Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/318298141
CITATIONS READS
0 135
5 authors, including:
Some of the authors of this publication are also working on these related projects:
Performance of autoclaved areated concrete masonry elements and strengthening methods under
blast loads View project
Integrated Health Monitoring Systems for Infrastructure Structures in Operational Environments View
project
All content following this page was uploaded by Jian Cui on 11 July 2017.
A R T I C L E I N F O A B S T R A C T
Keywords: The objective of this study is to characterize the damage evolution behavior of concrete under hydrostatic
Concrete pressures varying from 30 MPa to 500 MPa for better understanding the concrete material properties at complex
Hydrostatic pressure stress states. A series of uniaxial tests were carried out to evaluate the damage degree of concrete after hy-
Damage drostatic tests and a number of microscopic observations were also provided to allow visualizing the changes of
True tri-axial test
microstructures of the specimen after and before hydrostatic tests. The eect of stress path on compressive
Stress path
meridian and tensile meridian of concrete was also studied. Experimental results indicate that concrete suers
obvious damage if the applied hydrostatic pressure is higher than the uniaxial compressive strength of concrete
specimen. The strength and Young's modulus of concrete decrease signicantly after hydrostatic tests and the
aggregate-mortar interfacial transition zone (ITZ) is the most obvious damage region. The stress path has in-
signicant eects on ultimate strength envelopes. But if the stress path involves a high hydrostatic pressure that
damages the concrete specimens upon unloading, the strength envelope shrinks because of the damage to
concrete due to high hydrostatic pressure.
1. Introduction materials will not suer any damage in whatever high hydrostatic
pressure. The commonly used concrete material models in hydrocodes
Concrete is a composite material consisting of dierent sized ag- such as K & C model [8] and RHT model [9] in LS-DYNA [10] therefore
gregates which are embedded in a cement paste matrix, and this com- do not consider the damage of concrete material in hydrostatic pressure
position makes it a heterogeneous material. The cement paste has a although concrete is not exactly a homogeneous material. However, the
high porosity, and these pores are lled with air and water. A large damage of concrete under high hydrostatic pressure inuences the
number of bond micro-cracks exist at the interfaces between coarse failure surface, damage evolution algorithm and equation of state (EOS)
aggregates and cement matrix. These properties strongly aect the of the concrete constitutive model under complex stress states [11].
behaviors of concrete under complex stress states. Gebbeken and Ruppert [12] also suggested considering this damage in
Concrete is the most commonly used construction material. Under dynamic concrete model as concrete is a highly inhomogeneous porous
extreme loading conditions such as near-eld explosions and projectile material. In the experimental study by Poinard et al. [13], it was ob-
penetrations, the material subjects to complex stress states. For ex- served that the bulk modulus of the concrete with a 29 MPa uniaxial
ample, when a projectile impacts a concrete structure, the stress state compressive strength decreased substantially after the specimen having
induced in the structure is very complex owing to wave propagation, been subjected to a hydrostatic pressure higher than 60 MPa. The au-
reection and refraction. The damage to concrete in complex stress thors attributed this drop to cement matrix damage. Pham et al. [14]
states, e.g. in high hydrostatic pressure, is still not clear. Most experi- found that in their FRP-conned concrete tests, the core concrete has
mental results available in the literature only address the damage and already suered serious damage although the strength of FRP-conned
destruction of concrete material under deviatoric stress [17], usually concrete still could signicantly increase a lot. The study on the eect
obtained with a cylindrical specimen subjected to a conning pressure of stress path on the ultimate strength envelopes of concrete is also very
and an axial loading. The study of concrete under hydrostatic pressure limited. Kotsovos [15] designed three dierent stress paths to get the
is limited owing to the diculty in applying the very high true tri-axial ultimate strength envelopes and found the ultimate strength envelopes
loads in tests. As it is well known that in theory the homogeneous were independent of the stress paths. Gerstle et al. [16] reported a
Corresponding author.
E-mail address: hong.hao@curtin.edu.au (H. Hao).
http://dx.doi.org/10.1016/j.cemconres.2017.06.005
Received 24 January 2017; Received in revised form 15 May 2017; Accepted 29 June 2017
0008-8846/ 2017 Elsevier Ltd. All rights reserved.
J. Cui et al. Cement and Concrete Research 100 (2017) 140152
similar observation. Gabet et al. [17] designed a pseudo tri-axial test by (50 mm) to avoid collision of the load transfer bars in dierent direc-
varying the conning pressure proportionally with the applied axial tions when the specimen experiences a large strain during the loading
stress during the loading process. The latter study also revealed that the process, as illustrated in Fig. 2. The axial loads were recorded by the
strength envelopes of concrete were independent of the loading paths. load cells, and the deformation of the specimen was measured by LVDT
These studies all revealed that concrete material behaves dierently sensors. The elastic deformation of the load transfer bar was removed
under complex stress state as compared to the uniaxial stress state, and from the measured deformation in the subsequent data analyses to
may be subjected to damage even under hydrostatic stress condition. obtain the strain of the tested specimen, as shown in Fig. 2.
