Construction and Building Materials 146 (2017) 283–296

Contents lists available at ScienceDirect

Construction and Building Materials

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

Influence of superabsorbent polymer on shrinkage properties of reactive

powder concrete blended with granulated blast furnace slag
Mo Jinchuan a,⇑, Ou Zhongwen a, Zhao Xuxin b, Liu Jinming a, Wang Yahui c
a
Department of Chemical and Material Engineering, Logistical Engineering University, China
b
China Aerodynamics Research and Development Center, China
c
Laixi Municipal Bureau of Quality and Technical Supervision, China

h i g h l i g h t s

Autogenous shrinkage of RPC decreased as the content of GBFS increasing.

IC by means of SAP further reduced autogenous shrinkage.

Drying shrinkage became serious as the total W/B and content of GBFS increasing.

The influence of IC on drying shrinkage was not obvious.

The total shrinkage of samples with GBFS and IC was lower than references.

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

Article history: This study involved investigating the contributions of granulated blast furnace slag (GBFS) and internal
Received 17 October 2016 curing (IC) by means of superabsorbent polymer (SAP) on shrinkage behaviour, hydration heat, and
Received in revised form 27 March 2017 mechanical strength of reactive powder concrete (RPC). The results indicated that autogenous shrinkage
Accepted 12 April 2017
decreased with an increase in GBFS content, and a combination with IC completely mitigated autogenous
Available online 21 April 2017
shrinkage and even led to a net expansion. The mechanical strength decreased with the addition of GBFS.
However, compressive strength still exceeded 100 MPa even when the replacement level of GBFS corre-
Keywords:
sponded to a maximum of 50%. Although IC by means of SAP slightly reduced the strength, this reduction
Reactive powder concrete
Superabsorbent polymer
was acceptable given the important role that it played on mitigating autogenous shrinkage. The drying
Shrinkage shrinkage and mass loss increased significantly with an increase in the GBFS content. The IC did not exhi-
Granulated blast furnace slag bit an obvious effect on drying shrinkage and mass loss. However, it counteracted high drying shrinkage
by significantly reducing autogenous shrinkage. The total shrinkage of RPC with IC and GBFS was signif-
icantly lower than that without IC and GBFS. The combination of GBFS and IC can adequately solve seri-
ous shrinkage deformation and provide an environmental friendly and low cost method to produce RPC.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction caused by a combination of high autogenous shrinkage and drying

RPC is a type of ultra-high performance concrete that is devel-
oped through microstructural enhancement techniques [1,2]. It
generally exhibits high mechanical properties, high impact resis-
tance, and high durability [3–6]. Furthermore, it also exhibits good
repair and retrofit potential for concrete structures [7]. However,
the production of RPC involves the consumption of large amount
of cements [2], and this increases cost but and leads to high energy
consumption and serious pollution. Additionally, high content of
silica fume makes RPC prone to cracking at early ages, and this is
⇑ Corresponding author.
E-mail address: 728500246@qq.com (J. Mo).
shrinkage. RPC typically experiences significant self-desiccation
exceeding those of ordinary concrete due to its extremely low
water to binder ratio (W/B). Autogenous shrinkage is significantly
related to self-desiccation [8,9]. Therefore, it is necessary to main-
tain internal relative humidity to alleviate self-desiccation to mit-
igate high autogenous shrinkage of RPC. However, RPC is
characterized by dense microstructure and extremely low perme-
ability [5]. A traditional external curing method is not suitable
for RPC. It is necessary to apply other curing options on RPC to
reduce shrinkage deformation.
An effective method involves implementing internal curing (IC)
by means of superabsorbent polymers (SAP). The SAP covalently
cross-links by copolymerization of acrylic acid and acrylamide,

http://dx.doi.org/10.1016/j.conbuildmat.2017.04.105
0950-0618/Ó 2017 Elsevier Ltd. All rights reserved.
284 J. Mo et al. / Construction and Building Materials 146 (2017) 283–296
which is proven as effective with respect to applications in con-
crete due to its good chemical stability and high swollen ability
in a strong alkaline saline concrete pore solution [10,11]. The SAP
can absorb mixing water and act as a water-reservoir. The water
stored in SAP is released as cement hydration occurs through
which the decrease in internal relative humidity is delayed and
self-desiccation is partially mitigated [12]. Schröfl et al. [13]
reported a reduction in the autogenous shrinkage of high perfor-
mance concrete due to the presence of SAP. Justs et al. [14] mea-
sured a maximum shrinkage of 650 lm/m for ultra-high
performance concrete at 28 d. The shrinkage reduced to approxi-
mately 150 lm/m after the addition of SAP. Wang et al. [15] argued
that the water released by SAP promotes hydration of its surround-
ing cement paste, reduces porosity, and enhances the micro hard-
ness of the affected zone. Jensen et al. [16] found that the
autogenous shrinkage of cement paste was successfully mitigated
by SAP and even exhibited an amount of net expansion.
The high content of cement causes high hydration heat and high
temperature differential between the surface and core of an element
[17]. Thermal cracking occurs due to thermal stress caused by the
differential. These problems can be solved by replacing partial
cement with mineral admixtures. Granulated blast furnace slag
(GBFS) is a common admixture that is used in the construction
industry. It is a by-product of the production of pig iron that exhibits
a fine, granular, and glassy form and possesses latent hydraulic
properties [18,19]. Concrete blended with GBFS often exhibits a
low temperature increase due to the low content of C3S and C3A in
the compound composition of the binder [20–22]. Additionally, it
can improve the workability of mixtures and act as a micro-filler
to optimize the pore structure of a hardening matrix [23]. Previous
studies also examined the influence of GBFS on the shrinkage of con-
crete. Snoeck [24] reported that early age autogenous shrinkage was
reduced by the addition of GBFS. A study by Huo indicated that the
shrinkage of high performance concrete containing GBFS was even
lower than that of normal concrete [25]. However, a few previous
studies reported that the drying shrinkage of mortar and concrete
increased with the addition of GBFS. The studies revealed that the
samples exhibited higher drying shrinkage strains when the content
and fineness of GBFS increased [26,27]. The fore-mentioned results
are normally obtained from an experiment on conventional con-
crete or high performance concrete. There is a paucity of extant
research examining the effect of GBFS on the shrinkage of RPC.
Moreover, studies examining the effects of SAP on the autogenous
and drying shrinkage of RPC containing GBFS are even more rare.
The present study involved designing an experimental program
to assess the influence of SAP on autogenous and drying shrinkage
of RPC containing GBFS that is cured at room temperature. The
hydration heat of pure paste and the mechanical properties of
RPC were also analysed in the study.
2. Experimental program
2.1. Materials
The raw materials used in this study included Portland cement
(PC), silica fume (SF), GBFS, quartz sand, quartz powder, SAP, and a
superplasticizer (SP). A brass-coated steel fibre (13 mm long with a
0.2 mm diameter) was employed to enhance the ductility of RPC.
A PC (cementⅡ42.5R) complied with China National Standards
GB175-2007 [28] was used, and its chemical composition is pre-
sented in Table 1. Additionally, SF with SiO2 content exceeding
94% was used in the study. It exists in the form of nattier blue pow-
der. The GBFS used in the study was produced in Jiangsu Province,
China. The particle size distribution and chemical composition of
the binders (cement, SF, and GBFS) are listed in Fig. 1 and Table 1,
respectively.
Quartz sand with a particle size distribution in the range of
0.075–0.6 mm was used as an aggregate. Quartz powder produced
in Guangdong, China was added as a micro-filler. The mean particle
diameter of the quartz powder corresponded to 5.47 lm, and 97%
of the particles were less than 18 lm. Additionally, SAP with a dry
bulk density of 0.85 g/cm3 was manufactured by employing a bulk
polymerisation technique and subsequently crushed in the form of
an irregular particle shape. The particle size varied from 150 lm to
180 lm. A commercial SP with high range water reducing ability
was employed in the study. In order to eliminate the influence of
SP on the setting time of RPC mixtures and hydration heat of RPC
pure pastes, its dosage in all the samples was maintained at a rate
of 2.4% of the binder mass.
2.2. Mix proportions and mixing procedure
A series of samples were prepared to study the effect of GBFS
and SAP on the properties of RPC. The W/B of reference C-17 corre-
sponded to 0.17. Furthermore, GBFS was added at four dosage
levels, and the amount of SAP was varied as 0.24%, 0.4% and
0.64% with respect to the weight of binders in the study. Corre-
spondingly, the ratios of water entrained by SAP to that if the bin-
der (W/B)e corresponded to 0.03, 0.05, and 0.08 respectively, and
this indicated that the total W/B ((W/B)T) of specimens that
employed IC varied from 0.20 to 0.25. For every samples employed
IC by means of SAP, a corresponding sample with the same total W/
B and same amount of GBFS was considered as a control. The
details of all mixture composition are presented in Table 2. In the
table, the mix nomenclature is as follows: when C-IC-20 is consid-
ered as an example, the first part denotes that the specimen does
not contain GBFS, the middle part denotes internal curing by
SAP, the last part denotes the (W/B)T. With respect to the B15-IC-
20, the first part denotes the mass replacement ratio with GBFS
and the last two parts denote the same as that in the C-IC-20.
The abbreviation without ‘‘-IC-” specifies that this specimen does
not involve conducting internal curing by SAP.
The mixing procedure was an important factor that significantly
influenced the properties of cementitious materials [29]. The mix-
ing method of RPC is different from that of normal concrete consid-
ering its special composition. Based on previous studies, the
mixing procedure adopted by this paper is as follows:
(1) All dry mixing powders (cement, SF, GBFS, quartz powder,
quartz sand) were added and mixed for 2 min (noticed that
for the samples employed IC, the SAP should be added in dry
form and dry mixed with powder).
(2) During the dry mixing process, the steel fibre was added.
(3) 50% of water containing 50% SP was added; and wet mixing
for 3 min.
(4) The rest of water and SP were added.
(5) Mixing for 5 min.
When the mixing sequences were completed, the mixture was
poured into specific moulds for the corresponding experiment.
The setting time was measured in RPC mixtures without fibres.
The isothermal calorimetry test was measured on pure pastes in
the study. The preparation of pure pastes was similar to that of
RPC with the exception of those for the sequence (1) and (2). The
quartz powder, quartz sand, and steel fibre did not necessitate pure
pastes.
2.3. Methods
2.3.1. Setting time
With respect to a restrained concrete element, the deformation
prior to initial setting exerts a negligible influence on stress
 J. Mo et al. / Construction and Building Materials 146 (2017) 283–296 285

