Abstract: The use of carbon fibre reinforced polymer (CFRP) wrapping to strengthen plain concrete prisms of square
cross section was investigated experimentally. The study was aimed at quantifying the increase in axial compressive
strength and ductility that can be achieved and assessing the effect of cross-sectional size on the increases. Thirty
prisms of three different square cross-sectional sizes (100 mm 100 mm 300 mm, 125 mm 125 mm 375 mm,
150 mm 150 mm 450 mm) were tested. Ten prisms were constructed in each size. Five prisms in each size were
left unwrapped as control specimens, and five were wrapped with two layers of unidirectional CFRP laminate. All
prisms were loaded in axial compression until failure. Significant increases in strength and ductility were achieved by
wrapping. The effectiveness of the wrap, as measured by the percentage increases in strength and peak axial strain, re-
duced with increasing cross-sectional size. These tests indicate that the use of CFRP wrapping is an effective technique
for strengthening and (or) rehabilitating concrete columns. Test results available in the literature by other authors are
also summarized. Although these results are highly scattered, they are consistent with the findings of the current tests.
Key words: concrete, prism, column, square, rehabilitation, strengthening, FRP, experimental.
Rsum : Lutilisation dun chemisage de polymres renforcs de fibres de carbone ( carbon fibre reinforced poly-
mer : CFRP ) pour renforcer des prismes en bton ordinaire de section carre a fait lobjet dune tude exprimentale.
Cette tude visait quantifier laugmentation de la rsistance en compression axiale et en ductilit qui peut tre at-
teinte et valuer leffet de la dimension de la section sur les augmentations. Trente prismes ont t tests, ayant trois
sections carrs diffrentes (100 mm 100 mm 300 mm, 125 mm 125 mm 375 mm et 150 mm 150 mm
450 mm). Dix prismes ont t construits dans chaque grosseur. Cinq prismes de chaque grosseur ont t laisss libres
de tout chemisage en tant que spcimens de contrle, et cinq ont t recouverts de deux couches de lamin CFRP uni-
directionnel. Tous les prismes ont t chargs en compression axiale jusqu la rupture. Le chemisage a permis
datteindre des augmentations importantes en rsistance et en ductilit. Lefficacit du chemisage, telle que mesure par
les augmentations de pourcentage en rsistance et en dformation maximale axiale diminuait en fonction de
laugmentation des dimensions de la section. Ces tests indiquent que lutilisation de chemisages en CFRP est une tech-
nique efficace pour renforcer et (ou) rhabiliter les colonnes de bton. Les rsultats des tests effectus par dautres au-
teurs et qui sont disponibles dans la littrature sont galement rsums. Bien que trs disperss, ces rsultats
concordent avec les rsultats des tests effectus.
Mots cls : bton, prisme, colonne, carr, rhabilitation, renforcement, FRP, exprimental.
[Traduit par la Rdaction] Masia et al. 13
Received 13 January 2003. Revision accepted 10 July 2003. Published on the NRC Research Press Web site at http://cjce.nrc.ca on
12 January 2004.
M.J. Masia.1 Centre for Infrastructure Performance and Reliability, School of Engineering, The University of Newcastle, University
Drive, Callaghan, NSW 2308, Australia.
T.N. Gale. Hunter Water Australia (Engineering), PO Box 5007 HRMC, NSW 2310, Australia.
N.G. Shrive. Department of Civil Engineering, The University of Calgary, 2500 University Drive NW, Calgary, AB T2N 1N4,
Canada.
Written discussion of this article is welcomed and will be received by the Editor until 30 June 2004.
1
Corresponding author (e-mail: mark.masia@newcastle.edu.au).
Can. J. Civ. Eng. 31: 113 (2004) doi: 10.1139/L03-064 2004 NRC Canada
I:\cjce\cjce3101\L03-064.vp
January 6, 2004 3:09:49 PM
Color profile: Disabled
Composite Default screen
tion from the research community and has found significant Fig. 1. Confined area (Micelli et al. 2001).
use in practice. Several studies on the performance of FRP
wrapped columns have been conducted, using both experi-
mental and analytical approaches. The majority of such stud-
ies have focused on the performance of columns of circular
cross section. The data available for columns of square or
rectangular cross sections have increased over recent years
but are still limited.
