Composite Structures xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

Composite Structures
journal homepage: www.elsevier.com/locate/compstruct

Experimental investigation on the confining effect of fibers in SHFRCC

⁎
Antroula V. Georgioua, , Stavroula J. Pantazopouloub
a
Department of Civil and Environmental Engineering, University of Cyprus, Kallipoleos 75, Nicosia 1678, Cyprus
b
Department of Civil Engineering, Lassonde Faculty of Engineering, York University, 4700 Keele Street, 11 Arboretum Ln, Toronto, ON M3J 1P3, Canada

A R T I C L E I N F O A B S T R A C T

Keywords: Strain Hardening Fiber Reinforced Cementitious Composites (SHFRCC) comprise of a mix design that differs
Strain hardening from normal concrete mixes. Fibers act as mass reinforcement leading to large deformation capacities, bridging
Compression of the cracks and transferring stresses through cracked regions, delaying or even entirely mitigating cover
Confinement splitting. This contribution is interpreted as an intrinsic “confining effect”. To characterize the confining effect,
an experimental program was undertaken, including a series of specimens comprising either SHFRCC or plain
cementitious matrix of the same composition (PC), subjected to various levels of confinement of known intensity
(steel stirrups and FRP jacketing). A triaxial state of stress is found to develop in the SHFRCC, where the internal
stresses supported by the fibers are related to classical constitutive models for plain concrete with FRP jackets or
steel stirrups. The lateral and volumetric behavior of the composites is investigated and related to the tensile
strength of SHFRCC, measured through a variety of tests. Overall, it was concluded that the fiber reinforced
composites respond in a manner that is entirely consistent with commonly understood confined concrete be-
havior.

1. Introduction right of the axis and decreased the slope of the descending branch.
Increase in ductility is attributed to the resistance against crack growth
Being heterogeneous, cementitious materials have a propensity for in the lateral direction by the fibers. The strain at maximum stress of the
failure due to localization of cracking followed by loss of load carrying specimens with fibers or stirrups was greater than that of plain concrete
capacity in directions orthogonal to the orientation of the cracks. whereas shifting of strain at maximum stress confirms that dilation is
Concrete under uniaxial compression is characterized by axial splitting resisted by the fibers.
and fast decay of strength for axial strains that exceed the range of Based on the 2008 Recommendations by the JSCE [4] for the design
0.002–0.003; failure is heralded by a significant increase in transverse of SHFRCC the most important parameter needed for design is the
strains which grow rapidly producing apparent values of the Poisson’s stress-strain law of the material under compression; a Hognestad type
ratio that exceed the limit of incompressibility beyond the peak parabola is used to represent the nonlinear ascending part of the stress-
(v = 0.5, see [1]). Recently developed self-consolidating SHFRCC with strain law, up to a peak stress of 0.85f′ck/γc and a corresponding strain
short randomly oriented synthetic fibers present the opportunity for ε′m, followed by a constant value of 0.85f′ck/γc:
innovative resilient materials where the brittleness and localization
′ εc′
0.85fck ε′
upon cracking is mitigated by the fibers bridging the cracks. σc′ = · ·⎜⎛2− c ⎟⎞ for εc′ ⩽ εm′
Fibers contribute to the behavior of the cementitious matrix by re- γc εm′ ⎝ εm′ ⎠ (1)
straining the crack propagation as shown in previous studies [2], Recommended value for the strain at attainment of peak stress is,
thereby increasing ductility of concrete. Shah and Rangan [3] have ε′m = 0.4%, (much greater than that of normal concrete, 0.2%), un-
studied the differences between fiber reinforced concrete and con- derscoring the intrinsic passive confinement effect imparted by the
ventionally confined concrete (using stirrups). Specimens reinforced network of fibers. The JSCE Recommendations [4] suggest that Young’s
with 0.5% volume of steel were compared, comprising either of ran- Modulus may be obtained by the same testing methods used for normal
domly distributed fibers or closed stirrups, with different spacing. Al- concrete, yielding values in the order of 1/2 to 2/3 of that of normal
though the differences on the ascending elastic part of the axial stress- concrete. Additionally value of the proposed Poisson’s ratio is taken
strain curve were not noticeable, it was found nevertheless that steel νECC = 0.226. Similarly, the new Model Code [5] states that “the
fibers increased the ductility by shifting the strain at peak stress to the compressive relations valid for plain concrete apply to FRC as well”

⁎
Corresponding author.
E-mail addresses: [email protected] (A.V. Georgiou), [email protected] (S.J. Pantazopoulou).

http://dx.doi.org/10.1016/j.compstruct.2017.09.110
Received 18 September 2017; Accepted 28 September 2017
0263-8223/ © 2017 Elsevier Ltd. All rights reserved.

