Construction and Building Materials 55 (2014) 379–389

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier.com/locate/conbuildmat

Effects of confinement level, cross-section shape and corner radius

on the cyclic behavior of CFRCM confined concrete columns
Piero Colajanni a,⇑, Marinella Fossetti b, Giuseppe Macaluso c
a
Dipartimento di Ingegneria Civile, Informatica, Edile, Ambientale e Matematica Applicata, Università di Messina, Via Nuova Panoramica dello Stretto, S. Agata,
98126 Messina, Italy
b
Facoltà di Ingegneria e Architettura, Università degli Studi di Enna Kore, Cittadella Universitaria, 94100 Enna, Italy
c
Dipartimento di Ingegneria Civile, Ambientale, Aerospaziale e dei Materiali, Università degli Studi di Palermo, Viale delle Scienze, 90128 Palermo, Italy

h i g h l i g h t s

Cyclic behavior of concrete columns wrapped with Fiber Reinforced Cementitious Matrix.
Experimental tests on monotonic and cyclic axial compression.
Investigation on effect of confinement level, cross-section shape and corner radius.
Gain in compressive strength, deformability and absorption energy capacity.

a r t i c l e i n f o a b s t r a c t

Article history: The main results of an experimental research aiming to investigate the behavior of medium-size low-
Received 19 September 2013 strength concrete columns wrapped with Carbon Fiber Reinforced Cementitious Matrix (CFRCM) under
Received in revised form 7 January 2014 monotonic and cyclic compressive axial loads are presented. Thirty columns with circular, square and
Accepted 11 January 2014
rectangular cross-sections were tested under monotonic and cyclic axial loads to investigate the effect
Available online 12 February 2014
of the confinement level, the cross-section shape and the corner radius on the stiffness, strength, and duc-
tility of CFRCM confined concrete columns under cyclic loads. The results prove that CFRCM confining
Keywords:
jackets provide substantial gain in compressive strength, deformability and absorbed energy.
Carbon fiber
Cementitious matrix
Ó 2014 Elsevier Ltd. All rights reserved.
Cyclic actions
Confinement
Experimental tests

1. Introduction However, some drawbacks occur when organic resins are used
to bind the fibers: for example, inapplicability on wet surfaces, lack
The use of Fiber Reinforced Polymer (FRP) composites as wraps of vapor permeability, high costs of epoxy resin and of specialized
or jackets for upgrading existing reinforced concrete columns has workers for application, inapplicability at temperatures of less than
become increasingly popular in recent years. This jacketing tech- 10 °C or more than 30 °C, sensitivity to elevated or low tempera-
nique has been developing rapidly as it makes it possible to obtain ture. Epoxy resin degrades quickly when temperature grows,
a large increment of the deformation capacity of the concrete in releasing toxic fumes; fiber embedded in matrix resin system
critical regions [1], where large plastic deformations are expected, undergoes liquid transition at very high temperatures, resulting
increase in concrete strength and reduction in loss of stiffness and in a weak bond between concrete and external reinforcement.
strength due to cyclic action, protection and prevention of corro- To solve these drawbacks, inorganic matrix systems have been
sion, with minimal change in the weight and geometry of the introduced as a replacement for the organic matrix systems [2–
structure. 11]. Unfortunately, as a consequence of the granularity of the mor-
tar, penetration and impregnation of fiber sheets is very difficult to
achieve since mortars cannot wet individual fibers. However, the
use of fiber textile, in conjunction with the development of new
⇑ Corresponding author. Tel.: +39 0903977161, mobile: +39 333 17 00 589; fax:
materials for the fiber [8,9,11] are able to ensure an adequate bond
+39 0903977457.
between textile and matrix.
E-mail addresses: [email protected] (P. Colajanni), marinella.fossetti@
unikore.it (M. Fossetti), [email protected] (G. Macaluso).

http://dx.doi.org/10.1016/j.conbuildmat.2014.01.035
0950-0618/Ó 2014 Elsevier Ltd. All rights reserved.
380 P. Colajanni et al. / Construction and Building Materials 55 (2014) 379–389

