Construction and Building Materials 121 (2016) 429–436

Contents lists available at ScienceDirect

Construction and Building Materials

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

Aggregate and slag cement effects on autogenous shrinkage in

cementitious materials
Zhichao Liu ⇑, Will Hansen
Department of Civil and Environmental Engineering, University of Michigan, USA

h i g h l i g h t s

Autogenous shrinkage is calculated by removing the thermal effects associated with cement hydration.

Aggregate content effect is studied by an improved Pickett’s model.

Slag cement effect is characterized by a scaling method.

Partial replacement with LWA is used to effectively mitigate autogenous shrinkage.

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

Article history: This study investigates the susceptibility of cementitious materials to internal moisture condition by the
Received 19 January 2016 autogenous deformation measurement in the 0.35 water-binder (w/b) ratio systems (paste, mortar and
Received in revised form 1 June 2016 concrete). Autogenous shrinkage is obtained by separating the thermal effects from the measured total
Accepted 4 June 2016
deformation using the maturity concept and the coefficient of thermal deformation (CTD). Three factors
Available online 11 June 2016
affecting the autogenous shrinkage are presented, that is the aggregate content, the partial replacement
of portland cement with slag cement and of normal weight fine aggregate with lightweight aggregate
Keywords:
(LWA). Autogenous shrinkage is clearly reduced by an increasing aggregate content and this effect is pre-
Concrete
Autogenous shrinkage
dicted by an improved Pickett’s model using a time-dependent aggregate restraining factor. A binary
Slag cement cementitious system of portland cement and slag cement increases the autogenous shrinkage in the long
Lightweight aggregate term. Contribution of slag cement can be characterized by the difference in autogenous shrinkage
Aggregate content between the binary system and the control system scaled down by the replacement ratio. The negative
effect of slag cement can be neutralized by the incorporation of LWA as a partial sand replacement.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction
Concrete is intrinsically susceptible to moisture from different
aspects. Moisture is the carrier of deleterious species into concrete,
such as chloride ions accelerating the corrosion of embedded rein-
forcements and sulfate ions causing internal cracking and disinte-
gration of the cementitious matrix [1]. Sub-freezing phase
transformation of the internal moisture is also the culprit for frost
damage in concrete [2]. This susceptibility can be minimized by the
advent of high performance concrete (HPC) which features a low
water-binder (w/b) ratio and the addition of supplementary
cementitious materials (or SCM, such as slag cement) [3]. These
characteristics reduce the porosity and re-configure the pore struc-
⇑ Corresponding author at: Department of Civil and Environmental Engineering,
2350 Hayward, 2330 GG Brown, Ann Arbor, MI 48109-2125, USA.
E-mail address: lzc@umich.edu (Z. Liu).
ture towards a more refined pore size [4,5], which results in a den-
ser matrix (thus limiting the penetration of the detrimental
chemicals) and a lower amount of freezable moisture (thus reduc-
ing the risk of frost damage).
At the same time, HPC experiences exacerbated internal pore
drying due to the secondary chemical reaction of SCM with the
moisture [6]. This self-desiccation effect [7], when coupled with a
finer pore size, generates an enhanced net compressive stress on
capillary pores on a micro-scale and more significant autogenous
volume contraction on a macro-scale [8,9]. This is exemplified by
the increased autogenous shrinkage in concrete with partial
replacement by slag cement in the long term [10–14]. As a result,
there is an increasing concern in the autogenous shrinkage-
induced cracking in HPC nowadays [15–17]. This creates the need
for accurate characterization of the autogenous shrinkage, which
has resulted in a number of prediction models for concrete shrink-
age based on the elastic properties of concrete and its components

http://dx.doi.org/10.1016/j.conbuildmat.2016.06.012
0950-0618/Ó 2016 Elsevier Ltd. All rights reserved.
430 Z. Liu, W. Hansen / Construction and Building Materials 121 (2016) 429–436

[18–22]. In addition, several mitigation strategies have been devel- Table 1

oped [23,24], including the use of chemical admixtures (such as Mix design (kg/m3).