Unfortunately the relevant studies, especially the true tri-axial tests, are
limited in literature owing to the constraints of available testing 2.2. Specimens
equipment.
In the present study, more comprehensive experimental tests are Two concrete mixes, i.e., LSC (low strength concrete) and HSC
carried out to further evaluate the performance and damage of concrete (High strength concrete) were considered in the study. The concrete
material after experiencing hydrostatic pressures. The uniaxial com- mix properties are given in Table 1. Natural river sand with a neness
pressive strength and the uniaxial tensile strength of concrete speci- modulus of 2.9 was used as ne aggregate. Cleaned gravels with a
mens after experiencing certain level of hydrostatic pressure are com- maximum size of 10 mm were used as course aggregate. The specimens
pared with the ones without experiencing hydrostatic pressure. were cast in a 50 mm cubic metal molds and removed from the mold
Electron microscope observations and ultrasonic examinations are also 24 h after casting. The specimens were moist-cured at 20 2 C and
made to study the damage mechanism and damage degree of concrete 100% relative humidity for 28 days. Before the test, the surfaces of
subjected to high hydrostatic pressure. The inuence of stress paths on specimen were smoothed by a polisher and coated with grease to re-
ultimate strength envelopes is also examined. Compressive meridian duce the friction between the specimen and the rigid loading platens.
and tensile meridian are determined by loading and unloading methods For the uniaxial compressive test, a 500 kN computer-controlled elec-
respectively. Based on the testing data the modied material model is tromechanical servo hydraulic pressure testing machine was used to
put forward to consider the damage to concrete material under high conduct the test. The loading rate was 0.03 mm/min and the de-
hydrostatic pressure. The results obtained in this study lead to better formation was recorded from LVDTs. The 28-day uniaxial compressive
understanding of the micro-damages to concrete material by hydro- strength of LSC and HSC were 33.6 and 61.6 MPa, respectively. How-
static pressure and its inuences on strength degradation, and hence ever, at the time of testing concrete was approximately half year old,
development of more accurate strength envelope and equation of state and the corresponding uniaxial compressive strengths were increased to
of concrete materials. 35.2 and 63.8 MPa. The test set-up for the uniaxial tensile test is de-
tailed in Section 3.2.2. It must be noted that the end friction inevitably
2. Experimental setup and specimens existed and might aect the testing results although methods have
taken to reduce it. To investigate the possible friction constraint, the
2.1. Equipment static friction coecients were measured between the greased spe-
cimen and steel surface. The average friction coecient was found to be
The experiments were conducted by a true tri-axial hydraulic servo- 0.105, which is relatively small. As is discussed later end friction only
controlled test system in Central South University in China [18,19]. The had minimum inuences on the tri-axial testing results, and its inu-
system consists of the main machine, servo-controller, data acquisition ence was limited to a small layer near the specimen surface.
system and measurement sensors, as shown in Fig. 1. It is a digital
closed-loop control system, which can control load and displacement 3. Concrete damage in high hydrostatic pressure
automatically.
This apparatus applies quasi-static loads along the three principal 3.1. Hydrostatic test procedure
stress directions through hydraulically driven pistons independently.
The maximum load capacities of the X, Y, and Z actuator are 2000, A hydrostatic load with one loading-unloading cycle was conducted
2000 and 3000 kN respectively. High-strength steel (20CrNiMo), which during the test, as is shown in Fig. 3. To ensure 1 = 2 = 3 (1, 2,
has a yield strength of 785 MPa and an elastic modulus of 210 GPa, is and 3 are major, intermediate, and minor principal stresses, respec-
used to transfer the applied loads from the actuators to the specimen as tively) during the loading-unloading process, the forces of X, Y and Z
shown in Figs. 1 and 2. The side length of the steel load transfer block is axes were applied by the force control mode at a rate of 1 kN/s
47 mm, 3 mm shorter than the side length of the cubic specimen (0.4 MPa/s) until reaching the desired stress level. Before unloading,
141
J. Cui et al. Cement and Concrete Research 100 (2017) 140152
Fig. 2. The test set-up: (a) overall view; (b) 2D section view.
Table 1
Mix proportions of the studied concrete.
500
LSC
400 HSC
Pressure(MPa)
Concrete type LSC HSC
800 remove the deformation of the block for deriving the accurate specimen
600 deformations.
Fig. 4 gives the hydrostatic pressure-volumetric strain curve of LSC
400 and HSC. The concrete starts to have a plastic deformation when the
200 pressure reaches about 90% of the uniaxial compressive strength. Then
pore crush and cement matrix damage may occur gradually which leads
0 to the reduction of the bulk modulus. Under 500 MPa pressure, the
0.0 0.2 0.4 0.6 0.8
volumetric strain of LSC is about 11% while that of HSC is about 9.5%
Time(h)
because the LSC has a larger porosity associated to a higher water/
Fig. 3. An example of the loading-unloading process. cement ratio and a lower bulk modulus than HSC. Note there is very
strong nonlinearity at the completion of unloading, and the residual
the desired stress level was maintained for about 6 min. From Fig. 3, volumetric strain of concrete is less than 3%. Poinard, et al. [13] also
one can nd that during the loading-unloading process, the forces of X, observed the same trend and attributed this sudden decrease in tangent
Y and Z axes were kept the same. Therefore there was no deviatoric modulus to cement matrix damage when the granular skeleton, which
stress produced during the whole test process. remained elastic, recovered its initial shape.