We, therefore, carried out an experimental study into the
use of carbon fibre reinforced polymer (CFRP) wrapping to
strengthen square-section concrete columns. The study used
plain concrete prisms and was aimed at quantifying the in-
crease in axial compressive strength and ductility that can be
achieved and, in particular, assessing the effect of
cross-sectional size on the increases. As far as the authors
are aware, no other experimental study has explicitly ad-
dressed the effect of column size on the increases in strength Fig. 2. Confined area (Thriault and Neale 2000).
and ductility achieved by FRP wrapping of square-section
concrete columns. The current paper, therefore, adds to the
limited data available for square-section columns against
which numerical models and design equations can be
checked and (or) refined.
In the assessment of the influence of column
cross-sectional size on the strengthening achieved by CFRP
wrapping, greater confinement of smaller columns was ex-
pected. The rationale for this expectation is developed in the
following paragraphs. The experimental program was de-
signed to investigate this effect systematically, and the re-
sults are reported in the following sections. fined area as shown in Fig. 2. Shrive et al. (2003) and Parvin
For FRP wrapped, axially loaded columns the design phi- and Wang (2001) draw similar conclusions regarding the
losophy relies on the wrap to carry tensile forces around the confined area using nonlinear finite element analyses for
perimeter of the column as a result of lateral expansion of square-section columns that have had the corners chamfered
the underlying column when loaded axially in compression. or rounded prior to wrapping.
Constraining the lateral expansion of the column confines When wrapping square- or rectangular-section concrete
the concrete and thereby increases its axial compressive ca- columns with FRPs it is usual first to round or chamfer the
pacity. It should be emphasized that passive confinement of column corners to prevent damage to the FRP fabric at the
this type requires significant lateral expansion of the con- sharp column corner. Thriault and Neale (2000) recom-
crete before the CFRP is loaded and confinement is initiated. mend minimum corner radii of b/6 or 35 mm, whichever is
Experimental results reported by several authors for CFRP larger, where b is the side length of the square-column cross
wrapped concrete columns indicate that tensile strains in the section.
CFRP sheets are negligible prior to reaching 60%70% of Rochette and Labossire (2000) have shown experimen-
the ultimate column strength (Purba and Mufti 1999; tally that for concrete columns the strength and ductility
Rochette and Labossire 2000; Demers and Neale 1999). gains achieved by wrapping increase with the radius of the
For columns square or rectangular in cross section the column corner, implying that as the corner radius is in-
confinement is effective at the column corners only with creased, the confined area increases. This concept is illus-
negligible resistance to lateral expansion being provided trated in Fig. 3. Conversely, for a fixed corner radius, as the
along the flat column sides (Thriault and Neale 2000; column section size is increased, the ratio of confined area
Demers and Neale 1999). The result is that not all the col- to total cross-sectional area will reduce. Pessiki et al. (2001)
umn cross section is effectively confined. The increased provide an equation for square- or rectangular-column cross
compressive strength due to confinement can only be acti- sections that can be used to illustrate this effect. They define
vated over the so-called confined area. Various authors a shape factor, s, as the ratio of the confined area Ac to
have discussed this effect, based mainly on the work of the total section area AT. Limiting the expression of Pessiki
Mander et al. (1988). Micelli et al. (2001) describe the con- et el. (2001) to square sections and plain concrete allows s
fined area as bounded by parabolas extending from the col- to be expressed as a function of the ratio of the column side
umn corners (Fig. 1). They confirmed this experimentally length b to corner radius r (eq. [1]).
from the observed cracking pattern over the cross sections of
natural stone masonry columns confined with CFRP wraps. Ac 2(b / r 2) 2
[1] s = =1
Micelli et al. (2001) state On the upper surface of the speci- AT 3(b / r) 2
men, parabolic-shaped cracks indicated the trajectories along
which the confinement stresses had spread along the cross The ratio Ac/AT or s is plotted against b/r in Fig. 4. The
section. Thriault and Neale (2000) conceptualize the con- plot clearly shows the decrease in the ratio of confined area
I:\cjce\cjce3101\L03-064.vp
January 6, 2004 3:09:50 PM
Color profile: Disabled
Composite Default screen
Masia et al. 3
Fig. 3. Effect of corner radius on confined area: (a) small radius Fig. 4. Confined area versus cross-sectional size.
and (b) large radius.