Please cite this article as: Georgiou, A.V., Composite Structures (2017), http://dx.doi.org/10.1016/j.compstruct.2017.09.110
A.V. Georgiou, S.J. Pantazopoulou Composite Structures xxx (xxxx) xxx–xxx

while acknowledging the confining beneficial effect of fibers in en- Table 1

hancing the coordinates of the peak point in the compressive stress- Identification of compression specimens for various types of confinement.
strain response of concrete.
Identification Core Material Confinement method Confinement ratio#
Results from a series of uniaxial compression tests conducted on
cylinders of cementitious composites, with variables of study being the PC Plain Composite No confinement
fibers (fiber length and coating) and the matrix confinement (with steel (PC)
M8 SHCC 8 mm PVA fibers 2% by volume
stirrups and FRPs) are presented in this paper. The main purpose is to
M8C SHCC 8 mm coated PVA 2% by volume
document and quantify the confining effect presented by the fibers fibers
through comparisons with the response of plain concrete under mea- M12 SHCC 12 mm PVA fibers 2% by volume
surable lateral confinement, whereby an effective passive confining M12C SHCC 12 mm coated PVA 2% by volume
stress imparted by the embedded fibers may be identified and docu- fibers
S3/5/8 Plain Composite Steel stirrups Stirrup spacing: 3, 5,
mented. Thus a triaxial state of stress is found to develop in fiber re-
(PC) 8 mm
inforced concrete, where the internal stresses supported by the fibers C1/2/3 Plain Composite Carbon FRP sheets No. of layers: 1, 2, 3
may be related to classical constitutive models for plain concrete with (PC)
FRP jackets or steel stirrups. G1/2 Plain Composite Glass FRP sheets No. of layers: 1, 2
(PC)

2. Description of the experimental program #

Here it is recognized that distributed fiber is a form of confinement.

2.1. Objectives and scope of the experimental investigation composition and coating procedure are given in Ref. [6]). The per-
centage of the fibers was constant in all cases at 2% per total volume.
Experimental results regarding the effect of PVA fiber’s length and Additionally the compressive behavior was compared to the mix
surface properties on the composite’s compressive strength and strain without any fibers referred to as plain mix PC.
are examined. Fiber lengths considered are 8 and 12 mm; in all cases Uniaxial compression tests were also carried out on the matrix (PC)
the matrix materials and volumetric ratios remain the same including with the use of stirrups, by varying their spacing (30, 50 and 80 mm)
the fibers’ diameter and content ratio per volume of the composite. By and therefore the volumetric ratio of confinement reinforcement. The
altering the surface properties, stress and strain response of the com- stirrups were steel bars of a diameter of 8 mm with a yield tensile
posites in compression are modified. Tensile strength of SHFRCC is strength of 500 MPa and 15 mm clear cover. Other than confinement by
measured through a variety of tests (direct tension, or indirectly steel stirrups the method of confinement with FRP sheets is examined.
through splitting tests); in this manner the net action of fibers in tension Two types of FRP sheets were used (carbon and glass) with various
is measured immediately. These results are then used to calibrate the layers for confinement as shown in Table 1.
effective confining pressure exerted by the fibers in the radial direction Three specimens were tested for each mix and for each confinement
of a compressed cylinder through the hoop tension that is resisted by method with a total of 13 × 3 = 39 specimens. Cylinder dimensions
the engaged fibers as a result of the Poisson’s effect. were 100 mm in diameter and 200 mm in height. Loading was applied
Quantifying the confining effect in practical terms such as an under displacement control at a rate of 1.5 μm/s for all specimens ex-
equivalent pressure that is owing to the action of fibers would enable cept the ones wrapped with FRP sheets that were tested at 3 μm/s.
the easy adaptation and extension of the design models of reinforced Additionally to the tests under compression uniaxial tensile tests and
concrete to steel-reinforced members comprising an SHFRCC matrix. split cylinder tests on SHFRCC were used to correlate the lateral con-
This development would be an essential step towards prequalification finement to the tensile strength of the composites. Note that splitting
of these materials for use in construction. To characterize the confining has the advantage of ease of execution while it bears greater relevance
effect through comparison, the experimental program was extended to to reality with regards the randomness of fiber orientation as compared
include a series of specimens of plain cementitious matrix (PC) sub- to flat dog bone specimens used in direct tension tests.
jected to various levels of confinement of known intensity (steel stirrups
and FRP jacketing). This served as a basis for deduction of the effective
confining pressure and its effects on the strength and strain capacity of 2.3. Materials, fabrication and curing of test specimens
SHFRCC. Thus a triaxial state of stress develops in the SHFRCC, where
the internal stresses supported by the fibers are related to classical In the present study the matrix used had a mix design similar to
constitutive models for plain concrete with FRP jackets or steel stirrups. ECC-M45 [7], being intended for large scale casting. The cement used
The lateral and volumetric behavior of the composites is investigated was a type of Blended Cement with a compressive strength at 28 days
with reference to results obtained from a circumferential extensometer equivalent to that of 42.5 R OPC. Silica sand (> 95% Si) used had a
used in the compression tests. maximum grain size of 300 μm. Fly ash was of type F (i.e., with poz-
zolanic properties [8]). In the present investigation 39 μm diameter
2.2. Specimen design PVA fibers were used, having a length of either 8 mm or 12 mm (Kur-
allon K-II); fibers were coated as described in Ref. [9] prior to mixing.
For the benefit of improved understanding of the performance of The fiber’s nominal tensile strength was 1600 MPa, Young’s Modulus
SHFRCC materials under compression, the experimental program is 40 GPa and density 1300 kg/m3. Details of the mix design and corre-
designed so as to quantify the confinement imposed by the fibers to the sponding identification code are listed in Table 2. Specimens were cast
compressive strength. To this end, results obtained from uniaxial
compression tests on SHFRCC composites are compared to those ob- Table 2
Mixture proportions for the cementitious composites (per weight unless stated otherwise).
tained from the plain cementitious mix (same composition but no fi-
bers) subjected to well-known confinement systems. Considering mixes Mix Cement Fly Ash Sand Water Super Plasticizer Fibers (% volume)!
with two different lengths of PVA fibers (8 and 12 mm) was intended to
illustrate the effect of fiber length on the strain hardening behavior of SHCC 1.00 1.20 0.80 0.60 0.02 2
PC 1.00 1.20 0.80 0.55 0.012 –
the SHCC. Two different ratios of coating on the fibers in the mix (0%,
6.6% per weight of fibers) were considered so as to demonstrate the !
For those specimens that contained fiber, which had already been coated as described
effect of surface specific bond properties on the final compressive in Ref. [9], an amount of 1.2 ml of defoamer was added to the mix for 1 m3 of composite
properties of the composite (details about coating’s chemical to cancel out the air-entraining action of the fiber coating substance.

