Sie sind auf Seite 1von 13

Mechanics of Materials 79 (2014) 45–57

Contents lists available at ScienceDirect

Mechanics of Materials
journal homepage: www.elsevier.com/locate/mechmat

Experimental investigation of the effect of coarse aggregate


shape and composition on concrete triaxial behavior
E. Piotrowska, Y. Malecot ⇑, Y. Ke
UJF-Grenoble 1, Grenoble-INP, CNRS UMR 5521, 3SR Lab, Grenoble F-38041, France

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

Article history: This article focuses on identifying concrete behavior under high triaxial loading. The study
Received 1 December 2013 is carried out within a more general context of understanding the behavior of concrete
Received in revised form 18 June 2014 under impact. The effect of coarse aggregate shape and composition on concrete under
Available online 3 September 2014
high triaxial compression is examined by means of a very high-capacity triaxial press.
The shape effect is further investigated using rolled aggregates, crushed aggregates and
Keywords: glass balls. The influence of aggregate composition is determined on concretes mixed with
Concrete
siliceous aggregates, glass aggregates and limestone aggregates. These analyses specifically
Coarse aggregate shape
Coarse aggregate composition
indicate that coarse aggregate shape exerts no influence on the concrete response at high
Triaxial test confinement. Moreover, concrete behaves like a non-cohesive granular stacking that seems
High confinement to be governed by the compacted cement matrix provided a sufficiently high aggregate
strength. Otherwise, a lower aggregate strength serves to weaken this granular stacking.
At an intermediate level of confinement, the shear strength of concrete is mainly controlled
by aggregates strength; however, irregularly-shaped coarse aggregates slightly increase
the overall strength of concrete.
Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction target is thin enough, then both simple tension and shear
stresses can be observed on the distal face of the concrete
Concrete is the world’s most widely used man-made specimen during the final penetration phase. A validation
material. More specifically, it is employed in the construc- of concrete behavior models, which simultaneously take
tion of complex infrastructure (high-rise buildings, dams, into account the phenomena of brittle damage and irre-
nuclear reactors, etc). A thorough understanding of the versible strains, requires test results that enable reproduc-
behavior of concrete under extreme loading situations, ing these complex and intense loading paths.
such as near-field detonations or ballistic impacts, is The static characterization of a constitutive model to
essential. In these situations, concrete undergoes very predict dynamic calculations represents a common prac-
high-intensity triaxial stress states (Zukas, 1992). When a tice in the study of geomaterials. The behavior of concrete
projectile strikes a concrete structure, various localized in confined compression test slightly depends on the strain
effects are generated: spalling on the front face of the rate for dried or wet specimens (Forquin et al., 2010). Static
structure can be associated with simple tension, while triaxial compression is the loading type discussed in this
penetration of the projectile into the structural core consti- article. Concrete behavior under such loading is now rela-
tutes the source of dynamic triaxial compression. If the tively well-known, especially at moderate levels of confin-
ing pressure (Pugh, 1970; Kupfer and Gerstle, 1973; Linhua
et al., 1991; lmran and Pantazopoulou, 1996; Sfer et al.,
⇑ Corresponding author. 2002). These studies have particularly revealed the transi-
E-mail address: yann.malecot@ujf-grenoble.fr (Y. Malecot). tion from brittle to ductile behavior that characterizes

http://dx.doi.org/10.1016/j.mechmat.2014.08.002
0167-6636/Ó 2014 Elsevier Ltd. All rights reserved.
46 E. Piotrowska et al. / Mechanics of Materials 79 (2014) 45–57

cohesive materials. Some triaxial tests at high confinement model of concrete (Dupray et al., 2009; Malecot et al.,
levels have also been performed, on mortar (Bazant et al., 2014). The present study completes these research efforts
1986; Burlion et al., 2001; William et al., 2005) and con- with the knowledge of the effect of coarse aggregate on
crete samples (Warren et al., 2004; Schmidt et al., 2009); the concrete triaxial behavior.
their results have shown the evolution in material behav- The experimental device used in this study will be
ior and limit states in the presence of confinement. described in Section 2. Test results concerning the effects
The results presented in this article refer to triaxial of coarse aggregate shape on concrete behavior will be pre-
compression tests conducted on concrete samples by sented in Section 3.1, followed by the effect of coarse
means of a high-capacity hydraulic triaxial press, called aggregate composition (Section 3.2). Concluding remarks
GIGA. This experimental device allows testing specimens will be provided in Section 4.
at stress levels in the order of 1 GPa with static, homoge-
neous and well-controlled loading paths. An extensive
experimental study on an ordinary concrete was previ- 2. Experimental set-up
ously carried out with the GIGA press. Gabet et al. (2008)
studied the influence of loading path on concrete behavior 2.1. The triaxial cell
and revealed that under high confinement, the concrete
limit state remains relatively independent of both the Triaxial tests were performed by means of a high-
loading path and Lode’s angle. To describe the overall capacity triaxial press, called GIGA. A full description of
behavior of concrete under high confinement in greater this testing procedure on concrete has been presented in
detail, Poinard et al. (2010) performed cyclic tests using detail in Vu et al. (2009c). The press, illustrated in Fig. 1,
several loading–unloading cycles on the same concrete. is able to generate a confining pressure of up to 0.85 GPa
These results indicated that the evolution of elastic fea- and an axial stress reaching 2.3 GPa on cylindrical concrete
tures (Young’s modulus, Poisson’s ratio) becomes less pro- specimens 7 cm in diameter and 14 cm long. Fig. 2 shows a
nounced as confinement increases. Using the same cross-sectional view of the confining cell, where the
baseline material, Vu examined the effect of the saturation concrete sample is to be placed. The confining fluid,
ratio on concrete behavior under high confinement diethylhexyl azelate, a non-volatile organic, inert and
(Vu et al., 2009a). This ratio exerts a major influence on slightly-compressible liquid, is injected into the cell
concrete behavior at high confinement: the deviatoric through an upper opening; it is then pressurized by means
strength of concrete considerably increases with confining of a pressure multiplying jack. The axial force is generated
pressure for dried samples and remains constant over a from a 13-MN jack located under the cell and then trans-
given confining pressure range for either wet or saturated mitted to the sample via a piston that passes through the
samples. Another study by Vu et al. (2009b) demonstrated lower plug of the confining cell.
that the water/cement ratio has no influence on concrete
strength under high confinement. Concrete behaves like a
2.2. Instrumentation and measurements
non-cohesive granular stacking and shows no sensitivity
to conventional compressive strength f c28 . A subsequent
The press is equipped with various sensors that serve to
study of the effect of cement paste volume and coarse
supervise the tests and provide information on the state of
aggregate size (Vu et al., 2011) revealed that at very high
the sample during loading. An axial displacement sensor is
confinement, both the paste volume and aggregate size
installed in order to control the axial jack displacement,
only had a slight influence on concrete deviatoric behavior
while an axial force sensor and pressure sensor positioned
though aggregates size was significantly correlated with
inside the confining cell yield the stress state on the
concrete strain limit states.
specimen.
This study is specifically concerned with the influence
of coarse aggregate type on concrete behavior at high
stress levels. Many experimental studies have already
examined this effect under uniaxial loading, in exposing
that the influence of coarse aggregate type on concrete
strength properties and failure mechanisms varies in mag-
nitude and depends on the water/cement ratio of the mix
(Kaplan, 1959; de Larrard and Belloc, 1997). To the best
of the authors’ knowledge, no results are available regard-
ing the effects of coarse aggregate type on concrete behav-
ior under high triaxial compression loading. This issue will
be addressed in the present article, from the standpoint of
both coarse aggregate shape and composition.
The results of these experiments are essential to meso-
scopic modeling of concrete under high confinement.
Indeed, the validation of predictive models for several
kinds of concrete formulations requires experimental data.
The GIGA press enabled identifying the behavior of mortar
under high stress levels, as part of developing a mesoscopic Fig. 1. Overview of the GIGA press.
E. Piotrowska et al. / Mechanics of Materials 79 (2014) 45–57 47