Despite a lower tight interaction, the inorganic matrix has a circular and square cross-section specimens made of low strength
number of advantages, such as full compatibility with the concrete concrete subjected to monotonic uniaxial compression to investi-
substrate, the same resistance to fire, applicability on wet surfaces gate the efficiency of a Polypara-phenylene-benzo-bisthiazole
and low costs of application (easy application even on rough or (PBO) Fiber Reinforced Cementitious Mortars (FRCM) system. The
irregular surfaces) and does not requires qualified workmen, and experimental results showed that confinement with PBO fiber pro-
is a sustainable and reversible system. duced a noticeable increment in strength and ductility, though the
During the past decade, extensive research has been conducted low mechanical ratios of fiber considered were not always able to
on the behavior of FRP confined concrete columns, but fewer ensure hardening behavior up to rupture. Trapko [11] investigated
studies have been concerned with the use of the inorganic matrix the efficiency of PBO fiber bidirectional warped mesh, with four
composite system as a confinement technique. times as many fibers in the direction of the warp as in the direction
Researches on the use of textile for reinforcement of concrete ele- of the weft, embedded in an inorganic matrix in confining concrete
ment started in the 1980s [12], while in the late 1990s fervent re- cylinders. He investigated the feasibility and the performance of
searches began on the use of fiber embedded in inorganic matrix multilayer jacketing, and the influence of external temperature, by
for design and strengthening of concrete elements [2,13]. The first comparison with CFRP embedded in epoxy resin, also testing speci-
studies regarding the use of carbon fiber sheet embedded in inor- mens exposed to temperatures ranging from 30 °C to 180 °C. He
ganic matrix for confinement of concrete element reported in the stressed that FRCM could give an even better strengthening effect
international literature is that of Wu and Teng [3], where small than CFRP, since the failure modes are quite different. CFRP speci-
cylinders were tested, and the inorganic matrix was found to ensure mens failed by sharp rupture of the jacketing, and the higher the
the same behavior, in terms of strength and deformability incre- number of sheet layers, the more the concrete was crushed. The
ment, as the traditional epoxy resin. Triantafillou et al. [4] investi- damage pattern in PBO confined elements was completely different,
gated the use of Textile Reinforced Matrix (TRM) jacketing as an having initiated at the external pleat, which was 70 mm for each
alternative to FRP jacketing for confinement of concrete elements. specimen, and the damage propagated slowly. With an increase in
They used carbon fiber textile, and investigated the influence on the number of FRCM layers the performance of the confining system
confinement efficiency of the use of inorganic mortar versus epoxy increased, though the author concluded that there is a limit thick-
resin and of the strength of the inorganic mortar, and the role of ness of composite jacket beyond which composite strengthening
the number of textile layers. The experimental test proved that ceases to have positive effects; moreover, for FRCM heat treatment
TRM jacketing was effective in providing a strength and deformabi- had little effect on intensity of strengthening.
lity increment, with reduced efficiency with respect to FRP jacketing It has to be emphasized that in the literature there is still a lack
of about 80% for strength and 50% for ultimate strain, with the excep- of knowledge on many aspects of FRCM confined concrete that
tion of square columns, where almost the same efficiency was found. have been clarified for FRP jacketing, such as the role of the round-
Both the strength of the mortar and the number of textile layers ness of the corner, as well as the effect of cyclic load in deteriorat-
were found to have great influence on the performance of the confin- ing the strength, deformability and toughness provided by the
ing system. Moreover, it was found that failure of TRM was less confining system.
abrupt compared with that of its resin impregnated counterparts, In this context, this paper reports the main results of an experi-
due to slowly progressing fracture of individual fiber bundles. In a mental research aiming to investigate the behavior of medium-size
subsequent paper, Bournas et al. [5] investigated the effectiveness low-strength concrete columns wrapped with Carbon Fiber Rein-
of the same confining system in providing strength and deformabi- forced Cementitious Matrix (CFRCM) under monotonic and cyclic
lity increases to reinforced concrete columns prone to buckling of compressive axial loads. Thirty columns with circular, square and
longitudinal reinforcement, due to poor detailing of the stirrups. rectangular cross-sections were tested under monotonic and cyclic
They found TRM confining jackets approximately 10% less effective axial loads to investigate the effect of the confinement level, the
than FRP ones when small-scale specimens subjected to an axial cross-section shape and the corner radius on the stiffness, strength,
load were considered, and more effective as the number of textile and ductility of CFRCM confined concrete columns under cyclic loads.
layers grows, while no reduction in effectiveness was found in tests
on nearly full-scale columns under cyclic flexure. Ombres [6] evalu- 2. Materials and mechanical characterization
ated the effectiveness of Carbon Fiber Reinforced Cement Mortar
(CFRCM) in confining columns with circular, square and rectangular 2.1. Plain concrete
cross-sections under axial compressive loads with variation in the
The columns were cast with a concrete typical of buildings that require struc-
amount of fiber reinforcement and at temperatures ranging from tural retrofitting (low-strength concrete). Portland cement (ASTM International
25 °C to 90 °C. They found that the constitutive behavior of the con- Type I) was used. The cement:sand:gravel proportions in the concrete mixture were
fined concrete changes with variation in the number of confining roughly 1:1.9:2.33 by weight and the water/cement ratio was 0.35. The maximum
layers of CFRCM, ranging from strain softening behavior for a low size of the coarse aggregate was 10 mm. In order to characterize the mechanical
behavior of the concrete, six cylindrical specimens measuring 150 300 mm were
amount of fiber layers to strain hardening for heavily confined spec-
prepared and tested in monotonic and cyclic compression after 28 days of curing.
imens; the increase in the temperature produced a reduction in the The average compressive strength was 16.66 N/mm2 while the average ultimate
strength of the confined elements, while the influence on the ulti- strain was 0.406%. Fig. 1 shows the experimental results of the three tests in terms
mate strain value was negligible. De Caso y Basolo et al. [7] explored of r–e curves.
different types of grouts as inorganic matrices along different types
of fiber architecture (unidirectional and bidirectional meshes of 2.2. Carbon fiber mesh
low-density and high-density glass fiber sheet) and three types of
As a consequence of the granularity of the mortar impregnation of fiber sheets
grout, namely acrylic-modified Portland cement based matrix,
is very difficult to achieve. To facilitate the degree of penetration of the mortar ma-
hydraulic cement based matrix with high water retention, and a sin- trix in the fibers, as also reported in [4,5,10,11], a fiber mesh was used. A commer-
gle-component magnesium-phosphate-based matrix. They found cial carbon fiber mesh measuring 10 10 mm (RUREDIL XMESHC10) with equal
that the use of low-density glass fiber allowed for more effective fi- quantities of high-strength carbon fiber in the two orthogonal directions was used
ber impregnation, and was the most effective reinforcement with (Fig. 2a). The width and the nominal thickness (based on the equivalent smeared
distribution of fibers) of each layer of carbon fiber were 4 mm and 0.047 mm
both types of acrylic and hydraulic cementitious matrix, with higher respectively, while the weight of carbon fibers was 168 g/m2. The tensile strength
levels of ultimate strength reached with the hydraulic base grout and the elastic modulus of the carbon fibers were 4800 MPa and 240 GPa, respec-
and similar levels of ultimate strain. Colajanni et al. [8,9] tested small tively. All values were provided by the manufacturer.
P. Colajanni et al. / Construction and Building Materials 55 (2014) 379–389 381

20 specimens with rectangular cross-section shape having dimensions

monotonic 200 400 mm (series R). All specimens were 600 mm in height.
cyclic In order to investigate the effect of medium and high lateral
confinement pressure, the cylindrical specimens were wrapped
15 with 2 and 3 layers of bonded textile, while the prismatic speci-
mens were wrapped with 2 and 4 layers of bonded textile (denoted
as F2 and F4 respectively), in order to check whether the increment
in the number of textile layers was able partially to compensate for
σ [MPa]

