Vous êtes sur la page 1sur 17

Construction and Building Materials 225 (2019) 256–272

Contents lists available at ScienceDirect

Construction and Building Materials


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

Stress-strain relationship for plain and fibre-reinforced lightweight


aggregate concrete
Xi Liu, Tao Wu ⇑, Yang Liu
School of Civil Engineering, Chang’an University, Xi’an, China

h i g h l i g h t s

 The mechanical properties of fibre-reinforced LWAC are experimentally studied.


 Both steel fibre and carbon fibre provide an improvement on the ductility of LWAC.
 The regression stress-strain models for plain LWAC, steel-fibre and carbon-fibre reinforced LWAC are presented.
 A statistically stochastic damage constitutive model is established.
 The proposed models provide improved predictions of stress-strain behavior over existing models.

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

Article history: Lightweight aggregate concrete (LWAC) shows good performance in structural applications owing to its
Received 20 February 2019 expected lower contribution to the deadweight of a structure, it nonetheless exhibits significant reduc-
Received in revised form 9 June 2019 tion in certain mechanical properties compared with the normal weight aggregate concrete. The inclusion
Accepted 16 July 2019
of fibres could provide some positive changes to its mechanical properties and stress–strain behaviour.
This study experimentally investigated the mechanical properties and full stress–strain curve of LWAC
reinforced with steel and carbon fibres, involving specimens of strength classes ranging from LC40 to
Keywords:
LC60 and fibre volume fractions from 0% to 0.9%. The test results showed that steel fibres could signifi-
Lightweight aggregate concrete
Fibre-reinforced concrete
cantly improve the flexural strength and splitting tensile strength. Both the steel and carbon fibres
Stress-strain relationship strengthened the crack resistance of LWAC, while the compressive strength was not obviously enhanced.
Statistically stochastic damage constitutive A general stress–strain model adapted to both plain LWAC and fibre-reinforced LWAC was proposed by
model comparing and modifying the existing models, and a statistically stochastic damage constitutive model
was derived. The validity of the proposed models was examined by experimental results and test data
from other studies.
Ó 2019 Elsevier Ltd. All rights reserved.

1. Introduction such as steel fibre, provided improvements to LWAC on flexural


strength, splitting tensile strength, and toughness. However, it
The application of structural lightweight aggregate concrete was noted that the workability and density of LWAC could be neg-
(LWAC) is versatile in modern construction practices owing to its atively affected by introducing steel fibres. The inclusion of steel
low density and high specific strength [1]. The low density of fibres could form some hoops around the porous lightweight
LWAC permits a saving dead load for the design of structures and aggregate during the mixing procedure of LWAC, whereby aggre-
foundations. However, LWAC also exhibits more brittleness than gates would be confined and enhanced. Thus, the compressive
normal weight concrete (NWC) at comparable strength, which is strength of LWAC with a proper volume of steel fibres would
a serious shortcoming affecting both the application and the increase. The action mechanism of the steel fibres nonetheless
long-term structural performance of LWAC. would be undermined when their volume fraction exceeded the
Using fibres as an additive into the mixes has proven to be a appropriate extent. Hence, some previous studies achieved an
method that is beneficial for certain engineering mechanical prop- increase in the compressive strength of steel-fibre-reinforced
erties of LWAC. In some accomplished works [2–4], metallic fibre LWAC [4,5] and some did not [6,7].
LWAC reinforced by non-metallic fibres presented an equally
⇑ Corresponding author. good behaviour in terms of flexural strength, splitting tensile
E-mail address: wutaochd0922@yahoo.com (T. Wu). strength, and toughness, compared with LWAC reinforced by steel

https://doi.org/10.1016/j.conbuildmat.2019.07.135
0950-0618/Ó 2019 Elsevier Ltd. All rights reserved.
X. Liu et al. / Construction and Building Materials 225 (2019) 256–272 257

fibres. However, different effects on the compressive strength of relevant Chinese standard GB/T 17431.2–2010 are presented in
LWAC were obtained by using different types of non-metallic Table 1 [24]. Ordinary Portland cement 42.5 was used for all mixes.
fibres. For instance, carbon fibres embedded in the cement matrix Its specific chemical compositions and physical properties are
could provide confinement of transverse deformation; as a result, given in Table 2.
the compressive strength is increased. Kayali et al achieved Two types of fibres were selected to enhance the LWAC, namely,
improved indirect tensile strength and modulus of rupture by add- steel fibres and carbon fibres. Steel fibres having a length of 13 mm
ing polypropylene fibres into LWAC [6], but no significant enhance- and a diameter of 200 lm were selected, and carbon fibres having
ment in compressive strength was observed. Li et al noted that a length of 6 mm and a diameter of 7 lm were used. The mechan-
when the volume content of polypropylene fibres increased ical and physical properties of these fibres are listed in Table 3.
beyond 0.95%, the dispersion of these fibres becomes very difficult, Scanning electron microscope (SEM) images of the aggregates,
and thus, the compressive strength is hardly or even negatively as shown in Fig. 1, were used to analyse the macro properties of
influenced [8]. Despite the different effects of non-metallic fibres the LWAC in terms of the microstructure of shale ceramsites. As
on compressive strength, the non-metallic fibres exactly improve can be seen in Fig. 1(c), the surface of the shale ceramsite aggregate
the mechanical properties of LWAC, and the density of the concrete is relatively smooth and dense, which would produce a negative
did not appear to be influenced by the lower specific weight of effect on the interlocking action between the aggregates and
non-metallic fibres. cement matrix and, contrary to their surface, the inner structure
Although the guidelines for LWAC had been included in the of the shale ceramsite is porous, as shown in Fig. 1(d). This porosity
some codes from European and American, most of the mechanical explains the good performance of the shale ceramsite in water
properties of fibre-reinforced LWAC are poorly categorized, and the absorption, low density, and coefficient of thermal conductivity.
stress–strain relationships of both plain and fibre-reinforced LWAC However, the cavities shown in Fig. 1(c) and (d) in the shale ceram-
are rarely reported [1,9,10]. The stress–strain curves of LWAC typ- site also represent weak faces in LWAC, where cracks could directly
ically yield linear ascending part approaching 90% of the compres- pass through.
sive strength [11–13], while it was reported that the ascending
part of the stress–strain curve was not significantly affected by
2.2. Mix proportions
the inclusion of fibres. In addition, the elastic modulus of fibre-
reinforced LWAC was not obviously distinct from that of plain
The mix proportions of LWAC are listed in Table 4. Water/binder
LWAC, especially with a minor volume content of fibres [14]. Yet
(W/B) ratios of 0.4, 0.3, and 0.26 were chosen to achieve strength
the presence of fibres in the LWAC could effectively modify its
classes of 40 MPa, 50 MPa, and 60 MPa, respectively. The volume
rapidly descending portion [13], and the fibres having high elastic
fraction of fibres was altered to study its effect on the LWAC prop-
modulus contributed more improvement to the brittleness of
erties. Three of the mix proportions had 0.3%, 0.6%, and 0.9% by vol-
LWAC comparative to the fibres with lower elastic modules [4].
ume of steel fibres. Similarly, the volume percentages of mixes
Fibre-reinforced LWAC generally shows a similar performance in
containing carbon fibres varied from 0.3% to 0.9%, and the rest of
the ascending part of the stress–strain relation, but higher ductility
the series was cast without fibres. Furthermore, supplementary
in the descending branches [15]. In the past, researchers [16–19]
cementitious materials, including EM920U silica fume, grade Ⅰ fly
have proposed widely adaptable constitutive models of concrete,
ash, S95 ore fines, and BKS-199 polycarboxylic acid superplasti-
and the model for fibre-reinforced NWC and plain LWAC are also
cizer (SP), were used to produce the specimens. Tap water was
extensively available in the literatures [20–23]. Nevertheless, the
used as mixing water.
stress–strain model for fibre-reinforced LWAC was rarely investi-
gated owing to the complex action mechanism and the differences
among various types of fibres. 2.3. Test details
In this study, an experimental program on the cubic compres-
sive strength, axial compressive strength, and flexural and splitting To study the mechanical properties of LWAC and fibre-
tensile strength of plain LWAC and fibre-reinforced LWAC was car- reinforced LWAC, nine series of specimens were prepared by vary-
ried out. Then, the concrete failure processes, combined with the ing amount of fibres, type of fibres, and concrete strength, respec-
microstructure of lightweight aggregates obtained from SEM tively. The oven-dry density of nine mixtures were tested
images, were used to analyse the effect of fibres on the macro per- according to GB/T 17431.2-2010 [24]. Specific specimens used in
formance of LWAC. Furthermore, the mechanical properties and tests are listed in Table 5.
stress–strain relations of plain and fibre-reinforced LWAC were The concretes were mixed in a forced mixer, and the shale cer-
discussed. Based on the experimental results, expressions for fea- amsite were wetted with the mixing water before the process of
ture points of the constitutive curve and the elastic modulus of mixing. During the mixing process, the fibres dispersed in the mix-
plain and fibre-reinforced LWAC were established. Finally, to ing water were added after the addition of aggregates. All speci-
develop the stress–strain curve of plain LWAC and fibre- mens were demolded after 24-h curing time, and then were
reinforced LWAC, a stress–strain relationship was proposed by cured in water. All specimens were tested after a 28-day curing
modifying the existing models. A statistically stochastic damage time.
constitutive model was derived based on the Weibull distribution A displacement-control procedure was applied during the axial
and the log-normal distribution. compression test. To obtain continuous and steady curves, a slow
displacement rate, 0.03 mm/min, was employed. A load-control
procedure was adopted in the cubic compression strength test.
2. Experimental program The loading rate of the cubic specimens, for LC40 and LC50, was
6 kN/s, while a loading rate of 10 kN/s was employed in series
2.1. Materials properties using a concrete strength of 60 MPa.
The compression tests were conducted with an electric-
Medium sands passing through the 4 mm sieve with bulk den- hydraulic servo universal test machine having a maximum load
sity of 1510 kg/m3 were used as the fine aggregate in all mixtures. capacity of 1000 kN. In order to avoid local crushing of concrete,
The coarse LWAC used in this work is a grade 900 shale ceramsite, steel ferrules were installed at the ends of the specimens, respec-
and the physical properties of the selected LWAs determined by tively, as shown in Fig. 2. The axial displacements were measured
258 X. Liu et al. / Construction and Building Materials 225 (2019) 256–272