Dierent levels of hydrostatic pressures were applied in the tests, as
indicated in Table 2, to investigate the damage of the specimens after
3.2. Damage study of concrete after hydrostatic test
experiencing dierent levels of hydrostatic pressures. The minimum
hydrostatic pressure considered was equal to the uniaxial compressive
3.2.1. Uniaxial compressive strength of the specimen after hydrostatic test
Uniaxial compressive tests were carried out to evaluate the residual
Table 2 compressive strength and the Young's modulus of the tested specimens.
Loading scheme of hydrostatic pressures. Compression tests were conducted under the displacement control
Group 1 2 3 4 5
mode with a constant rate of 0.03 mm/min. Grease was applied at the
end of the specimen to reduce the inuence of end friction connement.
Pressure 35 MPa (fc) 70 MPa (2fc) 175 MPa (5fc) 350 MPa 500 MPa Table 3 gives residual uniaxial compressive strengths of the LSC spe-
of (10fc) (14fc) cimens after the hydrostatic test while Table 4 gives those of HSC
LSC
(MP-
specimens. From Tables 3 and 4, it can be concluded that the residual
a) compressive strength of the specimen decreases as the preloaded hy-
Pressure 64 MPa (fc) 128 MPa (2fc) 320 MPa (5fc) 500 MPa drostatic pressure increases although the data are scattered. As can be
of (8fc) noted, for each case, four specimens were tested to obtain the averaged
HSC
results to minimize the uncertainties. It should be noted that several
(MP-
a) data are missing either because of the malfunctioning of the data sto-
rage program during the tests or because of the intensive damage to the
142
J. Cui et al. Cement and Concrete Research 100 (2017) 140152
Table 3
Residual uniaxial compressive strength of the LSC specimen after the hydrostatic tests.
Residual strength of specimen no.1 (MPa) 32.3 32.7 21.5 19.4 18.4
Residual strength of specimen no.2 (MPa) 29.6 31.4 31.9 18.8 16.8
Residual strength of specimen no.3 (MPa) 36.4 32.5 25.2 25.1
Residual strength of specimen no.4 (MPa) 40.1 33.9 33.4 19.3
Average (MPa) 34.6 32.6 28.0 20.7 17.6
Table 4
Residual uniaxial compressive strength of the HSC specimen after the hydrostatic tests.
Residual strength of specimen no.1 (MPa) 58.2 46.2 45.4 23.8 Fig. 6. Eect of end friction force (2D section).
Residual strength of specimen no.2 (MPa) 66.1 43.3 27.8 26.0
Residual strength of specimen no.3 (MPa) 49.2 42.6 37.5
Residual strength of specimen no.4 (MPa) 51.5 37.8 eect on the testing data. The detailed results are not shown here for
Average (MPa) 57.8 45.9 37.1 24.9 brevity. It was found that the end friction had insignicant eect on tri-
axial tests. Its inuences were limited to a small layer near the specimen
surface, and could damage specimen near the surface. The Young's
specimen by the 500 MPa hydrostatic pressure. Some specimens easily modulus is low at low strain level is due to this damaged thin layer of
ruptured after removing the hydrostatic loads, which made the reading mortar. After this thin layer mortar was compacted, the Young's mod-
of the subsequent uniaxial tests unreliable. ulus of specimen increases and represents the true elastic modulus of
Fig. 5 shows typical stress-strain curves of the tested LSC specimens the specimen. For the tensile tests described in Section 3.2.2, this thin
and HSC specimens under the uniaxial compression. From the gure, it layer of mortar is removed because the grease on the surface of spe-
is clear that as the preloaded hydrostatic pressure increases, the re- cimen made it dicult to glue the specimen to the steel end cap for
sidual strength and Young's modulus of the concrete decrease, in- tensile tests, therefore about 0.51.0 mm thick was polished o from
dicating application of hydrostatic pressure has caused damage to the the specimen before tensile tests.