10 the loss of confinement efficiency with a change from a circular to

a prismatic section shape.
Half the prismatic specimens were cast with a corner radius of
15 mm and the remaining half with a corner radius of 30 mm. To
5 evaluate the influence of cyclic loading on the behavior of CFRCM
confined concrete columns, half the specimens were tested under
monotonic compression and the remaining half under cyclic com-
pression. In addition to the confined specimens, for each series of
0 specimens unconfined concrete columns were prepared as control
0 0.004 0.008 0.012 0.016
specimens.
ε
The specimens are identified by a code in the format X J F#K-m,
Fig. 1. Stress–strain curves for plain concrete. where X refers to the cross-section shape (X = C for cylindrical spec-
imens, X = S for square specimens, X = R for rectangular specimens),
2.3. Binding mortar J (if present) refers to the corner curvature radius (J = 15 or J = 30),
The stabilized inorganic matrix was a commercial pozzolanic hydraulic binder
the digit F# refers to the confinement system (F0 for unconfined
provided by the manufacturer of the fiber mesh (RUREDIL). The mechanical charac- specimens, F# for specimens confined with # layers of textile), K re-
teristics were obtained through flexural and compression testing on six fers to the application load mode (K = M for monotonic load, K = C
160 40 40 mm mortar prisms hardened for 28 days. The average of the results for cyclic load), and the code letter m (m = a or m = b) was used
obtained was consistent with the manufacturer’s instructions: compressive
to distinguish the samples for identical specimens. For example,
strength 31.17 MPa, flexural strength 9.46 MPa and secant elastic modulus
>15,000 MPa. the specimen S15F4M is a square column with a corner radius of
15 mm, wrapped by four layers of carbon fiber and subjected to a
monotonic load.
3. Experimental program The cylindrical and prismatic specimens were cast in stiff plas-
tic and steel moulds respectively. All specimens were capped with
3.1. Preparation of test specimens special self-leveling high-strength mortar. Application of the
CFRCM jacket required soaking in water, application of the binding
In addition to investigating the effects of cyclic load, this exper- mortar in a layer about 3 mm thick and burial of the carbon fiber
imental program included studies on the following column param- mesh in it. Afterwards it was necessary to apply a second layer
eters: shape of cross-section; percentage of confinement fibers of binding mortar about 3 mm thick so as to cover the mesh com-
(confinement volumetric ratio); corner radius in the prismatic pletely. When several layers of fiber mesh were required, the oper-
sections. All specimens were axially unreinforced since jacket- ation described above was repeated, wet on wet, by applying the
ing–reinforcement interaction was outside the scope of the present next layer when the previous layer was not completely hardened.
study. At the end of the fiber mesh a 100 mm overlap was added, as rec-
To examine the aforementioned tests variables, the following ommended by the FRCM system manufacturer at the date at which
were cast: 10 specimens with circular cross-section shape having the experimental survey started.
diameter U = 200 mm (series C), 10 specimens with square For unconfined and confined columns, the key geometrical
cross-section shape with side 200 200 mm (series S) and 10 characteristics are presented in Tables 1 and 2 respectively. The

Fig. 2. (a) Carbon fiber mesh; and (b) application of CFRCM jacket.
382 P. Colajanni et al. / Construction and Building Materials 55 (2014) 379–389

Table 1
Geometrical details of unconfined columns and key results.

Specimen Cross Corner No of Loading Maximum stress Ultimate stress Axial strain at Ultimate axial Absorbed
section radius textile pattern fco (N/mm2) fuo (N/mm2) maximum stress eco strain euo (%) energy Eo
(mm) (mm) layers (%) (MJ/m3)
CF0M-a U200 – – Monotonic 16.80 13.44 0.2511 0.4292 0.057
CF0M-b U200 – – Monotonic 16.08 12.86 0.2027 0.3572 0.047
CF0C U200 – – Cyclic 16.55 13.24 0.1677 0.2528 0.042
S15F0M 200 200 15 – Monotonic 16.48 13.18 0.2037 0.3038 0.035
S30F0C 200 200 30 – Cyclic 15.90 12.72 0.1828 0.2724 0.030
R15F0M 200 400 15 – Monotonic 14.99 11.99 0.2513 0.4398 0.050
R30F0C 200 400 30 – Cyclic 14.62 11.70 0.1913 0.4113 0.046
Average 15.92 12.73 0.2072 0.3524 0.044
value

Table 2
Geometrical details of CFRCM confined columns and key results.

Specimen Corner No of qf (%) Loading Normalized Normalized axial strain at Normalized ultimate Normalized
radius textile pattern maximum stress maximum stress ecc/eco axial strain ecu/euo absorbed energy
(mm) layers fcc/fco E/Eo
CF2M-a – 2 0.188 Monotonic 1.24 5.58 3.81 5.06
CF2M-b – 2 0.188 Monotonic 1.28 6.63 4.40 5.88
CF2C – 2 0.188 Cyclic 1.24 5.23 3.45 4.42
CF3M-a – 3 0.282 Monotonic 1.41 4.67 3.41 5.04
CF3M-b – 3 0.282 Monotonic 1.49 4.34 2.79 4.10
CF3C-a – 3 0.282 Cyclic 1.48 6.42 3.93 5.88
CF3C-b – 3 0.282 Cyclic 1.44 5.12 3.48 5.15
S15F2M 15 2 0.188 Monotonic 1.20 1.04 2.07 2.62
S15F2C 15 2 0.188 Cyclic 1.16 1.62 2.04 2.23
S15F4M 15 4 0.376 Monotonic 1.41 2.04 2.36 3.24
S15F4C 15 4 0.376 Cyclic 1.54 3.61 2.59 4.31
S30F2M 30 2 0.188 Monotonic 1.16 1.44 2.06 2.51
S30F2C 30 2 0.188 Cyclic 1.18 1.34 2.03 2.44
S30F4M 30 4 0.376 Monotonic 1.40 4.22 2.77 4.26
S30F4C 30 4 0.376 Cyclic 1.48 1.61 2.69 5.13
R15F2M 15 2 0.094 Monotonic 1.29 1.69 2.42 3.10
R15F2C 15 2 0.094 Cyclic 1.14 2.27 2.54 2.99
R15F4M 15 4 0.188 Monotonic 1.39 2.25 2.17 2.94
R15F4C 15 4 0.188 Cyclic 1.40 2.00 2.00 2.35
R30F2M 30 2 0.094 Monotonic 1.26 2.04 2.55 3.31
R30F2C 30 2 0.094 Cyclic 1.34 1.86 2.01 2.58
R30F4M 30 4 0.188 Monotonic 1.45 1.92 2.96 4.40
R30F4C 30 4 0.188 Cyclic 1.43 2.35 3.08 4.58