Table 1
Properties of lightweight aggregate used.

Types Size of aggregate (mm) Dry density (kg/m3) Bulk density (kg/m3) Water absorption (%) Cylinder compressive strength (MPa)
Shale ceramiste 5–16 1512 860 2.2 6.9
Medium sand 4 2620 1510 1.9 –

Table 2 cess, and the load sensing system in the testing machine achieved
Chemical compositions and physical properties of ordinary Portland cement. the loads. The concrete cubes for the splitting tensile strength tests
Chemical composition (%) Ordinary Portland cement were placed on an arching steel block and loaded by a spherical
Calcium oxide (CaO) 62.81
hinge at a loading rate of 1 kN/s. The test results of the oven-dry
Silicon dioxide (SiO2) 20.36 density, cubic compressive strengths, flexural strengths, and split-
Aluminum oxide (Al2O3) 5.67 ting tensile strengths are summarized in Table 6.
Ferric oxide (Fe2O3) 3.84
Magnesium oxide (MgO) 2.68
Sulfur trioxide (SO3) 2.51 3. Test results and discussions
K2O 0.87
Na2O 0.19
Loss on ignition 1.07 3.1. Mechanical properties
Specific gravity (g/cm3) 3.14
Fineness (m2/kg) 329 3.1.1. Compressive strength
The specimens of LC60 and of LC60 reinforced with fibres were
used to study the compressive characteristics of the fibre-
by a quasi-static data acquisition system with the sampling fre- reinforced LWAC. In this study, both the steel and carbon fibres
quency of 10 Hz. provided a maximum 12.1% (carbon fibre) and 12.3% (steel fibre)
The concrete prisms for the flexural tests were subjected to a increase in cubic compressive strength with the volume fraction
two-point load with a distance of 100 mm between the two load- of 0.9% compared with the specimen without fibres. Meanwhile,
ing points, and the support span was of 300 mm, as shown in Fig. 3. a maximum 5.7% reduction in concrete strength was observed by
A loading rate of 0.3 kN/s was employed during the loading pro- using a 0.6% volume fraction of steel fibres. This decrease in com-

Table 3
Properties of fibres used.

Types of fibres Density (kg/m3) Length (mm) Diameter (lm) Elongation (%) Elastic modulus (GPa) Tensile strength (MPa)
Steel fibre 7800 13 200 3.2 200 >3000
Carbon fibre 1760 6 7 1.8 240 >4000

Fig. 1. The surface and internal microstructure of shale ceramsite aggregates used: (a) Appearance of LWAs used (b) profiles of LWAs used (c) outer surfaces of LWAs used
(1000, scale 50 lm) (e) Microstructure of shale ceramsite aggregate (300, scale 100 lm).

Table 4
Mix proportions of fibre-reinforced LWAC.

Groups W/B Cement Sand Ceramsite Fly ash Silica fume Water SP kg/m3 Steel fibre Carbon fibre Fibre volume fraction/(%)
(kg/m3) (kg/m3) (kg/m3) (kg/m3) (kg/m3) (kg/m3) (kg/m3) (kg/m3)
Steel fibre Carbon fibre
LC40 0.40 360 688 606 90 0 180 2.6 – – –
LC50 0.30 400 700 616 100 0 150 4.0 – – – –
LC60 0.26 440 689 607 66 44 143 5.5 – – – –
LC60-0.3C 0.26 440 685 604 66 44 143 7 – 5.28 – 0.30
LC60-0.6C 0.26 440 680 599 66 44 143 8.5 – 10.56 – 0.60
LC60-0.9C 0.26 440 676 596 66 44 143 10 – 15.84 – 0.90
LC60-0.3S 0.26 440 686 604 66 44 143 5.7 23.4 – 0.30 –
LC60-0.6S 0.26 440 683 602 66 44 143 5.9 46.8 – 0.60 –
LC60-0.9S 0.26 440 680 599 66 44 143 6.1 70.2 – 0.30 –
X. Liu et al. / Construction and Building Materials 225 (2019) 256–272 259

Table 5
Amounts of specimens used in all tests.

Groups Cubic compressive tests Axial compressive tests Flexural strength tests Splitting tensile strength
(100  100  100 mm3) (100  100  300 mm3) (100  100  140 mm3) (100  100  100 mm3)
LC40 3 3 – –
LC50 3 3 – –
LC60 3 3 3 3
LC60-0.3C 3 3 3 3
LC60-0.6C 3 3 3 3
LC60-0.9C 3 3 3 3
LC60-0.3S 3 3 3 3
LC60-0.6S 3 3 3 3
LC60-0.9S 3 3 3 3
In total 27 27 21 21

Fig. 2. Set-up of compression tests: (a) testing machine (b) loading diagram.

Fig. 3. Loading diagrams of flexural and splitting tensile tests: (a) Flexural test (b) Splitting tensile tests.

Table 6
Results of oven-dry density qd, cubic compressive strength fcu, splitting tensile strength fst, and flexural strength ff.

Groups qd(kg/m3) fcu(MPa) ff(MPa) fst(MPa) Flexural/Compressive ratio Tensile/Compressive ratio


LC40 1824 – – – –
LC50 1832 – – – –
LC60 1871 76.7 3.53 6.48 0.046 0.084
LC60-0.3C 1849 73.3 3.72 7.31 0.051 0.100
LC60-0.6C 1814 72.3 5.39 7.75 0.075 0.107
LC60-0.9C 1845 86.1 5.88 10.04 0.068 0.117
LC60-0.3S 1846 78.8 4.78 8.16 0.061 0.104
LC60-0.6S 1857 81.0 5.04 9.45 0.062 0.117
LC60-0.9S 1897 86.0 5.15 10.44 0.060 0.121
260 X. Liu et al. / Construction and Building Materials 225 (2019) 256–272

pressive strength could be attributed to the unexpected dispersion that of plain LWAC. Additionally, the improvement was more sig-
of fibres, which would cause poor workability and incomplete nificant with steel fibres than that with carbon fibres.
compaction [25,26]. During the flexural strength tests, specimens with carbon fibres
exhibited similar failure modes to the plain LWAC specimens: the
breakup happened when the ultimate load was reached, and the
3.1.2. Splitting tensile and flexural strength crack extended through the specimen rapidly. It can be seen from
As can be seen in Table 6, both the flexural strength and split- Fig. 4 that a different failure mode from the other two types of
ting tensile strength of the specimens with fibres were higher than specimens was obtained in the steel fibre-reinforced specimens.
Specimens with steel fibres cracked before the ultimate load was
reached, which was not observed in other series of specimens.
Thus, the steel fibres across the crack resisted their propagation
and carried the load until the specimen was broken.
Fig. 5(a) shows the ratios that flexural strength to cubic com-
pressive strength. It can be clearly seen that the fibres provided a
promotion on the ratio but the effect was not consistently ascend-
ing. The addition of fibres can virtually improve the flexural/com-
pressive ratio of LWAC, and carbon fibres provide a better
improvement on this performance than steel fibres when the vol-
ume fraction exceeds 0.6%. Ratios that tensile strength to compres-
sive strength in Fig. 5(b) indicated that fibres made a significant
improvement on the tension–compression ratio of LWAC, and car-
bon fibres produced a better effect than steel fibres.