tests were controlled in displacement with a strain rate


of approx. 10-5/s.

2.4. Characteristics of the concrete samples

The present study was conducted on four concrete


samples distinguished by the type of coarse aggregate in
their mix design. The compositions of each of these four
concretes are listed in Table 1. The reference concrete
(SR) contains siliceous rolled aggregates originating from
a natural deposit. This reference is an ordinary concrete
in terms of both strength (30 MPa) and slump (7 cm). The
three modified concretes all contain the same cement
matrix volume, the same aggregate volume and an aggre-
gate size distribution as close as possible to the reference
Fig. 2. Cross-sectional view of the confining cell.
mix (see Fig. 3). They only differ by the type of coarse
aggregate in their composition: crushed aggregate
concrete (SC) contains crushed aggregates obtained from
Sample strain measurements are carried out using an
siliceous rock; limestone aggregate concrete (LC) is
LVDT (Linear Variable Differential Transformer) sensor,
produced with crushed aggregates obtained from soft
an axial gauge and two circumferential gauges. The LVDT
limestone rock; in the glass ball concrete (GB), coarse
sensor outputs the length variation of the specimen. Each
aggregates are replaced by spherical glass aggregates.
part of the sensor is fastened onto a loading head, yielding
The choice of specimen size (i.e. 7 cm in diameter and
a global measurement of the axial strain. The axial gauge,
14 cm long) makes it possible to test the specimens under
bonded to the middle of the concrete sample, measures
high stress levels and, in considering concrete a homoge-
the local strain. A comparison of the two output values
neous material, with respect to maximum aggregate size
allows evaluating sample strain homogeneity. The circum-
(8 mm). Setting the length/diameter ratio equal to 2, i.e. a
ferential strain is measured using two gauges, in order to
value commonly used to characterize concrete and other
increase the probability of maintaining at least one
geomaterials, both limits the influence of boundary condi-
measurement until the end of the test protocol and verify
tions on specimen behavior and prevents buckling.
the strain homogeneity. These gauges allow for strain
After pouring, the concrete test blocks were conserved
measurements of up to 15%.
for 1 month in water, then machined and kept for at least
The porous nature of concrete requires developing a
3 months in a drying oven at 50 °C, which practically dried
multilayer protective membrane around the sample,
the concrete by the time of testing (i.e. a concrete satura-
composed of 8 mm of latex and 2 mm of neoprene; this
tion ratio equal to approximately 11%). Mass stabilization
membrane prevents the confining fluid from infiltrating
was achieved after 1 month in the oven, once the daily var-
the specimen (Gabet et al., 2006).
iation did not exceed 0.1%. Moreover, all test samples were
conserved in the oven long enough to assume negligible
2.3. Loading path effect of age on concrete strength.
The manufacturing procedure for concrete samples was
In this paper, compressive stresses and contractive developed with the aim of ensuring minimal variability in
strains are assumed to be positive. rx is the principal axial the mechanical properties of tested concretes. Several tests
stress, p the confining pressure, rm the mean stress and performed on SR concrete show a very good reproducibil-
q the principal stress difference: ity of the results (Vu et al., 2009b,c): the dispersion in con-
rx þ 2p crete deviatoric strength is less than 6% at confining
rm ¼ ð1Þ pressure levels greater than 50 MPa.
3

q ¼ rx  p ð2Þ
3. Results
The triaxial compression loading is applied in two phases:
hydrostatic and deviatoric. The hydrostatic phase consists In this section, the results of triaxial tests performed on
of applying a confining pressure all around the specimen the four test concretes, SR, SC, LC and GB, at various levels
at a constant stress rate of 1.67 MPa/s up to a desired of confinement will be discussed.
value. Once the intended confinement has been reached, For the unconfined compression test, volumetric behav-
the specimen is loaded axially at a constant displacement ior (mean stress rm vs. volumetric strain ev ) and axial
rate of 20 lm=s (i.e. a strain rate of approx. 10-4/s) while behavior (axial stress rx vs. axial strain ex and circumferen-
keeping the confining pressure constant. The unloading tial strain eh ) are output. For tests carried out under a con-
phase proceeds in the reverse order. fining pressure, volumetric behavior and the deviatoric
Uniaxial compression tests were performed on a differ- part of axial behavior (deviatoric stress q vs. axial strain
ent laboratory press, i.e. the 100-ton Schenk press. These ex and circumferential strain eh ) are determined. Each
48 E. Piotrowska et al. / Mechanics of Materials 79 (2014) 45–57

Table 1
Composition and porosity of the studied concretes; x – values not measured.