concrete specimens. On the other hand, the strain at the maximum
stress increases, indicating the loading-unloading process not only
3.2.2. Uniaxial tensile strength of the specimen after the hydrostatic test
softens the concrete material but also makes it more ductile. After ex-
Uniaxial tensile tests were also carried out to evaluate the inuence
periencing hydrostatic pressure, the axial strain of the cubic concrete
of damage caused by hydrostatic pressure on the tensile strength of the
specimen under uniaxial compression is larger at the same stress level,
tested specimens. The specimens were glued to end-caps with an epoxy
and this becomes more prominent with the increase in the applied
24 h prior to the test. Once the glue dried, the specimen and end-caps
hydrostatic pressure level. As shown, the strain at the maximum stress
were attached to the machine (500 kN MTS/647 Hydraulic Wedge
of the specimen experienced 500 MPa hydrostatic pressure is about two
Grip) through a exible chain to ensure there was no bending moment
times of the specimen without experiencing tri-axial test. This is be-
during the loading stage, as shown in Fig. 7. A deformation sensor was
cause the tri-axial loading-unloading process damages the cement ma-
set between the upper and lower end-caps to record the deformation of
trix, resulting in lower stiness and strength of the concrete although
the specimen. The tests were conducted under the displacement control
the elastic granular skeleton largely recovers its initial shape of the
mode at a constant rate of 0.015 mm/min. The test result was con-
specimen after removing the tri-axial load. It can also be noted that in
sidered valid only if the fracture location was near the middle region of
the low strain range in the compressive stress-strain curve, the spe-
the specimen, as shown in Fig. 8. More than four hydrostatic tested
cimen shows a relatively low modulus. However, the modulus increases
specimens were prepared in one group for uniaxial tensile tests because
quickly with strain. This characteristic is more obvious in the specimen
many tested specimens failed with the rupture section very close to the
experienced higher tri-axial loadings. This is because of the damage
specimen ends, which were therefore considered invalid. To save time
caused by shear stresses induced near the specimen surface owing to
and cost, if tests from two specimens in each group were valid, the test
end friction connement. Although specimen was polished and a thin
stopped. Therefore there are only two test data for each case, instead of
layer of grease applied to the specimen surface before the tests to re-
four in the compression tests.
duce friction eect, the end friction eect was inevitable as illustrated
Tables 5 and 6 give the residual uniaxial tensile strength of the LSC
in Fig. 6. A numerical model was established to analyze the end friction
specimen and HSC specimen after the hydrostatic tests. Fig. 9 shows the
Stress (MPa)
500MPa pressure
40
20
30
20
10
10
0 0
0.000 0.002 0.004 0.006 0.008 0.000 0.002 0.004 0.006 0.008
Strain Strain
(a) (b)
143
J. Cui et al. Cement and Concrete Research 100 (2017) 140152
Table 5
Residual uniaxial tensile strength of the LSC specimen after the hydrostatic tests.
Residual strength of specimen no.1 (MPa) 2.90 1.33 0.7 0.4 0.34
Residual strength of specimen no.2 (MPa) 2.31 1.89 0.48 0.49
Average (MPa) 2.61 1.61 0.59 0.45 0.34
Table 6
Residual uniaxial tensile strength of the HSC specimen after the hydrostatic tests.
144
J. Cui et al. Cement and Concrete Research 100 (2017) 140152
Stress(MPa)
2
1
1
0 0
0.0000 0.0001 0.0000 0.0001
Strain Strain
(a) (b)
0.8 HSC-Tension
produced inside the concrete, its tensile strength drops sharply. This also called the interfacial transition zone (ITZ) is considered to be the
also explains why the ratios of tensile strength degradation for both the weakest link inside the concrete and have a signicant inuence on the
HSC and LSC are similar while they are very dierent for the com- failure mode and the macro-mechanical properties of concrete [23,24].
pressive strength. For the tensile Young's modulus, the similar trend is This study also conrms this conclusion. From Figs. 13 and 14 one can
found. nd that most of the damaged areas are on the ITZ or in the cement
matrix near the ITZ. The micro-cracks between the cement matrix and
the course aggregates are very clear. There are almost no damaged
3.3. Microstructures of the concrete after hydrostatic tests
course aggregates observed in the LSC specimen, but for the HSC spe-
cimen some damaged aggregates can be found near the ITZ, as shown in
Electron microscope provides a direct observation of the damages of
Fig. 14(c). This is because the strength of course aggregate is much
the tested specimens, and hence helps to better understand the damage
higher than the strength of cement matrix for LSC, however, this
mechanism of concrete subjected to hydrostatic pressure. In this study,
strength dierence is less signicant for HSC. Therefore some ag-
for better electron microscope scanning, typical virgin specimens and
gregates in HSC were also damaged by the application of hydrostatic
the specimens after the application of 500 MPa hydrostatic pressure
pressures.
were sliced to several 20 mm 20 mm 12 mm blocks by a Wire
Saw. The corresponding micrographs of LSC and HSC are shown in
Figs. 10 and 11 respectively.
In the meso-scale analysis, the cement matrix/aggregate interface,
145
J. Cui et al. Cement and Concrete Research 100 (2017) 140152
Fig. 13. Electron microscope photos of LSC: (a) virgin concrete; (b), (c) concrete after application of 500 MPa hydrostatic pressure.