3.2. Stress–strain curves

3.2.1. Typical stress–strain curve


To research the mechanical behaviour of fibre-reinforced LWAC,
the typical stress–strain diagram shown in Fig. 6 was used to
explain the observations acquired in tests. The points O, A, B, and
C on the stress–strain curve were selected corresponding to the
failure processes of the specimens. Thus, the stress–strain curve
can be divided into four stages.
The stress increased linearly from the point O to point A. The
initial stiffness is reflected by the slope of the ascending part and
is considerably smaller than that of normal concrete. When point
A was reached, the specimens appeared to internally crack. The
stress steadily increased, while the strain had a faster increase than
before, meaning that the stiffness of the specimen was decreased.
At the portion between the points A and B, the concrete contin-
ually cracked, and the cracking was considered to be more frequent
than NWC owing to the lower-strength of porous lightweight
aggregate. Then, when the peak point B was reached, the stress
of concrete decreased rapidly to approximately 50% of the peak
Fig. 4. Failure modes of specimens in Flexural strength test: (a) Plain and carbon- stress, which suggests the significant brittleness of LWAC. After
fibre reinforced LWAC (b) Steel-fibre-reinforced LWAC. the point C, the frictional resistance and the residual stress primar-

0.10 0.14
Reinforced with carbon fibres Reinforced with carbon fibres
Reinforced with steel fibres Reinforced with steel fibres
0.08
0.12
fst/fcu

0.06
ff/fcu

0.10
0.04

0.08
0.02

0.00 0.06
0.0 0.3 0.6 0.9 0.0 0.3 0.6 0.9
vs vs
(a) (b)
Fig. 5. Flexural to compressive and tension to compression ratio of fibre-reinforced LWAC: (a) Comparison of flexural/compressive ratio (b) Comparison of
tensile/compressive ratio.
X. Liu et al. / Construction and Building Materials 225 (2019) 256–272 261

15 mm near the side. Spalling of concrete attributed to cracks


60 B LC50
Internal cracking
extending into the corner was observed in the specimens without
fibres. In contrast, as shown in Fig. 7, the concrete specimens with
50 fibres did not appear to spall, but to expand, owing to the lateral
Elastic A Visible cracking
constraints provided by the fibres.
Stress (MPa)

deformation
40 The response of the fibre-reinforced LWAC subjected to axial
C monotonic compression was recorded and analysed to illustrate
30 its difference from the plain LWAC. The microstructure of the light-
Failure
weight aggregates was observed in the SEM pictures to explain the
Micro-crack
20 performance of the specimens.
LWA
Firstly, the load increased linearly with an increase in the dis-
10 placement, and the only deformation observed was due to the elas-
tic deformation of the aggregates and cement crystals. At this
O point, internal cracks had not formed yet, and the surface of con-
0
0 2000 4000 6000 8000 crete had no obvious deformation characteristics. The deformation
Strain (×10 -6 mm/mm) characteristics among the test groups were virtually similar at this
point, where the fibres had not worked yet.
Fig. 6. Typical stress–strain curve for specimen. As the loading continued, the micro-cracks gradually bridged to
each other and extended into the gel, and the specimens yielded
ily provided the capacity across cracks, so the stress decreased cracking sounds. In this case, the fibres played a certain role in mit-
slowly with the increase of strain. igating the propagation of micro-cracks. Nonetheless, the fibres did
not completely exerted the resistance to the cracking of concrete
3.2.2. Test behaviours owing to that fibres near around the aggregates assumed relatively
All the experimental results from the axial compression tests more parallel distribution to aggregates’ boundary. For the plain
are presented in Table 7, namely, peak stress rc, strain correspond- LWAC, when the maximum load was reached, the first crack paral-
ing to peak stress ec, strain corresponding to 85% of peak stress lel to the load direction occurred at the middle height of specimen
e0.85, initial elastic modulus Ec, and secant elastic modulus Esec. and then rapidly extended through the specimen. The number of
During the cubic compressive strength tests, both plain LWAC cracks markedly increased and slight spalling of concrete could
and fibre-reinforced LWAC exhibited no obvious changes at the be observed at this point.
earliest loading period. As the loading continued, crackling sounds However, in the case of specimens with fibres, the fibres
were heard in all specimens, while the fibre-reinforced LWAC pos- delayed the formation of the first crack across the crack, and these
sessed a higher load at that moment. Then, the first crack, located fibres retarded the visible cracking. In addition, the spalling
at 20 mm near the side on the plain LWAC, could be observed. In decreased with an increase of the fibres, by which the ductility of
comparison, the fibre-reinforced LWAC firstly cracked at about high-strength LWAC was improved.

Table 7
Results of axial compressive tests and the calculation.

groups rc(MPa) ec(10-6) e0.85(10-6) Ec(MPa) Esec(MPa) qd(kg/m3) Rf Re RE


LC40 49.8 2210 2621 28,613 22,545 1859 0.95 1.10 0.79
LC50 59.0 2478 2757 29,557 23,806 1874 1.06 1.17 0.86
LC60 80.8 3056 3273 32,159 26,437 1894 1.00 1.13 0.89
LC60-0.3C 61.9 3133 3919 33,566 19,747 1866 1.09 1.09 0.84
LC60-0.6C 65.0 3433 5528 34,379 18,926 1883 1.08 1.02 0.84
LC60-0.9C 60.3 3500 5521 29,036 17,235 1890 0.96 0.88 0.92
LC60-0.3S 78.5 3553 4289 32,211 22,101 1910 0.87 0.97 0.90
LC60-0.6S 65.5 3780 6589 32,644 17,316 1925 1.10 0.95 0.91
LC60-0.9S 67.3 3688 4643 32,154 18,262 1945 0.96 0.90 0.89
Average 1.01 1.02 0.87
variance 0.006 0.011 0.002

Fig. 7. Failure modes of specimens in cubic compressive strength test: (a) Plain LWAC (b) Steel-fibre reinforced LWAC (c) Carbon- fibre reinforced LWAC.
262 X. Liu et al. / Construction and Building Materials 225 (2019) 256–272

Finally, when the load reduced by 50–70% of the peak stress, for
all specimens, the axial deformation continued to increase while
the bearing capacity declined slowly. As for the steel fibre-
reinforced LWAC, the sound of fibre pull-out could be heard obvi-
ously during the test, whereas the carbon fibre-reinforced LWAC
showed no obvious difference in the test response compared with
plain specimens. It was observed that the carbon fibre distributed
cross the crack was ruptured. Fibres in the failure surface uni-
formly distributed into the specimen, whereby the amount of split
aggregates was decreased. The specimens with fibres acquired a
higher remnant capacity when the target displacement was
obtained. The typical failure modes of plain LWAC and fibre-
reinforced LWAC are shown in Fig. 8.

Fig. 10. Feature points of toughness value.

Table 8
Toughness and brittleness indices of specimens.

Groups A1(Nmm/mm3) A2(Nmm/mm3) B I


LC40 0.066 0.118 0.556 2.798
LC50 0.084 0.145 0.580 2.726
LC60 0.138 0.225 0.610 2.638
0.3C 0.123 0.185 0.668 2.498
0.6C 0.150 0.325 0.461 3.168
0.9C 0.135 0.305 0.444 3.253
0.3S 0.171 0.337 0.506 2.975
0.6S 0.164 0.410 0.401 3.496
Fig. 8. Failure modes obtained in axial compressive strength test: (a) Plain LWAC
0.9S 0.162 0.324 0.502 2.994
(b) Carbon-fibre reinforced LWAC (c) Steel-fibre reinforced LWAC.

LC40 LC60 LC60


80 LC50 80 LC60-0.3C 80 LC60-0.3S
LC60 LC60-0.6C LC60-0.6S
LC60-0.9C LC60-0.9S
Stress (MPa)

60 60 60
Stress (MPa)
S tre s s (M P a )

40 40 40

20 20 20

0 0 0
0 2 4 6 8 10 0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16
-3 -3
Strain (×10 mm/mm) Strain (×10 mm/mm) Strain (×10-3 mm/mm)
(a) (b) (c)
Fig. 9. Comparisons of the stress–strain curves for LWAC: (a) Comparison of different strengths (b) Comparison of different carbon fibre volume fractions (c) Comparison of
different steel fibre volume fractions.
X. Liu et al. / Construction and Building Materials 225 (2019) 256–272 263

0.70 2.9 0.8 3.6 0.7 4.0


Brittleness indice Brittleness indice Brittleness indice
Toughness indice Toughness indice Toughness indice 3.8

Brittleness indice
3.4
Brittleness indice

0.65

Toughness indice
Brittleness indice
Toughness indice

Toughness indice
2.8 0.7 0.6 3.6
0.60 3.2
3.4
0.55 2.7 0.6 3.0 0.5 3.2
0.50 2.8 3.0
2.6 0.5 0.4
2.6 2.8
0.45
2.6
0.40 2.5 0.4 2.4 0.3
40 50 60 0 0.3 0.6 0.9 0 0.3 0.6 0.9
Strength (MPa) Volume fraction of carbon fibres (%) Volume fraction of steel fibres (%)
(a) (b) (c)
Fig. 11. Brittleness and toughness indices of specimens: (a) Toughness values and brittleness values of LC40, LC50 and LC60 (b) Brittleness indices and toughness indices of
specimens containing 0%, 0.3%, 0.6%, and 0.9% of carbon fibres (c) Brittleness and toughness indices of specimens containing 0%, 0.3%, 0.6% and 0.9% of steel fibres.