Concrete SR SC GB LC
Gravel/Crushed rock D 0.5/8 [kg/m3] 1008 1008 – 891
Glass balls D 1.5/9 [kg/m3] – – 980 –
Sand Dmax 1800 lm [kg/m3] 838 838 838 838
Cement CEM I 52.5 N PM ES CP2 (Vicat) [kg/m3] 263 263 263 263
Water [kg/m3] 169 167.5 169 169
Superplasticizer ‘‘Sikafluid’’ [kg/m3] 0 1.8 0 0
Density [kg/m3] 2278 2278 2250 2161
Porosity accessible to water [%] 11.8 10.8 12.6 x
Porosity accessible to mercury [%] 12.6 11.9 9.7 13.8

Fig. 3. Granulometry curves of coarse aggregates used in the studied concretes.

curve is offset such that strain (e0h;x ) is equal to 0 when q images of these two concretes displayed in Fig. 4 show the
equals zero. strong shape difference.
Axial strain is obtained from the LVDT sensor, while
circumferential strain is derived as the mean of the two 3.1.1. Unconfined compression test
circumferential gauge readings. Volumetric strain is Fig. 5 shows the results of unconfined compression tests
calculated from these two measurements, in assuming that performed on the two studied concretes. The correspond-
the concrete radial strain is similar to the circumferential ing elastic properties and strengths are listed in Table 3.
strain: It can be observed that the two concretes display identical
elastic properties, though slight differences in the ultimate
ev ¼ ex þ 2eh ð3Þ stress are noted.
The uniaxial strength of an ordinary concrete is
controlled by the strength at the mortar/aggregate
3.1. Rolled vs. crushed aggregates: the influence of shape interface, which is the weakest zone in concrete provided
strong enough aggregates (see typical crack pattern in
A variety of aggregate properties, which are often diffi- Fig. 12(a)). A hypothesis can be made that the nearly
cult to measure, complicate the task of studying the effect identical composition of rolled and crushed aggregates
of a single aggregate characteristic on concrete behavior. yields a similar microscopic surface roughness and, conse-
This section will attempt to evaluate the influence of quently, a comparable interfacial bond strength. In con-
coarse aggregate shape. Triaxial tests were performed on trast, higher angularity of crushed aggregates results in
the concrete samples made with two aggregate shapes: the higher strength of concrete, most likely due to the
rolled and crushed. Both aggregates are siliceous with a mechanical interlocking of the aggregate and the cement
very similar silica content (see Table 2). X-ray tomography matrix. Similar observations can be found in the literature
in case of uniaxial compression tests on ordinary con-
cretes: higher compressive strength of concretes made
Table 2
Properties of rolled and crushed aggregates. with crushed aggregates than those made with gravel
(Özturan and Çeçen, 1997; Giaccio and Zerbino, 1998).
Aggregate type Rolled Crushed
Silica content [%] 99 97 3.1.2. Triaxial compression at 50 MPa of confinement
Density [g/cm3] 2.6 2.6
Fig. 6 presents the results of triaxial tests conducted at
Bulk density [g/cm3] 1.6 1.6
50 MPa of confinement on concretes samples with rolled
E. Piotrowska et al. / Mechanics of Materials 79 (2014) 45–57 49

(a) Rolled aggregate concrete (b) Crushed aggregate concrete


Fig. 4. Tomographic cross-sectional view of three concretes varying in their coarse aggregate shape.

45
14 SR
SR
SC
40 SC
12
35
10 30
σx (MPa)
σm (MPa)

8 25
20
6
15
4
10
2 5
0 0
−0.04 −0.02 0 0.02 0.04 0.06 −0.4 −0.2 0 0.2 0.4 0.6
ε (%) εθ,x (%)
v

(a) (b)
Fig. 5. Unconfined compression tests on concretes with rolled (SR) and crushed (SC) aggregates: (a) mean stress rm vs. volumetric strain ev , (b) axial stress
rx vs. strain components eh and ex .

Table 3 3.1.3. Triaxial compression at 650 MPa of confinement


Unconfined compression test: identification of the main concrete Fig. 7 presents the results of triaxial tests at 650 MPa of
characteristics.
confinement conducted on the two studied concretes. The
Concrete SR SC volumetric behavior displayed in Fig. 7(a) suggests barely
Young’s modulus, E [GPa] 33 33 no difference in the compaction behavior. The deviatoric
Poisson’s ratio, m [–] 0.16 0.16 response shown in Fig. 7(b) indicates that under very high
Compressive strength, rmax [MPa] 40 43 confinement, the two concretes behave nearly the same
also in terms of ultimate stress and tangential stiffness.
Their respective failure patterns are also very similar. A
series of cracks perpendicular to the axial direction is
and crushed aggregates. The volumetric behavior is nearly observed (see Fig. 8). Most of aggregates are debonded
linear during the hydrostatic phase for both samples. around these cracks except a few that are broken. Such
Fig. 6(b) indicates the differences in maximum devia- observations were also made thanks to X-ray tomographic
toric stress. At this level of confinement the cement matrix images obtained for both types of concrete specimens
is still cohesive. Thus, as with the unconfined compression, (Poinard et al., 2012). Most probably, the observed cracks
the higher strength of SC is probably due to the higher do not open during the loading. The crack opening in con-
roughness of crushed aggregates leading to a stronger crete is here fully inhibited by the mean stress level. A pos-
aggregate/mortar bond at the mesoscopic scale. sible way to evaluate when these cracks open is to examine
Rolled aggregate concrete and crushed aggregate con- the unloading stiffness. The stiffness does not decrease sig-
crete both feature very similar crack patterns. As observed nificantly above a mean stress of 50 MPa whereas it tends
previously by X-ray CT (Poinard et al., 2012), the failure to zero at low confinement. This suggests that the crack
mechanism is highly localized in a few cracks inclined at opening occurs during the last part of unloading. On the
30° in the axial direction. Some large aggregates are bro- other hand, unloading–reloading cycles, which are revers-
ken, especially in the crushed aggregate sample. ible, suggest that the damage appears during the loading
50 E. Piotrowska et al. / Mechanics of Materials 79 (2014) 45–57