Table 1
Chemical composition of experiment materials (%).

CaO SiO2 Al2O3 Fe2O3 MgO Na2O MnO K2O SO3 Loss Pozzolanic activity indices (%) Fineness (kg/m2)
Cement 64.13 21.43 2.24 3.78 2.07 0.78 2.25 3.32 360
SF 0.52 94.51 0.61 0.22 4.14 112 470
GBFS 45.75 31.4 12.3 0.79 5.25 0.42 0.51 0.37 2.32 0.89 97 26,500

2.3.2. Autogenous shrinkage test

The linear autogenous shrinkage of RPC was measured by using
eddy-current displacement sensor (ECDS) based on the electro-
magnetic induction effect [30]. A schematic illustration of the
apparatus is shown in Figs. 2 and 3. The operation sequences are
as follows:

(1) The inner surfaces of steel mould

(100 mm  100 mm  515 mm) were smeared with grease
and then a Teflon sheet was subsequently placed to elimi-
nate friction between the inner surfaces of the mould and
RPC specimens.
(a) Cement (2) Two steel target seats were placed on the proper position of
the bottom surface of moulds, and the distance of the two
steel targets was set as 455 mm. Standard targets were mag-
netically attached to the steel target seats.
(3) The mixtures were cast into moulds and the steel target
seats were embedded into RPC specimens to ensure that it
deformed simultaneously with the RPC. Subsequently, the
thermocouples were embedded at the centre of RPC mix-
tures to separate the autogenous shrinkage from the mea-
sured total deformation by monitoring the temperature
change within RPC.
(4) The top surface of RPC mixture was sealed with two layers of
polyethylene sheets to avoid the influence of exterior drying.
This was followed by fastening the sensor support on the top
of the mould and regulating the fitting screw to place the
(b) Silica fume ECDS at the proper position.
(5) The test then commenced. The interval of measurement data
acquisition was set as 1 min.

During the complete testing period, the temperature was main-

tained at 20 ± 1 °C. After the test, the measured autogenous shrink-
age was zeroed at the initial setting time as determined by
corresponding mixtures. The deformation results were obtained
by measuring three specimens in parallel to ensure a better esti-
mation of autogenous shrinkage.

2.3.3. Isothermal calorimetry

(c) GBFS
Fig. 1. Particle size distribution of powder materials.
development. Therefore, the initial setting time is typically consid-
ered as the start point of autogenous shrinkage [30,31]. The initial
setting time of RPC mixtures was measured by a penetration resis-
tance method prescribed in ASTM C403 [32]. The accuracy of initial
setting time is very important for the results of the autogenous
shrinkage test due to the high deformation rate of specimens
around this point. However, the penetration resistance test is unre-
liable around this point, and thus in order to obtain a relatively
accurate result, it is necessary to perform the setting time test in
triplicate and to carefully compare the results. In order to eliminate
the influence of water evaporation on the setting time, the surface
of samples was covered by polyethylene sheets during the
measurement.
The hydration heat flow of pure paste was measured by an
eight-channel TAM Air heat conduction calorimeter. The very low
W/B necessitated high mixing energy to ensure that the paste is
uniform. Hence, the paste in the study was mixed externally, and
as a result, the initial heat peak immediately after the addition of
water to binder was missed. The mixing was followed by weighing
approximately 10 g pure paste and pouring it into a plastic
ampoule. The ampoule was then sealed and placed into a calorime-
ter. In order to minimize the influence of an increase in tempera-
ture caused by external mixing and handling of samples, the
collection of hydration heat data commenced 1 h after mixing.
The test lasted for 72 h, and the operating temperature was main-
tained at 20 °C throughout procedure. Finally, the results were nor-
malized by a gram of paste.
2.3.4. Drying shrinkage and mass loss
Three RPC prisms with dimensions corresponding to
100 mm  100 mm  515 mm were produced to measure drying
286 J. Mo et al. / Construction and Building Materials 146 (2017) 283–296

Table 2
The mix proportion of RPC.

Sample Binder (mass fraction) Quartz powder Quartz sand W/Ba (W/B)ea (W/B)Ta SAPb SPb STc%
Cement GBFS SF
C-17 0.8 0 0.2 0.32 0.88 0.17 0.17 0.024 1
C-20 0.8 0 0.2 0.32 0.88 0.2 0.2 0.024 1
C-IC-20 0.8 0 0.2 0.32 0.88 0.17 0.03 0.2 0.0024 0.024 1
C-22 0.8 0 0.2 0.32 0.88 0.22 0.22 0.024 1
C-IC-22 0.8 0 0.2 0.32 0.88 0.17 0.05 0.22 0.004 0.024 1
C-25 0.8 0 0.2 0.32 0.88 0.25 0.25 0.024 1
C-IC-25 0.8 0 0.2 0.32 0.88 0.17 0.08 0.25 0.0064 0.024 1
B15-17 0.68 0.12 0.2 0.32 0.88 0.17 0.17 0.024 1
B15-20 0.68 0.12 0.2 0.32 0.88 0.2 0.2 0.024 1
B15-IC-20 0.68 0.12 0.2 0.32 0.88 0.17 0.03 0.2 0.0024 0.024 1
B15-22 0.68 0.12 0.2 0.32 0.88 0.22 0.22 0.024 1
B15-IC-22 0.68 0.12 0.2 0.32 0.88 0.17 0.05 0.22 0.004 0.024 1
B15-25 0.68 0.12 0.2 0.32 0.88 0.25 0.25 0.024 1
B15-IC-25 0.68 0.12 0.2 0.32 0.88 0.17 0.08 0.25 0.0064 0.024 1
B30-17 0.56 0.24 0.2 0.32 0.88 0.17 0.17 0.024 1
B30-20 0.56 0.24 0.2 0.32 0.88 0.2 0.2 0.024 1
B30-IC-20 0.56 0.24 0.2 0.32 0.88 0.17 0.03 0.2 0.0024 0.024 1
B30-22 0.56 0.24 0.2 0.32 0.88 0.22 0.22 0.024 1
B30-IC-22 0.56 0.24 0.2 0.32 0.88 0.17 0.05 0.22 0.004 0.024 1
B30-25 0.56 0.24 0.2 0.32 0.88 0.25 0.25 0.024 1
B30-IC-25 0.56 0.24 0.2 0.32 0.88 0.17 0.08 0.25 0.0064 0.024 1
B50-17 0.4 0.4 0.2 0.32 0.88 0.17 0.17 0.024 1
B50-20 0.4 0.4 0.2 0.32 0.88 0.2 0.2 0.024 1
B50-IC-20 0.4 0.4 0.2 0.32 0.88 0.17 0.03 0.2 0.0024 0.024 1
B50-22 0.4 0.4 0.2 0.32 0.88 0.22 0.22 0.024 1
B50-IC-22 0.4 0.4 0.2 0.32 0.88 0.17 0.05 0.22 0.004 0.024 1
B50-25 0.4 0.4 0.2 0.32 0.88 0.25 0.25 0.024 1
B50-IC-25 0.4 0.4 0.2 0.32 0.88 0.17 0.08 0.25 0.0064 0.024 1
B80-17 0.16 0.64 0.2 0.32 0.88 0.17 0.17 0.024 1
B80-20 0.16 0.64 0.2 0.32 0.88 0.2 0.2 0.024 1
B80-IC-20 0.16 0.64 0.2 0.32 0.88 0.17 0.03 0.2 0.0024 0.024 1
B80-22 0.16 0.64 0.2 0.32 0.88 0.22 0.22 0.024 1
B80-IC-22 0.16 0.64 0.2 0.32 0.88 0.17 0.05 0.22 0.004 0.024 1
B80-25 0.16 0.64 0.2 0.32 0.88 0.25 0.25 0.024 1
B80-IC-25 0.16 0.64 0.2 0.32 0.88 0.17 0.08 0.25 0.0064 0.024 1
a
Not including water from SP.
b
Fraction by binder mass.
c
Volume percentage.