4.0 3.2.3. Effect of strength


Carbon fibre By comparing the curves in Fig. 9(a), it can be concluded that
3.5 Toughness indice of LC60 Steel fibre the slope of the ascending part and the strain corresponding to
the peak point both increase with an increase in concrete strength,
T o u g h n e s s i n d ic e

3.0
while the difference between the initial elastic modulus and the
2.5 peak elastic modulus gradually decreases, i.e., the elastic deforma-
tion stage is extended. Moreover, as can be seen in the figures, the
2.0 curve of LC60 has the most rapid slope in the descending parts,
sequentially the LC50 and LC40, which means that a specimen with
1.5 a higher strength exhibits more brittle behaviour.
1.0
3.2.4. Effect of fibres
0.5 Fig. 9(b) shows the favourable effect of carbon fibres on the
stress–strain behaviour of LWAC in compression. As can be seen,
0.0
0.3 0.6 0.9 for the specimens with 0.6% and 0.9% by volume of fibres, the
Volume fraction of fibres (%) descending part of the curves in Fig. 9(b) becomes slow and the
ultimate strain significantly increased. It indicates that the ductil-
Fig. 12. Comparison between brittleness indices of carbon-fibre reinforced LWAC ity of the specimens can be improved effectively by using carbon-
and steel- fibre reinforced LWAC. fibre contents from 0.6% to 0.9%. However, in this case, the com-

Table 9
Literatures in the database.

Literatures Ec f’c qcf/qd* numbers


(MPa) (MPa) (kg/m3)
Ahmad and Shah [28] 15240–18960 29.8–51.7 1545–1860 7
Almusallam and Alsayed [29] 9620–16800 16.6–52.3 1300 2
Balaguru and Foden [30] 14800–20100 22.4–35.1 1684–1810 5
Chi et al [31] 13300–23100 21.3–48.2 1899–2195 36
Cui et al [32] 13300–23100 21.3–48.2 1899–2195 12
Hanson [33] 10480–34956 18.6–73.6 1464–1951 40
Haque et al [34] 21991–29040 38–64.5 1775–1800 12
Hossain [35] 10000–14500 18–36 1734–2291 8
Kayali et al [6] 24,000 65 1939 1
Ke et al [36] 15733–28845 24.6–43.6 1513–2071 25
Kluge et al [37] 579–19354 0.7–48.6 689–1922 29
Nassif [38] 13200–17400 22.6–38.4 1756 12
Richart and Jensen [39] 5792–34956 3.2–90.1 1450–2244 271
Shah et al [40] 13353–19593 29–43 1882–2046 6
Shannag [41] 17457–22477 22.5–43.2 2025–2066 11
Shideler [42] 5723–23511 5.4–63.6 1443–1831 217
Slate et al [43] 10570–19060 19–56.7 1300 3
Topçu and Uygunoglu [12] 6639–18258 15.9–24.6 1711–1877 11
Wang et al [44] 11780–18640 23.4–55.5 1869–2029 6
Wilson and Malhotra [45] 23800–27000 33.6–60.8 1870–1950 6
Yang and Huang [46] 15800–24660 31.5–49.9 2023–2221 12
Zhang and Gjorv [47] 17800–25900 49.7–90.1 1595–1880 9
Pingjiang Li [48] 18.5–68.1 21–32.7 1750–1960 10(10)
Zhiqing Cheng [49] 52–56.1 33.3–34.4 1870–1950 3(3)
Hailong Wang [50] 23.33–33.98 23.51–27.7 1935–1950 9(9)
Jiajun Ye [51] 48.5–72.1 26.3–29.3 1960–1980 3(3)

(continued on next page)


264 X. Liu et al. / Construction and Building Materials 225 (2019) 256–272

Table 9 (continued)

Literatures Ec f’c qcf/qd* numbers


(MPa) (MPa) (kg/m3)
Ling Cheng [52] 61.7–63.5 31.5–34.7 – 3(0)
Yan Chen [53] 39.2–51.8 – 1836–1920 9(0)
Xiao Yu [54] 51–58.8 29.4–32.8 1832–1934 5(5)
Wenbin Li [55] 40.8–57.9 – 1630–1913 7(2)
Shuqiang Qu [56] 38.9–42.1 21–25 – 2(0)
Fazhou Wang [57] – 25.2–30.8 1930–1960 4(0)
Lianfa Chen [58] 51.9–57.8 33.2–34.5 – 3(0)
Zhixing Zeng [59] 22.4–30.7 16.3–17.6 – 5(0)
Aijun Zhang [60] 31.52–48.03 26.9–38.4 1848–1981 12(12)
Ze Shen [61] 38.8–53.4 16–23.7 1657.6–1784.4 11(11)
Lili Xu [62] 38.7–50.84 25.1–35.5 – 9(0)
Jingjun Li [63] 39.61–43 – – 6(0)
Summary 579–34956 0.7–90.1 689–2291 741

pressive strength of the specimens is reduced by about 20% to 25%. It can be found in Fig. 9(c) that the addition of steel fibres does
This can be explained considering that during the mixing of con- not affect the ascending part of the curves. Only the secant modu-
crete the fibres appear to congregate, which prevents the bubbles lus of the peak point decreases with an increase in the volume frac-
from floating upward, resulting in the porosity of concrete. More- tion of steel fibres. The slowest slope occurred in specimens made
over, the weak interfaces between the fibres and concrete produce with 0.6% by volume of steel fibres. In this case, the strain at peak
part of the strength loss. stress is 23.69% higher than that of the LC60 group, and the ulti-
mate strain is 101.31% higher than that of LC60. This suggests that
the volume fraction of 0.6% provides the most significant improve-
ment in the LWAC. In addition, the steel fibres decrease the peak
2.0 100 stress by approximately 3% to 19%.
Cubic compressive strength Experimental results have shown that the presence of fibres
fc/fcu
produces a significant increase in the ductility and crack resistance
80 of LWAC. Furthermore, the improvement provided by steel fibres is
1.5
Strength ratio fc/fcu

greater due to less strength loss and increased ductility caused by


Strength (MPa)

the incorporation of the steel fibre.


60
Axial compressive strength
1.0 3.2.5. Compression toughness
40 The importance of achieving a certain degree of ductility in the
elements of concrete structures has been widely acknowledged [6].
0.5 In this study, the compression toughness was used to assess the
20 ductility and the energy-absorption capacity during deformation
under compression.
0.0 0
The calculation recommended by ASTM C1018 [27] was
40 50 60 0.3C 0.6C C 3S 6S 9S employed to obtain the toughness values of the specimens,
LC LC LC - 0.9 -0. 60-0. 60-0.
60- C60 60- C60 C
L C L L C L L C L whereas in this work, the first feature point was replaced by the
Groups strain at peak stress, as follows:
R 3ec R
Fig. 13. The ratio of axial compressive strength to cube compressive strength of A1 þ A2 rde ðrc  ec Þ 03 ydx
LWAC. I¼ ¼ R0ec ¼ R ð1Þ
A1 0
rde ðrc  ec Þ 1 ydx 0

90 90
Test results Test results
80 2 80 2
y=0.78x R =0.92
y=0.92x R =0.96
70 70

60 60
50 50
fc ( M P a )
fc (MPa)

40 40
30 30
20 20
10 10
0 0
0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90 100
fcu (MPa) fcu(MPa)
(a) (b)
Fig. 14. Relations between fcu and fc of plain LWAC and fibre-reinforced LWAC: (a) Plain LWAC (b) Fibre-reinforced LWAC.
X. Liu et al. / Construction and Building Materials 225 (2019) 256–272 265

R ec R1 9
rde ðrc ec Þ ydx > The toughness and brittleness values were achieved by using
A1
R
B ¼ A2 ¼ 3ec 0
¼ R03 =
r de ðrc ec Þ ydx ð2Þ the stress–strain curves tested in this study. Table 8 contains the
ec 1
>
; toughness and brittleness indices of all specimens.
I ¼1þB 1
Fig. 11(a) shows the toughness and brittleness values of speci-
mens with strengths of 40 MPa, 50 MPa, and 60 MPa. It can be seen
where I represents the toughness indice, B represents the brittle- that the brittleness indice increased with an increase in concrete
ness indice, A1 and A2 represent the area enclosed by the stress– strength, but with a decrease in toughness indice.
strain curve and X-axis, as shown in Fig. 10. The brittleness and toughness values of specimens containing
0%, 0.3%, 0.6%, and 0.9% by volume of carbon fibres in Fig. 11(b)
were used to compare the effects of fibres on compression tough-
ness. The toughness indice was slightly reduced when a content of
0.3% of carbon fibres was included, and the toughness indice obvi-
ously increased with the increase in carbon-fibre volume fraction.
In contrast, the brittleness indice exhibits an opposite trend, where
firstly a slight increase and then a descending part can be observed.
Changes in both the brittleness and toughness values become slow
after 0.6% of fibre content was reached, whereby the brittleness
indice was decreased by 24.43% and the toughness indice was
increased by 20.1% compared with the plain LWAC.
The brittle and toughness values of the specimens with steel
fibres are shown in Fig. 11(c). A maximum improvement on the
toughness indice of specimens with steel fibres, up to about
32.52% compared with the plain LWAC, was obtained when 0.6%
of fibre content was used. Similarly, a maximum reduction on
the brittleness indice, up to about 34.26%, was achieved at 0.6%
of steel fibre content.
As can be seen in Fig. 12, the steel fibre produces a more signif-
icant improvement on the toughness value for specimens mixed
Fig. 15. Calculation of the elastic modules model (Ec) for LWAC.
with 0.3% and 0.6% by volume of fibres. However, with 0.9% of fibre
content, the effect of the carbon fibre is better than that of the steel
fibre.