identification of each specimen is shown in the first column. In All specimens were loaded up to failure with a rate displace-
Table 2 the fiber volumetric ratios qf = 4ntf/dmax are also reported, ment of 0.5 mm/s. In cyclic test displacement steps offset of
where n denotes the number of fiber layers, tf is the nominal thick- 0.2 mm for the unconfined specimens and of 0.5 mm for the con-
ness of the fiber mesh and dmax is the maximum transverse dimen- fined specimens were fixed.
sion of the specimens.

3.2. Test setup 4. Test results and discussion

The compression tests on all columns were carried out with a 4.1. Control specimens (unconfined columns)
hydraulic press (Zwick/Roell & Toni Technik) with a maximum load
capacity of 4000 kN, which enables tests in controlled displace- The monotonic and cyclic stress–strain curves of the control
ment mode. The test type and the equipment were regulated by specimens (without wrapping) are shown in Fig. 4. The dashed line
an electronic control unit, user interface via personal computer curves represent axial strain evaluated as the average value
and using the software TestXpertv.7.10. To measure the applied deduced from the electronic gauges, and the solid line curves the
displacements of the samples a linear voltage displacement trans- values derived by means of the displacements recorded by the
ducer (LVDT) was placed between the press plates. Moreover, elec- LVDT transducer.
tronic gauges with gauge length 200 mm were installed in the axial The figure plots typical stress–strain curves for unconfined con-
direction at the middle height of each specimen, namely four crete specimens under axial compression characterized by an
gauges in the prismatic column, one on each side of the specimen, ascending branch up to the peak and a rapidly descending post-
and three equally spaced gauges, 120° apart, in the circular col- peak branch until collapse (brittle failure). The comparison
umn. Fig. 3 shows a column during the test. The electronic gauges between monotonic and cyclic testing confirms that the envelope
provide an accurate evaluation of the strain in the pre-peak zone, curve, representing the boundary of the responses in cyclic
the LVDT average values of the deformation in the post peak compression, is identical to the corresponding stress–strain curve
branch, where cracking and concentration of ‘‘plastic’’ deformation obtained from a test with monotonically increasing strain [14].
can make the lecture of the gauges unreliable. The curves recorded by the electronic gauges are characterized
P. Colajanni et al. / Construction and Building Materials 55 (2014) 379–389 383

Fig. 3. Tested specimen and details of equipment.

20
Circular CF0M-a (gauges mean) Square Rectangular
CF0M-b (gauges mean) S15F0M (gauges mean) R15F0M (gauges mean)
S15F0M (LVDT) R15F0M (LVDT)
S30F0C (LVDT) R30F0C (LVDT)
15
Axial Stress [MPa]

10 CF0M-a (LVDT)
CF0M-b (LVDT)
CF0C (LVDT)

0
0 0.002 0.004 0.006 0 0.002 0.004 0.006 0 0.002 0.004 0.006 0.008
Strain

Fig. 4. Axial stress–strain curves of unconfined columns (control specimens).

by greater stiffness than those recorded by the LVDT. It has to be Figs. 5–7 respectively. The axial stress was calculated by the ratio
remarked that the curves of the displacements recorded by the of the axial load to the concrete cross-section area, i.e. neglecting
electronic gauges are interrupted at the maximum stress, because the area increment due to mortar.
of the loss of accuracy in the post-peak branch due to the local Apart from column S15F4M, which in the post-peak branch
damage occurring in the specimen. Therefore, in the following showed abnormal behavior (probably due to ineffective position-
sections only the displacements recorded by the LVDT transducer ing of the confining jacketing), in general in CFRCM confined
are shown. columns there is no experimental evidence of strength degradation
In Table 1 the characteristic response parameters are reported, due to cyclic loads up to the peak stress.
namely: – the stress and strain values at the peak stress fco, and eco; The results are consistent with previous studies [15–18], which
– the ultimate values corresponding to failure fuo, and euo, (the con- verified that in FRP confined columns the monotonic stress–strain
ventional ultimate strain was fixed as the strain value at 80% of the curve can be considered as the upper boundary of the response of
maximum stress); – the energy absorption capacity, measured for any different cyclic stress–strain curve in the entire deformation
all specimens as the total area under the r–e curve up to failure. It range. A recent study [19] indicated that these observations are
can be observed that none of these values are significantly affected also valid regardless of cross-section aspect ratio, concrete strength
by the geometrical test variables (shape of cross-section and corner and confinement level.
radius in the prismatic sections); the mean value of unconfined The results of the present tests clarify that in the case of CFRCM
concrete strength was 15.92 N/mm2, slightly lower than the value confined columns too, the monotonic stress–strain curve is tangent
of 16.66 N/mm2 obtained by the small cylinders used for mechan- to the upper boundary of the cyclic stress–strain curve and the
ical characterization of the material. hypothesis that the cyclic envelope curve is coincident with the
monotonic curve is valid for all specimens up to the peak stress.
4.2. Confined columns: effect of cyclic loading However, it is noteworthy that in all the previously mentioned pa-
pers the FRP confinement was efficient enough to ensure a harden-
The monotonic and cyclic stress–strain curves of specimens ing behavior up to rupture. This was not observed in many of the
with circular, square and rectangular cross-sections are shown in tested specimens, and in some of them (CF2C, CF3Ca, S15F2C,
384 P. Colajanni et al. / Construction and Building Materials 55 (2014) 379–389

25 25
CF2C CF3C-a
R15F2C R15F4C
CF2M-a CF3M-a
R15F2M R15F4M
20 20

Axial Stress [MPa]

Axial Stress [MPa]

15 15

10 10

5 5

0 0
0 0.004 0.008 0.012 0.016 0.004 0.008 0.012 0.016 R30F2C R30F4C

Strain R30F2M R30F4M

20

Axial Stress [MPa]

Fig. 5. Monotonic and cyclic axial stress–strain curves for circular columns.