at 20 ± 2 °C. The specimens were removed from the water and

placed in curing chamber with 20 ± 2 °C and 60% ± 5%RH. The ini-
tial length and mass were recorded after 3 h, and then the samples
were cured in a dry condition in the curing chamber. The evolution
of length and mass were measured at 1 d, 3 d, 7 d, 14 d, 21 d, 28 d,
42 d, 56 d, 70 d, and 100 d.

2.3.5. Strength test

The strength test was carried out according to GB/T 17671-1999
Fig. 2. Non-contact concrete shrinkage deformation testing system. 1 – Steel
mould; 2 – Sensor support; 3 – Displacement sensor; 4 – Standard target. [34] on the computer controlled electro-hydraulic testing machine.
The RPC prisms with the dimension of 40 mm  40 mm  160 mm
were produced to measure flexural and compressive strength. After
demoulding, the specimens would be cured in water at room tem-
perature for 28 days. The loading rate of flexural strength of speci-
mens was set as 50 ± 10 N/s. The specimen was loaded at its mid-
Sample 1
span and the distance between two cylindrical supports was
100 mm. The test should be performed in triplicate. After the flexu-
Data Acquisition Computer
Sample 2 ral test, the remaining six broken pieces were subjected to compres-
System Data Processing
sive test. The loading rate of all specimens was set as 2.4 ± 0.2 KN/s.
Sample 3

3. Results and discussion

Fig. 3. Schematic diagram of the RPC.
shrinkage for each composition based on GB/T 29417-2012 [33].
The mixtures were poured into moulds, and the specimens were
then covered by a plastic film to prevent water evaporation. After
24 h, the specimens were demoulded and cured in water for 7 d
3.1. Setting time
The initial setting time of RPC mixtures are shown in Table 3.
It should be noted that even the shortest setting time of C-17
used in the study exceeded 10 h. This significantly exceeded that
 J. Mo et al. / Construction and Building Materials 146 (2017) 283–296 287

Table 3
Initial setting time (h:min).

Specimen Initial setting time Specimen Initial setting time Specimen Initial setting time
C-17 10:05 B15-25 13:05 B50-22 18:15
C-20 10:50 B15-IC-25 12:50 B50-IC-22 17:50
C-IC-20 10:35 B30-17 12:10 B50-25 19:05
C-22 11:10 B30-20 13:00 B50-IC-25 18:50
C-IC-22 11:00 B30-IC-20 12:30 B80-17 19:20
C-25 12:20 B30-22 13:50 B80-20 20:55
C-IC-25 11:50 B30-IC-22 13:25 B80-IC-20 20:45
B15-17 10:50 B30-25 15:40 B80-22 25:15
B15-20 11:05 B30-IC-25 14:50 B80-IC-22 21:25
B15-IC-20 10:50 B50-17 16:30 B80-25 25:50
B15-22 12:20 B50-20 17:35 B80-IC-25 23:35
B15-IC-22 11:40 B50-IC-20 16:50

of the ordinary concrete. The phenomenon was attributed to the eAS ðtÞ ¼ eOD ðtÞ  a½TðtÞ  T 0  ð1Þ
high content of SP used in mixtures. With respect to specimens
In Eq. (1), eAS ðtÞ is the autogenous shrinkage of RPC at time t
without GBFS, the initial setting time increased with increases in
(1 lm/m); eOD ðtÞ is the original data obtained by instrument at
the W/B. Conversely, the setting time of RPC mixtures that con-
time t (1 lm/m); a specifies the thermal expansion coefficient of
tained SAP occurred slightly earlier than the pastes without SAP
RPC mixtures, and is approximately regarded as constant value of
under the condition of same total W/B.
105/°C in this study; TðtÞis the temperature of RPC mixtures at
The setting time increased with increases in the GBFS content
time t (°C); T 0 means the ambient temperature and is set as
with a fixed total W/B. As a kind of latent hydraulic materials, GBFS
20 °C in this study.
mainly acts as filler at early age when blended with cement. GBFS
The autogenous shrinkage of all RPC specimens can be obtained
particles can act as nucleation site to function seeding effect and
according to equation 1, as shown in Figs. 4 and 5.
provide extra space for the hydration products of the clinker
The autogenous shrinkage curves of specimens with and with-
phases [35], by which the reaction of cement is significantly pro-
out entrained water by SAP and without GBFS are shown in
moted at early age. Previous studies indicated that the amount of
Fig. 4. The positive and negative shrinkage strain represent the
portlandite (mainly generated from the hydration of Portland
shrinkage and expansion of specimens, respectively.
cement) in PC-GBFS system was similar or even higher than that
The results indicated that the samples with lower total W/B
in pure PC system during the first hours or days of hydration
possessed higher autogenous shrinkage due to higher hydrostatic
[36,37]. However, only in high alkalinity environment can the
tension forces [24]. The phenomenon verified that autogenous
latent hydraulic of GBFS be activated. When blended with cement,
shrinkage is an important problem for high performance concrete
the reactivity of GBFS highly depends on the alkalinity of cement
with very low W/B. As shown in Fig. 4(a), the process of autoge-
pore solution [38]. Although the GBFS promotes the hydration of
nous shrinkage can be approximately divided into four stages
cement, the high alkalinity of the pore solution still needs few days
especially for the mixtures without IC. The sedimentation of parti-
to build up [35]. Therefore, the amount of reacted GBFS during the
cles was slow after the initial setting, and it still influenced the
first few days is very limited. For example, in Escalante’s research
shrinkage of mixtures. Thus, in stage 1, the mixtures exhibited
[36], the percentage of reacted GBFS in blended system was less
short but rapid shrinkage due to the combination of cement hydra-
than 10% at 3 days with the W/B of 0.35 and GBFS replacement
tion and sedimentation. In stage 2, the shrinkage showed a contin-
level of 30% at 30 °C; as the W/B increasing to 0.50, still only less
uously downward trend prior to approximately 12 h due to early
than 15% of GBFS had been reacted. In this study, the W/B of RPC
age expansion. Common wisdom suggests that this expansion
was no more than 0.25, the reacted percentage of GBFS will be
was caused by the production of portlandite formation [40,41].
even more less. This can attribute to that the cement reaction
When the expansion phase was completed, the shrinkage stage
reduced with decreases of W/B which may bring weakening of
grew very fast until approximately 24 h in stage 3. Following stage
cement-activating environment for the reaction of GBFS. Thus,
3, the growth rate of shrinkage decreased significantly in stage 4
the initial setting time prolonged due to that most of GBFS didn’t
and this led to a stable shrinkage stage. It should be noted that
react at early age. The alkaline activating environment provided
most of the autogenous shrinkage strain up to 7 days occurred dur-
by the hydration of cement reduced continuously as the GBFS
ing the first day for the specimens without SAP and GBFS. At this
replacement level increasing, which led to further reduction of
stage, the tensile strength and strain capacity were too low to
reactivity of GBFS and further prolonged of setting time.
endure shrinkage stress [30]. That indicates that the conventional
RPC is more prone to cracking at early age without proper curing.
3.2. Autogenous shrinkage test Autogenous shrinkage strain was clearly reduced when the
entrained water was introduced by SAP. The reduction was
The ambient temperature was set as 20 °C throughout experi- ascribed to the limitation of self-desiccation by means of IC [14].
ment. However, the temperature of mixtures was not continuously Autogenous shrinkage reduced continuously with increases in
constant especially during the first day after casting. This was the content of SAP and entrained water. The reason for this reduc-
because the hydration of cement caused an increase in tempera- tion can be explained in Fig. 4(b). As shown in Fig. 4(b), the dura-
ture at an early stage. The increase in temperature led to thermal tion of stage 1 of specimens with IC was shorter when compared
strain, and this procedure was always coupled with the autogenous with those of the specimens without IC. In contrast, the expansion
shrinkage of RPC. The data obtained by an instrument included phase of specimens with IC was longer when compared with those
thermal strain caused by the temperature increase. A study by of the specimens without IC. Furthermore, when compared with
Chu [39] indicated that it was necessary to subtract the thermal the specimens without IC, the autogenous shrinkage rate of speci-
strain from original data as shown in equation 1 to demonstrate mens with IC was extremely low in stage 3. The three fore-
the development of autogenous shrinkage with age. mentioned differences caused the IC specimens to exhibit low
288 J. Mo et al. / Construction and Building Materials 146 (2017) 283–296