4. Stress–strain model

4.1. Database

To develop a complete stress–strain model for plain and fibre-


reinforced LWAC, 842 groups of test results regarding the light-
weight aggregate concrete were organized in the present study.
The relevant information of literatures contained in the database
are given in Table 9.

4.2. Elastic modulus and peak conditions

4.2.1. Compressive strength


LWAC exhibits a different relationship that between its cubic
and axial compressive strength compared to NWC. Fig. 13 exhibits
Fig. 16. Calculation of the peak strain model for LWAC.
the trend lines of the axial compressive strength and cubic com-

Table 10
Existing stress–strain relationships for concrete in compression.

Models Fitting expressions Feature points Crucial parameters


 0 1=3 h i
Carreira (1985) bx   0
y¼ b1þxb Ec ¼ 10200 f c ec ¼ 0:71f 0c þ 168  105 x = e/ec, y = fc/f0c b ¼ 1= 1  ðf c =ec Ec Þ
 0 1=3 h i
Wee (1996) k1 bx  1=4 0
y¼k
1 b1þx 2
k b Ec ¼ 10200 f c ec ¼ 0:00078 f 0c b ¼ 1= 1  ðf c =ec Ec Þ When fc’50 MPa,k1 = k2 = 1;
When 50 MPa  fc’120 MPa, k1=(50/fc’)3,
k2=(50/fc’)1.3
 0 1=3
Yang (2014) y ¼ ðb 1 þ 1Þx
b þxb1 þ1 Ec ¼ 8470 f c ðwc =2300Þ1:17
Ascending:

1
0 0:67
h 0
i b1 ¼ 0:20exp 0:73ð10=f c Þ ðwc =2300Þ1:17
e0 ¼ 0:0016exp 240ðf c =Ec Þ h i
h  1:75 i 0 0:67
e0:5 ¼ 0:0035exp 1:2 ð10=f 0c Þðwc =2300Þ Descending: b1 ¼ 0:41exp 0:77ð10=f c Þ ðwc =2300Þ1:17

Guo (1982) Ascending: – Parameter a is a ratio that initial elastic modulus to secant
y ¼ ax þ ð3  2aÞx2 þ ða  2Þx3 elastic modulus.; Parameter b is determined according to
Dscending: y ¼ x
2
the concrete strength and constraint condition.
bðx1Þ þx
266 X. Liu et al. / Construction and Building Materials 225 (2019) 256–272

Fig. 17. Comparison between theoretical and experimental curves: (a) LC40 (b) LC50 (c) LC60 (d) LC60-0.3C (e) LC60-0.6C (f) LC60-0.9C (g) LC60-0.3S (h) LC60-0.6S (i) LC60-
0.9S.
X. Liu et al. / Construction and Building Materials 225 (2019) 256–272 267

pressive strength referring to different groups. In the absence of 4.2.2. Modelling of elastic modulus
fibres, the fc/fcu varied from 0.870 to 0.971, which was slightly Table 7 shows the initial elastic modulus obtained by the exper-
higher than that of normal concrete, which is about 0.7 to 0.92. imental results. It can be seen that the elastic modulus of LWAC
In contrast, the presence of fibres caused a clear difference can be improved by an appropriate amount of fibres. However, test
between the two trends. The fc/fcu of carbon-reinforced LWAC results of LC60-0.9C and LC60-0.9S show that an excessive content
increased with an increase of fibres, from approximately 0.716 to of fibres may lead to a reduction of the elastic modulus, which is
0.810, whereas the fc/fcu of concretes mixed with steel fibres attributed to the reduction of the bond performance between con-
decreased firstly and then increased, and the ratio was approxi- crete and fibres.
mately 0.712–0.892, which is close to that of normal concrete. Several studies have pointed out that the elastic modulus of
In order to obtain an accurate conversion relationship between LWAC is significantly influenced by factors such as the concrete
axial compressive strength and cubic compressive strength of strength and fresh density, but indistinctively influenced by factors
LWAC, a regression analysis using collecting test data and data in such as section size and curing methods. Accordingly, based on 842
this study was conducted, and the results are presented in test data of plain LWAC and fibre-reinforced LWAC, the influences
Fig. 14. The relationships between cubic and axial compressive of the cylinder compressive strength (fc’) and concrete density (qcf)
strength are expressed as on the elastic modulus (Ec) of LWAC were statistically quantified
Plain concrete: using a multivariable regression analysis, which resulted in the
expression given in Eq. (5)
f c ¼ 0:92f cu ð3Þ
 1:146
Fibre-reinforced LWAC: Ec ¼ 5681:67ðf c 0 Þ
0:403
qcf =2250 ð5Þ

f c ¼ 0:78f cu ð4Þ where qcf represents the concrete fresh density as presented by
2 Shannag [41], which is taken as 0.91qd + 344.1.
The correlation coefficients R of Eqs. (3) and (4) are 0.96 and
This expression is applicable when fc’90 MPa and 689 kg/
0.92, respectively, which means that the predictions of the pro-
m3  qcf  2291 kg/m3. Fig. 15 shows that the predictions of the
posed expressions are in good agreement with the experimental
proposed expression (Eq. (5)) are in good agreement with the
results. The difference between two conversions were considerably
experimental results.
attributed to the effect that fibres intended to parallel to speci-
The elastic modulus predicted by Eq. (5) was compared with the
men’s longer side when the axial specimens were cast by horizon-
test results, as listed in Table 7, and it can be seen that ratios that
tal, which causes reduction in the confinement of transvers
calculation to the test results RE are greater than 0.79 and variance
deformation provided by fibres. The Table 7 compares the axial
is 0.002, respectively, which indicates that the predictions are in a
compressive strength predicted by Eqs. (3) and (4) with the exper-
reasonable agreement with the test results.
imental results, in which the mean of ratio that calculations to the
test results Rf and variance are 1.01 and 0.006, respectively; thus,
the predictions are in good agreement with the experimental 4.2.3. Modeling of strain corresponding to peak compressive stress
results. As one of crucial parameters in the concrete uniaxial compres-
sive constitutive model, the strain at peak stress is influenced by
several factors such as concrete strength, size of section, loading
Table 11 rate, and constraint condition. The strain of LWAC is greater than
Expressions of parameters in modified stress–strain model.
that of normal concrete at comparable strength levels, which is
Materials Ascending branch Descending branch attributable to the lower elastic modulus of LWAC, and thus, the
Plain LWAC a ¼ 1:797Ec =Ep  1:264 3:062
k1 ¼ ð37:15=f c Þ
existing expressions for normal concrete are no longer applicable.
1:544
k2 ¼ ð41:25=f c Þ
Therefore, strain at compressive strength of LWAC was regressed
Carbon-fibre k1 ¼ 1:343r  þ0:108f c  6:811 as
reinforced k2 ¼ 0:715r þ 0:048f c  2:538
LWAC ec ¼ 3:496  10 - 5  f 0c þ 0:001 ð6Þ
Steel-fibre k1 ¼ 0:220r  0:017f c þ 1:758
reinforced k2 ¼ 0:299r þ 0:014f c  0:505 where fc’ represents the cylinder compressive strength, and is taken
LWAC as 0.8fcu(1 0 0), presented by Shannag [41]. Fig. 16 shows that the
*
Ep and r represent the secant elastic modulus and the volume fraction of fibre, predictions of the proposed expression (Eq. (6)) are in good agree-
respectively ment with the experimental results.

Table 12
Parameters regressed by statistically stochastic damage constitutive model.