2
A.V. Georgiou, S.J. Pantazopoulou Composite Structures xxx (xxxx) xxx–xxx

Table 3
Characteristics of the FRP sheets.

SikaWrap Type ft (MPa) E (GPa) Weight (gr/m2) Depth (mm) Type of wrap Type of application

SikaWrap230 C/45 Carbon 4300 238 230 0.131 Knotted single direction Dry
SikaWrap 430 G/25 Glass, Type E 2300/3400 76 445 0.172

without any vibration due to the self-compacting properties of the conducted under displacement control based on the displacement of the
mixtures. The specimens were cured under water for 100 days while plates at a constant stroke rate of 0.50 mm/min on 100 mm diameter
specimens with steel stirrup reinforcement were wrapped in burlap and cylinders having a length of 100 mm.
were moist-cured daily for 100 days. The steel stirrups had a diameter
of 8 mm and a yield strength of 500 MPa. Details of the characteristics 3. Test results and discussion
for the carbon and glass FRPs used in this study are listed in Table 3.
3.1. Tensile tests
2.4. Test setup and procedures
3.1.1. Uniaxial tension
2.4.1. Uniaxial compression In the tests, strain was calculated as the elongation measured by the
Uniaxial compression tests were performed well into the post-peak equipment mounted on the specimens’ two opposite sides and divided
range under displacement control using a closed-loop, servo hydraulic by the gauge length (100 mm). During formation of multiple cracks, the
controlled testing machine. Axial deformations were measured using composite matrix-fiber system appears to strain-harden. Due to stress,
two linear variable differential transducers (LVDTs) mounted on op- fibers bridging the cracks elongate but they also pull-out partially from
posite sides of two rings that were attached on the specimen over a the matrix at the crack locations. In the dog-bone specimens the for-
gauge length of 140 mm. An additional LVDT measured the deforma- mation of cracks saturated the full measuring length (narrow part of the
tion over the entire length of the specimen, through the displacement of specimen) and extended beyond that towards the wider ends of the
the platens. Circumferential elongation was measured using a cir- specimen when coated fibers were used (mixes M8C and M12C).
cumferential extensometer (chain), placed at the middle of the height of Beyond a limiting strain, new cracks cease to form and crack localiza-
the coupon. Additionally 3 stain gauges were fixed (2 parallel to the tion occurs. The average stress-strain curves from the uniaxial tension
loading direction and 1 perpendicular to the load) for the estimation of tests for the 4 mixes are shown in Fig. 2(a).
the Poisson’s ratio as shown in Fig. 1(c) (a combination of the two According to previous research [6], the estimated number of fibers
measuring devices was necessary to measure lateral deformations due crossing a unit area in all tested samples is the same, i.e.:
to the limited range of the strain gauges at post-cracking lateral strains 1 4
and the relative insensitivity of the chain at very low strain levels). nf = ·ρ ·
2 v πdf2 (2)
2.4.2. Uniaxial tension and split cylinder test Here, nf = 4·2%·0.5/(π·0.039 ) = 8.37 fibers per mm with any differ-
2 2