(a) Replacement level of GBFS is 15%

(a) Autogenous shrinkage of specimens in 7 days

(b) Replacement level of GBFS is 30%

(b) Autogenous shrinkage of specimens in 1 day

Fig. 4. Autogenous shrinkage for specimens without GBFS.
(c) Replacement level of GBFS is 50%

autogenous shrinkage at an early age. As a result, the autogenous

shrinkage was reduced and shifted to a later age when the strength
of RPC was sufficiently strong to confront the shrinkage stress. This
effect of IC on autogenous shrinkage has significant implications
for controlling early age cracking of RPC. However, it should be
noted that the autogenous shrinkage of specimens with IC by
means of SAP was not completely mitigated even for the C-IC-25.
The consumption of entrained water led to self-desiccation regain-
ing the upper hand and induced substantial shrinkage. Therefore,
autogenous shrinkage at 7 days is still significant.
Fig. 5(a–d) show the autogenous shrinkage curve of C-17 and
specimens with GBFS. The replacement levels of cement by GBFS
were 15%, 30%, 50% and 80% respectively.
First, the autogenous shrinkage of RPC reduced with the addi- (d) Replacement level of GBFS is 80%
tion of GBFS when compared with the specimens without the addi-
Fig. 5. Autogenous shrinkage of specimens with GBFS.
tion of GBFS. As shown in Fig. 5(a)–(d), a high content of GBFS
generated low autogenous shrinkage with the same total W/B.
For example, the autogenous shrinkage at 7 days reduced from latent-hydraulic material with a pozzolanic reaction that is slower
806 lm/m to 327 lm/m as the replacement ratio increased from than the hydration of cement at the initial stage [42,43]. It acted
15% to 80% with a W/B corresponding to 0.17. GBFS is a type of more like filler at an early age, and an increase in GBFS content
 J. Mo et al. / Construction and Building Materials 146 (2017) 283–296 289

decreased the amount of cement and led to less water consump-

tion. The drop in internal relative humidity slowed as the GBFS
replacement level increased, and thus autogenous shrinkage was
reduced at an early age. However, in this study, the shrinkage
growth rate of specimens containing GBFS and without IC evi-
dently increased after 2 d especially when the replacement level
corresponded to 15%, 30%, and 50% respectively. Given this situa-
tion, it is necessary to further reduce the autogenous shrinkage
of RPC at an early age.
Second, IC by means of SAP further reduced the autogenous
shrinkage of RPC with GBFS. The autogenous shrinkage decreased
significantly when the entrained water was introduced by SAP.
For instance, the shrinkage strains only corresponded to 400 lm/
m, 258 lm/m, and 51 lm/m for B15-IC-20, B15-IC-22, and B15-
IC-25, respectively at 7 d as shown in Fig. 5(a). This was attributed
to the delay in the declination of internal relative humidity due to
IC. As widely known, water is consumed during hydration process (a) Influence of W/B on the autogenous shrinkage of specimens
and leads to empty pores and self-desiccation. Nevertheless, the IC without GBFS and SAP
can replenish water from SAP to re-fill the pores and maintain the
internal relative humidity for a specified period until the entrained
water is used up. As a result, the reduction of shrinkage deforma-
tion is expected. With respect to the specimens with a replacement
level of 15%, the autogenous shrinkage was almost completely mit-
igated with a total W/B of 0.25. When the replacement level
increased, the IC by means of SAP played a more important role
in reducing the autogenous shrinkage deformation. When the
replacement level reached 30%, the shrinkage strains were further
reduced, and it was interesting to note that the specimen of B30-
IC-25 exhibited a slight net expansion at 7 d. When the replace-
ment level increased to 50% and 80%, all the specimens with IC
exhibited net expansion during 7 d. Although the inclination of
the shrinkage curve also slightly increased at later stage, the
shrinkage growth rate was still significantly less than that without
IC. As reported by previous studies, the reaction degree boundary
of GBFS in the blended cement only corresponded to approxi-
(b) Heat flow and cumulative heat of C-22 and C-IC-22
mately 60–70% [20]. The entire GBFS was not consumed during
the hydration process, and more unhydrated GBFS acted as a filler
when the replacement level increased. Thus, the IC by means of
SAP could effectively mitigate the autogenous shrinkage of RPC
containing GBFS and especially for the samples with a high GBFS
volume.
Finally, the shrinkage strain also reduced when the total W/B
increased with the same content of GBFS as that shown in Fig. 5
(a–d), and the reason for this was similar to that for the specimens
without the GBFS.