Groups Ec/MPa fc/MPa ec/le m g c Ascending branches Descending branches


h i h   e  2 i
LC40 27,982 49.8 2210 4.63 0.0031 0.607 r ¼ 27982eexp 0:216ðe=0:002210Þ4:63 49:8exp 0:5 ln 0:00221 =0:607
h i h   e  2 i
LC50 29,769 59.0 2478 4.48 0.0035 0.546 r ¼ 29769eexp 0:223ðe=0:002478Þ4:48 59:0exp 0:5 ln 0:002478 =0:546
h i h   e  2 i
LC60 32,122 80.8 3056 5.12 0.0042 0.527 r ¼ 32122eexp 0:195ðe=0:003056Þ5:12 80:8exp 0:5 ln 0:003056 =0:527
h i h   e  2 i
0.3C 33,117 61.9 3133 1.94 0.0044 0.533 r ¼ 33117eexp 0:517ðe=0:003133Þ41:94 61:9exp 0:5 ln 0:003133 =0:533
h i h   e  2 i
0.6C 34,806 65.0 3433 1.64 0.0046 0.883 r ¼ 34806eexp 0:609ðe=0:003433Þ1:64 65:0exp 0:5 ln 0:003433 =0:883
h i h   e  2 i
0.9C 29,180 60.3 3500 1.90 0.0049 0.876 r ¼ 29180eexp 0:528ðe=0:003500Þ1:90 60:3exp 0:5 ln 0:003500 =0:876
h i h   e  2 i
0.3S 32,315 78.5 3552 2.63 0.0051 0.794 r ¼ 32315eexp 0:380ðe=0:003552Þ2:63 78:5exp 0:5 ln 0:003552 =0:794
h i h   e  2 i
0.6S 32,662 65.5 3780 1.58 0.0050 1.317 r ¼ 32662eexp 0:634ðe=0:003780Þ1:58 65:5exp 0:5 ln 0:003780 =1:317
h i h   e  2 i
0.9S 32,327 67.3 3688 1.75 0.0051 0.944 r ¼ 32327eexp 0:572ðe=0:003688Þ1:75 67:3exp 0:5 ln 0:003688 =0:944
268 X. Liu et al. / Construction and Building Materials 225 (2019) 256–272

Eq. (6) is used to predict the peak strain. The ratios of predicted in the present study. It was found that the models proposed by
to test results are included in Table 7. It is observed that the ratio Guo and Wee presented the values of R2a and R2d greater than
that calculations to the test results Re are within 0.88–1.17 and the 0.989 and 0.84 for ascending and descending parts of the curves,
variance is 0.011, respectively; thus, the predictions are in agree- whereas the models suggested by Carreira and Yang yielded R2a
ment with the test results. Furthermore, the proposed expression and R2d less than 0.97 and 0.67. Given the better interpretation
satisfies the requirement of actual engineering owing to its about the experimental data, as shown in Fig. 17, the models
effectiveness. proposed by Wee and Guo were adopted and modified to estab-
lish the new analytical model for plain and fibre-reinforced
4.3. Stress–strain relation LWAC.
Considering the discussions above and the characteristics of
Extensive research has been conducted to study the mechan- fibre-reinforced LWAC, a modified stress–strain model of LWAC
ical properties of LWAC. Table 10 presents equations for con- was developed by modifying parameters in existing models. In
crete in compression developed by Carreira, Wee, Yang, and order to obtain a good accuracy and a simpler apparent form, the
Guo, respectively, and the models selected in Table 10 were equation for the ascending branch of the stress–strain curve is
used to predict the stress–strain response of the specimens used expressed as

60 1.2
Test results Damage parameter of LC40
Proposed model
50 1.0

Damage parameter
40 0.8
Stress (MPa)

30 0.6

20 0.4

10 0.2

0 0.0
0 1000 2000 3000 4000 5000 6000 7000 8000 0 1000 2000 3000 4000 5000 6000 7000 8000
-6
Strain (×10 mm/mm) (a) Strain (×10-6 mm/mm) (b)
70 1.2
Test results Damage parameter of LC50
60 Proposed model
1.0
Damage parameter

50
0.8
Stress (MPa)

40
0.6
30
0.4
20

10 0.2

0 0.0
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 0 2000 4000 6000 8000
-6 -6
Strain (×10 mm/mm) (c) Strain (×10 mm/mm) (d)
90 1.2
Test results Damage parameter of LC60
80 Proposed model
1.0
70
Damage parameter

60 0.8
Stress (MPa)

50
0.6
40
30 0.4
20
0.2
10
0 0.0
0 2000 4000 6000 8000 10000 0 2000 4000 6000 8000 10000
-6 -6
Strain (×10 mm/mm) (e) Strain (×10 mm/mm) (f)
Fig. 18. Stress–strain curves of plain LWAC and the evolution of damage parameter: (a) Stress–strain curves of LC40 (b) The evolution of damage parameter of LC40 (c)
Stress–strain curves of LC50 (d) The evolution of damage parameter of LC50 (e) Stress–strain curves of LC60 (f) The evolution of damage parameter of LC60.
X. Liu et al. / Construction and Building Materials 225 (2019) 256–272 269

y ¼ ax þ ð3  2aÞx2 þ ða  2Þx3 ð7Þ the ascending and descending branches were over 0.98; thus, the
proposed model is capable of accurately predicting the uniaxial
which is obtained by the model proposed by Guo (1982), and the compressive stress–strain curves of fibre-reinforced LWAC.
equation for the descending branch is suggested by Wee (1996)
4.4. Statistically stochastic damage constitutive model
k1 bx
y¼ ð8Þ
k1 b  1 þ xk2 b Since Dougill introduced the conception of damage into the
where x, y, and b are determined according to Table 8. The peak concrete theory [64], many studies [65–70] focusing on the dam-
strain ec is determined according to Eq. (6) and the initial elastic age mechanics have been carried out. Based on the strain equiva-
modulus could be given by Eq. (5). lence assumption obtained by Benallal et al [71], the uniaxial
Based on the equations proposed above, expressions of the stress–strain of concrete is depicted in this work as
parameters (a, k1 and b) used in Eq. (7) and Eq. (8) were achieved
r ¼ ð1  DÞe ð9Þ
by using a regression analysis, as shown in Table 11. It can be con-
cluded that the predictions of the proposed model are in good where the parameter D represents the damage parameter, and
agreement with the test results. The correlation coefficients of both r and e represent the stress and the strain, respectively.

70 1.2
Test results Damage parameter of 0.3C
60 Proposed model
1.0

Damage parameter
50
0.8
Stress (MPa)

40
0.6
30
0.4
20

10 0.2

0 0.0
0 2000 4000 6000 8000 10000 0 4000 8000 12000 16000
-6 -6
Strain (×10 mm/mm) (a) Strain (×10 mm/mm) (b)
80 1.2
Test result Damage parameter of 0.6C
70 Proposed model
1.0
D a m a g e p a ra m e te r

60
0.8
S tr e s s ( M P a )

50

40 0.6

30
0.4
20
0.2
10

0 0.0
0 4000 8000 12000 16000 0 4000 8000 12000 16000
-6 -6
Strain (×10 mm/mm) (c) Strain (×10 mm/mm) (d)
70 1.2
Test results Damage parameter of 0.9C
60 Proposed model
1.0
D am ag e p a ram et e r

50
0.8
S tre s s ( M P a )

40
0.6
30
0.4
20

10 0.2

0 0.0
0 2000 4000 6000 8000 10000 12000 14000 16000 0 4000 8000 12000 16000
-6
Strain (×10 mm/mm) (e) Strain (×10-6 mm/mm) (f)
Fig. 19. Stress–strain curves of carbon-fibre-reinforced LWAC and the evolution of damage parameter: (a) Stress–strain curves of LC60-0.3C (b) The evolution of damage
parameter of LC60-0.3C (c) Stress–strain curves of LC60-0.6C (d) The evolution of damage parameter of LC60-0.6C (e) Stress–strain curves of LC60-0.9C (f) The evolution of
damage parameter of LC60-0.9C.
270 X. Liu et al. / Construction and Building Materials 225 (2019) 256–272

4.4.1. Ascending branch of constitutive model where m and g represent the shape and scale parameters, respec-
Before the strain corresponding to the peak stress was reached, tively, and both are greater than zero. According to Eqs. (10) and
the damage parameter D could be related to the strain by the Wei- (11), the parameter D could be expressed as
bull distribution as Z e 
m
e
D¼ uðxÞdx ¼ 1  exp  ð12Þ
dD
¼ uðeÞ ð10Þ 0 g
de
Substituting Eq. (12) into Eq. (9), the expression for the ascend-
where ing branch of the stress–strain curve would be written as

m1 
m 
m
m e e e
uðeÞ ¼ exp  ð11Þ r ¼ Eeexp  ; e  ec ð13Þ
g g g g

90 1.2
Test results Damage parameter of 0.3S
80 Proposed model
1.0
70

Damage parameter
60 0.8
Stress/Mpa

50
0.6
40
30 0.4
20
0.2
10
0 0.0
0 4000 8000 12000 16000 0 4000 8000 12000 16000
-6 -6
Strain (×10 mm/mm) (a) Strain (×10 mm/mm) (b)
80 1.2
Test results Damage parameter of 0.6S
70 Proposed model
1.0
60
Damage parameter
Stress (MPa)