shrinkage reducing admixture, expansive additives) [25], the Mix Portland Slag Gravel Normal LWA Water
entrainment of internal curing agents (such as lightweight aggre- cement cement sand
gate and super absorbent polymer) [26] or the application of a lean 035-0S-0NA 1497 0 0 0 0 524
mix [27]. 035-0S-20NA 1198 0 0 530 0 419
Autogenous shrinkage is a result of an exothermic chemical 035-0S-40NA 899 0 0 1059 0 314
035-25S-40NA 667 222 0 1059 0 311
reaction between cementitious materials and water, which can 035-50S-40NA 440 440 0 1059 0 308
lead to a temperature rise and thermal dilation in the specimen 035-0S-60NA 599 0 941 649 0 210
in the early age. However, the typical measurement techniques 035-0S-78NA 329 0 1223 843 0 115
for monitoring the deformation of a sealed homogenous specimen 035-25S-78NA 245 82 1223 843 0 114
035-50S-78NA 161 161 1223 843 0 113
indiscriminately include both the autogenous and thermal defor-
035-0S-30NA-10LWA 899 0 0 795 177 314
mations. As a result, there have been many efforts in separating 035-0S-20NA-20LWA 899 0 0 530 353 314
the two components [28–31]. 035-50S-30NA-10LWA 440 440 0 794 177 308
In this paper, autogenous shrinkage was calculated from the 035-50S-20NA-20LWA 440 440 0 530 353 308
measured total deformation on cementitious systems of different
compositions by accounting for the thermal effects. Factors affect-
ing the autogenous shrinkage were studied such as the aggregate
content, the addition of slag cement and/or lightweight aggregate cast while the fresh paste and mortar mixes were stored in a sealed bucket. Bleed-
(LWA). This work will hopefully result in improved understanding ing is a common unwanted feature and is especially severe at a low aggregate con-
of autogenous shrinkage characteristics and the mitigation tent, which may interfere the displacement measurement [33]. Regular stirring was
accordingly carried out to create a homogenous mix prior to casting. When no clear
strategies.
sign of bleeding water was observed, the mix was thoroughly agitated before cast-
ing. The standing period ranged from 3 to 9 h, depending on the aggregate content.
Double polystyrene films were used to seal the specimen to prevent external dry-
2. Experimental
ing. Two layers of the 2-mm thick foam rubber were used to separate the sealed
specimen and the rig to minimize the friction. While there was a short period of
2.1. Materials and mix characteristics
temperature fluctuation in the early age due to cement hydration, an isothermal
condition at 20 ± 1 °C was achieved in the long run by circulating water through
Type I portland cement was used as the cementitious material. Fine aggregate
chambers embedded into the lateral sides and bottom of the rigs. Specimen temper-
was silica sand with a fineness modulus of 2.43 and an absorption capacity of
ature was monitored by a thermocouple inserted into the mid-depth of the speci-
1.6%. Coarse aggregate was crushed lime stone with a 19 mm nominal maximum
men. The specimen was positioned with one end fixed to the rig and the other
size and an absorption capacity of 1.2%. A commercially available LWA was used
end connected to a movable plate in contact with a LVDT of a 0.1 lm resolution.
as a partial replacement for fine aggregate. LWA was in saturated surface dry
The specimen displacement and temperature were both measured every 5 min
(SSD) with a SSD specific gravity of 1.77 and an absorption capacity of 10%. The gra-
and the displacement was automatically converted to strain.
dation curves are shown in Fig. 1 for each aggregate.
Mixes consisted of one w/b ratio (0.35), five different aggregate fractions (0%,
20%, 40%, 60% and 78% by volume). In the case of the mixes with 40% and 78% aggre-
2.3. Coefficient of thermal deformation (CTD) measurement
gate contents, two replacement ratios (25% and 50%) were used for portland cement
with slag cement by weight; in addition, two replacement ratios (25% and 50%) for
One 100  200 mm cylinder was prepared for each system the same way as the
normal sand with LWA by volume were used for the mix with 40% aggregate con-
shrinkage test. The mixing date was one day apart between each mix, such that the
tent. Mix design is listed in Table 1. The mix nomenclature is as follows: the first
test age was the same. The cylinders were sealed cured in the mold at 20 °C for
part is the w/b ratio, the middle part is the mass replacement ratio with slag cement
28 days. Then a concrete saw was used to cut two small-scale prisms
and the last part specifies the volume fraction of aggregate (either normal weight
(10  10  90 mm) out of each cylinder. Extra moisture was removed from the
aggregate denoted as NA or lightweight aggregate denoted as LWA, if included).
specimen by a damp cloth and the plastic wrap was used to seal the four lateral
Compressive strength at different ages was tested on the sealed-cured concrete
sides of the specimen to minimize moisture loss during the test.
specimens (0.35-0S-78NA, 0.35-25S-78NA and 0.35-05S-78NA) according to ASTM
CTD of the cementitious systems was measured by a high-precision dilatometer
C 39 [32].
capable of simultaneous displacement and temperature measurements (Fig. 3(a)).
The dilatometer has a length-change resolution of 1.25 nm/digit and a temperature
2.2. Autogenous deformation measurement precision of 0.1 K. The sample holder was customized to accommodate specimens
up to 12 mm  12 mm in cross section and 100 mm in length (Fig. 1(b)) and a liquid
Autogenous shrinkage and temperature were measured simultaneously on nitrogen dewar was equipped for low temperature control. The specimen was
duplicate cementitious systems of 60  100  1000 mm, as shown in Fig. 2. Imme- placed in the sample holder with the thermocouple touching the lateral side for
diately after mixing, the concrete mixes at 60% and 78% aggregate volumes were temperature measurement. The pushrod connecting the LVDT system was in con-
tact with one specimen end for displacement recording (Fig. 3(b)). The specimen
was exposed to a specific temperature profile fluctuating between 10 °C and
40 °C, as shown in Fig. 3(c). Both the cooling and heating rates were 10 °C/h.
100
90
Percentage passing, %