I:\cjce\cjce3101\L03-064.vp
January 6, 2004 3:09:52 PM
4
Table 1. Test results from other sources for square-section prisms and columns.
Strength (MPa) Peak axial strain (%)
3 a
Source fc (MPa) Specimen size, b h r (mm ) FRP/layers (properties) U W % inc U W % inca
Color profile: Disabled
Composite Default screen
29
I:\cjce\cjce3101\L03-064.vp
fuw = 958 MPa
Ew = 73 GPa
uw = 0.0133
tw = 1.0 mm/layer
53 200 1200 20 chamfer CFRP/2 (as above) 53 70.3 33
Feng et al. 2002 21.1 200 600 20 (steel rebar also) GFRP/1 20.7 26.9 30 0.27 1.48 448
fuw = 900 MPa
Ew = 65 GPa
tw = 0.304 mm/layer
21.1 200 600 20 (steel rebar also) GFRP/2 (as above) 20.7 27.0 30 0.27 1.88 596
21.1 200 600 20 (steel rebar also) GFRP/3 (as above) 20.7 29.0 40 0.27 1.93 615
21.1 200 600 20 (steel rebar also) CFRP/1 20.7 26.3 27 0.27 1.18 337
fuw = 3550 MPa
Ew = 235 GPa
tw = 0.111 mm/layer
Mirmiran et al. 1998 40.6 152.5 305 6.35 GFRP/1.45 mm thick 40.6 43.6 7
fuw = 2186 MPa
Ew = 69.6 GPa
40.6 152.5 305 6.35 GFRP/2.21 mm thick (as above) 40.6 45.0 11
40.6 152.5 305 6.35 GFRP/2.97 mm thick (as above) 40.6 45.0 11
Parvin and Wang 2001 21.4 108 305 8.26 CFRP/1 22.6 34.7 54 0.17 1.32 695
fuw = 3800 MPa
Ew = 189 GPa
uw = 0.016
tw = 0.165 mm/layer
21.4 108 305 8.26 CFRP/2 (as above) 22.6 45.2 100 0.17 2.03 1120
Pessiki et al. 2001 26.4 152 610 38 CFRP/1 26.4 41.4 57 0.28 1.33 375
fuw = 580 N/mm/ply
Ew = 38.1 kN/mm/ply
uw = 0.015
26.4 152 610 38 CFRP/2 (as above) 26.4 55.1 109 0.28 1.70 507
26.4 457 1830 38 (steel rebar also) CFRP/3 (as above) 31.5 37.4 19 0.21 0.23 10
26.4 457 1830 38 (steel rebar also) GFRP/3 31.5 35.5 13 0.21 0.35 67
fuw = 383 N/mm/ply
Ew = 21.6 kN/mm/ply
uw = 0.019
Rochette and Labossire 2000 42 152 500 5 CFRP/3 42.6 39.5 7 0.69
fuw = 1265 MPa
Ew = 82.7 GPa
uw = 0.015
tw = 0.30 mm/layer
% inca
Fig. 5. Specimen dimensions.
900
2050
Peak axial strain (%)
2.00
4.30
W
0.20
0.20
a
U
% inc
20
91
Strength (MPa)
28.0
44.8
W
23.4
23.4
U
work stripped from them, and then they were cured in a fog
room for 17 d. The specimens were then removed from the
fog room and allowed to dry in the laboratory. In each of the
three size categories, five of the prisms were wrapped with
CFRP (Sika Wrap Hex-230) and five were left unwrapped
fc (MPa)
13.5
I:\cjce\cjce3101\L03-064.vp
January 6, 2004 3:09:53 PM
Color profile: Disabled
Composite Default screen
Masia et al. 7
Fig. 6. Detail for CFRP wrapping. Fig. 7. Typical unwrapped prism specimen.