0.8
50

40 0.6

30
0.4
20
0.2
10

0 0.0
0 2000 4000 6000 8000 10000 12000 14000 0 4000 8000 12000 16000
-6 -6
Strain (×10 mm/mm) (c) Strain (×10 mm/mm) (d)
80 1.2
Test results Damage parameter of 0.9S
70 Proposed model
1.0
60
D a m a g e p a ra m e te r
S tre s s (M P a )

0.8
50

40 0.6

30
0.4
20
0.2
10

0 0.0
0 4000 8000 12000 16000 0 4000 8000 12000 16000
-6 -6
Strain (×10 mm/mm) (e) Strain (×10 mm/mm) (f)
Fig. 20. Stress–strain curves of steel-fibre-reinforced LWAC and the evolution of damage parameter: (a) Stress–strain curves of LC60-0.3S (b) The evolution of damage
parameter of LC60-0.3S (c) Stress–strain curves of LC60-0.6S (d) The evolution of damage parameter of LC60-0.6S (e) Stress–strain curves of LC60-0.9S (f) The evolution of
damage parameter of LC60-0.9S.
X. Liu et al. / Construction and Building Materials 225 (2019) 256–272 271

The shape parameter m and the scale parameter g in Eq. (13) 4. A modified piecewise stress–strain model was developed based
can be solved by boundary conditions at the origin and the peak on the existing models. The proposed model is applicable to
point. plain LWAC, steel-fibre-reinforced LWAC, and carbon-fibre-
The m and g in Eq. (12) were calculated as follows reinforced LWAC, and it showed good match to the test results.
) 5. A statistically stochastic damage constitutive model based on
m ¼ 1=lnðEec =f c Þ
ð14Þ the Weibull distribution and the log-normal distribution was
g ¼ ð1=mecÞ1=m established in this study, and the parameter c in its descending
part was fitted with the test results. This model exhibits a good
agreement with the experimental results.
4.4.2. Descending branch of constitutive model
The logarithmic normal distribution was chosen to describe the
descending branch of the constitutive model:
"   2 !# Declaration of Competing Interest
ln be
r ¼ aexp 0:5 ð15Þ The authors declare there is no conflicts of interest regarding the
c
publication of this paper.
It was found that a and b in Eq. (15) are equal to the stress and
strain at the peak point of the stress–strain curve (fc, ec). Then, the Acknowledgements
statistically stochastic damage constitutive model is expressed as
8 h  m i
> The authors gratefully acknowledge the financial supports for
>
>
< E e exp  1
m
e
ec ; e  ec
this research by the National Natural Science Foundation of China
"
2 #
r¼ lnðeec Þ
ð16Þ (51878054, 51578072 and 51708036) and the Fundamental
>
> 0:5 ; e  ec
> f
: c exp c Research Funds for the Central Universities (300102288401).

The damage parameter is represented by References


8 h  m i
>
> e ; e  ec
< 1  exp  m ec
1 [1] Manu S. Nadesan, P. Dinakar, Influence of type of binder on high-performance
D¼  ln e 2 ð17Þ sintered fly ash lightweight aggregate concrete, Constr. Build. Mater. 176
>
> =Ee; e  ec
(2018) 665–675.
: 1  f c exp 0:5 cec [2] D.D. Theodorakopoulos, Narayan Swamy, Contribution of steel fibers to the
strength characteristics of lightweight concrete slab-column connections
failing in punching shear, Struct. J. 90 (4) (1993) 342–355.
[3] Daneti Saradhi Babu, Mechanical and deformational properties, and shrinkage
4.4.3. Comparison between statistically stochastic damage constitutive cracking behaviour of lightweight concretes. Diss. 2009.
model and the test results [4] Nicolas Ali Libre et al., Mechanical properties of hybrid fiber reinforced
lightweight aggregate concrete made with natural pumice, Constr. Build.
The parameter c in the descending branch of the stress–strain
Mater. 25 (5) (2011) 2458–2464.
curve was fitted using SPSS software and the experimental results. [5] Yining Ding, Wolfgang Kusterle, Compressive stress–strain relationship of steel
The stress–strain expressions calculated by the statistically fibre-reinforced concrete at early age, Cem. Concr. Res. 30 (10) (2000) 1573–
stochastic damage constitutive model for the specimens tested in 1579.
[6] O. Kayali, M.N. Haque, B. Zhu, Some characteristics of high strength fiber
this work are included in Table 12. reinforced lightweight aggregate concrete, Cem. Concr. Compos. 25 (2) (2003)
Figs. 18, 19, and 20 show the comparisons between the test 207–213.
results and the statistically stochastic damage constitutive model. [7] H.T. Wang, L.C. Wang, Experimental study on static and dynamic mechanical
properties of steel fiber reinforced lightweight aggregate concrete, Constr.
The proposed model shows a good agreement with the test results, Build. Mater. 38 (2013) 1146–1151.
and the damage parameter D in the proposed model is capable of [8] Jing Jun Li et al., Investigation on mechanical properties and microstructure of
predicting the process of damage evolution in concrete. high performance polypropylene fiber reinforced lightweight aggregate
concrete, Constr. Build. Mater. 118 (2016) 27–35.
[9] ACI 318–14, Building code requirements for reinforced concrete, American
5. Conclusions Concrete Institute, Farmington Hills, 2014.
[10] CEB-FIP, C. E. B. F. I. P., Model code 2010, Comite Euro-International du beton
(2010).
1. The porous microstructure of the LWAC aggregates causes more [11] S. Chandra, L. Berntsson, Lightweight aggregate concrete: science, technology,
weak-faces than in NWC, and thus, more cracks and split aggre- and applications, Noyes/William A Pub, United States, 2002.
gates were observed in the tests. For the fibre-reinforced LWAC, [12] I.B. Topcu, T. Uygunoglu, Effect of aggregate type on properties of hardened
self-consolidating lightweight concrete (SCLC), Constr. Build. Mater. 24 (2010)
the remnant capacity at the target displacement was higher 1286–1295.
owing to the fibre effect, and the ductility was certainly [13] Lucyna Domagała, Modification of properties of structural lightweight
improved by the fibres. concrete with steel fibres, J. Civ. Eng. Manag. 17 (1) (2011) 36–44.
[14] Mahmoud Hassanpour, Payam Shafigh, Hilmi Bin Mahmud, Lightweight
2. The carbon fibre produced a greater improvement on the ductil- aggregate concrete fiber reinforCem.–a review, Constr. Build. Mater. 37
ity of LWAC compared with steel fibres. The failure of plain (2012) 452–461.
LWAC and carbon-fibre reinforced LWAC under flexural load [15] Oğuz Akın Düzgün, Rüstem Gül, Abdulkadir Cüneyt Aydin, Effect of steel fibers
on the mechanical properties of natural lightweight aggregate concrete, Mater.
occurred suddenly, while the specimens with steel fibres exhib- Lett. 59 (27) (2005) 3357–3363.
ited a more ductile failure. The most optimal content of 0.6% [16] Domingo J. Carreira, Kuang-Han Chu, Stress-strain relationship for plain
was observed for both steel and carbon fibres. concrete in compression, J. Proc. 82 (6) (1985).
[17] T.H. Wee, M.S. Chin, M.A. Mansur, Stress-strain relationship of high-strength
3. Expressions for the prediction of peak stress, peak strain, and
concrete in compression, J. Mater. Civil Eng. 8 (2) (1996) 70–76.
elastic modulus were established by using a large set of exper- [18] Keun-Hyeok Yang et al., Stress-Strain Model for Various Unconfined Concretes
imental data. Furthermore, the accuracy of the proposed in Compression, ACI Struct. J. 111 (2014) 4.
expressions was verified using the results of the tests conducted [19] Z.H. Guo, X.Q. Zhang, D.C. Zhang, Experimental study on full stress-strain
curves of concrete, J. Build. Struct. 3 (1) (1982) 3–14 (in Chinese).
in this study. The proposed expressions are applicable to plain [20] Jian C. Lim, Togay Ozbakkaloglu, Stress–strain model for normal-and light-
LWAC, steel-fibre-reinforced LWAC, and carbon-fibre- weight concretes under uniaxial and triaxial compression, Constr. Build.
reinforced LWAC. Mater. 71 (2014) 492–509.
272 X. Liu et al. / Construction and Building Materials 225 (2019) 256–272

