Professional Documents
Culture Documents
Abstract: Recent earthquakes have shown the vulnerability of unreinforced masonry 共URM兲 buildings, which have led to an increasing
demand for techniques to upgrade URM buildings. Fiber reinforced polymer 共FRP兲 can provide an upgrading alternative for URM
buildings. This paper presents results of dynamic tests investigating the in-plane behavior of URM walls upgraded with FRP 共URM-FRP兲.
These tests represent pioneer work in this area 共dynamic and in-plane兲. Five half-scale walls were built, using half-scale brick clay units,
and upgraded on one face only. Two moment/shear ratios 共1.4 and 0.7兲, two mortar types 共M2.5 and M9兲, three composite materials
共carbon, aramid, and glass兲, three fiber structures 共plates, loose fabric, and grids兲, and two upgrading configurations 共diagonal “X” and full
surface shapes兲 were investigated. The test specimens were subjected to a series of synthetic earthquake motions with increasing
intensities on a uniaxial earthquake simulator. The tests validate the effectiveness of the one side upgrading: the upgrading technique
improved the lateral resistance of the URM walls by a factor ranging from 1.3 to 2.9; however, the improvement in the lateral drift was
less significant. Moreover, no uneven response was observed during the test due to the single side upgrading. Regarding the upgrading
configurations, the bidirectional surface type materials 共fabrics and grids兲 applied on the entire surface of the wall 共and correctly anchored兲
can help postpone the three classic failure modes of masonry walls: rocking 共“flexural failure”兲, step cracking, and sliding 共“shear
failures”兲. Additionally, in some situations, they will postpone collapse by “keeping the bricks together” under large seismic deformations.
On the other hand, the diagonal “X” shape was less successful and premature failure was developed during the test.
DOI: 10.1061/共ASCE兲1090-0268共2005兲9:6共524兲
CE Database subject headings: Walls; Masonry; Dynamic tests; Seismic effects; Retrofitting; Composite materials.
Upgrading of Unreinforced Masonry Walls with the in-plane lateral resistance of URM walls by a factor of 1.7,
Composite Materials while Abrams and Lynch 共2001兲 increased the in-plane lateral
resistance by a factor of 3.
Recently, several researchers have explored the potential use of In addition, the FRP upgrading technique could increase the
fiber reinforced polymer 共FRP兲 on unreinforced masonry 共URM兲 lateral drift. In the case of upgrading for out-of-plane failure,
walls. Ehsani et al. 共1999兲 increased the out-of-plane resistance of drifts as much as 2.5% were reached 共Ehsani et al. 1999兲. These
high drifts have been consistently explored by others 共e.g., Albert
URM walls, during static cyclic tests, as much as 32 times the
et al. 2001兲. For in-plane, Schwegler 共1994兲 increased the lateral
weight of the wall using FRP. The increase in out-of-plane resis-
in-plane drift of URM-FRP by a factor of 3, while Abrams and
tance of URM walls has been demonstrated by others 共e.g., Albert
Lynch 共2001兲 indicated that the FRP has no effect on the lateral
et al. 2001兲. However, the increase in the in-plane lateral resis- drift of URM walls. Recently, Holberg and Hamilton 共2002兲
tance of URM walls upgraded with FRP 共URM-FRP兲 was deter- achieved a lateral drift of 1.7%. This high level of lateral drift was
mined to be less significant than the increase in the out-of-plane achieved by using a ductile connection between the URM-FRP
resistance. During static cyclic tests, Schwegler 共1994兲 increased and its footing. However, in some cases drift was limited to 0.6%
due to the eccentricity caused by this connection, which resulted
1
Postdoctoral Research Fellow, Dept. of Civil and Environmental in out-of-plane failure. An extensive literature review for using
Engineering, The Univ. of Auckland, Private bag 92019, New Zealand; FRP for retrofitting of masonry walls maybe found in ElGawady
formerly, PhD Candidate, Applied Computing and Mechanics et al. 共2004兲
Laboratory 共IS-IMAC兲, School of Architecture, Civil and Environmental
Engineering, Swiss Federal Institute of Technology, 1015 Lausanne
共EPFL兲, Switzerland 共corresponding author兲. E-mail: melg003@ec.
auckland.ac.nz Experimental Program
2
Lecturer, IS-IMAC-ENAC-EPFL, 1015 Lausanne, Switzerland.
