, an increment of 2 kips
high-modulus±high-strength±medium-viscosity epoxy as (8.9 kN) per cycle was used. An increment of 0.25 in.
the polymer matrix for the composite lay-ups. Prior to the (63.50 mm) was used for the displacement-control portion
application of the composite systems, two methods using of all tests. The loading frequency was selected at 0.25 Hz.
epoxy were used to ®ll the cracks. For specimen SP-1, a Initially, the load-control regime was used to capture the
manual epoxy gravity-feeding method was used, while steel reinforcement yielding point of specimens SP-1 and
specimen SP-2 was repaired using powered epoxy injection SP-2. The displacements at yield were recorded and used as
technique. All concrete surfaces were ground smooth, wiped the calculation baselines for the ductility comparison.
clean, and dried completely prior to the application of the Subsequent tests were performed using displacement-
composite systems. During the application of the polymer control regime up to the ultimate failure load of the connec-
composites, care was taken to ensure full impregnation of tion specimens. Fig. 3 depicts the typical load-control
A.S. Mosallam / Composites: Part B 31 (2000) 481±497 485

Fig. 3. Typical load control regime history (for SP-1 and SP-2) and displacement control regime history for all specimens.

Fig. 4. Specimen SP-1 load-controlled test results: (a) P/d ; (b) M/Q ; (c) P/e (tension); (d) P/e (compression).

regime history for specimens SP-1 and SP-2 and the typical 5. Experimental results
displacement regime history for all the specimens. After
repairing pre-cracked specimens SP-1 and SP-2, these two 5.1. Control specimen SP-1
specimens were re-designated as SP-1R and SP-2R.
Specimens SP-3 and SP-4 (undamaged specimens) were This specimen represents an existing old structure. Fig. 4
retro®tted with the same laminate architecture of E-glass shows the load-control test results of this specimen. This
and carbon composites, respectively. Displacement controlled initial test was performed to characterize the specimen up
regime test were performed on SP-1R, SP-2R, SP-3, and SP-4. to its yielding point. The load corresponding to the steel
486 A.S. Mosallam / Composites: Part B 31 (2000) 481±497

Fig. 5. Crack distributions of control specimen SP-1.

yielding was found experimentally from the strain data that of specimen SP-1. Fig. 9 shows the result of the load-
and was approximately 25 kips (111.20 kN) with the control experimental results of specimen SP-2. As for
corresponding vertical displacement of 0.5 in. (12.70 mm). specimen SP-1, this test was conducted up to the yielding
Flexural and shear cracks were observed close to the panel point of the control specimen. The load corresponding to the
zone and their size ranged from 0.0625 in. (1.6 mm) to steel yielding was found experimentally from the strain data
0.125 in. (3.2 mm) in width. Larger cracks were observed (approximately 26 kips (115.65 kN) with the corresponding
and the propagation of the cracks was well into the joint. vertical displacement of 0.4 in. (10.16 mm)). The size and
Fig. 5 shows a close-up view of the damaged area. Samples distributions of cracks were similar to ®rst control specimen
of the displacement-control experimental plots are shown in SP-1.
Fig. 6. Fig. 10 shows a sample of the experimental results of
control specimen SP-2. Severe cracks were developed in the
5.2. E-glass±epoxy repaired specimen SP-1R beam at approximately 12 in. (305 mm) from the face of the
vertical member. Fig. 11 shows the crack distribution of speci-
The surface of the pre-cracked control specimen SP-1
men SP-2. Large size cracks and concrete spalling were
was cleaned and loose spalled concrete pieces were
observed at the left side of the horizontal member (see Fig. 12).
removed. All the cracks were gravity-®lled with Sikadur 35
high-modulus±low-viscosity epoxy (product of Sika Corp.).
The epoxy was allowed to cure for 72 h after application. The 5.4. Carbon±epoxy repaired specimen SP-2R
concrete surfaces were then ground smooth and quasi-isotro-
Unlike the E-glass±epoxy repaired specimen, SP-1R,
pic (08/908/^458) E-glass composite laminates were applied
where gravity feed method was used in applying the
with Sikadur 32. In this case, SP-1 became SP-1R after
epoxy to the cracks, pressure epoxy injection method was
the applying the E-glass±epoxy repair system.
used. The epoxy resins used were the same types as for SP-
The repaired specimen was tested to failure under dis-
1R specimen. The injected epoxy was allowed to cure for
placement-control regime as shown in Fig. 3. This test
48 h after application. The concrete surfaces were then
was terminated at about 30 s before total collapse. The
ground smooth, cleaned, and a quasi-isotropic carbon±
initial failure was initiated at a distance of about 4 in.
epoxy laminates were applied.
(101.6 mm) from the right side of the outer edge of the
Similar to the ®rst repaired specimen, the control
connection panel zone. No surface cracks were observed
specimen SP-2 was re-designated SP-2R after the
within the joint. Fig. 7 shows the failure mode of specimen
application of the carbon±epoxy composite repair system.
SP-1R. As shown in the ®gure, the composite laminate
The same displacement-control regime, as in specimen
failed in shear along the vertical boundary of the joint. A
SP-1R, was used. At the third compressive stroke, a loud
subsequent examination revealed cohesive failure at the
crackling sound was heard indicating the initiation of a local
concrete±composite interface. Fig. 8 depicts the data plots
failure of the composite laminate. The test sequence was
for SP-1R.
terminated after 13 s due to visual observation of large
5.3. Control specimen SP-2 spalling from the left side of the beam. It was observed
that large deformation occurred at the same area that had
The ®rst part of testing specimen SP-2 was identical to large severe cracks in the control specimen SP-2 (the same
A.S. Mosallam / Composites: Part B 31 (2000) 481±497 487