[21] Bing Han, Tian-Yu Xiang, Axial compressive stress-strain relation and Poisson [46] Chung-Chia Yang, Ran Huang, Approximate strength of lightweight aggregate
effect of structural lightweight aggregate concrete, Constr. Build. Mater. 146 using micromechanics method, Adv. Cem. Based Mater. 7 (3-4) (1998) 133–
(2017) 338–343. 138.
[22] M.C. Nataraja, N. Dhang, A.P. Gupta, Stress–strain curves for steel-fiber [47] Min Hong Zhang, Odd E. Gjvorv, Mechanical properties of high-strength
reinforced concrete under compression, Cem. Concr. Compos. 21 (5-6) lightweight concrete, Mater. J. 88 (3) (1991) 240–247.
(1999) 383–390. [48] Pingjiang Li, Xunbo Liu, Fundamental mechanical properties of concrete with
[23] M.A. Mansur, M.S. Chin, T.H. Wee, Stress-strain relationship of high-strength high strength expanded shale, J. Build. Mater. 7 (1) (2004) 113–116 (in
fiber concrete in compression, J. Mater. Civil Eng. 11 (1) (1999) 21–29. Chinese).
[24] CS (Chinese Standard) GB/T 1743.2-2010, Lightweight aggregates and its test [49] Zhiqing Cheng, Experimental research on high performance shale light
methods——part 2: test methods for lightweight aggregates, Ministry of aggregate concrete, Central South University, 2007 (in Chinese).
Construction of the People’s Republic of China and Quality Supervision [50] Hailong Wang, Experimental research on early mechanics and freeze
Inspection and Quarantine of the People’s Republic of China, 2010 (in Chinese). resistance of light aggregate concrete, Inner Mongolia Agricultural
[25] Jianming Gao, Wei Sun, Keiji Morino, Mechanical properties of steel fiber- University, 2009 (in Chinese).
reinforced, high-strength, lightweight concrete, Cem. Concr. Compos. 19 (4) [51] Jiajun Ye, Optimum design and performance study of high strength
(1997) 307–313. lightweight aggregate concrete components, Wuhan University of
[26] Chenkui Huang, Guofan Zhao, Properties of steel fibre reinforced concrete Technology, 2005 (in Chinese).
containing larger coarse aggregate, Cem. Concr. Compos. 17 (3) (1995) 199– [52] Ling Cheng, Mix design and performance study of LC50 lightweight aggregate
206. concrete, Changsha University of Technology, 2013 (in Chinese).
[27] ASTM, Standard test method for flexural toughness and first-crack strength of [53] Yan Chen, Study on mix proportion design and performance of high strength
fiber-reinforced concrete (using beam with third-point loading). C-1018 lightweight aggregate concrete, Jilin University, 2007 (in Chinese).
(1997). [54] Yu Xiao, Preparation of high strength shale ceramsite and study on its concrete
[28] Shuaib H. Ahmad, Surendra P. Shah, Behavior of hoop confined concrete under properties, Chongqing University, 2004 (in Chinese).
high strain rates, J. Proc. 82 (5) (1985). [55] Wenbin Li, Experimental research on high performance concrete with ceramic
[29] T.H. Almusallam, S.H. Alsayed, Stress-strain relationship of normal, high- light aggregate, Xi’an University of Architectural Science and Technology, 2012
strength and lightweight concrete, Mag. Concr. Res. 47 (1995) 170. (in Chinese).
[30] Perumalsamy Balaguru, Andrew Foden, Properties of fiber reinforced [56] Qu Shuqiang, Experimental research on mechanical properties and stress
structural lightweight concrete, ACI Struct. J. 93 (1) (1996) 62–78. behavior of fly ash ceramic concrete slab, Inner Mongolia University of Science
[31] J.M. Chi et al., Effect of aggregate properties on the strength and stiffness of and Technology, 2012 (in Chinese).
lightweight concrete, Cem. Concr. Compos. 25 (2) (2003) 197–205. [57] Fazhou Wang, Research and Application of High Performance Lightweight
[32] H.Z. Cui et al., Experimental investigation and development of analytical Aggregate Concrete, Wuhan University of Technology, 2003 (in Chinese).
model for pre-peak stress–strain curve of structural lightweight aggregate [58] Falian Chen, Yue Chen, Long Li, Study on mechanical properties of high
concrete, Constr. Build. Mater. 36 (2012) 845–859. performance lightweight aggregate concrete, Silicate Bulletin 34 (10) (2015)
[33] J.A. Hanson, Shear strength of lightweight reinforced concrete beams, J. Proc. 2822–2828 (in Chinese).
55 (9) (1958). [59] Zhixing Zeng, Experimental study on mechanical properties and damage
[34] M.N. Haque, H. Al-Khaiat, O. Kayali, Strength and durability of lightweight fracture analysis of steel fiber lightweight aggregate concrete, Tianjin
concrete, Cem. Concr. Compos. 26 (4) (2004) 307–314. University, 2003 (in Chinese).
[35] Khandaker M. Anwar Hossain, Properties of volcanic pumice based cement and [60] Aijun Zhang, Experimental research on the physical mechanics and toughness
lightweight concrete, Cem. Concr. Res. 34 (2) (2004) 283–291. of steel fiber light aggregate concrete, Inner Mongolia Agricultural University,
[36] Y. Ke et al., Influence of volume fraction and characteristics of lightweight 2008 (in Chinese).
aggregates on the mechanical properties of concrete, Constr. Build. Mater. 23 [61] Ze Shen, Experimental study on basic mechanical properties of steel fiber
(8) (2009) 2821–2828. lightweight aggregate fly ash concrete, North China University of Water
[37] R.W. Kluge, M.M. Sparks, E.C. Tuma, Lightweight aggregate concrete, J. Proc. 45 Resources and Electric Power, 2015 (in Chinese).
(5) (1949) 625–642. [62] Xu Lili, Experimental study on mechanical properties and microstructure of
[38] Hani H. Nassif, Husam Najm, Nakin Suksawang, Effect of pozzolanic materials fiber lightweight aggregate concrete, Inner Mongolia Agricultural University,
and curing methods on the elastic modulus of HPC, Cem. Concr. Compos. 27 (6) 2012 (in Chinese).
(2005) 661–670. [63] Jingjun Li, Experimental research on mechanical properties and
[39] Frank Erwin Richart, Vernon Peter Jensen, Tests of plain and reinforced microstructures of plastic steel fiber light aggregate concrete, Inner
concrete made with haydite aggregates; a report of an investigation conducted Mongolia University of Science and Technology, 2015 (in Chinese).
by the Engineering Experiment Station, University of Illinois in coo peration [64] John W. Dougill, On stable progressively fracturing solids, Zeitschrift für
with the Western Brick Company. University of Illinois at Urbana Champaign, angewandte Mathematik und Physik ZAMP 27 (4) (1976) 423–437.
College of Engineering. Engineering Experiment Station. 1931. [65] S. Kandarpa, D.J. Kirkner, B.F. Spencer Jr., Stochastic damage model for brittle
[40] Surendra P. Shah, A.E. Naaman, J. Moreno, Effect of confinement on the materials subjected to monotonic loading, J. Eng. Mech. 122 (8) (1996) 788–
ductility of lightweight concrete, Int. J. Cem. Compos. Lightweight Concr. 5 (1) 795.
(1983) 15–25. [66] Jie Li, Xiaodan Ren, Stochastic damage model for concrete based on energy
[41] M.J. Shannag, Characteristics of lightweight concrete containing mineral equivalent strain, Int. J. Solids. Struct. 46 (11-12) (2009) 2407–2419.
admixtures, Constr. Build. Mater. 25 (2) (2011) 658–662. [67] Jacky Mazars, A description of micro-and macroscale damage of concrete
[42] Joseph J. Shideler, Lightweight-aggregate concrete for structural use, J. Proc. 54 structures, Eng. Fract. Mech. 25 (5-6) (1986) 729–737.
(10) (1957). [68] Jacky Mazars, Gilles Pijaudier-Cabot, Continuum damage theory—application
[43] Arthur H. Nilson, Salvador Martinez, Mechanical properties of high-strength to concrete, J. Eng. Mech. 115 (2) (1989) 345–365.
lightweight concrete, J. Proc. 83 (4) (1986). [69] J.C. Simo, J.W. Ju, Strain-and stress-based continuum damage models—I.
[44] P.T. Wang, S.P. Shah, A.E. Naaman, Stress-strain curves of normal and Formulation, Math. Comp. Model. 12 (3) (1989) 378.
lightweight concrete in compression, J. Proc. 75 (11) (1978). [70] J.C. Simo, J.W. Ju, Strain-and stress-based continuum damage models—II.
[45] H.S. Wilson, V.M. Malhotra, Development of high strength lightweight Computational aspects, Int. J. Solids. Struct. 23 (7) (1987) 841–869.
concrete for structural applications, Int. J. Cem. Compos. Lightweight Concr. [71] A. Benallal, R. Billardon, J. Lemaitre, Failure analysis of structures by
10 (2) (1988) 79–90. continuum damage mechanics, Fract. (1984) 3669–3676.

Vous aimerez peut-être aussi