15

25
S15F2C S15F4C 10
S15F2M S15F4M
20
5
Axial Stress [MPa]

15
0
0 0.004 0.008 0.012 0.016 0.004 0.008 0.012 0.016
10 Strain

Fig. 7. Monotonic and cyclic axial stress–strain curves for rectangular columns.
5

(point A), ere and fre the axial strain and corresponding stress at the
0
S30F2C S30F4C
beginning of the reloading (point b), fnew is the stress in the reload-
S30F2M S30F4M ing branch at the deformation value eun,env (point c) where the
20 unloading started, eret,env and fret,env the strain and corresponding
stress at the return on the envelope curve (point d), and epl the
Axial Stress [MPa]

15
residual plastic strain at complete unloading (f = 0).
Fig. 9a and b shows the correlation between the strain eun,env
where the unloading path starts and b1, for specimens with
10 medium and high lateral confinement pressure respectively. These
figures show that for small envelope unloading strains
5 (eun,env < 0.001), bi is almost equal to 1, and the strength degrada-
tion is negligible. As emax,i increase, bi tends to decreases and
reaches a constant value of about 0.93 for both specimens with
0 medium and high lateral confinement pressure.
0 0.004 0.008 0.012 0.016 0.004 0.008 0.012 0.016
Strain Thus it can be concluded that the stress deterioration is inde-
pendent of the confinement parameters that were assumed as test
Fig. 6. Monotonic and cyclic axial stress–strain curves for square columns. variables in the present study.
The result is consistent with the ratio proposed in [20], where
R30F2C) the cyclic response in the post-peak deformation range
for FRP-confined concrete cylinders and strain eun P 0.0035 was
was below the monotonic curve, perhaps due to extensive cracking
found a value of 0.92 for a stress deterioration ratio, and with
of the confining mortar.
the value of 0.90 suggested in [16,19].
In order to investigate this result in more detail, stress deterio-
ration, plastic strains and loss of tangent stiffness under unloading/
4.2.2. Plastic strains
reloading cycles was also evaluated for all specimens tested in the
In Fig. 10a and b the curves of the plastic strains epl versus enve-
present study.
lope unloading strains are shown for columns with medium and
high lateral confinement pressure respectively. The values of epl
4.2.1. Effect of cyclic loading on stress deterioration were not directly recorded, since the reloading started at a stress
In order to evaluate the degree of stress deterioration for each value corresponding to an axial force of 25 kN approximately. Thus
cycle, similarly to what was done in [16,17,19,20] a stress deterio- the plastic strain was estimated from the experimental stress–
ration ratio b1 was defined as follows: strain curve by extending the unloading path smoothly to the zero
stress point according to the following approximate simple rela-
fnew
b1 ¼ ð1Þ tion [15]
fun;env

where referring to the symbols in Fig. 8, eun,env and fun,env are the ax- fre
epl ¼ ere
ial strain and the corresponding axial stress at the unloading point Esec 0
P. Colajanni et al. / Construction and Building Materials 55 (2014) 379–389 385

(a) (b)

Fig. 8. Key parameters for evaluation of stress deterioration and stiffness loss under unloading/reloading cycles: (a) hardening branch; and (b) softening branch.

CF2C CF3C-a
1.1 1.1 CF3C-b
S15F2C
S30F2C S15F4C
R15F2C S30F4C
1 R30F2C 1 R15F4C
R30F4C
β1

β1
0.9 0.9

(a) (b)
0.8 0.8

0 0.0035 0.007 0.0105 0.014 0 0.0035 0.007 0.0105 0.014

ε un,env ε un,env

Fig. 9. Stress deterioration under unloading/reloading cycles: (a) columns with medium lateral confinement pressure; and (b) columns with high lateral confinement
pressure.

where Esec 0 is the secant elasticity modulus of the unconfined con- fnew frel
D½eun;env ¼ ð2Þ
crete (Esec 0 = fc0/ec0). The figures show that the plastic strain versus eun;env erel
envelope unloading strain law are almost independent of the level
Fig. 11 shows the stiffness loss during the cyclic test, normal-
of confinement [21]; this is consistent with the assumption in
ized with respect to the initial secant value D1 = D[e = 0.0005], for
[17,20] though the former suggested that the plastic strain is inde-
columns CF2C, S15F2C, R30F2C and CF3C-a, S30F4C, R15F4C. The
pendent of the unconfined concrete strength, while by contrast the
specimens were chosen so as to consider different cross-sections,
latter shows that a strong dependence exists.
different levels of lateral confinement and different corner radiuses
in the prismatic sections.
4.2.3. Effect of cyclic loading on stiffness loss This figure shows that for small strains (e 6 0.001), the ratios Di/
In the cyclic axial stress–strain curves, the loss of stiffness in the D1 are equal to 1, and there is no stiffness loss. As e increases, the
reloading branch can be evaluated by the slope decrease of the initially constant stiffness decreases as the test progresses and the
straight line b-c in Fig. 8a and b. For this purpose, the stiffness of rate of decrease is greater in the initial branch. However, it can be
the reloading branch D[eun,env] at the unloading strain eun,env, observed that, according to Eqs. (1) and (2) fnew = b1 fun,env, and
characterizing the generic i-th cycle of reloading was defined as (eun,env–ere) = eun,env–(epl + fre/Esec,0); since fre = 25 kN and b1 tends
follows: to the constant value 0.93, the stiffness decay for a given value of

0.014 0.014
CF2C R30F4C
S15F2C CF3C-a
0.0105 S30F2C 0.0105 CF3C-b
R15F2C S15F4C
R30F2C S30F4C
ε pl
ε pl