Fig. 6. Specimen SP-1 experimental test results: (a) P/d ; (b) M/Q ; (c) P/e (tension); (d) P/e (compression).

test specimen before repair). The ultimate failure mode was a ultimate failure of specimen SP-2R. It is anticipated that if
combination of a shear failure (shear failure was initiated at wrapping around the cross-section was used at this location,
the top of the vertical face of the subassembly and was propa- this type of failure would have been either avoided or
gated at about 458 along the depth of the horizontal member), limited by the added con®nement of the composite
and a concrete cohesive failure at the top and the bottom of wraps. Samples of the experimental results are shown
the beam due to that lack of con®nement. Fig. 13 shows the in Fig. 14.

Fig. 7. Ultimate failure mode of repaired specimen SP-1R.


488 A.S. Mosallam / Composites: Part B 31 (2000) 481±497

Fig. 8. Specimen SP-1R experimental test results: (a) P/d ; (b) M/Q ; (c) P/e (bottom steel).

Fig. 9. Specimen SP-2 load-control test results: (a) P/d ; (b) M/Q ; (c) P/e (tension); (d) P/e (compression).
A.S. Mosallam / Composites: Part B 31 (2000) 481±497 489

Fig. 10. Specimen SP-2 experimental test results: (a) P/d ; (b) M/Q ; (c) P/e (tension); (d) P/e (compression).

5.5. E-glass±epoxy retro®tted specimen SP-3 load, a loud crackling noise was heard. The initiation of
failure occurred at about 2 in. (50 mm) from the vertical
The ªundamagedº beam-column specimen, SP-3, was member face. The E-glass±epoxy laminates failed in shear
retro®tted with the same lay-up of E-glass±epoxy quasi- at the right side of the beam (see Fig. 15). At the ultimate
isotropic laminates as for the repaired specimen SP-1R. failure load, no bond line failure was observed, instead a
The same displacement controlled loading regime was concrete cohesive failure was the major mode of failure.
used. After about 7 s from the application of the cyclic Samples of the experimental results are shown in Fig. 16.

Fig. 11. Ultimate failure of control specimen SP-2.


490 A.S. Mosallam / Composites: Part B 31 (2000) 481±497

Fig. 12. Large size cracks and concrete spalling of specimen SP-2 at ultimate load.

Fig. 13. Ultimate failure mode of repaired specimen SP-2R.

5.6. Carbon±epoxy retro®tted specimen SP-4 duration of the test. Initially, no surface cracks were
observed. However, at about 80% of the ultimate load, the
The undamaged specimen SP-4 was retro®tted with the carbon composite laminates suffered from a local shear fail-
same lay-up of carbon±epoxy lamination as used in the ure which was initiated at the right side of the beam and
second repaired specimen SP-2R. The same displacement- followed by a concrete cohesive failure at the top side of the
control loading regime was used. Loud crackling sound was horizontal member (see Fig. 17). Samples of the experimen-
heard at early stage of loading and continued through out the tal results are shown in Fig. 18.