Fig. 14. Electron microscope photos of HSC: (a) virgin concrete; (b), (c) concrete after application of 500 MPa hydrostatic pressure.
evaluate the damage degree of concrete. It is also used in this study to Hydrostatic pressure Ultrasonic velocity - No.1 Ultrasonic velocity - No.2
evaluate the damage of the specimens before and after hydrostatic test. (MPa) (km/s) (km/s)
The development of micro-cracks inside the concrete specimen makes virgin/damaged virgin/damaged
the propagation velocity of ultrasonic wave decrease; therefore the
64 5.05/5.01 4.92/4.88
change of the ultrasonic velocity can reect the damage degree of the 128 5.12/3.40 4.95/3.52
concrete. The following Tables 7 and 8 give the comparisons of ultra- 320 4.96/3.10 5.05/2.97
sonic velocity in the virgin specimen and that in the specimen after the 500 4.88/2.01 5.10/2.16
hydrostatic test. From the two tables, it can be seen that the ultrasonic
velocity of the tested concrete decreases gradually along with the in-
crease in hydrostatic pressure level. Under the hydrostatic pressure of where V is the ultrasonic velocity; E is the average Young's modulus
500 MPa, the ultrasonic velocity of the tested specimen decreases by of concrete dened as the ratio of strength and strain at the maximum
about 55% compared with the virgin one for the LSC, and about 60% stress; and v are the density and the Poisson's ratio of concrete, re-
for the HSC. These results further demonstrate the damage to concrete spectively. The reduction of wave propagation velocity is therefore
specimens by application of hydrostatic pressure. related to the degradation of Young's modulus E. Density can be as-
The relationship between ultrasonic velocity and the Young's mod- sumed unchanged before and after the tests (Fig. 4). Poisson's ratio
ulus of concrete is dened as: changes with material damage, according to the reference [25], a
simplied linear change from 0.2 (virgin concrete) to 0.4 (fully da-
(1 v ) E maged concrete) is used in this study. The damage D can be dened as:
V=
(1 + v )(1 2v ) (9)
Etested
D=1 (method 1)
E virgin (10)
Table 7
Comparisons of ultrasonic velocities in the virgin concrete and damaged concrete for LSC based on the change in Young's modulus; or equivalently based on
specimens. change in wave velocity as:
Hydrostatic pressure Ultrasonic velocity - No. Ultrasonic velocity - No. 2 2
Vtested (1 + v)(1 2v)(1 v )
(MPa) 1 (km/s) (km/s) D=1 2
( method 2)
virgin/damaged virgin/damaged
Vvirgin (1 + v )(1 2v )(1 v) (11)
35 4.37/4.37 4.41/4.13 where Etested, Vtested and v are the average Young's modulus, ul-
70 4.32/3.33 4.35/3.56 trasonic velocity and Poisson's ratio of tested specimens, respectively;
175 4.29/2.78 4.38/2.66
Evirgin, Vvirgin, and v are properties of virgin concrete.
350 4.28/2.05 4.35/2.33
500 4.39/1.70 4.37/2.14 The results derived from Eq. (10) and Eq. (11) are shown in Fig. 15.
It can be found from the gure that the damage of concrete derived
146
J. Cui et al. Cement and Concrete Research 100 (2017) 140152
0.6 0.6
Damage
Damage
0.4 0.4
Method 1 for compression Method 1 for compression
0.2 Method 1 for tension 0.2 Method 1 for tension
Method 2 Method 2
0.0 0.0
0 4 8 12 0 4 8
Nominal pressure Nominal pressure
(a) (b)
from the reduction of Young's modulus (method 1) and the reduction of the LSC, and at conning pressure 20 MPa the failure strength ratio is
ultrasonic velocity (method 2) are similar. For compression test, the 2.58 for the HSC in the present study, while a 16.4 MPa conning
estimated damage values from method 1 are a little smaller than those pressure corresponds to a failure strength ratio about 2.8 in the latter
from method 2 while for tension this observation reverses. This is be- study by Ansari and Li [1]. This dierence in the failure strength ratio
cause the tensile Young's modulus decreases more sharply than com- obtained in the two studies can certainly be attributed to the un-
pressive Young's modulus owing to tensile modulus of concrete is more conned strength of the concrete, and probably can also be attributed
sensitive to the micro-cracks caused by loading and unloading of the to the dierent sizes and shapes of specimens used in the two studies,
hydrostatic pressure. implying the size and shape eect on the tri-axial compression tests.
4. Eect of stress path on the behavior of concrete 4.1.2. Stress path 2: unloading program
In this case the hydrostatic pressure (1 = 2 = 3 > 0) was ap-
4.1. Eect of stress path on the compressive meridian plied rst to a desired value at a constant rate of 1 kN/s (0.4 MPa/s),
then 1 was kept unchanged but 2 and 3 were gradually reduced by
The inuence of stress path on the strength envelope of concrete is displacement control mode until the specimen was destroyed
investigated in this study by subjecting the specimens to dierent stress (1 > 2 = 3 > 0). In the unloading stage, the unloading rate was
states. carefully controlled from 0.3 mm/min to 0.5 mm/min for the both di-
rections to keep 2 = 3. Fig. 17 shows an example of the loading
4.1.1. Stress path 1: loading procedure process. As shown, 2 = 3 was maintained during the unloading pro-
In this case, the specimen was subjected to a hydrostatic pressure cess. The test results are given in Table 10.