I:\cjce\cjce3101\L03-064.vp
January 6, 2004 3:10:02 PM
Color profile: Disabled
Composite Default screen
for both batches. The concrete compressive strengths were more direct comparison between unwrapped and wrapped
tested using 100 mm diameter, 200 mm high cylinders, in stiffnesses. The unusually large axial stiffness calculated for
accordance with the Australian standard AS 1012.9 (Stan- specimen WS3 (Table 2) was ignored in calculating the
dards Australia 1999). The cylinders were subjected to the mean stiffness for the WS prisms.
same curing conditions as the prism specimens, and the The peak CFRP strain is the largest tensile strain re-
compressive strength tests were conducted at the same time corded in the CFRP laminate, averaged between the two
as testing of the prism specimens. The mean compressive strain gauges on each specimen.
strength for batch 1 was 27.2 MPa (standard deviation
1.4 MPa) and for batch 2, 26.8 MPa (standard deviation Unwrapped prisms
1.3 MPa). The consistency achieved between the two The unwrapped prisms displayed behaviour typical of
batches allowed direct comparison of results between prism plain concrete loaded in compression (Fig. 8). In the failed
specimens from each batch assuming the same concrete state, the prisms displayed vertical and diagonal cracking
properties. and diagonal cone or pyramid cracking near the confined
The CFRP fabric (Sika Wrap Hex-230) had the follow- ends, similar to that observed in cylindrical compression test
ing properties: thickness, 0.13 mm; tensile strength, specimens.
3500 MPa; elastic modulus, 230 GPa; and elongation at The mean values of peak axial stress, peak axial strain,
break, 1.5% (information sourced from manufacturer data and axial stiffness appear to be independent of specimen size
sheet). The fabric was unidirectional with a nonstructural (Table 2).
weave in the secondary direction to hold the fabric together.
The fabric was laminated to the prisms using Sikadur 330 Carbon fibre reinforced polymer wrapped prisms
epoxy. Carbon fibre reinforced polymer was selected in pref- The axial stressstrain responses for the wrapped prisms
erence to glass fibre reinforced polymer (GFRP) because its are shown in Fig. 9. In each case, the response was essen-
higher stiffness provides a higher level of confinement for a tially linear until the compressive strength of the unconfined
given thickness of wrap. concrete was achieved (around 25 MPa). At that point the
curves flattened significantly and then followed another es-
Specimen naming convention sentially linear branch until ultimate failure occurred when
The prism specimens and associated tests are identified the CFRP wrap ruptured. Other than the changing shape of
using a three character naming code. The first letter indi- the stressstrain curve, the only signs of distress prior to the
cates unwrapped U or wrapped W, the second indicates final sudden failure were faint audible creaking sounds.
specimen size (small S, medium M, or large L), and the It is believed that the sudden loss of axial stiffness at the
third is the specimen number (1 to 5). For example, WM3 unconfined compressive strength is due to the concrete be-
is a 125 mm 125 mm 375 mm wrapped prism, specimen coming significantly cracked. Prior to that point the CFRP
number 3 of 5. tensile strains are very small (Figs. 1012), indicating insig-
nificant contribution from the wrap. As the concrete cracks
Results and discussion it dilates but is confined by the CFRP wrap that begins to
develop stress in hoop tension. The prism continues to sup-
The results for all tests are summarized in Table 2 and port more load by transferring more hoop stress into the
Figs. 814. The linear potentiometer data for test US2 were CFRP wrap. Final failure occurs when the brittle CFRP
unfortunately lost. wrap suddenly ruptures.