3.3. Isothermal calorimetry test

The heat flow and cumulative heat of pastes up to 72 h are pre-

sented in Figs. 6–8.
Theoretically, the cement is completely hydrated only when the
W/C exceeds 0.38. However, the total W/B of the RPC in this study
ranged from 0.17 to 0.25. The hydration exothermic process of
pastes was significantly influenced by the low W/B. Fig. 6 presents
the heat flow and cumulative heat of pastes without GBFS and with
or without SAP. As mentioned in Section 2.3.3, the initial heat peak
was missing, and the entire heat exothermic process was generally
divided into the five stages: a dormant period, an acceleration per-
iod, a period after approximately 24 h that enters the deceleration
period, a period following the main peak in which all the pastes
showed a shoulder during the period approximately ranging from
36 h to 45 h with a significantly lower heat flow, and finally the
steady period in which the heat flow decreased steadily. As shown
in Fig. 6(a), the results indicated that when the total W/B increased,
the main heat peaks of pastes increased but appeared slightly later.
The retardation of the main heat peak was explained by the
(c) Heat flow and cumulative heat of C-25 and C-IC-25
Fig. 6. Heat flow and cumulative heat of pastes without GBFS.
dilution effect of the alkali in the pore solution due to an increase
in W/B [44]. Additionally, the cumulative heat increased with an
increase in W/B. This result agreed well with that obtained in a
study by Zhang et al. [45] that reported that the hydration heat
of pastes used in high performance concrete increased with
increases in W/B.
Fig. 6(b) and (c) show the pastes without GBFS that vary with
respect to the total W/B and in the amount of SAP, respectively.
The results visualized that the addition of SAP slightly reduced
290 J. Mo et al. / Construction and Building Materials 146 (2017) 283–296
(a) Influence of GBFS on the heat flow and
cumulative heat with the W/B of 0.17
(b) Influence of GBFS on the heat flow and
cumulative heat with the W/B of 0.20
Fig. 7. Heat flow and cumulative heat of pastes containing GBFS and varying in the total W/B.
the main heat peak and cumulative heat. This is mainly because a
certain amount of water and alkali ions were absorbed by the SAP
at an early stage. Thus, time is required to release the entrained
water and alkali ions from the SAP.
The heat flow and cumulative heat of pastes with different
amount of GBFS and different total W/B are presented in Fig. 7
(a–d). The figures showed that the main heat peak and cumulative
heat decreased continuously when the amount of GBFS increased.
The hydration activity of GBFS is much lower than that of cement,
and thus the GBFS primarily acted as filler at early age, and its
hydration is required to consume alkali that result from the hydra-
tion of cement. Though the GBFS particles can act as nucleation site
deposition and growth of hydration products during the process of
nucleation and crystal growth [46], however, the amount of
cement reduced as the dosage of GBFS increased, and this caused
a short supply of alkali ions. Therefore, the main hydration heat
of pastes reduced from 15.5% to 90.4% and the cumulative heat
reduced from 10.9% to 85.5% as the dosage of GBFS increasing from
15% to 80% with the W/B of 0.17. The similar situation is also
shown in Fig. 7(b–d). What’s more, the occurrence of heat peaks
also delayed due to the addition of GBFS as shown in figures. For
example, in Fig. 8(a) the main heat peaks were appeared at
22.5 h, 25.5 h, 26.9 h, 34.2 h and 43.9 h respectively with the con-
tent of GBFS increasing from 0% to 80%. As shown in figure, the dor-
mant and acceleration period were longer for the pastes with GBFS
than that without, which was responsible for the retardation of
(c) Influence of GBFS on the heat flow and cumulative
heat with the W/B of 0.22
(d) Influence of GBFS on the heat flow and cumulative
heat with the W/B of 0.25
main heat peak. This phenomenon testified the results of
Section 3.1 that the setting time of mixtures became longer when
partial cement was replaced by GBFS, and as the replacement level
increasing the setting time increased constantly.
Fig. 8(a) illustrates the influence of the total W/B on the heat
flow and cumulative heat of pastes with the GBFS replacement
ratio of 30%. When the total W/B increases, the main heat peak
increased but its occurrence was continuously retarded. The cumu-
lative heat also increased with an increase in total W/B at 72 h.
However, the exothermic rate of pastes with lower total W/B
was initially higher, and thus the cumulative heat of B30-17 was
the highest before approximately 33 h. Subsequently, the trend
was reversed. The reason for this phenomenon was similar to that
of pastes without GBFS. Fig. 8(b–d) visualizes the effect of IC by
means of SAP on the heat flow and cumulative heat of pastes con-
taining GBFS. In order to save space, some typical figures were
selected to elaborate this problem. As shown in the figures, under
the condition of same total W/B, the main heat peak appeared ear-
lier with the addition of SAP. This was attributed to the increase in
the pore solution concentration that was caused by the swelling of
SAP [47]. When the total W/B was the same, there were almost no
significant differences in cumulative heat curves between the
pastes with and without SAP. The cumulative heat of pastes with
SAP was only slightly lower than that without SAP, and this indi-
cates that most of water entrained in SAP was released and con-
sumed for hydration during first 3 days.
 J. Mo et al. / Construction and Building Materials 146 (2017) 283–296 291

(a) Influence of W/B on the heat flow and cumulative heat with (c) Heat flow and cumulative heat of B30-22 and B30-IC-22
the GBFS replacement ratio of 30%

(b) Heat flow and cumulative heat of B30-20 and B30-IC-20 (d) Heat flow and cumulative heat of B30-25 and B30-IC-25
Fig. 8. Heat flow and cumulative heat of pastes with GBFS, varying in the amount of SAP and in the total W/B.