E-mail: pierino.lestuzzi@epfl.ch The dynamic experimental program investigates the effectiveness
3
IS-BETON-ENAC-EPFL, 1015 Lausanne, Switzerland; formerly, of composite materials as externally bonded upgrading materials
Professor. E-mail: marc.badoux@epfl.ch for the in-plane upgrading of URM walls. The experimental tests
Note. Discussion open until May 1, 2006. Separate discussions must
included two phases: the first phase consists of testing five-
be submitted for individual papers. To extend the closing date by one
month, a written request must be filed with the ASCE Managing Editor. reference URM specimens until a predefined degree of damage
The manuscript for this paper was submitted for review and possible occurred; the second phase consists of upgrading these reference
publication on December 11, 2003; approved on February 15, 2005. This specimens, using one layer of FRP on one face, then retesting
paper is part of the Journal of Composites for Construction, Vol. 9, No. them. Table 1 summarizes the tested specimens. The results of the
6, December 1, 2005. ©ASCE, ISSN 1090-0268/2005/6-524–535/$25.00. first phase are reported in ElGawady et al. 共2003兲 and the results
of the second phase are the object of this paper. During the second and S1-LAMI-C-X. Specimen L1-LAMI-C-I was especially de-
phase, one of the squat walls was upgraded using two different signed to investigate the shear behavior of a slender URM wall by
materials 共S1-LAMI-C-X and S1-WRAP-G-F兲; this results in a increasing its flexural resistance with minimal increase of shear
total of six upgraded specimens 共Table 1兲. The paper aims to resistance by the addition of two vertical Carbon FRP 共CFRP兲
present effects of different upgrading parameters 共configuration, plates 共50 mm wide by 1.2 mm thick兲 on one side. At the end of
material, and structure兲 on specimens’ behavior. In addition, the testing L1-LAMI-C-I, the vertical plates of CFRP were taken off
experimental results could serve as a data base for development and the specimen reupgraded using diagonal bands of glass FRP
of a future design model. 共GFRP兲. For the other two specimens 共L1-WRAP-G-X, and S1-
LAMI-C-X兲, Fig. 4 shows the upgrading details. A test specimen
was constructed on a precast reinforced concrete 共RC兲 footing.
Test Specimens
After allowing the specimen to cure 共from 3 to 7 days兲, a precast
Test specimens were intended to represent structures built in the RC head beam was fixed at the top of the specimen using strong
mid-20th century in Central Europe 共Fig. 1兲; half-scale single mortar 共M20兲. Anchorage failure of the FRP was prevented by
wythe walls were constructed using half-scale hollow clay ma- clamping the FRP ends to the specimen’s footing and head beam
sonry units 共Fig. 2兲. The test specimens had two aspect ratios using steel plates and screw bolts. In addition, the FRP had an
共Fig. 3兲: slender walls “L family” with effective moment/shear anchorage length 共over the RC head beam and foundation兲 of
ratio of 1.4 and squat walls “S family” with effective moment/ approximately 150 mm.
shear ratio of 0.7; also, two mortar types were used: Type 1 was a
strong mortar 共M9兲 and Type 2 was a weak mortar 共M2.5兲. The
Test Setup
average compressive strengths were 7.2 and 5.7 MPa for masonry
assemblages built using mortar Types 1 and 2, respectively. In The specimens were tested on the uniaxial earthquake simulator
addition, different types of FRP 共Table 2兲 and upgrading configu- of the Swiss Federal Institute of Technology in Zurich 共ETHZ兲
ration 共Table 1兲 were used to upgrade the specimens. As shown in 共Fig. 5兲. The test setup was designed to simulate the structural
Table 1, all the specimens were upgraded on the entire surface of features of the building shown in Fig. 1 共Lestuzzi et al. 1999 and
a single side except Specimens L1-LAMI-C-I, L1-WRAP-G-X, ElGawady et al. 2003兲. In this typical Swiss building, the normal
force is carried by slender columns and bearing walls, while the
bearing walls resist all the lateral force. The tributary areas of the
floors belonging to the bearing wall for gravity load are different
from the areas for horizontal mass inertia forces. To apply the
same structural concept the calculated gravity load 共due to the hibit the same acceleration response spectrum. The walls were
reinforced concrete slab, flooring, and live load兲 was applied di- subjected to dynamic excitations 共test runs兲 of nominal increasing
rectly to the test specimens by two external post-tension bars, the intensity; the increment was usually 10% of acceleration. In the
inertia force will be generated by a 12 Mg mass. This mass was following sections, each test run is followed by brackets which
placed on rails mounted on a separate steel structure. include the percentage of the applied acceleration in this test run.