0.007 0.007 R15F4C

0.0035 0.0035

(a) (b)
0 0
0 0.0035 0.007 0.0105 0.014 0 0.0035 0.007 0.0105 0.014
ε un εun

Fig. 10. Plastic strains versus envelope unloading strains: (a) columns with medium lateral confinement pressure; and (b) columns with high lateral confinement pressure.
386 P. Colajanni et al. / Construction and Building Materials 55 (2014) 379–389

propagates rather slowly in the neighboring bundles, resulting in

CF3C-a
a more ductile failure mechanism compared with FRP. This phe-
CF2C
1.25 nomenon is due to the fact that the fiber mesh in the inorganic ma-
S30F4C
S15F2C trix is usually not well impregnated, and loading of the fibers is
1 R15F4C done with non-uniform distribution of forces, leading to so-called
R30F2C ‘‘telescopic failure’’ [5]. Furthermore, multiple cracking in the ma-
Di/D1

0.75 trix of CFRCM jackets at low tensile stress levels leads to pullout
failure at higher stress levels, rather than sudden fracture of the
0.5 fibers.

0.25 4.4. Effect of the cross-section shape

0 0.0035 0.007 0.0105 0.014

As is well known, the effectiveness of the FRP jacket is maxi-
ε mum for columns with circular cross-sections (radial symmetry)
for which the confining pressure is uniform in the plane of the sec-
Fig. 11. Loss of stiffness at reloading branches. tion, decreases for columns with square cross-sections due to con-
centration of confinement action at the corners, and is reduced for
columns with rectangular cross-sections, the more so the larger the
the unloading strain eun,env depends on the fun,env value, i.e. on the ratio depth/width of the cross-section.
efficiency of the confinement system. The observation that can be deduced from Figs. 11–13 and from
the key results of the tests reported in Table 2 shows that for
4.3. Effectiveness of CFRCM jacket CFRCM jackets these phenomena are also present, but are charac-
terized by reduced amplitude when compared with FRP jackets.
In Fig. 12 only the stress–strain curves obtained in tests with Comparing the peak stress (Fig. 13) it can be recognized that
monotonically increasing strains are shown and in Table 2, the jacketing effectiveness in strength enhancement is similar in
peak and ultimate stress and strain, and absorbed energy values circular and square columns, while lower efficiency is exhibited
normalized to the unconfined ones are presented. They prove that by rectangular columns with a large number of textile layers.
confinement with CFRCM jackets can enhance the performance of Moreover, circular columns also exhibit, both for qf = 0.188 and
unconfined columns with a substantial gain in strength, ductility for qf = 0.282, hardening behavior up to failure, revealed by a sharp
and energy absorption capacity. corner followed by a falling branch with high slope.
All the stress–strain curves in Fig. 12 exhibit an ascending The stress–strain curves of square specimens are characterized
branch, which almost coincides with that for unconfined columns, by a flat branch after the peak stress, having amplitude depending
followed by a second one, close to linear, up the formation of the on the number of layers of textile, followed by a steep descending
first cracks; then the specimens exhibit marked nonlinear behavior branch; specimen S15F4M, as already mentioned in the previous
with a shape depending on the effectiveness of the confining sys- section, has a distinct softening branch probably due to the ineffec-
tem. It should be noted that for some specimens, e.g. for CF3M-a, tive positioning of the jacketing.
a plateau in the stress–strain behavior is exhibited for stress values Lastly, in rectangular specimens, although the jacketing is able
slightly higher than those of the maximum stress of the unconfined to produce a significant increment in maximum stress, the post-
concrete, which highlights the need for a phenomenon of dilatancy peak behavior is always characterized by a softening branch with
to be activated before the jacketing confining action is effectively a gentle slope, and the ultimate condition defined by a reduction
activated. in stress at 80% of the peak stress is always reached for a large
Another important aspect of the response is that, contrary to strain value.
FRP jackets, CFRCM jackets did not fail abruptly. As also reported All these different shapes of stress–strain relationship can be
in [4,5], in the case of inorganic matrix composite jackets, fracture reproduced analytically by the modified Sargin’s model proposed
initiates from a limited number of fiber bundles and then in [22].

25
Circular Square S15F0M Rectangular R15F0M
S15F2M R15F2M
S30F2M R30F2M
20
S15F4M R15F4M
S30F4M R30F4M
Axial Stress [MPa]

15

10 CF0M-a
CF0M-b
CF2M-a

5 CF2M-b
CF3M-a
CF3M-b

0
0 0.004 0.008 0.012 0.016 0 0.004 0.008 0.012 0.016 0 0.004 0.008 0.012 0.016
Strain

Fig. 12. Monotonic axial stress–strain curves.

P. Colajanni et al. / Construction and Building Materials 55 (2014) 379–389 387

28
Circular Square Rectangular
23.00 23.00 22.96
22.15 22.00
20.70
21
19.00 18 ,72 18.57 19.00
16.48
16.00 15.90

fcc [MPa]
14.99
14.62
14

0
CF0 mean

CF2 mean

CF3 mean

R30F0C
S15F0M

R15F0M
S30F0C

R15F2 mean

R15F4 mean

R30F2 mean

R30F4 mean
S15F2 mean

S15F4 mean

S30F4 mean
S30F2 mean
Fig. 13. Effectiveness of CFRCM jacket on the peak stress.

0,3
Circular Square Rectangular

0.225 0.221
0.206 0.197
0,2
E [MJ/m3]

0.166

0.134 0.129
0.116
0.107 0.109
0,1

0.050 0.046
0.044
0.035 0.030

0
CF0 mean

CF2 mean

CF3 mean

R15F4 mean

R30F2 mean

R30F4 mean
S15F4 mean

S30F2 mean

S30F4 mean
S15F2 mean

R30F0C
S30F0C

R15F0M
S15F0M

R15F2 mean

Fig. 14. Effectiveness of CFRCM jacket on the energy absorption capacity.