3.4. Mechanical strength

The strength of specimens without GBFS is shown in Fig. 9. The

compressive and flexural strength decreased with increases in the
total W/B. When compared with reference C-17, the compressive
strength of samples without IC decreased by 11.7%, 21.0%, and
29.7% when the total W/B increased from 0.20 to 0.25. The strength
of specimens with SAP was slightly lower than that without SAP
when the total W/B was same. The decent degree corresponded
to 4.9%, 2.4%, and 4.3% for the total W/B corresponding to 0.20,
0.22, and 0.25 respectively. The trend of flexural strength was sim-
ilar to that of compressive strength.
It is easy to understand the decline in strength due to the
increase of total W/B. The RPC is a type of ultra-high performance
concrete, and its super high strength is obtained by its extremely
compacting structure. However, the addition of SAP introduced
some initial porosity. The introduced pores were partially re-
Fig. 9. The strength of specimens without GBFS at 28 days.
filled by calcium hydroxide during further hydration of binders
when the entrained water was released from SAP. However, the
pores were never completely compensated. After the entrained 120 MPa at 28 d as shown in the figure above. The reduction in
water was completely released from SAP, the residual pores still strength is acceptable considering the substantial reduction of
corresponded to the largest defect that existed in the RPC autogenous shrinkage due to SAP.
microstructure [14]. Thus, the reduction in strength caused by The strength of groups containing different dosages of GBFS is
SAP was inevitable. Nevertheless, the reduction was slight, and presented in Fig. 10(a–d). For the group with same GBFS replace-
for example, even sample C-IC-25 with highest content of SAP ment level, the strength declined with the W/B increasing which
exhibited a compressive strength that ranged from high to was similar to that of without GBFS. The results also indicated that
292 J. Mo et al. / Construction and Building Materials 146 (2017) 283–296
the addition of GBFS reduced the strength of RPC, and as the
replacement level increasing the strength declined continuously.
It can be ascribed to the different hydration mechanism of GBFS.
The GBFS is a kind of latent-hydraulic mineral admixtures which
need to react with the calcium hydroxide generated from the
cement hydration to form C-S-H gel. As a result, the development
of strength is slowed down for all specimens containing GBFS at
early age. However, it should be noted that the compressive
strength of specimens was all over 100 MPa at 28 days except for
the group with GBFS replacement level of 80%. This finding means
that producing RPC with high volume of GBFS is feasible. The sim-
ilar conclusion was obtained by Yazıcı et al. [2], and the strength of
RPC containing GBFS was even more higher in their research due to
the autoclave curing.
The effect of SAP on the strength performance of RPC with GBFS
was analogous to that of without. Under the condition of same
total W/B, the IC by means of SAP further reduced the strength
for all the groups with different replacement levels of GBFS. For
example, the compressive strength reduced 6.2%, 1.2% and 5.9%
as the total W/B increasing from 0.20 to 0.25 respectively when
the GBFS replacement level was 30%. Based on these results, we
can get a conclusion that the strength of RPC containing GBFS is
also mainly influenced by the total W/B, and the effect of SAP is
negligible. In view of the significant reduction of autogenous
shrinkage, applying IC by means of SAP on RPC containing GBFS
is feasible and advantageous.
(a) Compressive and flexural strength of specimens with 15% GBFS
(b) Compressive and flexural strength of specimens with 30% GBFS
Fig. 10. The strength of specimens containing GBFS at 28 days.
3.5. Drying shrinkage and mass loss
Dry shrinkage is a type of volume reduction that is caused by
the loss of water from pores due to the humidity difference
between the interior and exterior of concrete [48]. The results of
drying shrinkage and mass loss of RPC without GBFS are presented
in Fig. 11. The figure mainly visualized the influence of different
total W/B and dosages of SAP on the drying shrinkage and mass
loss for specific days.
The total W/B is an important factor for the drying shrinkage of
RPC. As shown in Fig. 11, the drying shrinkage for RPC without SAP
increased when the total W/B increased up to 100 days. The drying
shrinkage strains corresponded to 400 lm/m, 604 lm/m, 664 lm/
m, and 710 lm/m when the total W/B increased from 0.17 to 0.25.
Prior to 28 d, the drying shrinkage rates were high for all mixtures,
and this was followed by a gradual reduction. The same trends
were also observed in the results of mass loss. It was demonstrated
that a strong correlation exists between drying shrinkage and mass
loss. When the total W/B increased, the disequilibrium of relative
humidity between internal and outside of RPC became more signif-
icant. Thus, more water was evaporated and this led to increased
mass loss and higher drying shrinkage. Additionally, the evapora-
tion of water formed meniscus in the pores, and this caused surface
tension and shrinkage of the RPC. With the same total W/B, the
addition of SAP slightly reduced the drying shrinkage and mass
loss. This was because some residual water stored in SAP could
(c) Compressive and flexural strength of specimens with 50% GBFS
(d) Compressive and flexural strength of specimens with 80% GBFS
 J. Mo et al. / Construction and Building Materials 146 (2017) 283–296
Fig. 11. Mass loss and drying shrinkage of RPC without GBFS, varying in W/B and in
the amount of SAP.
partially re-fill the pores and reduce surface tension. However, as
mentioned in chapter 3.3, most of the entrained water was con-
sumed during first 3 days, and thus the reduction of drying shrink-
age and mass loss caused by SAP was very limited. Therefore, the IC
by means of SAP did not evidently affect the drying shrinkage of
RPC.
Although the specimens with higher total W/B exhibited larger
drying shrinkage strains, it’s should be noted that the total shrink-
age was still lower than that with low total W/B. The main reason
was because the advantage of less drying shrinkage of mixtures
with a low total W/B was offset by significant autogenous shrink-
age at early age. In addition to the autogenous shrinkage, the total
shrinkage of specimens without SAP corresponded to 1508 lm/m,
1471 lm/m, 1453 lm/m, and 1262 lm/m, when the total W/B
increased from 0.17 to 0.25. When the IC by means of SAP was
employed, the total shrinkage reduced to 1068 lm/m, 1081 lm/
m, and 996 lm/m for C-IC-20, C-IC-22, and C-IC-25, respectively.
Therefore, although the SAP exerted a negligible influence on dry-
ing shrinkage, it reduced the total shrinkage by significantly
decreasing the autogenous shrinkage at early age.
The development of experimentally obtained drying shrinkage
strains and mass loss for the RPC mixtures with different replace-
ment levels of GBFS and different amounts of SAP are presented in
Fig. 12(a–d).
As shown in the figures, the slope of drying shrinkage and mass
loss curve during first 42 d was obviously higher than those at later
ages for all the mixtures. This indicated that the drying shrinkage
and mass loss increased with time but decreased with respect to
the rate. The maximum period with a relatively high shrinkage at
an increasing rate corresponded to 42 d, and this was longer than
that of the samples without GBFS. This was attributed to the sec-
ondary hydration reaction of GBFS in which the microstructure
was more compacting and drying shrinkage also increased.
Although the shape of curves was similar for every single group
with the same amount of GBFS and different total W/B, the drying
shrinkage and mass loss increased with increases in total W/B. As
reported by extant studies [49], higher W/B can lead to lower stiff-
ness and higher volume of evaporable water, and this can result in
increased drying shrinkage and increased mass loss. For example,
the drying shrinkage up to 100 d corresponded to 460 lm/m,
583 lm/m, 683 lm/m, and 822 lm/m for the mixtures with 15%
GBFS and without SAP as the total W/B increased from 0.17 to
293
0.25. As shown in Fig. 12(a–d), the drying shrinkage and mass loss
also increased with an increase in the GBFS replacement level
under the condition of same total W/B. In a manner similar to
the RPC mixtures with a total W/B of 0.22, the drying shrinkage
at 100 days corresponded to 664 lm/m, 700 lm/m, 750 lm/m,
955 lm/m, and 1080 lm/m when the dosage of GBFS increased
from 0% to 80%. There are two main reasons for this phenomenon.
First, with the same total W/B, the cement content reduced due to
the substitution of GBFS. As a result, there were significant reduc-
tions in the water required for the hydration of cement and the
reaction of calcium hydroxide produced by cement hydration with
GBFS. Thus, more evaporable water existed in the RPC that con-
tained GBFS, and evidently the drying shrinkage increased when
water evaporation increased. Consequently, an increase in the
amount of GBFS contained by the RPC increased drying shrinkage
and mass loss. Furthermore, the drying shrinkage associated with
the compressive strength and its rate and magnitude increased
with the reduction in compressive strength according to extant
studies [26]. As reported in Section 3.4, an increase in the GBFS
content decreases the compressive strength with the same total
W/B. In this section, for every group in Fig. 12(a–d), the drying
shrinkage strains continuously increased when the replacement
level of GBFS increased. This result is in conformity with the results
of compressive strength, and was used to verify the fore-
mentioned hypotheses.
Although the compressive strength of RPC containing GBFS with
SAP was slightly lower than that without SAP, Fig. 12(a–d) still
indicated that the IC by means of SAP could slightly reduce the dry-
ing shrinkage and mass loss. This was attributed to the residual
water stored in SAP. However, the IC by means of SAP had a negli-
gible effect on long-term drying shrinkage because almost all
entrained water was consumed during first 3 days.
The property of drying shrinkage of RPC was discussed as stated
above. The drying shrinkage can be classified into reversible drying
shrinkage and irreversible drying shrinkage [50,51]. Reversible
drying shrinkage is the part of total shrinkage that is repeatability
after rewetting; and the irreversible drying shrinkage is the part
which cannot be repeated after rewetting [50]. Zhang et al. [52]
reported that the drying shrinkage of normal concrete was period-
ically changed during dry-wet cycles, exhibiting shrinkage as dry-
ing and reversible shrinkage as wetting. It can be explained by that
the interior humidity of concrete deceases as concrete undergoes
drying and increases as concrete undergoes wetting. The extent
of variation of interior humidity under dry-wet cycles is depend
on the permeability of concrete, which means that the reversible
drying shrinkage is highly relevant to permeability. For concrete
with high permeability such as normal concrete or recycled con-
crete, the drying shrinkage reversibility is usually significant. As
Tam et al. reported, the reversible percentage of recycled concrete
reached over 30% [50]. However, as abovementioned, RPC is char-
acterized by extremely low permeability and high elasticity mod-
ulus. The influence of dry-wet cycles on its interior humidity can
be neglected at late age. Consequently, it is reasonable to think that
the most part of drying shrinkage of RPC is irreversible, and the
reversible part is negligible.
The total shrinkage of all mixtures is grouped and presented in
Fig. 13. In the figure, W/B-20 denotes all the mixtures with a W/B
of 0.2, and W/B-IC-20 denotes mixtures with a W/B of 0.20 and
with IC by means of SAP and so on. The autogenous shrinkage of
RPC exhibited a sustained downward trend when the substitution
rate of GBFS increased. The total shrinkage first decreased and then
increased for every group with same total W/B and same amount
of SAP. This indicated that the higher drying shrinkage was caused
by the addition of GBFS. Conversely, the total shrinkage of mixtures
that employed IC was lower than those without of IC due to the
mitigation of autogenous shrinkage at an early age. The minimum
294 J. Mo et al. / Construction and Building Materials 146 (2017) 283–296

(a) Drying shrinkage and mass loss of specimens with 15% GBFS (c) Drying shrinkage and mass loss of specimens with 50% GBFS

(b) Drying shrinkage and mass loss of specimens with 30% GBFS (d) Drying shrinkage and mass loss of specimens with 80% GBFS

Fig. 12. Mass loss and drying shrinkage of RPC with GBFS, varying in the W/B and in the amount of SAP.

value of total shrinkage of every group is labelled in the figure. The

total shrinkage of specimen B50-IC-20 at 100 days corresponded to
601 lm/m, which is only 39.9% of that for C-17. A more meaningful
term with respect to the application of IC is that it shifts a more
dangerous early age shrinkage to a later age, and at the later age
the RPC are sufficiently strong to resist the shrinkage stress.
In conclusion, the deformation of RPC induced by autogenous
and drying shrinkage is a major hindrance to its application in con-
struction. However, a combination of GBFS and IC by means of SAP
can adequately solve this problem as indicated by the fore-
mentioned results and analysis.