In Table 2, peak axial strain is the largest value of the Typically the wrap ruptured over a height between 50 and
mean of the four transducer readings divided by the appro- 100 mm (Fig. 15). The end confining brackets ensured fail-
priate gauge length. Note that peak axial strain does not nec- ure did not occur at the prism ends. In most cases the rup-
essarily coincide with peak axial stress, although this was ture occurred at a corner. For a small number of specimens
the case for many of the CFRP wrapped prisms. the rupture was mid-face. The measured peak CFRP strains,
The axial stiffness for each prism was calculated based which ranged from 0.77% to 1.22%, were less than the ulti-
on the procedure for determining the elastic modulus E of mate breaking strain of 1.5% reported by the CFRP manu-
concrete described in AS 1012.17 (Standards Australia facturer. These strains were measured mid-face, and it is
1997). For the unwrapped prisms this involved determining likely that the strains were higher at the location of the rup-
the slope of a chord drawn between two points on the plot of tures. The strains recorded indicate that much of the CFRP
axial stress versus axial strain corresponding to (i) a strain of laminate was not able to achieve its ultimate strain. It is be-
50 106 and (ii) a stress equal to 40% of the peak axial lieved that one source of stress concentration in the CFRP
stress. This chord falls well within the elastic response re- laminate arises where the laminate bridges underlying cracks
gion of the stressstrain curve for the unwrapped prisms. For in the concrete (Pessiki et al. 2001). This could be a possible
the wrapped prisms 40% of the peak axial stress approaches explanation for the laminate rupturing at mid-face for some
the transition point where the prism behaviour is becoming specimens. Another source is associated with the corner de-
inelastic and nonlinear. Therefore, the wrapped stiffnesses tail. The recorded strains highlight the need to provide a sig-
were calculated using a chord terminating at 40% of the un- nificant corner radius to prevent stress concentrations and
wrapped peak axial load, averaged for all unwrapped speci- (or) damage in the CFRP laminate. For the current tests, a
mens in each size category. This approach ensured that the larger corner radius of 35 mm as suggested by Thriault and
unwrapped and wrapped stiffnesses were calculated over Neale (2000) may have helped prevent the laminate punch-
similar portions of the elastic response region and allowed a ing observed at the corners for most specimens.
I:\cjce\cjce3101\L03-064.vp
January 6, 2004 3:10:03 PM
Masia et al.
Color profile: Disabled
Composite Default screen
Fig. 8. Axial stress versus axial strain for unwrapped prisms. Fig. 10. Axial stress versus CFRP strain and axial strain for
small wrapped prisms.
Fig. 11. Axial stress versus CFRP strain and axial strain for me-
dium wrapped prisms.
Fig. 12. Axial stress versus CFRP strain and axial strain for
large wrapped prisms.
I:\cjce\cjce3101\L03-064.vp
January 6, 2004 3:10:10 PM
Color profile: Disabled
Composite Default screen
Masia et al. 11
Fig. 13. Effect of size on peak axial stress. Fig. 15. Typical failure for the wrapped prisms.
I:\cjce\cjce3101\L03-064.vp
January 6, 2004 3:10:17 PM
Color profile: Disabled
Composite Default screen
I:\cjce\cjce3101\L03-064.vp
January 6, 2004 3:10:17 PM
Color profile: Disabled
Composite Default screen
Masia et al. 13
forced polymer wraps. Canadian Journal of Civil Engineering, Aeff confined area as defined by Micelli et al. (2001)
27: 10111020. AT total area of square prism or column cross section
Tsai, K.C., and Lin, M.L. 2002. Seismic jacketing of rc columns b side length for a square-section prism or column
for enhanced axial load carrying performance. Journal of the Ew elastic modulus of the FRP wrap
Chinese Institute of Engineers, 25(4): 389402. fcaverage compressive strength of the unconfined concrete
Wang, Y.C., and Restrepo, J.I. 2001. Investigation of concentrically fcccompressive strength of the confined concrete
loaded reinforced concrete columns confined with glass fi- fconf ultimate confining pressure provided by the FRP wrap
ber-reinforced polymer jackets. ACI Structural Journal, 98(3): fuw ultimate strength of the FRP wrap
377385. h height of a prism or column specimen
r rounding radius at the corners of a square prism or col-
umn cross section
List of symbols
tw thickness of the FRP wrap
Ac confined area of square or rectangular prism or column uw ultimate tensile strain of the FRP wrap
cross section s shape factor (Pessiki et al. 2001) equal to Ac/AT
I:\cjce\cjce3101\L03-064.vp
January 6, 2004 3:10:17 PM