(1 = 2 = 3 > 0) applied slowly up to a given level at a constant
rate of 0.5 kN/s (0.2 MPa/s), then 2 = 3 were kept unchanged but 1 4.1.3. Comparison of the results obtained from the two stress paths
(1 > 2 = 3 > 0) was further increased at a constant rate of Fig. 18 shows the compressive meridians of concrete obtained from
0.3 mm/min until the specimen was destroyed. the above two stress paths. The horizontal coordinate is the mean
The results are given in Table 9. It can be found that the axial pressure, given as p = (1 + 2 + 3)/3. The vertical coordinate is the
strength of concrete increases prominently with conning pressure. The deviatoric stress, given as q = 1 3. Fig. 18 shows that the envel-
increase in the conned strength with the conning pressure for LSC is opes seem independent of the stress paths, especially when the mean
more prominent than that for HSC, i.e. if the conning pressure is the pressure p < 1.2fc. When p > 1.2fc, the envelope obtained from the
same, the LSC has a larger failure strength ratio 1u/fc than the HSC. stress path 1 starts to become slightly thinner than the envelope ob-
Fig. 16 (a) shows the axial stress-strain relations of LSC in this study tained from the stress path 2. This is because the concrete suered
and Fig. 16 (b) is the test results obtained by Ansari and Li [1] using certain level of damages when it was subjected to a high hydrostatic
100 200 cylindrical specimens with unconned uniaxial compressive pressure in the rst step of stress path 2 before reducing the conning
strength 42 MPa. From the results in Fig.16 and Table 9, it can be seen stress. Therefore it still can be concluded that the strength envelope is
that at conning pressure 17.5 MPa, the failure strength ratio is 4.1 for independent of the loading path. The LSC and the HSC follow the si-
milar trend.
Table 9
Fig. 19 shows the failure mode under dierent stress states for LSC.
Test results of stress path 1 for compressive meridian.
The specimen was taken out of the tri-axial testing equipment when the
Specimen Conning No. of Averaged Averaged test was over and at this time the specimen was still intake without
pressure specimens failure strength failure strength obvious fragments. However, with a slight strike by a rubber hammer
2 = 3 (MPa) tested 1u (MPa) ratio 1u/fc the specimens broke quickly as shown in Fig. 19. From the gure, one
LSC 0 4 35.2 1.0 can nd that the failure mode of the specimen depends on the applied
7 (0.2fc) 3 79.8 2.27 stress level. When the conning pressure was low, the concrete spe-
12.3 (0.35fc) 3 114.7 3.26 cimen failed with a dominant crack penetrating through the entire
17.5 (0.5fc) 3 144.0 4.09 specimen. This failure was more like a brittle failure with a short
35 (fc) 3 237.5 6.75
70 (2fc) 2 342.0 9.72
softening stage. When the applied conning pressure was high, more
HSC 0 4 63.8 1.0 intensive damage to cement matrix was created, therefore led to more
5 (0.08fc) 3 96.9 1.52 number of damage surfaces and hence more number of smaller frag-
10 (0.16fc) 3 120.1 1.88 ments. The concrete specimen was destroyed because of many micro-
20 (0.32fc) 3 164.5 2.58
cracks developed inside the specimen. This is a ductile failure char-
40 (0.64fc) 2 242.3 3.78
80 (1.28fc) 1 365.1 5.72 acterized by a very long softening stage reected in the stress-strain
curve. The failure modes of specimens corresponding to stress path 2
147
J. Cui et al. Cement and Concrete Research 100 (2017) 140152
and stress path 1 have these similar features. The results of HSC are
300 similar to LSC, therefore they are not shown here for brevity.
x-axis
y-axis
4.2. Eect of stress path on the tensile meridian
Force(kN)
z-axis
200
4.2.1. Stress path1: loading program
In this case the hydrostatic pressure (1 = 2 = 3 > 0) was ap-
100 plied rst to a desired level at a constant rate of 0.5 kN/s (0.2 MPa/s).
Then 3 was unchanged but 1 and 2 were further increased by dis-
placement control mode until the specimen was destroyed
0
0.0 0.2 0.4 0.6 (1 = 2 > 3 > 0). The loading rate was regulated from 0.3 mm/
Time(h) min to 0.5 mm/min from time to time to keep 1 = 2. Fig. 20 shows an
example of this loading process. The results are given in Table 11. In the
Fig. 17. Unloading process for producing compressive meridian.
rst group of this stress path, the conning pressure 3 = 0, indicating
the specimen was under biaxial compression. The failure strength ratio
Table 10 1u/fc = 1.29 for LSC and 1.16 for HSC which are similar to the results
Test results of stress path 2 for compressive meridian.
obtained by other researchers [2628].
Specimen Unchanged stress 1 No. of specimens Averaged failure stress
(MPa) tested 2f = 3f (MPa) 4.2.2. Stress path 2: unloading program
In this case the hydrostatic pressure (1 = 2 = 3 > 0) was ap-
LSC 52.5(1.5 fc) 3 2.9
70 (2 fc) 3 4.9 plied rst to a desired level at a constant rate of 0.5 kN/s (0.2 MPa/s),
87.5 (2.5 fc) 3 8.4 then 1 = 2 were kept unchanged and 3 (1 = 2 > 3 > 0) was
105 (3 fc) 3 12.0 gradually reduced at a constant rate of 0.3 mm/min until the specimen
140 (4 fc) 2 22.7
was destroyed. Table 12 shows the test results.