Displacement controlled direct tensile tests were carried out on dog ence in response linked to the bond properties of the fiber’s contact
bone specimens, having the dimensions shown in Fig. 1(a). Setup and surface. Tensile strength of each composite appears to be mainly de-
mounting of specimens are described in detail in Ref. [9] with the test pendent on the length of the fiber, estimated at 3.5 MPa (1 MPa greater)
performed at a displacement rate of 0.0025 mm/s. Cylinder splitting is in the case where the 12 mm length fibers are used whereas it is only
a more straightforward method to determine the tensile strength of 2.3 MPa in the case where 8 mm length fibers are used.
normal-strength concrete [10] while in the case of strain hardening In their original state PVA fibers have a hydrophilic surface with
composites multiple cracks were expected to open parallel to the free hydroxyls creating chemical bonds with the surrounding cement
loading axis, according to previous research [9]. Splitting tests were products. These chemical bonds are so strong that are very difficult to

Fig. 1. (a) Dimensions of uniaxial tensile test coupon, (b) Gripping mechanism and testing setup for uniaxial tension (c) Uniaxial compression test setup with measuring devices (d)
Loading setup for split cylinder test.

3
A.V. Georgiou, S.J. Pantazopoulou Composite Structures xxx (xxxx) xxx–xxx

Fig. 2. (a) average stress-strain diagrams from uniaxial tensile tests for all mix designs and (b) average stress-diametric elongation from splitting tests measured perpendicular to the
direction of cracks for all mix designs.

break leading fibers to rupture rather than pull-out [11]. The rupture 3.2. Uniaxial compression
strength of a single PVA fiber may be calculated based on material
characteristics as Ff1 = fyf·Af = 1600·π·0.0392/4 = 1.91 N. A corollary 3.2.1. Fiber reinforced composites
to this is the observation that if all the fibers could reach their ultimate The crack formations after the uniaxial compression tests of the
strength by rupture (e.g. in the case of uncoated fibers), then the ulti- various mixes are shown in Fig. 3. A great discrepancy is observed
mate stress of the composite based on fiber rupture would be in the between the results of the unreinforced and the fiber reinforced speci-
range of =1.91·5231.9/(25·50) = 8 MPa. Note that based on the test mens. Specimens without any fibers cracked and suddenly collapsed. In
results presented in Fig. 2(a) none of the materials has attained this the absence of fibers deterioration of the specimens and fragmentation
ideal tensile strength value, since fibers at an angle with the loading into small pieces occurred. Cracks were vertical, parallel to the load and
direction may rupture prematurely due to combined cleavage at the extended far inside the core of the specimen. After sudden collapse the
crack surface. specimens seemed unable to carry any remaining loads. In the case of
the fiber reinforced composites multiple cracking is displayed all
around the specimens. Cracking was visible close to 70% of the max-
3.1.2. Split cylinder test results imum load, with formation of multiple parallel cracks that increased in
The cylinders tested using the standard splitting procedure develop number up to the attainment of the peak load. Up till the end of the test
a biaxial compressive state of stress immediately under the loading line no spalling of the specimen was observed. Post peak strength reduction
which quickly reverses to tensile (positive) stress in the central region was slower than in the case of the PC and the specimens after the test
[12]. A constant tensile stress distribution occurs over the central 90% seemed solid enough and able to withhold a residual amount of load. A
of the diameter of the cylinder, the intensity of which may be estimated major crack was formed in all cases of SHFRCCs at the end of the test
from Ref. [10]: that had a steeper slope in the cases of the fibers with coating (in Fig. 3
specimens M8 and M12 the crack extended from the bottom of the
fθ = 2P /(π·l·d ) (3) specimen to the side surface while in specimens M8C, M12C the crack
extended from the bottom face to the upper face of the specimen).
It has been demonstrated that the expression used for the calcula- Through these major cracks, the fibers seemed stretched, pulled or
tion of the maximum elastic tensile stress [Eq. (3)] is still valid for ruptured.
various plastic models [13]. The direct tensile strength may be esti- Fig. 3 plots the axial stress-axial strain (right) and axial stress-lateral
mated from the spit cylinder test results after properly reducing by the strain (left) obtained from the compression tests of the SHFRCC mate-
multiplier Asp (based on [5], a value of Asp = 1.0 has been selected). rials and the results for the same matrix without fibers (PC) as the
This equation was used to derive the tensile strength for the diagrams of average values from three specimens each. The compressive stress of
Fig. 2(b). If all the fibers could reach their ultimate strength followed by the material without the use of fibers reached an ultimate stress value of
rupture the ultimate tensile strength of the composite ought to be almost 50 MPa (average of three tested specimens), but with the use of
fctm = 8 MPa. In the splitting tests this value is closest to the tensile uncoated fibers, this stress value was increased. An increase in the axial
strength of the SHFRCC containing 12 mm long fibers without coating deformation of SHFRCC at peak load was observed, and a stable des-
(M12). Through comparison with the direct tension test results it is cending branch marked the improved performance. Note that lateral
concluded that indirect tension strength values obtained from splitting expansion is responsible for the steepness in the post peak branch in the
are an upper bound estimate, whereas the uniaxial tensile test yields the compressive stress-strain response of concrete (Zanganeh and Panta-
lower bound values (mostly owing to the fact that fiber pullout controls zopoulou [14]). However, in the presence of fibers the behavior was
owing to the fiber coating). The nature of each experiment also affects improved and the post-peak decay was much milder as fibers were
the result: for example, in the case of direct uniaxial tension the spe- mobilized in the lateral direction bridging the cracks, transferring load
cimen is under plane stress, whereas in the case of cylinder splitting a and limiting lateral expansion of the cylinder under compression, as
state of plane strain occurs with the longitudinal extension owing to evidenced by the restricted growth of post-peak lateral strain as com-
Poisson’s ratio affecting the final capacities. Also in the split cylinder pared to the plain concrete specimens which undergo excessive, un-
test the ultimate stress occurs at a single cross section (cross section controlled lateral expansion (Fig. 3). The stress-strain curves suggest
directly under loading) whereas in the case of direct tensile test stress it that past the peak load, concrete with fibers behaves as if passively
is uniform, occurring over a wider range of the specimen. This leads to confined. The intensity of the effective confinement owing to the role of
unavoidable concentration of deformations in one cross section and fibers is directly related to the materials' characteristic stress-strain
does not permit the full strain hardening properties of the material to be behavior under tension.
developed. The Modulus of Elasticity (determined from the compression stress-
strain diagram at 45% of peak load) was low, in the range of 20 GPa due