120 200

100
150
80

σm (MPa)

q (MPa)
60 100

40
50
20
SC SR
SR SC
0 0
0 0.5 1 1.5 2 2.5 0 5 10
εv (%) ε’θ,x (%)

(a) (b)
Fig. 6. Triaxial test at 50 MPa of confinement on concretes with rolled (SR) and crushed (SC) aggregates: (a) mean stress rm vs. volumetric strain ev , (b)
stress deviator q vs. strain components e0h and e0x .

1000 1000

800 800
σ (MPa)

600 600
q (MPa)
m

400 400

200 200
SR SR
SC SC
0 0
0 2 4 6 8 10 12 0 5 10
ε (%) ε’θ,x(%)
v

(a) (b)
Fig. 7. Triaxial test at 650 MPa of confinement on concrete with rolled (SR) and crushed (SC) aggregates: (a) mean stress rm vs. volumetric strain ev , (b)
stress deviator q vs. strain components e0h and e0x .

but is inhibited at this stage (no crack opening) (Poinard crushed limestone aggregates (LC), both of which exhibit
et al., 2010). very distinct physical and mechanical properties (see
At this level of confinement, concrete behaves like a non- Table 4 and Fig. 9), will be analyzed.
cohesive granular stacking. Indeed, the strain levels reached In order to better understand the response of concretes,
during hydrostatic compression strongly exceed the dam- the triaxial behavior of rocks was investigated: tests were
age threshold of the cement matrix and lead to the rupture performed on cylindrical samples, 50 mm in diameter
of cohesive links. Moreover, the shape of the hydrostatic and 105 mm high, instrumented in the same manner as
curve (stiffness reduction and the subsequent stiffening) for the previously described concrete specimens (see Figs. 9
is explained by the porosity closure, which is confirmed and 10). A reduced sample size was preferred in order to
by the X-ray tomography images (Poinard et al., 2012). reach high stress levels and avoid technical problems with
Thus, under high confinement, the cement matrix losses use of the triaxial press. Opting for smaller samples is fully
both its cohesion and its porosity. Consequently, the initial justified when taking into account the very small size of
cohesion, directly related to the initial porosity of concrete, heterogeneities in these materials.
has no more influence on the concrete behavior (Vu et al., Fig. 10 presents the axial behavior of limestone and sili-
2009b). On the other hand, when the aggregates are much ceous rocks tested at 50 MPa of confinement. Both rock
stronger than the matrix, they do not break and the behav- types exhibit very different triaxial behavior. A much
ior of concrete is likely being controlled by the compacted higher strength and stiffness were found for the siliceous
cement matrix. Due to the presence of this destroyed rock compared with limestone. Siliceous rock undergoes
cement matrix between the aggregates, their shape has no brittle failure after the peak stress, at nearly 1000 MPa.
effect on the behavior of this granular assembly. On the contrary, limestone rock displays ductile behavior
with a stress plateau at 250 MPa: this behavior is similar
3.2. Siliceous vs. limestone aggregates: the influence of to that of cement-matrix composites, which are character-
composition ized by both an increase in loading capacity with confine-
ment and a transition from quasi-brittle to ductile
This section pertains to the effect of coarse aggregate behavior. Moreover, at th same confining pressure, the
composition on triaxial behavior of concrete. The behaviors deviatoric strength of limestone rock is comparable to that
of concretes with crushed siliceous aggregates (SC) and of the cement matrix (Dupray, 2008).
E. Piotrowska et al. / Mechanics of Materials 79 (2014) 45–57 51

3.2.1. Unconfined compression test


Fig. 11 shows the results of uniaxial compression tests
performed on concretes with siliceous aggregates and
limestone aggregates. Two tests carried out on limestone
aggregate concrete have provided consistent results. The
Young’s modulus and Poisson’s ratio evaluated from the
linear part of the axial behavior are respectively 22 GPa
and 0.13; both these values are less than those correspond-
ing to the siliceous aggregate concrete (see Table 5). The
Young’s modulus values of concrete remain consistent
with those of aggregates making up 40% of the concrete
volume. The uniaxial strength, which lies between 55
and 59 MPa, is significantly higher than the corresponding
value for siliceous aggregate concrete. Moreover, cracks
pass through the limestone aggregates, whereas they tend
to circumvent the siliceous aggregates (see Fig. 12). These
observations confirm the importance of the interfacial
transition zone in the uniaxial behavior of concrete. The
stronger interfacial bond derived through the use of lime-
stone aggregates is explained by both high porosity and
absorption. In addition, some chemical bonding may take
place between the limestone and the cement matrix due
to a reaction between calcite in the aggregate and calcium
hydroxide in the hydrated cement (Monteiro and Mehta,
1986).