Comparing the ultimate strain, it can be recognized that the increasing the maximum stress, though the larger the number of
increment in deformation capacity is always large in circular textile layers is, the less the increment in strength is. Referring to
columns (ecu/euo ranging from 2.79 to 4.40), and smaller in square the ultimate strain, the increment in circular columns is negligible,
(2.03 6 ecu/euo 6 2.77) and rectangular columns (2.00 6 ecu/euo while in square and rectangular columns it is considerable.
6 3.08), though the latter two series have a greater maximum Consequently, also taking into account the previous consider-
number of layers of textile (#4) than circular ones (#3). ation on the shape of the stress–strain curve, for circular columns
Thus, the energy absorption capacity E of confined specimens is the increases in absorbed energy is negligible, while for square and
always much larger than that of the unconfined specimen E0 rectangular columns the ultimate absorbed energy increases 70%
(4.10 6 E/E0 6 5.88) as shown in Fig. 14, while in square and rect- and 20% when four instead of two layers of textile respectively
angular columns the increment is smaller, 2.23 6 E/E0 6 4.31, and are used (Fig. 14).
2.35 6 E/E0 6 4.58 respectively, and directly related to the number
of confining textile layers and the corner radius dimensions.
4.6. Effect of corner radius in prismatic sections

4.5. Effect of the number of CFRCM layers In FRP confined square and rectangular columns, the corner ra-
dius significantly affects the performance and efficiency of the FRP
It should be emphasized that the strength increase in series C, S jacket. The tests that have been reported in the literature [23,24]
and R having the two fiber layers is roughly the same (see Fig. 13), have established that the increase in deformation capacity and
although the parameter qf for the rectangular section is half that confined concrete strength is proportional to the corner radius
for circular and square sections. By contrast, when the number of value.
fiber layers is increased, in order to obtain the same strength incre- In CFRCM confined prismatic columns, the inorganic matrix en-
ment as obtained in circular columns with three layers of textile sures stress smearing action at the corners; therefore the Authors
(16%), in both square and rectangular ones four layers of textile believe that the values of the corner radius have less influence
are necessary. However, it is possible to conclude that the increase on the failure mode of the columns and on confined concrete
in the number of CFRCM layers is in general quite effective in strength than in CFRP confined elements.
388 P. Colajanni et al. / Construction and Building Materials 55 (2014) 379–389

Circular Square Rectangular

Stiffness/Initial Stiffness
1.25
1 2 3 1 2 3 1 2 3

Stress/Peak Stress
1

0.75

0.5

0.25

0 0.004 0.008 0.012 0.016 0.004 0.008 0.012 0.016 0.004 0.008 0.012 0.016
Strain

Fig. 15. Stress–strain and stiffness–strain curves for circular, square and rectangular columns.

Fig. 16. Failure mode for columns with cross section: (a) circular; (b) square; (c) rectangular. (d) Tensile fracture of the carbon fiber mesh at the corner.

Only for the rectangular columns (series R) with four CFRCM A transition zone follows in region 2, characterized by nonlinear
layers was recorded a significant improvement of the ultimate ascending stress–strain curves and stiffness decreases. In this re-
strain and the absorbed energy, 70% and 45% respectively, with gion the cracked concrete core fully interacted with the elastic
an increase in the corner radius (comparison between R15F4# CFRCM jacket with a redistribution of internal forces across the
and R30F4#). For square columns (series S) with four CFRCM lay- mortar. At peak load, due to expansion of the concrete core, multi-
ers, there was an increment in ultimate strain only (comparison ple sub-vertical cracking in the CFRCM jacket occurred. It is ob-
between S15F4# and S30F4#). served that the rupture of CFRCM jacket was always preceded by
However, due to the limited number of specimens tested, the sound of carbon fiber bursting. In all the cases region 2 had
further experimental investigations of the effect of corner radius large extensions since the damage process grew slowly.
for a wider range of geometric and material parameters are In region 3, beginning at the point where the slope of the stress–
required to confirm these conclusions. strain curves and of the stiffness–strain curves dropped suddenly,
the failure mechanism was activated. Failure occurred when the
CFRCM jacket could not restrain the unstable expansion of the con-
4.7. Failure mode and crack patterns crete core. In all cases the failure mechanism was due to jacket col-
lapse at the corner (Fig. 16).
In order to describe and discuss the main damage patterns and Moreover, it is observed that, as also reported in Trapko (2013),
the failure mechanisms for CFRCM confined columns, in Fig. 15 a under failure load the concrete core of the specimens was crushed;
comparison is shown between the stress–strain curve, normalized the higher the number of CFRCM layers, the more the core of con-
with respect to the peak value, and the corresponding decay in the crete disintegrated.
normalized secant stiffness–strain curves for some specimens Finally it has been emphasized that, even if nowadays an over-
tested. Only one specimen for each type of cross-section was con- lapping length longer than that used in the specimens (100 mm) is
sidered, because all CFRCM confined specimens always collapsed advised by the manufacturer, during the tests the rupture occurred
by rupture of the jacket and because there was no apparent differ- outside the overlapping zone for all the tested specimens.
ence between the failure modes of the columns with medium or
high lateral confinement pressure. 5. Conclusions
In the figure, three distinct regions can be recognized. At the
early stages of loading (region 1), the stress–strain curves are char- Experimental tests were performed to investigate the effects of
acterized by a linear ascending branch and the secant stiffness– confinement level, cross-section shape and corner radius on the
strain curves are characterized by a constant branch. This state cor- monotonic and cyclic behavior of concrete columns wrapped with
responds to the condition in which the concrete core and CFRCM Carbon Fiber Reinforced Cementitious Matrix (CFRCM). The results
jacket form a single resistant system. prove that:
P. Colajanni et al. / Construction and Building Materials 55 (2014) 379–389 389