4
A.V. Georgiou, S.J. Pantazopoulou Composite Structures xxx (xxxx) xxx–xxx

Fig. 3. Compression test crack formation on various mix

designs and average stress-axial/lateral strain of various mix
designs with and without fibers.

to the elimination of aggregates in the mix design while Poisson’s ratio, 3.2.3. Plain matrix confined with FRP sheets
ν, differed between the FRCs (≈0.22) and the plain matrix (≈0.33). When FRP sheets were used for confinement, the behavior changed
from softening to hardening, with a sudden collapse when the FRP
sheets ruptured, upon attainment of their rupture strain (Fig. 5). The
3.2.2. Plain matrix confined with steel stirrups results of the uniaxial compression tests on the 200 × 100 mm cylinders
The use of confinement changed the type of failure of the PC spe- are plotted in the diagrams of Fig. 5. Confinement due to jacketing
cimens. Spalling of the concrete cover was observed in all cases but increased the load carrying capacity, the axial deformation at maximum
strength deterioration decreased with denser stirrup spacing. Results of load as well as the lateral expansion of the specimens. In the case of
the uniaxial compression tests on the 200 × 100 mm cylinders are carbon FRP, even for only 2 layers wrapped around the specimen (C2),
plotted in Fig. 4. Each curve represents the average of three specimens the compressive strength is doubled, and after a point of apparent
and is marked by the respective batch code name. On the horizontal yielding associated with cracking of the matrix inside the FRP, there is
axis strains parallel and orthogonal to the direction of load are pre- an increase in the stress capacity and a plateau up to a ductility of 2.5.
sented (right and left side of the figure respectively). The vertical axis In this range of hardening or plateau in the axial stress – axial strain
represents the applied stress. The additional confinement changes the diagram, the lateral strain was steeply increasing but it always re-
load carrying capacity, the axial deformation at maximum load, as well mained lower in magnitude than half the axial strain (i.e. apparent
as the lateral expansion of the specimens. Evidently there is little effect dilation ratio < 0.5), mainly owing to the increasing magnitude of
by the use of stirrups at 80 mm center to center spacing. By decreasing lateral confining pressure exerted by the linearly elastic FRP material.
the center to center distance to 50 mm the post-peak behavior changes At a lateral strain of 1% the FRP ruptured and the bearing capacity
to a more ductile one, whereas reducing the center to center stirrup suddenly collapsed.
spacing to 30 mm caused further improvement to the post-peak des- Both types of FRP sheets increased the occurrence of first cracking
cending branch converting it almost to a plateau.

Fig. 4. Stress-axial/lateral strain and failure cracking for cylinders confined with steel stirrups.