210(6 fc) 2 39.9
HSC 80 (1.25 fc) 3 3.17
100 (1.57 fc) 3 5.9 4.2.3. Comparison of the results obtained from the two stress paths
120 (1.88 fc) 3 9.2 Fig. 21 shows the tensile meridians of concrete obtained from the
150 (2.35 fc) 2 17.5
240 (3.76 fc) 1 47.5
above two stress paths. From the gure, it can be concluded that the
stress paths have little eect on the strength envelopes. The envelopes
corresponding to the stress path 1 seems slightly thinner than that
from the stress path 2 at low hydrostatic pressure, but this reverses at
high hydrostatic pressure. This is again because of the dierent damage
degrees to the concrete specimen subjected to dierent hydrostatic
pressures before unloading 3 instead of the inuences of stress path.
(MPa)
150 150
100 100
50 50
0 0
0 40 80 120 160 0 40 80 120 160
(MPa) (MPa)
(a) (b)
148
J. Cui et al. Cement and Concrete Research 100 (2017) 140152
D2 = bln (p a) p 1 + a (12)
5. Eect of damage under high hydrostatic pressure on the
concrete constitutive model where a and b are parameters that control the damage degree. a
(> 0) denes the value of nominal hydrostatic pressure that D2 is larger
From the above experimental results, it is clear that the commonly than zero, or hydrostatic pressure is high enough to cause damage to
used concrete constitutive models overestimate the concrete strength concrete. a = 0 means hydrostatic pressure damages concrete material
under conning pressure, and ignores the damage to concrete by high when it is equal to or larger than the uniaxial compressive strength of
149
J. Cui et al. Cement and Concrete Research 100 (2017) 140152
150 150
(MPa)
(MPa)
100
100
50
50
150
J. Cui et al. Cement and Concrete Research 100 (2017) 140152
According to the study of Hartmann et al. [30] and Gebbken and The authors gratefully acknowledge the support from National
Hartmann [31], The total damage can be set to D = 1 (1 D1) Natural Science Foundation of China under grant number 51227006,
(1 D2). and the Australian Research Council under grant number DP160104557
It should be noted that theoretically, there should also exit a cap on for carrying out this research.
the failure surface because concrete could be destroyed by very high
hydrostatic pressure. Unfortunately, no testing data is available yet to References
derive this cap on failure surface because no hydrostatic test has
reached the ultimate strength of concrete. Therefore the above pro- [1] F. Ansari, Q. Li, High-strength concrete subjected to triaxial compression, ACI
posed modication only considers the damage that could be induced by Mater. J. 95 (1998) 747755.
high hydrostatic pressure, instead of a complete failure surface that [2] M. Attard, S. Setunge, Stress-strain relationship of conned and unconned con-
crete, Mater. J. 93 (1996) 432442.
denes the ultimate strength of concrete at high hydrostatic pressure. [3] D. Candappa, J. Sanjayan, S. Setunge, Complete triaxial stress-strain curves of high-
Within hydrocodes, the constitutive relations for the volumetric and strength concrete, J. Mater. Civ. Eng. 13 (2001) 209215.
the deviatoric material behavior are treated separately [31]. Usually [4] D. Candappa, S. Setunge, J. Sanjayan, Stress versus strain relationship of high
strength concrete under high lateral connement, Cem. Concr. Res. 29 (1999)
EOS is an independent equation used to model the relationship between 19771982.
volumetric strain and pressure p of concrete material, which is illu- [5] K.A. Harries, G. Kharel, Experimental investigation of the behavior of variably
strated in Fig. 25. Concrete starts to have plastic deformation (compact) conned concrete, Cem. Concr. Res. 33 (2003) 873880.
[6] M. Petkovski, Experimental detection of damage evolution in concrete under mul-
when the pore crush pressure Pcrush is reached, below which the model tiaxial compression, J. Eng. Mech. 139 (2012) 616628.
is elastic. When the pressure reaches the compaction pressure Pcomp the [7] X.H. Vu, Y. Malecot, L. Daudeville, E. Buzaud, Experimental analysis of concrete
material is assumed fully compacted. If the damage produced by hy- behavior under high connement: eect of the saturation ratio, Int. J. Solids Struct.
46 (2009) 11051120.
drostatic pressure D2 is not considered, unloading beyond the pore [8] L.J. Malvar, J.E. Crawford, J.W. Wesevich, D. Simons, A plasticity concrete material
crush pressure occurs along the initial elastic stiness and will result in model for DYNA3D, Int. J. Impact Eng. 19 (1997) 847873.
a permanent volumetric strain for zero pressure, subsequent reloading [9] W. Riedel, K. Thoma, S. Hiermaier, E. Schmolinske, Penetration of reinforced
concrete by BETA-B-500 numerical analysis using a new macroscopic concrete
occurs along the unloading curve. If the eect of damage D2 is con-
model for hydrocodes, Proceedings of the 9th International Symposium on the
sidered to the bulk modulus degradation, the slope of unloading/re- Eects of Munitions with Structures: sn, 1999.
loading curve will be lower than that of initial elastic loading curve, as [10] LS-DYNA version ls971 d R7.0.0 [Computer software]. Livermore Software
is shown in Fig. 25. Comparing with the EOS gure shown in Fig. 4 Technology Corporation. Livermore, CA.