3.2.2. Triaxial compression at 50 MPa of confinement


Fig. 13 provides the results of triaxial testing conducted
at 50 MPa of confinement on siliceous aggregate concrete
(SC) and limestone aggregate concrete (LC). The elastic vol-
umetric stiffness is slightly higher for limestone aggregate
concrete, which is consistent with the elastic properties of
Fig. 8. Sketch of the failure patterns in a concrete sample at 650 MPa of aggregates.
confinement for rolled aggregate concrete (SR) and crushed aggregate The stress–strain curve of the limestone aggregate con-
concrete (SC).
crete displayed in Fig. 13(b) exhibits a plateau, whereas a
peak stress occurs for the siliceous aggregate concrete.
The maximum deviatoric stress reached by the siliceous
Table 4
Physical and mechanical properties of coarse aggregates. aggregate concrete is about 11% higher than the maximum
stress in limestone aggregate concrete. This finding is
Aggregate type Siliceous Limestone Glass
attributed to the composition of aggregates. Due to the
Density [g/cm3] 2.6 2.3 2.5 strong interfacial bond and low aggregate strength in the
Young’s modulus [GPa] 78 51 70
LC sample, a large quantity of the aggregates is broken
Poisson’s ratio [–] 0.12 0.28 0.22
Compressive strength [MPa] 330 150 agg. strength
under shear loading; nonetheless, the decrease in strength
Porosity [%] <2 9.4 0 due to the use of soft limestone aggregates remains quite
limited.

160 400
140
120 300
100
σ (MPa)

σ (MPa)

80 200
x

60
40 100
20
0 0
−0.2 −0.1 0 0.1 0.2 0.3 0.4 −0.2 0 0.2 0.4 0.6
ε (%) ε (%)
θ,x θ,x

(a) (b)
Fig. 9. Unconfined compression test on (a) limestone rock and (b) siliceous rock: axial stress rx vs. strain components eh and ex .
52 E. Piotrowska et al. / Mechanics of Materials 79 (2014) 45–57

250 1000

200 800

150 600

σ (MPa)

σx (MPa)
100 400
x

50 200

0 0
−4 −2 0 2 4 6 8 10 −0.5 0 0.5 1 1.5
ε (%) ε (%)
θ,x θ,x

(a) (b)
Fig. 10. Triaxial test at 50 MPa of confinement on (a) limestone rock and (b) siliceous rock: axial stress rx vs. strain components eh and ex .

20 60
SC
LC−1
50 LC−2
15
40
σ (MPa)

σx (MPa)
10 30
m

20
5
SC 10
LC−1
LC−2
0 0
−0.05 0 0.05 0.1 0.15 −0.4 −0.2 0 0.2 0.4 0.6 0.8
ε (%) ε (%)
v θ,x

(a) (b)
Fig. 11. Simple compression test on concretes with siliceous aggregates (SC) and limestone aggregates (LC): (a) mean stress rm vs. volumetric strain ev , (b)
axial stress rx vs. strain components eh and ex .

Table 5 additional large plastic deformations due to their own


Unconfined compression test: identification of the main concrete porosity closure.
characteristics. During the deviatoric phase, the behavior of siliceous
Concrete LC SC aggregate concrete is considerably stiffer. Nevertheless,
Young’s modulus [GPa] 22 33 both concretes reach the same stress level. Aggregate
Poisson’s ratio [–] 0.13 0.16 strength does not affect the maximum deviatoric stress
Compressive strength [MPa] 57 43 at this level of confinement.

3.2.4. Triaxial compression at 650 MPa of confinement


3.2.3. Triaxial compression at 200 MPa of confinement The results of triaxial compression conducted at
Due to a lack of results for siliceous crushed aggregate 650 MPa of confinement on concretes with siliceous
concrete at 200 MPa of confinement, the limestone aggre- aggregates (SC) and limestone aggregates (LC) are
gate concrete (LC) will be compared with siliceous rolled shown in Fig. 15. The limestone aggregate concrete was
aggregate concrete (SR). It was showed in the previus sec- first loaded hydrostatically up to a pressure of 650 MPa.
tion that aggregate shape has no more influence on con- Due to a technical problem, the subsequent triaxial part,
crete behavior at very high confinement. Fig. 14 shows i.e. after unloading and hydrostatic reloading, was per-
the results of triaxial testing performed at 200 MPa of con- formed at 600 MPa of confining pressure. The volumetric
finement on both samples. In agreement with test results at strain is much higher for limestone aggregate concrete, a
50 MPa, volumetric stiffness is slightly higher for limestone finding that may be explained by the high porosity of
aggregate concrete at the beginning of the loading step. the rock.
Beyond a mean stress of 50 MPa, a reduction in stiffness Fig. 15(b) shows that tangential stiffness during the
can be observed and thereafter the volumetric strain deviatoric phase is nearly the same for both samples.
becomes higher for LC. This stiffness reduction is attributed A large proportion of the limestone porosity was probably
to a gradual damage and compaction process (porosity closed during the hydrostatic part of loading, which would
closure) of the cement matrix in both concretes. The stiffen the limestone aggregates to a greater extent than for
reduction is even more pronounced in LC since at this the 200-MPa test. To compare the maximum deviatoric
level of mean stress, soft limestone aggregates undergo stresses which are highly dependent on confining pressure,
E. Piotrowska et al. / Mechanics of Materials 79 (2014) 45–57 53

deviatoric stress reached by the limestone aggregate


concrete (730 MPa).
Aggregates composition strongly influences the failure
pattern in concrete samples at high confinement (see
Fig. 16). In limestone aggregate concrete, a single crack
passing through both mortar and aggregates appears,
while few horizontal cracks along with predominantly
debonded aggregates are observed in siliceous aggregate
concrete specimens.