5
A.V. Georgiou, S.J. Pantazopoulou Composite Structures xxx (xxxx) xxx–xxx

Fig. 5. Failure of cylinders and average stress-axial/lateral

strain diagrams for specimens confined with FRP wraps.

strain of the matrix from 55 MPa to approximately 80 MPa. The glass represents the applied stress. By adding fibers to the mix (F1, F2, F3,
FRP of one and two layers, as well as the one layer of carbon FRP, did F4) the behavior obtained is similar to that obtained from stirrup
not show an increase in strength after first cracking, whereas in the confinement (F5-S8/S5/S3) but with an increase of compressive capa-
cases of two and three layers of carbon FRP strain hardening was ob- city and a shift of strain at peak load at 5‰ strain, in comparison to the
served in the axial stress – axial strain diagram up to values of 120 MPa plain cementitious matrix (PC) at 3‰.
strength. While the transition from one to two layers of each FRP type The stress-strain curve shows a gradual increase in supported axial
plays a significant role in the overall response, no important difference stress up until the sudden rupture of the FRP followed by complete loss
appears between the two and three layers of carbon FRP, indicating that of strength. If the responses of otherwise identical concretes confined
important role in the strength capacity plays the restraint to lateral either by stirrups or FRP jackets are compared, it is seen that the ef-
expansion of the cylinders (limited by the strain capacity of the FRP fective Modulus of Elasticity increased from the unconfined value of
material) and not the stress that is provided by the number of sheets. 20 GPa to 27 GPa. In the case of the FRP confinement the stress-strain
curve presents a bilinear behavior for a large range of deformation,
reaching significantly higher stresses than the stirrup confined and the
4. Comparison of axial-lateral strains between different
FRC mixes. The first peak is identified by a sharp change in slope.
confinement methods
Strength and the corresponding strain at first peak did not increase
proportionately with the addition of more jacket layers. However, in-
4.1. Overall response
creasing the number of FRP layers caused a commensurate increase in
the maximum load and delayed volumetric expansion. Failure of the
The results of all uniaxial compression tests on cylinders are plotted
specimens happened in all cases by rupture of the FRP at a lateral strain
in Fig. 6. Each curve represents the average of three specimens. On the
of 5.5% (the nominal rupture strain of carbon and glass FRPs used were
horizontal axis strains parallel to (right hand side) and orthogonal (left
1.8% and 3% respectively).
hand side) to the direction of load are presented. The vertical axis

Fig. 6. Stress-axial/lateral strain of compression test for all specimens

with various confining methods.

6
A.V. Georgiou, S.J. Pantazopoulou Composite Structures xxx (xxxx) xxx–xxx

4.2. Lateral confinement

The abscissa in all diagrams of Fig. 7 is the axial strain in the di-
rection of loading. The respective coordinates from the top diagram to
bottom are the lateral strain, the volumetric strain εv (=εaxial + 2 εlat)
and the applied axial stress. The diagrams refer to the fiber reinforced
mixes obtained by varying the PVA length and coating properties
(mixes F1, F2, F3 and F4). It is evident that the addition of fibers in-
fluences the postpeak response of the matrix by improving its de-
formation capacity and reduces the strength loss from its peak value.
Addition of fibers did not affect the Modulus of Elasticity as it was
approximately the same as that of the plain matrix (20 GPa).
Failure of concrete is usually affiliated with the reversal of volu-
metric strain from net contraction to dilation as cracking progresses
[1,15]. Also the rate of volumetric strain reversal has been shown to
characterize deformation capacity and the rate of strength degradation.
A low rate of expansion is evident in the case of fibers, as they bridge
the cracks, transferring stresses across the cracks, therefore applying
passive confinement to the cylinder in the full extension of its cross
section. The onset of strength loss is delayed from 3.1‰ for the plain
matrix to 4.4‰ for the PVA-fiber reinforced composites. The vertical
dotted lines mark the point of maximum stress connecting it through
strain with the point of zero volumetric expansion point.