Table 1
Test result summary

Specimen ID Composite system: Ultimate loadÐkips (kN) Ultimate momentÐkips (kN m)


repair (R), retro®t
(RT)

SP-1 None 37.09 (164.97) 911.42 (102.98)


SP-1R E-glass±epoxy (R) 44.42 (197.58) 1091 (123.30)
SP-2 None 28.87 (128.41) 709 (80.18)
SP-2R Carbon±epoxy (R) 30.11 (133.93) 740 (83.62)
SP-3 E-glass±epoxy (RT) 41.50 (184.59) 1018 (115.03)
SP-4 Carbon±epoxy (RT) 42.00 (186.82) 1031 (116.50)
A.S. Mosallam / Composites: Part B 31 (2000) 481±497 491

Load vs. Displacement Moment vs. Θ


50 1000
40 800
30 600

Joint Moment (in-kips)


20 400
Load (kips)

10 200
0 0
-10 -200
-20 -400
-30 -600
-40 -800
-50 -1000
-1 -0.5 0 0.5 1 -0.02 -0.01 0 0.01 0.02
Displacement (inches) Rotation (radians)

Load vs. Strain (tension stroke) Load vs. Strain (compression stroke)

35 0

30 -5

-10
25
Load (kips) -15
Load (kips)

20
-20
15
-25
10 -30
5 -35

0 -40
-1.20E-03 -8.00E-04 -4.00E-04 0.00E+00 -1.20E-03 -8.00E-04 -4.00E-04 0.00E+00
Strain Strain

Fig. 14. Specimen SP-2R experimental test results: (a) P/d ; (b) M/Q ; (c) P/e (tension); (d) P/e (compression).

6. Hysteric behavior specimen before (SP-1) and after the repair with E-glass±
epoxy laminates (SP-1R) is shown in Fig. 20. The ®gure
The column tip load±displacement envelope plots for all shows that the use of the composite system did not only
connection specimens are shown in Fig. 19. A comparison restore the original capacity of the damaged specimen, but
between the load±displacement envelops of ®rst control also upgraded the ultimate load capacity by an additional

Fig. 15. Failure mode of retro®tted connection specimen SP-3.


492
A.S. Mosallam / Composites: Part B 31 (2000) 481±497
Fig. 16. Specimen SP-3 experimental test results: (a) P/d ; (b) M/Q ; (c) P/e (tension); (d) P/e (compression).
A.S. Mosallam / Composites: Part B 31 (2000) 481±497 493

Fig. 17. Failure mode of specimen SP-4.

53% in compression and about 18% in tension. Also, the The same comparison is presented in Fig. 21, between the
displacement at ultimate load was increased by about 29% in second control specimen before (SP-2), after repair with
tension and 36% in compression as compared to the ultimate carbon±epoxy laminates (SP-2R). There was a slight
displacement of the pre-damaged control specimen SP-1. increase in stiffness and strength of this connection repair

Fig. 18. Specimen SP-4 experimental test results: (a) P/d ; (b) P/e (tension).
494 A.S. Mosallam / Composites: Part B 31 (2000) 481±497

Fig. 19. Load±displacement envelope plots for all connection specimens.

detail. As it is shown in the ®gure, a rapid degradation in the A load±displacement behavior comparison between the
strength occurred at a load of 28 kips (125 kN). The lack of control ªundamagedº specimen (SP-2) and the E-glass±
ductility of this detail can be attributed to lower ultimate epoxy retro®tted specimen (SP-3) is shown in Fig. 22.
elongation of the carbon ®bers (about 1%) as well as the From the ®gure, one can see that a strength gain of about
quasi-isotropic lay-up of the applied composite laminates. 44% as compared to the unreinforced specimen was

Fig. 20. Load±displacement envelope plots for SP-1 and SP-1R.


A.S. Mosallam / Composites: Part B 31 (2000) 481±497 495

Fig. 21. Load±displacement envelope plots for SP-2 and SP-2R.

achieved. A slight increase in the ductility was also a moderate increase in both the connection strength and
observed. The force±displacement behavior of unreinforced stiffness was achieved. For example, the increase in the
detail SP-1 and the carbon±epoxy retro®tted connection strength was about 10% while the stiffness increase was
detail SP-4 is presented in Fig. 23. As shown in this ®gure about 20% as compared to the control ªunreinforcedº

Fig. 22. Load±displacement envelope plots for SP-2 and SP-3.