Table 3 summarizes the earthquake type and the corresponding
number of test runs as well as the maximum nominal earthquake
Loading System
intensity for each test specimen.
The head beam was connected to the movable mass and the foot-
ing pad was clamped to the shaking table platform 共Fig. 6兲. For
normal force, superimposed gravity load of approximately 30 kN Experimental Results
was simulated using two external post-tensioning bars of 13 mm
diameter. This was in addition to 12 kN of self-weight from steel The averages of the absolute maximum and minimum lateral
elements at the wall top 共due to the test setup兲, RC head beam, forces and drifts as well as the mode of failure for each test
and masonry panel weight; this normal force corresponded to a specimen are presented in Table 3. In addition, the peaks mea-
compression stress of 0.35 MPa. In addition, springs were used sured forces, displacements, and drifts in the south 共negative兲 and
with the post-tensioning bars. These springs were used since dur- north 共positive兲 directions are given in Tables 4 and 5.
ing testing L1-WRAP-G-F and L1-WRAP-G-X and due to in-
crease of the wall height as a result of opening of flexural cracks
the post-tensioning force increased many times. Table 3 indicates
the initial and maximum post-tensioning force for each test speci-
men. In addition, Fig. 7 shows the change in the post-tensioning
force before and after the use of the springs.
Dynamic Excitations
The displacement inputs of the shaking table were based on syn-
thetic acceleration time-histories compatible with Eurocode 8
共CEN 1994兲 for rock soil Type A and with a peak ground accel-
eration of 1.6 m / s2, the earthquake lasted approximately 14 s.
Three main types of synthetic earthquakes were used for the tests.
The first 共UG1, Fig. 8兲 was for lower accelerations 共up to about
120% of the reference spectrum兲 and used larger table displace-
ments. For higher accelerations and because of the limitations on
the table displacements, other synthetic earthquakes with smaller
table displacements had to be used 共UG1R and UG1RR兲. These Fig. 4. Diagonal upgrading 共millimeters兲: 共a兲 L1-WRAP-G-X and
supplementary earthquakes were derived from the first and ex- 共b兲 S1-LAMI-C-X
In general, the FRPs improved the lateral resistance of the 1. The wall started to rock, during test Run 10 共130%兲, at the
retrofitted specimens. In order to evaluate such improvement, the wall footing and a small opening roughly 1 mm width was
lateral resistances of the upgraded specimens are compared to observed between the masonry wall and its footing. This
those of the appropriate reference specimens 共Fig. 9兲. Neverthe- crack corresponded to a drift of 0.12% and a lateral load of
less, it is difficult to fairly evaluate the enhancement in the ulti- 60% of the specimen lateral resistance. The delamination
mate drift due to the upgrading, as the URM reference specimens process began in a few points during test Run 12 共150%兲 and
did not reach their ultimate drift. Different failure modes occurred continued until test Run 17 共200%兲.
during the tests: flexural 共slender specimens with full surface cov- 2. The grid started to rupture in tension at the bottom northern
erage, i.e., L1-WRAP-G-F and L2-GRID-G-F兲; shear-flexural side of L2-GRID-G-F during test Run 18 共210%兲. The grid
共slender specimen diagonal upgrading, i.e., L1-WRAP-G-X兲; completely ruptured at the bottom sides of L2-GRID-G-F
FRP rupture due to debonding 共squat specimen with diagonal
upgrading, i.e., S1-LAMI-C-X兲; or no failure at all 共squat speci-
mens with full surface coverage, i.e., S1-WRAP-G-F and
S2-WRAP-A-F兲. Fig. 10 shows these different modes of failure.
In the following sections detailed descriptions of two specimens
共a slender and a squat兲 are given followed by a brief description
of the other specimens. More details maybe found in ElGawady
et al. 共2003兲.