4.3. Analysis of the uniaxial compression strength based on lateral

expansion

4.3.1. Decay of strength based on lateral expansion

As axial compressive strain increases after the peak load (Fig. 7) a
rapid increase in lateral tensile strain occurs at a rate much faster than
what is estimated by the original Poisson’s phenomenon. The curvature
of the slope of lateral to axial strain curve is reversed exhibiting a
slower post-peak decay for the coated-PVA composite beyond an axial
strain of 0.5% and a lateral strain of 0.8% (Point O-Inflection Point in
Fig. 7(b) lateral to axial strain). In the volumetric expansion to axial
strain diagram (Fig. 7(c)) there is a linear elastic volumetric response
that contracts at a rate of (1 − 2v) up to the onset of cracking. The
composites stop to decrease in volume at the point where the curves
reach the horizontal axis and thereafter volumetric expansion begins.
The composites without coating reach zero volumetric change at a
higher axial strain suggesting a lower rate of lateral expansion (due to
stronger bonding of the fibers). In normal concrete volumetric stain
reaches a zero value at a level of uniaxial strain ε1∗ of about 80–100% of
the strain at peak stress, whereas in the case of fiber reinforced com-
posites ε1∗ occurs at the starting point of strength degradation sug-
gesting that fiber confinement was effective prior to that level (see
Fig. 7(a) for the definition of ε1∗).
The rate of increase between the lateral strain curves of confined PC
and SHCFRC concretes changes at Point O, with the composites of
coated-PVA fibers showing the lowest slope and therefore the lower
lateral expansion to axial contraction ratio. Comparing coated and un-
coated fiber mixes it is observed that the maximum strength coincides
with the point of zero volumetric expansion at an apparent Poisson’s
ratio of 0.5 for the un-coated fibers whereas for coated fibers this
milestone point occurs at lower values of the apparent Poisson ratio
(the apparent Poisson’s ratio is the ratio of absolute lateral to axial
strain). After Point O the apparent Poisson’s ratio ceases to increase
rapidly turning almost to a stable value of approximately 1.5 for the un-
coated fibers while it maintains lower values for the coated ones.
Reflecting the previous information in Fig. 7, a Poisson’s ratio of 0.5
Fig. 7. From top to bottom (a) Lateral strain-axial strain, (b) Volumetric strain-axial seems to be reached at the end of the axial stress plateau for the
strain, (c) Apparent Poisson’s ratio-axial strain and (d) Stress-axial strain for all fiber composites with un-coated fibers, marking the onset of the degrading
reinforced mixes and (e) Physical significance of area strain and definition of volumetric segment of the curve and Point O marks the initiation of the residual
strain plot [15]
stress plateau.
Based on previous experimental data by Imran [16] there is a cri-
tical level of (contractive) volumetric strain in the range of 0.75–1.25%

7
A.V. Georgiou, S.J. Pantazopoulou Composite Structures xxx (xxxx) xxx–xxx

Fig. 8. Passive confinement function in directions orthogonal to the compression stress field qC, under a transverse tension strain field εlat owing to Poisson- effect and estimation of
equivalent lateral confining pressure, σlat (Archontas and Pantazopoulou [18]).

corresponding to the well known “transition” point in the stress co- Values of lateral strain of the S5 specimens are not considered since the
ordinates. Beyond that level of contraction alterations in the micro- extensometer that was placed on the perimeter of the specimen mea-
structure are associated with gel failure. In the case of fiber composites, sures the total lateral deformation that occurred both in the confined
this point is no longer a failure point but a change of the mechanism core and the cover of the specimen. The equivalent confining pressure
transferring the stresses, mainly focusing on the properties of the fibers provided by the stirrups in the S5 specimens is calculated based on Eq.
bridging the cracks and their interfacial bond relation to the matrix. (4).
This transition point marked before as Point O is a limit point beyond σlat = αs·ρv ·f y,st /2 (5)
which the magnitude of the area strain decreases with increasing axial
strain due to the increasing lateral pressure exerted by the fibers. The where αs = (1 − s/(2d′)) = (1 − (42/2/62)) = 0.43, ρv = 4·50·3.14·62/
2 2