(a) The stress–strain curve of CFRCM confined concrete columns References

under monotonic loading is always the same as the envelope
curve of the cyclic stress–strain curve in the pre-peak [1] Campione G, Colajanni P, La Mendola L, Spinella N. Ductility of R.C. members
externally wrapped with frp sheets. J Comp Constr ASCE 2007;11:279–90.
branch; in the post-peak range when softening behavior is [2] Kurtz S, Balagru P. Comparison of inorganic and organic matrices for
exhibited, the curve for the cyclic load is somewhat lower strengthening of RC beams with carbon sheets. J Struct Eng, ASCE
than the monotonic one. 2001;127:35–42.
[3] Wu HC, Teng J. Concrete confined with fiber reinforced cement based thin
(b) The strength degradation for small strains is negligible. As sheet composites. In: Tan KH, editor. FRPRCS-6 Fiber Reinforced Polymer
emax,i increases (emax,i P 0.0035), strength degradation tends reinforcement for concrete structuree, vol. 1. World Scientific Publishing Co.;
to increase and reaches a constant value of about 0.93 for 2003. p. 591–600.
[4] Triantafillou TC, Papanicolaou CG, Panagiotis Zissimopoulos P, Laourdekis T.
both specimens with medium and high lateral confinement
Concrete confinement with textile-reinforced mortar jackets. ACI Struct J
pressure. 2006;103:28–37.
(c) The loss of stiffness for small strains (e < 0.001) is negligible. [5] Bournas DA, Lontou PV, Papanicolaou CG, Triantafillou TC. Textile-reinforced
As the strain increases, the stiffness for columns subject to mortar versus fiber-reinforced polymer confinement in reinforced concrete
columns. ACI Struct J 2007;104:740–8.
monotonic and cyclic loading decreases, depending on the [6] Ombres L. Confinement effectiveness in concrete strengthened with fiber
efficiency of the confinement system. reinforced cement based composite jackets. FRPCS-8. Patras, Greece; 2007.
(d) CFRCM confining jackets provide substantial gain in com- [7] De Caso y Basalo FJ, Matta F, Nanni A. Fiber Reinforced Cementitious Matrix
composites for infrastructure rehabilitation. Proc Compos Poly (Tampa, FL,
pressive strength, deformability and absorbed energy. An USA) 2009.
increase in the number of CFRCM layers is in general quite [8] Colajanni P, De Domenico F, Maugeri N, Recupero A, Spinella N, Mantegazza G.
effective in increasing the maximum stress and less effective Experimental results of RC columns strengthened with fiber reinforced
cementitious mortars. In: 3rd International conference on concrete repair,
in increasing ultimate deformation and energy absorption rehabilitation and retrofitting (Cape Town, South Africa); 2012.
capacity. [9] Colajanni P, De Domenico F, Recupero A, Spinella A. Concrete columns confined
(e) The investigations conducted on the effect of the cross-sec- with fiber reinforced cementitious mortars: experimentation and modelling.
Construction and Building Materials 2014;52:375–84.
tion shape confirmed that the greater jacketing effectiveness [10] Colajanni P, Di Trapani F, Fossetti M, Macaluso G, Papia M. Cyclic axial testing
is obtained, as expected, for circular columns. However, it is of columns confined with fiber reinforced cementitiuos matrix. In:
also to be noted that in rectangular columns with a ratio Proceedings of the 6-th international conference on FRP composites in civil
engineering (Roma, Italy); 2012.
depth/width of the cross-section equal to 2, the loss of jac-
[11] Trapko T. Fiber Reinforced Cementitious Matrix confined concrete elements.
keting effectiveness with respect to square columns is not Mater Des 2013;44:382–91.
large. [12] Gardiner T, Currie B. Flexural behavior of composite cement sheets using
(f) The effect of the corner radius in prismatic sections had a wowen polypropylene mesh fabric. Int J Cem Compos Lightweight Concr
1983;5:193–7.
smaller influence on the performance of CFRCM confined [13] Curbach M, Jesse F. High performance textile-reinforced concrete. Struct Eng
columns compared to FRP confined columns, probably due Int, IABSE 1999;4:289–91.
to stress smearing action at the corners. However, for series [14] Karsan ID, Jirsa JO. Behavior of concrete under compressive loadings. J Struct
Eng ASCE 95(ST12):2543–63.
R columns, the corner radius had a non-negligible influence [15] Lam L, Teng JG, Cheung CH, Xiao Y. FRP-confined concrete under axial cyclic
on ultimate strain value, and consequently, on the absorbed compression. Cem Concr Compos 2006;28:949–58.
energy capacity. [16] Shao Y, Zhu Z, Mirmiran A. Cyclic modelling of FRP-confined concrete with
improved ductility. Cem Concr Comp 2006;28:959–68.
(g) Failure of CFRCM confined columns, for both monotonic and [17] Sakai J, Kawashima K. Unloading and reloading stress–strain model for
cyclic loads, is less abrupt compared with that of FRP con- confined concrete. J Struct Eng ASCE 2006;132:112–22.
fined columns due to the slowly progressing fracture of indi- [18] Abbasnia R, Ziaadiny H. Behavior of concrete prisms confined with FRP
composites under axial cyclic compression. Eng Struct 2010;32:648–55.
vidual fiber bundles. [19] Abbasnia R, Ahmadi R, Ziaadiny H. Effect of confinement level, aspect ratio and
concrete strength on the cyclic stress–strain behavior of FRP-confined
concrete prisms. Composites B 2012;43:825–31.
[20] Lam L, Teng JG. Stress–strain model for FRP-confined concrete under cyclic
Acknowledgements
axial compression. Eng Struct 2009;31:308–21.
[21] Buyukozturk O, Tseng TM. Concrete in biaxial cyclic compression. J Struct Eng
This work has been carried out within the 2010–2013 Research ASCE 1984;110(3):461–76.
Project ‘‘DPC-ReLUIS (Dipartimento Protezione Civile – Rete dei [22] Colajanni P, Papia M, Spinella N. A stress–strain law for confined concrete with
hardening or softening behavior. Adv Civ Eng 2013;804904.
Laboratori Universitari di Ingegneria Sismica)’’, AT 1, Task 1.1.2. [23] Wang ML, Wu YF. Effect of corner radius on the performance of CFRP-confined
The related financial support was greatly appreciated. The Author square concrete columns: test. Eng Struct 2008;30:493–505.
thanks Dott. G. Mantegazza, Technical Director of Ruredil S.P.A., [24] Wu YF, Wei YY. Effect of cross-sectional aspect ratio on the strength of CFRP-
confined rectangular concrete columns. Eng Struct 2010;32:32–45.
for providing the materials for the experimental program.