[11] J. Cui, H. Hao, Y. Shi, Discussion on the suitability of concrete constitutive models
obtained in the present study, a modied unloading/reloading curve for high-rate response predictions of RC structures, Int. J. Impact. Eng. 106 (2017)
that takes into consideration the stiness degradation is suggested. 202216.
According to the relationship between bulk modulus and Young's [12] N. Gebbeken, M. Ruppert, A new material model for concrete in high-dynamic
hydrocode simulations, Arch. Appl. Mech. 70 (2000) 463478.
modulus, the bulk modulus degradation can be dened as [13] C. Poinard, Y. Malecot, L. Daudeville, Damage of concrete in a very high stress state:
experimental investigation, Mater. Struct. 43 (2010) 1529.
ED
KD = and ED = (1 D2 ) E0 [14] T.M. Pham, M.N. Hadi, T.M. Tran, Maximum usable strain of FRP-conned con-
3(1 2v) (15) crete, Constr. Build. Mater. 83 (2015) 119127.
[15] M.D. Kotsovos, Eect of stress path on the behavior of concrete under triaxial stress
where KD is the bulk modulus of damaged concrete under high states, J. Proc. (1979) 213226.
hydrostatic pressure, v is the simplied damaged Poisson's ratio [16] K.H. Gerstle, D.L. Linse, P. Bertacchi, Strength of concrete under multiaxial stress
states, Special Publication, 55 (1978), pp. 103132.
varying from 0.2 (D2 = 0) to 0.4 (D2 = 1) linearly [25]. Tests data [17] T. Gabet, Y. Malcot, L. Daudeville, Triaxial behaviour of concrete under high
about bulk modulus degradation are still very limited, therefore to stresses: inuence of the loading path on compaction and limit states, Cem. Concr.
study it more thoroughly, more specially designed tests should be car- Res. 38 (2008) 403412.
[18] K. Du, M. Tao, X. Li, J. Zhou, Experimental Study of Slabbing and Rockburst
ried out.
Induced by True-Triaxial Unloading and Local Dynamic Disturbance, Rock Mech.
The above discussions on possible modications of the concrete Rock. Eng. 49 (2016) 34373453.
material models are preliminary only and are based on the testing data [19] X. Li, K. Du, D. Li, True triaxial strength and failure modes of cubic rock specimens
with unloading the minor principal stress, Rock Mech. Rock. Eng. 48 (2015)
obtained in this study. More thorough analyses including theoretical
21852196.
derivations and more properly designed and executed tests are deemed [20] J.G.M. van Mier, S. Shah, M. Arnaud, J. Balayssac, A. Bascoul, S. Choi, et al., Strain-
necessary for developing a more comprehensive concrete material softening of concrete in uniaxial compression, Mater. Struct. 30 (1997) 195209.
model that takes into consideration damages at high hydrostatic pres- [21] J.G.M. van Mier, Multiaxial strain-softening of concrete, Mater. Struct. 19 (1986)
179190.
sures. [22] M. Sato, H. Okuda, Fracture Mechanism of Plain Concrete under Uniaxial Tension,
(1994).
[23] W.A. Tasong, C.J. Lynsdale, J.C. Cripps, Aggregate-cement paste interface: part I.
6. Conclusion Inuence of aggregate geochemistry, Cem. Concr. Res. 29 (1999) 10191025.
[24] J. Xiao, W. Li, D.J. Corr, S.P. Shah, Eects of interfacial transition zones on the
This paper presents testing data of concrete specimens subjected to stressstrain behavior of modeled recycled aggregate concrete, Cem. Concr. Res. 52
(2013) 8299.
tri-axial tests. It was found that the strength envelope of concrete was
151
J. Cui et al. Cement and Concrete Research 100 (2017) 140152
[25] Y. Bouaa, R. Smahi, H. Dumontet, M.S. Kachi, Modeling the behavior of concrete Advances in Materials Science Research, 1 2011, pp. 145182.
by damage mechanics with a Poisson's ratio variable, Proc. Mater. Sci. 3 (2014) [29] T. Borrvall, W. Riedel, The RHT concrete model in LS-DYNA, Proceedings of The 8th
714719. European LS-DYNA user conference, 2011.
[26] H.B. Kupfer, K.H. Gerstle, Behavior of concrete under biaxial stresses, J. Eng. Mech. [30] T. Hartmann, A. Pietzsch, N. Gebbeken, A hydrocode material model for concrete,
Div. 99 (1973) 853866. Int. J. Protect. Struct. 1 (2010) 443468.
[27] S.-K. Lee, Y.-C. Song, S.-H. Han, Biaxial behavior of plain concrete of nuclear [31] N. Gebbeken, T. Hartmann, A new methodology for the assessment of the EoS data
containment building, Nucl. Eng. Des. 227 (2004) 143153. of concrete, Int. J. Protect. Struct. 1 (2010) 299317.
[28] D. Yan, G. Lin, G. Chen, Dynamic properties of concrete under multi-axial loading,
152