problem of multiaxial stress state (compression due to loading and (3.14·622·50) = 6.4% and fy,st = 500 MPa. Therefore the lateral confine-
lateral confinement due to fibers) is not considered a pressure-depen- ment provided by the 8 mm stirrups at a center to center spacing of 50 mm
dent problem, but as restraint-dependent one according with Pantazo- is equal to σlat = 0.43·6.4%·500/2 = 7.2 MPa. This is the equivalent con-
poulou [17]. fining stress that is also provided by the fibers to the SHFRCC specimens
The εv-εaxial plot is an indicator of internal damage and the area in (M12). Revisiting Fig. 2 which plots the tensile strength of the split cy-
the graph that is bound between the curve and a 45° (εv = ε1 in that linder tests on the specimens with the M12 mix design one can find that
case) line represents the area increase per unit area of cross section the average splitting strength of the F2 mix specimens was calculated as
supporting the load [15] as shown in Fig. 7. In this case volumetric 7.7 MPa. In conclusion it is found that the confining pressure provided by
strains of up to −3% (minus symbol indicates contraction) are reached the 12 mm uncoated PVA fibers equals to its splitting strength by con-
for the 8 mm PVA un-coated fibers mix and up to −2% for PVA mix sidering that its lateral confinement matches the confinement provided by
designs where usual values for unconfined concrete are in the order of the stirrups with a center to center spacing of 50 mm.
−0.5%.
5. Conclusions
4.3.2. Confining role of fibers (restraint to transverse expansion)
Within this paper the behavior of fiber reinforced strain hardening
Archontas and Pantazopoulou [18] interpreted the internal con-
composites under uniaxial compression was studied, and a quantitative
fining effect generated by fibers as shown in Fig. 8. Normal compression
comparison of the confining effect of the fibers was obtained through
stress generates transverse tensile strains εlat; due to these strains, the
parallel tests on the plain matrix, the SHFRCC and the plain matrix
fibers crossing an arbitrary plane B-B′ parallel to the direction of uni-
confined by stirrups or FRP sheets. The lateral and volumetric behavior
axial compression develop tensile stresses and forces ff,i: To maintain
of the composites was investigated with the results obtained from a
equilibrium in direction orthogonal to compression a passive confining
circumferential extensometer used in the experimental study. Lateral
pressure σlat is required in order to counteract the stress resultant of the
strain is a meaningful quantifier of the internal cracking that is influ-
fiber forces normal to the B-B′ section. The equivalent transverse con-
enced past the inflection point by the effective fiber confinement. Based
finement in this case is,
on the lateral confinement provided by one of the well established
σlat = −λ eff ·nf ·Af ·f f ,i where, f f ,i = Ef ·εlat ⩽ fb,f ·0.5ℓf ·π·df (4) confinement methods (stirrups with 50 mm center to center spacing)
that had similar stress-strain curves to those obtained by one of the
where nf is the number of fibers crossing an arbitrary unit area of the SHFRCC mixes (M12), it was estimated that the effective intrinsic
bulk material (given in Ref. [6] – Eq. (2)), Af is the fiber cross sectional confinement provided by the fibers is equal to the splitting strength of
area, ff,i is the stress of the arbitrary fiber, fb,f the bond strength of the the material, whereas the strain and strength capacity increase resulting
fibers, ℓf the fiber length and df is the fiber diameter. from this confinement follows the established confinement models of
Lateral restraint in compression is reflected in the material stiffness regular concrete.
by delaying or mitigating damage due to transverse expansion normal
to the direction of principal compression, leading to a commensurate References
increase in compressive strain capacity at peak stress, εc. Although
other effects (entrapped air and disruption of the gel structure) prevent [1] Imran I, Pantazopoulou SJ. Experimental study of plain concrete under triaxial
an evident increase of compressive strength where cohesion is of stress. ACI Mater J 1997;93:589–601.
[2] Shah S, Rangan B. Fiber reinforced concrete properties. ACI J Proc 1971:126–37.
paramount importance), yet the effect of the confining pressure is seen [3] Shah SP, Rangan BV. Effects of reinforcements on ductility of concrete. J Struct Div
in the sustained axial strain with no loss of load carrying capacity for a 1970;96:1167–84.
prolonged range exactly because the lateral pressures, σlat, prevent the [4] Concrete Committee JSCE. Recommendations for design and construction of high
performance fiber reinforced cement composites with multiple fine cracks
uncontrolled lateral growth of the material under compression.
(HPFRCC); 2008.
Fig. 7 shows that the SHFRCC resembles the performance of the [5] FIB Model Code. Model Code 2010. Federation Internationale du Beton (fib); 2010.
specimens confined with stirrups at a center to center spacing of 50 mm. [6] Georgiou AV, Pantazopoulou SJ. Effect of fiber length and surface characteristics on

8
A.V. Georgiou, S.J. Pantazopoulou Composite Structures xxx (xxxx) xxx–xxx

the mechanical properties of cementitious composites. Constr Build Mater [12] Timoshenko S, Woinowshy-Krieger S. Theory of plates and shells. 2nd ed. New
2016;125. http://dx.doi.org/10.1016/j.conbuildmat.2016.09.009. York: McGrawl-Hill; 1959.
[7] Lepech M, Li V. Large-scale processing of engineered cementitious composites. ACI [13] Chen WF. Plasticity in reinforced concrete. New York: McGrawl-Hill; 1982.
Mater J 2008;105(4):358–66. [14] Zanganeh M, Pantazopoulou SJ. Triaxial tests of fiber-reinforced concrete. J Mater
[8] ASTM C618. Standard specification for coal fly ash and raw or calcined natural Civ Eng 2001;15:340–8.
pozzolan for use; 2008. http://dx.doi.org/10.1520/C0618. [15] Pantazopoulou S, Mills R. Microstructural aspects of the mechanical response of
[9] Georgiou A, Pantazopoulou S. Effect of fiber length and surface characteristics on plain concrete. ACI Mater J 1995;M26:605–16.
the mechanical properties of cementitious composites. Constr Build Mater [16] Imran I. Applications of non-associated plasticity in modelling the mechanical re-
2016:1216–28. sponse of concrete. Canada: University of Toronto; 1994.
[10] ASTM C496. Standard test method for splitting tensile strength of cylindrical con- [17] Pantazopoulou SJ. Role of expansion on mechanical behavior of concrete. J Struct
crete specimens. ASTM Int 2006:1–5. http://dx.doi.org/10.1520/C0496. Eng 1995;121:1795–805.
[11] Georgiou A, Pantazopoulou S, Petrou M. Increasing ductility of FRCC with pro- [18] Archontas ND, Pantazopoulou SJ. Microstructural behavior and mechanics of nano-
prietary coating agents. In 10th fib Int. PhD Symp. Civ. Eng. Quebec City; 2014. modified cementitious materials. Adv Concr Constr 2015;3:15–37.