496 A.S. Mosallam / Composites: Part B 31 (2000) 481±497

Fig. 23. Load±displacement envelope plots for SP-1 and SP-4.

connection detail. The increase in the stiffness can be concrete connections. In cases where strength is the
attributed to the high modulus of carbon composites. major design criterion, test results indicated that the use
Table 1 presents a summary of the test results. of carbon±epoxy quasi-isotropic laminates is recom-
mended over the E-glass±epoxy composites. However,
if the ductility is the major criterion, E-glass±epoxy will
7. Conclusions offer better performance as compared to carbon±epoxy
composites. This conclusion is based on the speci®c lay-
Based on the results of this experimental program, the
up, and geometry used in this pilot project.
following general conclusions were reached:
6. In all tests, a cohesive failure was achieved. This
1. The experimental results indicated that the use of quasi- failure mode is desirable to ensure complete bond
isotropic polymer composite laminates increases both the between both the sound concrete and the composite
rotational stiffness and the ultimate strength of the rein- laminate(s).
forced concrete moment frame connections.
2. The control ªun-retro®ttedº specimens exhibited rapid 8. Recommendations and future research
degradation in both stiffness and strength.
3. The use of E-glass±epoxy quasi-isotropic composites Although the results of this pilot research project have
contributed in an appreciable increase in stiffness, provided valuable information regarding the overall full-
strength, and ductility of the repaired connection. For reversed cyclic loading behavior of reinforced concrete
example, an increase of 42% in the ductility of repaired moment frame connections repaired and retro®tted with
specimen, SP-1R, was achieved (3.40 as compared to 2.4 polymer composites, there are several related aspects that
for the control specimen SP-1). need further investigation. For example, additional experi-
4. The use of E-glass±epoxy quasi-isotropic laminates to mental work is needed to further isolate several variables in
retro®t undamaged connections in specimen SP-3 has order to fully characterize their effects on the repairing and
contributed in an increase of about 20% of the initial retro®tting of the reinforced concrete moment frame
stiffness, as compared to the initial stiffness of the control connections. This includes: (i) the effect of changing the
specimen SP-2. ®ber architecture of the laminate; (ii) the effect of adding
5. Due to the inherent lower stiffness of E-glass±epoxy con®nement to the column regions (in this study, and for
composites, this system could be used to enhance the practicality reasons, no con®nement was added to the beam
ductility of both repaired and retro®tted reinforced due to the interference of the ¯oor slab in a typical existing
A.S. Mosallam / Composites: Part B 31 (2000) 481±497 497