4. Conclusion

The influence of GBFS and SAP on the properties of RPC was

studied by employing an extensive experimental program. Based
on the above results and discussion, the following main conclu-
Fig. 13. The total shrinkage of RPC. sions are obtained:
 J. Mo et al. / Construction and Building Materials 146 (2017) 283–296
(1) The setting time of RPC mixtures was prolonged by the addi-
tion of GBFS. Although GBFS can act as seeds to accelerate the
hydration of cement at early age, the blended system still
needs first few days to build up high alkalinity of pore solu-
tion. Thus, most of GBFS was still unreacted during the first
day or so, which was responsible for the delay in setting time.
(2) The autogenous shrinkage reduced continuously when the
replacement level of GBFS increased. This was attributed to
the fact that the GBFS mainly acted as a filler because of
its low activity at early age, and this caused a decrease in
the water consumption of specimens that contained GBFS.
The decrease in internal relative humidity was slower when
compared with that of the specimens without GBFS, and this
led to the reduction of autogenous shrinkage. Following the
IC by means of SAP employed, the autogenous shrinkage of
RPC significantly declined. With respect to a few samples
containing GBFS, the autogenous shrinkage was completely
mitigated and even exhibited net expansion at early age.
The results proved that applying IC by means of SAP to mit-
igate the autogenous shrinkage of RPC was effective irre-
spective of whether the pure cement system or GBFS
blended system was applied.
(3) The occurrence of heat peak was delayed and cumulative
heat was reduced when the replacement level of GBFS
increased, and this was ascribed to the special hydration
mechanism of GBFS. The results coincided well with the
results of the setting time test. With the same total W/B
and same amount of GBFS, the cumulative heat of pastes
containing SAP was slightly lower than that of the pastes
without SAP. The phenomenon indicated that most of the
entrained water was released over the first 3 days.
(4) The strength of RPC decreased when the GBFS content
increased. However, the compressive strength still exceeded
100 MPa at 28 d even when the replacement level of GBFS
reached 50%, and this indicated that it was feasible to pro-
duce RPC with a high volume of GBFS. However, although
the IC by means of SAP further reduced the strength, it
was still acceptable to apply the same on RPC by considering
its autogenous shrinkage mitigation effect.
(5) The RPC containing GBFS exhibited higher drying shrinkage
and water loss when compared to those of plain RPC. This
was because of the increase in evaporable water and
decrease in strength that was caused by the addition of
GBFS. The effect of SAP on the drying shrinkage and water
loss was negligible when compared to those in the speci-
mens with same total W/B and same amount of GBFS. How-
ever, when coupled with autogenous shrinkage, the total
shrinkages of all specimens with SAP were lower than those
of the specimens without SAP. Moreover, the IC can shift
early age shrinkage (that is more dangerous) to a later stage.
All the above analyses verified that the combination of GBFS
and IC by means of SAP can solve the problem of serious
shrinkage in RPC.
References
[1] P. Richard, M. Cheyrezy, Composition of reactive powder concretes, Cem.
Concr. Res. 25 (7) (1995) 1501–1511, http://dx.doi.org/10.1016/0008-8846
(95)00144-2.
[2] H. Yazıcı, H. Yigiter, A.S. Karabulut, B. Baradan, Utilization of fly ash and ground
granulated blast furnace slag as an alternative silica source in reactive powder
concrete, Fuel 87 (2008) 2401–2407, http://dx.doi.org/10.1016/j.fuel.2008.
03.005.
[3] H. Yazıcı, M.Y. Yardımcı, S. Aydın, A.S. Karabulut, Mechanical properties of
reactive powder concrete containing mineral admixtures under different
curing regimes, Constr. Build. Mater. 23 (2009) 1223–1231, http://dx.doi.org/
10.1016/j.conbuildmat.2008.08.003.
295
[4] L. Huynh, S. Foster, H. Valipour, R. Randall, High strength and reactive powder
concrete columns subjected to impact: experimental investigation, Constr.
Build. Mater. 78 (2015) 153–171, http://dx.doi.org/10.1016/
j.conbuildmat.2015.01.026.
[5] C.M. Tam, Vivian W.Y. Tam, K.M. Ng, Assessing drying shrinkage and water
permeability of reactive powder concrete produced in Hong Kong, Constr.
Build. Mater. 26 (2012) 79–89, http://dx.doi.org/10.1016/
j.conbuildmat.2011.05.006.
[6] Y. Wang, M.Z. An, Z.R. Yu, S. Han, Absorption characteristic of reactive powder
concrete on NaCl solution before and after freeze-thaw cycles, J. Harbin Inst.
Technol. 47 (4) (2015) 71–75. http://dx.doi.org/10.11918/j.issn.0367-6234.
2015.04.012 (in Chinese).
[7] M.G. Lee, Y.C. Wang, C.T. Chiu, A preliminary study of reactive powder concrete
as a new repair material, Constr. Build. Mater. 21 (2007) 182–189, http://dx.
doi.org/10.1016/j.conbuildmat.2005.06.024.
[8] C. Hua, P. Acker, A. Ehrlacher, Analyses and models of the autogenous
shrinkage of hardening cement paste I. Modelling at macroscopic scale, Cem.
Concr. Res. 25 (7) (1995) 1457–1468, http://dx.doi.org/10.1016/0008-8846
(95)00140-8.
[9] H. Chen, M. Wyrzykowski, K. Scrivener, P. Lura, Prediction of self-desiccation in
low water-to-cement ratio pastes based on pore structure evolution, Cem.
Concr. Res. 49 (2013) 38–47, http://dx.doi.org/10.1016/j.
cemconres.2013.03.013.
[10] X.W. Ma, J.K. Zhang, J.H. Liu, Review on superabsorbent polymer as internal
curing agent of high performance cement-based material, J. Chin. Ceram. Soc.
43 (8) (2015) 1099–1110, http://dx.doi.org/10.14062/j.issn.0454-
5648.2015.08.12.
[11] X.P. Chen, G.R. Shan, J. Huang, Z.M. Huang, Z.X. Weng, Synthesis and properties
of acrylic-based superabsorbent, J. Appl. Polym. Sci. 92 (2004) 619–624, http://
dx.doi.org/10.1002/app.20052.
[12] D.J. Shen, T. Wang, Y. Chen, M.L. Wang, G.Q. Jiang, Effect of internal curing with
super absorbent polymers on the relative humidity of early-age concrete,
Constr. Build. Mater. 99 (2015) 246–253, http://dx.doi.org/10.1016/
j.conbuildmat.2015.08.042.
[13] C. Schröfl, V. Mechtcherine, M. Gorges, Relation between the molecular
structure and the efficiency of superabsorbent polymers (SAP) as concrete
admixture to mitigate autogenous shrinkage, Cem. Concr. Res. 42 (2012) 865–
873, http://dx.doi.org/10.1016/j.cemconres.2012.03.011.
[14] J. Justs, M. Wyrzykowski, D. Bajare, P. Lura, Internal curing by superabsorbent
polymers in ultra-high performance concrete, Cem. Concr. Res. 76 (2015) 82–
90, http://dx.doi.org/10.1016/j.cemconres.2015.05.005.
[15] F.Z. Wang, J. Yang, S.G. Hu, X.P. Li, H. Cheng, Influence of superabsorbent
polymers on the surrounding cement paste, Cem. Concr. Res. 81 (2016) 112–
121, http://dx.doi.org/10.1016/j.cemconres.2015.12.004.
[16] O.M. Jensen, P.F. Hansen, Water-entrained cement-based materials II.
Experimental observations, Cem. Concr. Res. 32 (2002) 973–978, http://dx.
doi.org/10.1016/S0008-8846(02)00737-8.
[17] K.H. Yang, G.D. Moon, Y.S. Jeon, Implementing ternary supplementary
cementing binder for reduction of the heat of hydration of concrete, J. Clean.
Prod. 112 (2016) 845–852, http://dx.doi.org/10.1016/j.jclepro.2015.06.022.
[18] E. Özbay, M. Erdemir, H.I. Durmus, Utilization and efficiency of ground
granulated blast furnace slag on concrete properties – a review, Constr. Build.
Mater. 105 (2016) 423–434, http://dx.doi.org/10.1016/
j.conbuildmat.2015.12.153.
[19] P. Dinakar, K.P. Sethy, U.C. Sahoo, Design of self-compacting concrete with