3.3. Siliceous gravel vs. glass balls

The previously introduced concrete with rolled sili-


ceous aggregates (SR) will be compared herein with glass
ball concrete (GB) with emphasis placed on the effect of
aggregate external features as well as on the effect of their
mechanical properties.
The shape of these two types of aggregates, both
rounded, may be considered as comparable. However, the
(a) (b)
surface texture is much smother for glass beads. On the
Fig. 12. Sketches of failure patterns in concrete samples at an unconfined other hand, glass exhibits very attractive mechanical prop-
compression test: (a) siliceous aggregate concrete (SC, SR), (b) limestone erties, although they vary widely depending on mix design,
aggregate concrete (LC). production conditions, etc. Despite the very high theoreti-
cal compressive strength of glass, i.e. of the order of
104 MPa (Schott technical glasses physical, 2010), this
we analyzed a test conducted on rolled siliceous aggregate value cannot be used in practice since glass surface defects
concrete at a 600-MPa confining pressure: the resulting cause stress concentrations that generate widely varying
deviatoric stress of 830 MPa exceeds the maximum failure stresses. It is therefore considered herein that the

120 200

100
150
80
σ (MPa)

q (MPa)

60 100
m

40
50
20 SC SC
LC LC
0 0
0 0.5 1 1.5 2 2.5 −4 −2 0 2 4 6 8 10
ε (%) ε’ (%)
v θ,x

(a) (b)
Fig. 13. Triaxial test at 50 MPa of confinement on concretes with siliceous aggregates (SC) and limestone aggregates (LC): (a) mean stress rm vs. volumetric
strain ev , (b) stress deviator q vs. strain components e0h and e0x .

400 500

400
300
σ (MPa)

300
q (MPa)

200
m

200

100
100
SR SR
LC LC
0 0
0 2 4 6 8 10 −15 −10 −5 0 5 10 15 20 25
ε (%) ε’ (%)
v θ,x

(a) (b)
Fig. 14. Triaxial test at 200 MPa of confinement on concrete with siliceous aggregates (SR) and limestone aggregates (LC): (a) mean stress rm vs. volumetric
strain ev , (b) stress deviator q vs. strain components e0h and e0x .
54 E. Piotrowska et al. / Mechanics of Materials 79 (2014) 45–57

1000 1000

800 800

600

σ (MPa)
600

q (MPa)
400
m
400

200 200
SC SC
LC LC
0 0
0 2 4 6 8 10 12 14 16 18 −10 −5 0 5 10 15
εv (%) ε’θ,x (%)

(a) (b)
Fig. 15. Triaxial test at 650 MPa of confinement on concrete with siliceous aggregates (SC) and limestone aggregates (LC): (a) mean stress rm vs. volumetric
strain ev , (b) stress deviator q vs. strain components e0h and e0x .

3.3.2. Triaxial compression at 50 MPa of confinement


Fig. 18 shows the results of triaxial tests conducted at
50 MPa of confinement on concretes samples with rolled
siliceous aggregates and glass beads. The volumetric stiff-
ness of glass ball concrete is considerably greater than that
of rolled aggregate concrete. This result is attributed to the
elastic properties of aggregates and glass balls.
Fig. 18(b) presents the deviatoric behavior of concrete
with rolled siliceous aggregates (SR) and glass balls (GB).
The deviatoric strength of glass ball concrete is 24% higher
than that of siliceous aggregate concrete. In addition, glass
ball concrete exhibits stiffer behavior and a more pro-
nounced peak stress. This ultimate stress difference is
probably explained by the aggregate strength, which is
considerably higher for glass balls than for siliceous aggre-
gates. This strength difference protects glass ball aggre-
gates from shear failure and moreover leads to a crack
pattern completely different for glass ball concrete sample
than that observed for siliceous aggregate concrete (see
(a) (b) Fig. 19). Several cracks perpendicular to the axial direction
Fig. 16. Sketches of failure patterns in concrete samples at high confine- can be observed in the glass ball concrete sample; they are
ment (P200 MPa): (a) siliceous aggregate concrete (SC, SR), (b) limestone characterized by a glass ball debonding phenomenon.
aggregate concrete (LC).

3.3.3. Triaxial compression at 200 MPa of confinement


compressive strength of glass balls highly exceeds that of Fig. 20 displays the results of triaxial tests performed at
siliceous aggregates (330 MPa). 200 MPa of confinement on both siliceous aggregate con-
crete (SR) and glass ball concrete (GB). The volumetric stiff-
ness of glass ball concrete is greater than that of siliceous
3.3.1. Unconfined compression test aggregate concrete, an outcome due to the elastic proper-
Fig. 17 shows the results of unconfined compression ties of coarse aggregates, which make up some 40% of con-
tests performed on the two studied concretes. The corre- crete volume. The Poisson’s ratio m of glass is much higher
sponding elastic properties and strengths are listed in than that of siliceous rock, while their Young’s modulus
Table 6. The two concretes display very close elastic prop- values E are comparable. These two characteristics control
erties but considerable differences in the ultimate stress the elastic bulk modulus of the material K, such that:
are observed.
Glass ball concrete is less strong than concrete made E
with siliceous gravel. Very smooth spherical aggregates K¼ ð4Þ
3ð1  2mÞ
produce weaker interfacial bonds, which results in a lower
glass ball concrete strength. The weakness of the interface Fig. 20(b) reveals that the maximum deviatoric stress is
in glass ball concrete is confirmed by its tensile strength, as exactly the same for rolled aggregate concrete and glass
evaluated from the splitting tensile test at 2.7 MPa. This ball concrete. The tangential stiffness is nearly the same
value is notably less than the direct tensile strength of for both samples. At this level of confinement, the cement
rolled aggregate concrete, which equals 3.7 MPa. matrix loses its cohesive nature. The test has revealed that
E. Piotrowska et al. / Mechanics of Materials 79 (2014) 45–57 55

14 40 SR
SR GB
GB 35
12

10 30
25

σ (MPa)
σm (MPa)
8
20
6

x
15
4 10
2 5

0 0
−0.04 −0.02 0 0.02 0.04 0.06 −0.4 −0.2 0 0.2 0.4 0.6
εv (%) ε (%)
θ,x

(a) (b)
Fig. 17. Unconfined compression tests on concretes with siliceous rolled aggregates (SR) and glass ball concrete (GB): (a) mean stress rm vs. volumetric
strain ev , (b) axial stress rx vs. strain components eh and ex .

Table 6
Unconfined compression test: identification of the main concrete
characteristics.