reinforced concrete building); (iii) the effect of changing the [3] Wolfgram French C, Thorp GA, Tsai W. Epoxy repair techniques for
load history; (iv) the effect of changing the specimen size moderate earthquake damage. ACI Structural Journal 1990;July±
August:416±24.
and varying the beam and column cross-sections; (v) the [4] Flexural retro®t of rectangular reinforced concrete bridge columns by
case of exterior beam±column (knee) connections, (vi) the steel jacketing: experimental studies. Report No. SSRP-93/01.
case of eccentric beam connections; (vii) the effect of 3D Department of Applied Mechanics and Engineering Sciences, Univer-
elements such as beams and slabs on the connection beha- sity of California at San Diego, San Diego, CA, USA. 215 pp.
vior; (vii) the effect of the use of epoxy injection vs. surface [5] Biddah A, Ghobarah A, Aziz TS. Upgrading of nonductile reinforced
concrete frame connections. Journal of Structural Engineering, ASCE
epoxy repair; and (viii) the use of other types of composite
1997;123(8):1001±10.
systems including Kevlar±epoxy. Currently, the second [6] Alcocer SM, Jersa JO. Strength of reinforced concrete frame connec-
phase of this program is in progress at California State tions rehabilitated by jacketing. ACI Structural Journal (American
University at Fullerton (CSUF) and the results will be avail- Concrete Institute, Detroit, MI, USA) 1993;90(3):249±61.
able in the near future [19]. The ongoing program includes [7] Haroun, M. Seismic evaluation of composite repair systems for repair
and rehabilitation of bridge columns. Caltrans Report, 1998.
the effect of column full con®nement and beam partial
[8] NIST Workshop on Standards Development for the Use of Fiber
con®nement on the overall seismic behavior of beam± Reinforced Polymers for the Rehabilitation of Concrete and Masonry
column joints. Structures. In: Duthinh D, editor. Report No. NISTIR 6288. US
One of the critical issues that needs further investigation Department of Commerce, Washington DC. January 1998.
is the durability and the long-term performance of polymer [9] Mosallam AS, Chakrabarti PR, Arnold M. Making the connection.
Civil Engineering Magazine, ASCE 1999;69(4):56±9.
composites. This includes temperature (and ultimately ®re),
[10] Haroun MA, Ghoneam EH. Seismic performance testing of masonry-
humidity, chemical attacks, creep and creep rupture, and in®lled frames retro®tted by ®ber composite. Proceedings of 15th
other environmental factors on stressed and unstressed International Modal Analysis Conference, vol. 2. Orlando. February
specimens. Extensive durability work has been conducted 1997. p. 1650±6.
at the Aerospace Corporation under the sponsorship of [11] ACI Committee 52. Recommendations for design of beam±column
joints in monolithic reinforced concrete structures (ACI 352R-91).
California Department of Transportaion (Caltrans) [18].
American Concrete Institute, Detroit, MI, USA. 1991.
Recently, a joint durability research program has been [12] Lowes LN, Moehle JP. Seismic behavior of retro®t of older reinforced
initiated between CSUF and the Aerospace Corporation as concrete bridge T-joints. Report No. UCB/EERC-95/09. University of
a part of the ICBO (International Conference of Building California at Berkeley, Berkeley, California, USA. 1995.
Of®cials) Certi®cation Program for polymer composite [13] Bollo ME, Mahin SA, Moehle JP, Stephen RM, Qi X. Observations
and implications of tests on the Cypress Street Viaduct Test Structure.
repair systems.
Report No. UCB/SEMM-90/21. University of California at Berkeley,
Berkeley, CA. 1990.
[14] Parvin A, Granata P. Use of FRP overlays at beam±column connec-
Acknowledgements
tions. Proceedings of the Second International Conference on Compo-
sites in Infrastructures. Tucson, AZ. 1998 January 5±7. p. 213±18.
The effort and contribution of Mr E. Lau, Hughes Space [15] Sexsmith R, Anderson D, English D. Cyclic behavior of concrete
and Communications Co., and Dr Chakrabarti, and Mr bridge bents. ACI Structural Journal (American Concrete Institute,
J. Kiech, California State University, Fullerton are highly Detroit, MI, USA) 1997;94(2):103±13.
appreciated. The epoxy materials were donated by Sika [16] Gergely I, Pantelides CP, Nuismer RJ, Reaveley LD. Bridge pier
retro®t using ®ber-reinforced plastic composites. Journal of Compo-
Corp., and the epoxy injection work was conducted and sites for Construction, ASCE 1998;2(4):165±74.
donated by ACE Restoration and Waterproo®ng. [17] Mosallam AS, Chakrabarti PR, Lau EK. Concrete concertinos. Civil
Engineering Magazine, ASCE 1999;69(1):42±4.
[18] Steckel G, Bauer J, Hawkins G, Vanik R, Johnson E, Nokes J. Aero-
References space Report No. ATR-99 (7524)-2. Quali®cations for seismic retro-
®tting of bridge columns using composites, 3 volumes. January 1999.
[1] Advanced seismic repair and rehabilitation structural system. In: [19] Mosallam A, Chakrabarti P, Sim S, Elsanadedy M. Seismic response
Mosallam, A, editor. Proceedings, CSUF State-of-the-Art Conference of reinforced concrete moment connections repaired and upgraded
on Repair and Rehabilitation of Structures. Fullerton, CA. 1998 with FRP composites. In: Mossallam A (editor), Innovative systems
November 9. 201 pp. for seismic repair and rehabilitation of structures: design and applica-
[2] Marsh G. Seismic retro®t provides opportunities for FRP. Reinforced tion. Proceedings of SRRS2 Conference, Fullerton, CA, 21±22
Plastics Magazine 1998;42(3):38±43. March, 2000.

You might also like