ground granulated blast furnace slag, Mater. Des. 43 (2013) 161–169.
doi:10.1016/j.matdes.2012.06.049..
[20] E. Gruyaert, N. Robeyst, N.D. Belie, Study of the hydration of Portland cement
blended with blast-furnace slag by calorimetry and thermogravimetry, J.
Therm. Anal. Calorim. 102 (2010) 941–951, http://dx.doi.org/10.1007/s10973-
010-0841-6.
[21] Y. Ballim, P.C. Graham, The effects of supplementary cementing materials in
modifying the heat of hydration of concrete, Mater. Des. 42 (2009) 803–811,
http://dx.doi.org/10.1617/s11527-008-9425-3.
[22] C.J. Shi, D.H. Wang, L.M. Wu, Z.M. Wu, The hydration and microstructure of
ultra-high-strength concrete with cement-silica fume-slag binder, Cem. Concr.
Compos. 61 (2015) 44–52, http://dx.doi.org/10.1016/j.
cemconcomp.2015.04.013.
[23] C.C. Castellano, V.L. Bonavetti, H.A. Donza, E.F. Irassar, The effect of w/b and
temperature on the hydration and strength of blast-furnace slag cements,
Constr. Build. Mater. 111 (2016) 679–688, http://dx.doi.org/10.1016/
j.conbuildmat.2015.11.001.
[24] D. Snoeck, O.M. Jensen, N.D. Belie, The influence of superabsorbent polymers
on the autogenous shrinkage properties of cement pastes with supplementary
cementitious materials, Cem. Concr. Res. 74 (2015) 59–67, http://dx.doi.org/
10.1016/j.cemconres.2015.03.020.
[25] X.S. Huo, N. Al-Omaishi, M.K. Tadros, Creep, shrinkage, and modulus of
elasticity of high-performance concrete, ACI Mater. J. 98 (6) (2001) 440–448,
http://dx.doi.org/10.14359/10842.
[26] M. Shariq, J. Prasad, H. Abbas, Creep and drying shrinkage of concrete
containing GGBFS, Cem. Concr. Compos. 68 (2016) 35–45, http://dx.doi.org/
10.1016/j.cemconcomp.2016.02.004.
[27] W.Y. Zhang, Y. Hama, S.H. Na, Drying shrinkage and microstructure
characteristics of mortar incorporating ground granulated blast furnace slag
and shrinkage reducing admixture, Constr. Build. Mater. 93 (2015) 267–277,
http://dx.doi.org/10.1016/j.conbuildmat.2015.05.103.
296 J. Mo et al. / Construction and Building Materials 146 (2017) 283–296
[28] GB175-2007, Common Portland cement, Standardization Administration of
the People’s Republic of China, Beijing, 2007.
[29] M. Rahmat, R.M. Mohammad, G. Reza, S.S. Fatemeh, Assessment of factors
influencing mechanical properties of steel fiber reinforced self-compacting
concrete, Mater. Des. 83 (2015) 284–294, http://dx.doi.org/10.1016/
j.matdes.2015.06.024.
[30] C.H. Jiang, Y. Yang, Y. Wang, Y.N. Zhou, C.C. Ma, Autogenous shrinkage of high
performance concrete containing mineral admixtures under different curing
temperatures, Constr. Build. Mater. 61 (2014) 260–269, http://dx.doi.org/
10.1016/j.conbuildmat.2014.03.023.
[31] D.J. Shen, X.D. Wang, D.B. Cheng, J.Y. Zhang, G.Q. Jiang, Effect of internal curing
with super absorbent polymers on autogenous shrinkage of concrete at early
age, Constr. Build. Mater. 106 (2016) 512–522, http://dx.doi.org/10.1016/
j.conbuildmat.2015.12.115.
[32] ASTM C403, Standard Test Method for Time of Setting of Concrete Mixtures by
Penetration Resistance, ASTM International, West Conshohocken, PA, 2016.
[33] GB/T 29417-2012, Standard Test Methods for Drying Shrinkage Stress and
crackiNg Possibility of Cement Mortar and Concrete, Standardization
Administration of the People’s Republic of China, Beijing, 2013.
[34] GB/T 17671-1999, Method of Testing Cements-Determination of Strength, The
State Bureau of Quality and Technical Supervision, Beijing, 1999.
[35] B. Lothenbach, K. Scrivener, R.D. Hooton, Supplementary cementitious
materials, Cem. Concr. Res. 41 (2011) 1244–1256, http://dx.doi.org/10.1016/
j.cemconres.2010.12.001.
[36] J.I. Escalante, L.Y. Gomez, K.K. Johal, G. Mendoza, H. Mancha, J. Mendez,
Reactivity of blast-furnace slag in Portland cement blends hydrated under
different conditions, Cem. Concr. Res. 31 (2001) 1403–1409. (PII: S0008–8846
(01)00587–7).
[37] I. Pane, W. Hansen, Investigation of blended cement hydration by isothermal
calorimetry and thermal analysis, Cem. Concr. Res. 35 (2005) 1155–1164,
http://dx.doi.org/10.1016/j.cemconres.2004.10.027.
[38] M.H. Hubler, J.J. Thomas, H.M. Jennings, Influence of nucleation seeding on the
hydration kinetics and compressive strength of alkali activated slag paste,
Cem. Concr. Res. 41 (2011) 842–846, http://dx.doi.org/10.1016/j.
cemconres.2011.04.002.
[39] I. Chu, S.H. Kwon, M.N. Amin, J.K. Kim, Estimation of temperature effects on
autogenous shrinkage of concrete by a new prediction model, Constr. Build.
Mater. 35 (10) (2012) 171–182, http://dx.doi.org/10.1016/
j.conbuildmat.2012.03.005.
[40] V. Baroghel-Bouny, P. Mounanaga, A. Khelidj, A. Loukili, N. Rafai, Autogenous
deformations of cement pastes: Part II. W/C effects, micro-macro correlations,
and threshold values, Cem. Concr. Res. 36 (1) (2006) 123–136, http://dx.doi.
org/10.1016/j.cemconres.2004.10.020.
[41] G. Sant, B. Lothenbach, P. Juillant, G.L. Saout, W.J. Weiss, K. Scrivener, The
origin of early age expansions induced in cementitious materials containing
shrinkage reducing admixtures, Cem. Concr. Res. 41 (3) (2011) 218–229,
http://dx.doi.org/10.1016/j.cemconres.2010.12.004.
[42] Y. Wei, W. Hansen, J.J. Biernacki, E. Schlangen, Unified shrinkage model for
concrete from autogenous shrinkage test on paste with and without ground-
granulated blast-furnace Slag, ACI Mater. J. 108 (1) (2011) 13–20, http://dx.doi.
org/10.14359/51664211.
[43] K.M. Lee, H.K. Lee, S.H. Lee, G.Y. Kim, Autogenous shrinkage of concrete
containing granulated blast-furnace slag, Cem. Concr. Res. 36 (2006) 1279–
1285, http://dx.doi.org/10.1016/j.cemconres.2006.01.005.
[44] T. Zhang, Q. Yu, J. Wei, P. Gao, P. Zhang, Study on optimization of hydration
process of blended cement, J. Therm. Anal. Calorim. 107 (2012) 489–498,
http://dx.doi.org/10.1007/s10973-011-1531-8.
[45] Y.S. Zhang, W. Sun, S.F. Liu, Study on the hydration heat of binder paste in
high-performance concrete, Cem. Concr. Res. 32 (2002) 1483–1488, http://dx.
doi.org/10.1016/S0008-8846(02)00810-4.
[46] F.H. Han, Z.Q. Zhang, D.M. Wang, P.Y. Yan, Hydration heat evolution and
kinetics of blended cement containing steel slag at different temperatures,
Thermochim. Acta 605 (2015) 43–51, http://dx.doi.org/10.1016/j.
tca.2015.02.018.
[47] J. Justs, M. Wyrzykowski, F. Winnefeld, D. Bajare, P. Lura, Influence of
superabsorbent polymers on hydration of cement pastes with low water-to-
binder ratio, J. Therm. Anal. Calorim. 115 (2014) 425–432, http://dx.doi.org/
10.1007/s10973-013-3359-x.
[48] M. Valcuende, E. Marco, C. Parra, P. Serna, Influence of limestone filler and
viscosity-modifying admixture on the shrinkage of self-compacting concrete,
Cem. Concr. Res. 42 (2012) 583–592, http://dx.doi.org/10.1016/j.
cemconres.2012.01.001.
[49] J.A. Bogas, R. Nogueira, N.G. Almeida, Influence of mineral additions and
different compositional parameters on the shrinkage of structural expanded
clay lightweight concrete, Mater. Des. 56 (2014) 1039–1048, http://dx.doi.org/
10.1016/j.matdes.2013.12.013.
[50] V.W.Y. Tam, D. Kotrayothar, J.Z. Xiao, Long-term deformation behaviour of
recycled aggregate concrete, Constr. Build. Mater. 100 (2015) 262–272, http://
dx.doi.org/10.1016/j.conbuildmat.2015.10.013.
[51] J.Z. Xiao, Y. Fan, V.W.Y. Tam, On creep characteristics of cement paste, mortar
and recycled aggregate concrete, Eur. J. Environ. Civ. Eng. 19 (10) (2015) 1234–
1252, http://dx.doi.org/10.1080/19648189.2015.1008652.
[52] J. Zhang, Y. Gao, Y.D. Han, W. Sun, Shrinkage and interior humidity of concrete
under dry-wet cycles, Dry. Technol. 30 (6) (2012) 583–596, http://dx.doi.org/
10.1080/07373937.2011.653614.