Concrete SR GB
Young’s modulus, E [GPa] 33 32
Poisson’s ratio, m [–] 0.16 0.19
Compressive strength, rmax [MPa] 40 33

the greater strength of coarse aggregates does not increase


the deviatoric strength of concrete when the cement
matrix is heavily damaged.

3.3.4. Triaxial compression at 650 MPa of confinement


Fig. 21 presents the results of triaxial tests performed at
650 MPa of confinement on both siliceous aggregate con-
crete (SR) and glass ball concrete (GB). The volumetric
behavior displayed in Fig. 21(a) suggests the slightly (a) (b)
lower deformability of glass ball concrete. However, the
Fig. 19. Sketches of failure patterns in concrete samples at 50 MPa of
difference in volumetric strain is significant only at low
confinement: (a) siliceous aggregate concrete (SR, SC), (b) glass ball
confinement levels. Beyond 200 MPa of confinement, the concrete (GB).
tangential volumetric stiffness of the two test concretes
are nearly identical. For such a confinement range, the dif- The deviatoric behavior presented in Fig. 21(b) confirms
ferences in aggregate elastic properties are probably com- the observation made at 200 MPa of confinement:
pensated by compaction of the cement matrix. when the cement matrix is strongly damaged, the higher

250
140 SR
GB
120 200
100
150
q (MPa)
σ (MPa)

80

60 100
m

40
50
20 SR
GB
0 0
0 0.5 1 1.5 2 −4 −2 0 2 4 6 8 10
ε (%) ε’ (%)
v θ,x

(a) (b)
Fig. 18. Triaxial test at 50 MPa of confinement on concretes with siliceous rolled aggregates (SR) and glass ball concrete (GB): (a) mean stress rm vs.
volumetric strain ev , (b) stress deviator q vs. strain components e0h and e0x .
56 E. Piotrowska et al. / Mechanics of Materials 79 (2014) 45–57

350 500

300
400
250
300

σ (MPa)

q (MPa)
200

150 200
m
100
100
50 SR SR
GB GB
0 0
0 1 2 3 4 5 6 7 −4 −2 0 2 4 6 8 10 12 14
ε (%) ε’ (%)
v θ,x

(a) (b)
Fig. 20. Triaxial test at 200 MPa of confinement on concrete with siliceous rolled aggregates (SR) and glass ball concrete (GB): (a) mean stress rm vs.
volumetric strain ev , (b) stress deviator q vs. strain components e0h and e0x .

1000
900
800 800
700
600
σm (MPa)

600

q (MPa)
500
400 400
300
200 200
SR 100 SR
GB GB
0 0
0 2 4 6 8 10 12 −4 −2 0 2 4 6 8 10
ε (%) ε’ (%)
v θ,x

(a) (b)
Fig. 21. Triaxial test at 650 MPa of confinement on concrete with siliceous rolled aggregates (SR) and glass ball concrete (GB): (a) mean stress rm vs.
volumetric strain ev , (b) stress deviator q vs. strain components e0h and e0x .

1000 250

800
200
σm (MPa)

600
150
q (MPa)

400
100
SR
200 SC SR
LC 50 SC
GB LC
0 GB
0 5 10 15 0
ε (%) 0 50 100 150
v
σ (MPa)
m
Fig. 22. Volumetric behavior of concretes SR, SC, LC and GB: mean stress
rm vs. volumetric strain e0x . Fig. 24. Limit states of concretes SR, SC, LC and GB: stress q vs. mean
stress rm .
1000

800
strength of coarse aggregates does not increase the devia-
toric strength of concrete.
600
q (MPa)

400 3.4. Limit states


SR
SC
200 LC
This section provides a summary of tests performed on
GB the four concretes. Fig. 22 points up the influence of aggre-
Poly. (SR)
0 gate properties on the concrete compaction. Figs. 23 and
0 200 400 600 800 1000
σm (MPa) 24 plot the failure envelopes of concretes. This failure
surface, called strain limit states, corresponds to the stress
Fig. 23. Limit states of concretes SR, SC, LC and GB: stress q vs. mean stress rm . state associated with the transition point from contractant
E. Piotrowska et al. / Mechanics of Materials 79 (2014) 45–57 57