Specimen L2-GRID-G-F
L2-GRID-G-F is a slender specimen which was upgraded on the
entire surface of a single side using a grid of GFRP. The speci-
men’s average lateral resistance was 49.3 kN; under the same
normal forces, the average lateral resistance of the appropriate
reference specimen 共L2-REFE兲 was 15.4 kN. At the test end, the
average of the ultimate lateral resistance of L2-GRID-G-F re-
duced to 47.5 kN and for L2-REFE increased to 17.8 kN. This
means that the FRP enhanced the lateral resistance by a factor of
approximately 2.9. The specimen behavior during the tests was as
follows: Fig. 6. Slender specimen ready to test
关Fig. 11共a兲兴 during test Run 19 共220%兲. Also, a masonry flexural failure at the first brick course. Note that rocking was
failure was observed in the bottom southern side 关Fig. 11共b兲兴. responsible for the high drift of this specimen. This could be
The average peak drift during test Run 19 was approximately confirmed by examination of the vertical LVD Ts through the
1% which was approximately two times the drift for test specimen. The specimen was 1,570 mm long by 1,633 mm
Run 18, while the lateral force reduced slightly from 51 kN high. By multiplying the algebraic difference of the mea-
共test Run 18兲 to 47 kN 共test Run 19兲. By the end of test sured vertical displacement time-history by 1,633/ 1,570, an
Run 19, the specimen reached its ultimate capacity due to estimate of the horizontal displacement time-history caused
mately the same normal force the average lateral resistance of the Specimens S1-WRAP-G-F and S2-WRAP-A-F
appropriate reference specimen 共L1-REFE兲 was 16.9 kN. At the
Both specimens were squat; however, while S1-WRAP-G-F was
test end, when the normal force increased up to 99.5 kN the av-
built using mortar Type 1, S2-WRAP-A-F was built using mortar
erage lateral resistance slightly increased to 57.3 kN. This shows
Type 2. S1-WRAP-G-F and S2-WRAP-A-F were upgraded using
that the high normal force has an insignificant effect on the lateral
fabrics of glass and aramid FRP, respectively, where the upgrad-
resistance of URM-FRP; by contrast, the lateral resistance of the
ing was applied on the entire surface of one side. Both of the
reference specimen 共L1-REFE兲 increased to approximately
specimens did not reach their ultimate limit state, as the test was
30.6 kN. This means that the FRP enhanced the lateral resistance
interrupted because the maximum force capacity of the shaking
by a factor ranging from approximately 1.9 to 2.6 depending on
table hydraulic jack was reached. Within the range of the test, the
the normal force considered. The maximum drift was approxi-
behavior of both specimens was to a certain degree identical. The
mately 1%. At the test end, the specimen failed in flexural due to
average lateral resistance of S1-WRAP-G-F was 74.1 kN, while
FRP tensile failure and masonry compressive failure at the first
the average lateral resistance of S2-WRAP-A-F was 72.0 kN;
brick course.
under approximately the same normal force the average lateral
resistances of the appropriate reference specimens 共S1-REFE and
Specimen L1-WRAP-G-X S2-REFE兲 were 28.5 and 28.2 kN, respectively. This means that
Specimen L1-WRAP-G-X was a slender specimen which was the FRP enhanced the lateral resistance by at least a factor of 2.6.
upgraded in a diagonal shape 共X兲 on one side using fabrics of At the test end, S1-WRAP-G-F had a drift of 0.3% and
GFRP. In the first phase of the dynamic tests, this wall had been S2-WRAP-A-F had a drift of 0.1%. The main difference between
tested as a special reference specimen L1-LAMI-C-I 共ElGawady the behaviors of the two specimens was that during testing Speci-
et al. 2003兲. Due to the first test, there were many step cracks men S1-WRAP-G-F some delamination appeared. Such delami-
passing through the mortar joints; these cracks affected the be- nation did not appear during testing Specimen S2-WRAP-A-F.
havior of specimen L1-WRAP-G-X. The specimen’s average lat-
eral resistance, under a normal force of approximately 57 kN,
increased to 25.7 kN. Under approximately the same normal Discussion and Comparison
force, the average lateral resistance of the appropriate reference of Experimental Results
specimen was 16.9 kN. This means that the FRP enhanced the
lateral resistance by approximately a factor of 1.5. The maximum The FRPs increased the lateral resistance of the upgraded speci-
drift was approximately 1.1%. The GFRP failed at the masonry mens by a factor ranging from 1.3 to 2.9 times the lateral resis-
panel mid-height at the intersection between the two diagonal tance of the reference specimens. In addition, using mechanical
bands. anchorage 共steel plates兲 succeed in avoiding anchorage failure.
Fig. 10. Failure modes of specimens 共a兲 L2-GRID-G-F; 共b兲 L1-WRAP-G-F; 共c兲 L1-WRAP-G-X; 共d兲 S1-WRAP-G-F; 共e兲 S2-WRAP-A-F;
and 共f兲 S1-LAMI-C-X