to dilatant behavior of the material demonstrated in de Larrard, F., Belloc, A., 1997. The influence of aggregate on the
compressive strength of normal and high-strength concrete. ACI
Fig. 22. At low confinement (Fig. 24), the strain limit state
Mater. J. 94 (5), 417–426.
practically coincides with the maximum stress that the Dupray, F., 2008. Comportement du Béton Sous for Confinement: Étude
material can support. The evolution of concrete strength en Compression et en Extension Triaxiales à L’échelle Mésoscopique
with the increase of the confinement is observed for all (Ph.D. thesis). Université Joseph Fourier, Grenoble, France.
Dupray, F., Malecot, Y., Daudeville, L., Buzaud, E., 2009. A mesoscopic
tested concretes but it varies in magnitude depending on model for the behaviour of concrete under high confinement. Int. J.
the coarse aggregate type (Fig. 23). Numer. Anal. Methods. Geomech. 33 (11), 1407–1423.
Forquin, P., Safa, K., Gary, G., 2010. Influence of free water on the quasi-
static and dynamic strength of concrete in confined compression
4. Conclusion tests. Cem. Concr. Res. 40 (2), 321–333.
Gabet, T., Vu, X., Malecot, Y., Daudeville, L., 2006. A new experimental
technique for the analysis of concrete under high triaxial loading. In:
This experimental investigation examined the effects of Journal of Physics IV (Proceedings), vol. 134, pp. 635–640.
coarse aggregates on concrete behavior under high triaxial Gabet, T., Malecot, Y., Daudeville, L., 2008. Triaxial behaviour of concrete
under high stresses: influence of the loading path on compaction and
loading. Triaxial tests were performed at confining
limit states. Cem. Concr. Res. 38 (3), 403–412.
pressure levels between 0 and 650 MPa on four concretes Giaccio, G., Zerbino, R., 1998. Failure mechanism of concrete: combined
distinguished only by the type of coarse aggregate. The effects of coarse aggregates and strength level. Adv. Cem. Based
influence of aggregate shape was first evaluated on two Mater. 7 (2), 41–48.
Kaplan, M., 1959. Flexural and compressive strength of concrete as
of the concretes, containing rolled and crushed aggregates. affected by the properties of coarse aggregates. In: Proceedings of the
Next, concretes with siliceous aggregates and limestone ACI, vol. 55, pp. 1193–1208.
aggregates were introduced to study the influence of Kupfer, H., Gerstle, K., 1973. Behavior of concrete under biaxial stresses. J.
Eng. Mech. Div. 99 (4), 853–866.
aggregate composition. Finally, concrete with siliceous Linhua, J., Dahai, H., Nianxiang, X., 1991. Behavior of concrete under
rolled aggregates was compared with glass ball concrete triaxial compressive–compressive–tensile stresses. ACI Mater. J. 88
by taking into account overall aggregate properties. (2).
lmran, I., Pantazopoulou, S., 1996. Experimental study of plain concrete
The aggregate/mortar interface strength proved to be under triaxial stress. ACI Mater. J. 93 (6), 589–601.
the main factor governing the behavior of concrete in an Malecot, Y., Zingg, L., Piotrowska, E., Briffaut, M., Daudeville, L., 2014.
unconfined compression test. This phenomenon is well- Mesoscopic modeling of triaxial behavior of concrete. Comput. Model.
Concr. Struct., 199–206.
known for ordinary concrete, in which aggregate strength Monteiro, P., Mehta, P., 1986. Interaction between carbonate rock and
is much greater than mortar strength. Irregularly-shaped cement paste. Cem. Concr. Res. 16 (2), 127–134.
coarse aggregates slightly increase the strength generated Özturan, T., Çeçen, C., 1997. Effect of coarse aggregate type on mechanical
properties of concretes with different strengths. Cem. Concr. Res. 27
at the aggregate/mortar interface.
(2), 165–170.
At moderate confinement levels (50 MPa), the shear Poinard, C., Malecot, Y., Daudeville, L., 2010. Damage of concrete in a very
strength of concrete is mainly controlled by the aggregate high stress state: experimental investigation. Mater. Struct. 43, 15–
strength; however, along the same lines as what was 29.
Poinard, C., Piotrowska, E., Malécot, Y., Daudeville, L., Landis, E.N., 2012.
observed for unconfined compression, rough angular Compression triaxial behavior of concrete: the role of the
aggregates enhance the concrete strength. mesostructure by analysis of X-ray tomographic images. Eur. J.
At high confinement (P200 MPa), the shear strength of Environ. Civ. En. 16 (sup1), 115–136.
Pugh, H., 1970. Mechanical Behaviour of Materials Under Pressure.
concrete seems to be controlled by the compacted cement Elsevier Pub. Co.
matrix provided aggregate strength is sufficiently greater Schmidt, M., Cazacu, O., Green, M., 2009. Experimental and theoretical
than the mortar strength. Otherwise, a lower-strength investigation of the high-pressure behavior of concrete. Int. J. Numer.
Anal. Methods Geomech. 33 (1), 1–23.
aggregate reduces overall concrete strength at very high Schott technical glasses physical and technical properties, 2010. Tech.
confinement (600 MPa). The coarse aggregate shape does rep., SCHOTT.
not therefore seem to exert any influence under high Sfer, D., Carol, I., Gettu, R., Etse, G., 2002. Study of the behavior of concrete
under triaxial compression. J. Eng. Mech. 128 (2), 156–163.
confinement. Vu, X., Malecot, Y., Daudeville, L., Buzaud, E., 2009a. Experimental analysis
of concrete behavior under high confinement: effect of the saturation
ratio. Int. J. Solids. Struct. 46 (5), 1105–1120.
Acknowledgements
Vu, X., Malecot, Y., Daudeville, L., Buzaud, E., 2009b. Effect of the water/
cement ratio on concrete behavior under extreme loading. Int. J.
The GIGA press has been installed in the 3SR Laboratory Numer. Anal. Methods Geomech. 33 (17), 1867–1888.
Vu, X., Malecot, Y., Daudeville, L., 2009c. Strain measurements on porous
within the scope of a cooperative agreement with the
concrete samples for triaxial compression and extension tests under
CEA-Gramat. We would also like to thank Dr. Eric Buzaud very high confinement. J. Strain. Anal. Eng. Des. 44 (8), 633–657.
(CEA-Gramat) and Dr. Christophe Pontiroli (CEA-Gramat) Vu, X., Daudeville, L., Malecot, Y., 2011. Effect of coarse aggregate size and
for their sound technical and scientific advice. cement paste volume on concrete behavior under high triaxial
compression loading. Constr. Build. Mater. 25 (10), 3941–3949.
Warren, T., Fossum, A., Frew, D., 2004. Experimental investigation of size
References effect in concrete fracture under multiaxial compression into low-
strength (23 MPa) concrete: target characterization and simulations.
Int. J. Impact Eng. 30, 477–503.
Bazant, Z., Bishop, F., Chang, T., 1986. Confined compression tests of
William, E., Akers, S., Reed, P., 2005. Constitutive models for the triaxial
cement paste and concrete up to 300 ksi. In: ACI Journal Proceedings,
behaviour of concrete. Report no. erdc, Tech. rep., GSL TR-05-16,
vol. 83, pp. 553–560.
Geotechnical and Structures Laboratory, US Army.
Burlion, N., Pijaudier-Cabot, G., Dahan, N., 2001. Experimental analysis of
Zukas, J., 1992. Penetration and perforation of solids. Impact Dynamics.
compaction of concrete and mortar. Int. J. Numer. Anal. Methods
Krieger Publishing Co.
Geomech. 25 (15), 1467–1486.

Das könnte Ihnen auch gefallen