80 3. Results and discussion

70
60 The total deformation and corresponding temperature profile
50 were measured continuously on the 0.35 w/b ratio cementitious
40 systems (paste, mortar and concrete) with different aggregate con-
30 tents (0%, 20%, 40%, 60% and 78%), as shown in Fig. 4. In this study,
20 positive strain denotes shrinkage and negative strain denotes
10 expansion on the y-axis. The zero time on the x-axis represents
0 the mixing time.
25 19 12.5 9.5 4.75 2.36 1.18 0.60 0.30 0.15 <0.15 It can be seen there is good reproducibility between the two
Sieve size, mm specimens in the measured total strain. Specimen temperature is
Lime stone Silica sand LWA well maintained at around 20 °C during the test, except for the
presence of a small short-lived bump in the very early stage. This
Fig. 1. Gradation curves for the aggregates. is a result of the latent heat liberated during cement hydration
 Z. Liu, W. Hansen / Construction and Building Materials 121 (2016) 429–436 431

Fig. 2. Autogenous deformation measurement.

Fig. 3. CTD measurement set-up.

and the maximum temperature rise is reduced with the increase in
aggregate content (Fig. 4).
3.1. Calculation of autogenous shrinkage by accounting for the
temperature effect
The measured total deformation etotal consists of autogenous
and thermal deformations (eau and eth , respectively) (Eq. (1)). Sepa-
ration of the two components needs the knowledge of temperature
history and CTD development. This is further complicated by the
fact that autogenous deformation is a hydration-related property,
and thus temperature dependent as well. Fortunately, specimen
temperature fluctuating between 20 and 32 °C is not a major issue
in this test and the maturity concept (Eq. (2)) may be able to
account for the temperature effect on autogenous deformation
without introducing significant errors [31]. Fig. 5 represents the
total deformation development with the clock time converted to
maturity time at T s = 20 °C (abbreviated as M20 time) using the
equivalent age as the maturity function [34].
etotal ¼ eau þ eth ð1Þ
X Ea 1
te ¼ e R ðT a T s Þ Dt ð2Þ
where t e (days) is the equivalent age at a specified temperature
(20 °C in this case), Ea (J/mol) is the activation energy (normally
45,000 J/mol for type I cement [35]), R (J/mol/K) is the gas constant
(8.314 J/mol/K), T a (K) is the average temperature of the specimen
during the time interval Dt (days), T s (K) is the specified tempera-
ture (293.16 K in this case).
As for the thermal deformation eth , Eq. (3) can be used for its
determination, where CTD is obtained from the length-
temperature measurement on the small-scale specimens (Fig. 6
(a)). Calculated CTD at 28 days is shown in Fig. 6(b) for a range
of aggregate contents. There is a slight difference between the cool-
ing and heating stages, thus an average value is used for the follow-
ing calculation. The development of CTD with time is derived by a
model from [29] (Eq. (4)) with the asymptotic value being the mea-
sured 28-day CTD (Fig. 7). As shown in Fig. 8, the calculated ther-
432 Z. Liu, W. Hansen / Construction and Building Materials 121 (2016) 429–436

1200 Tmax=31.1 ºC 35 1200 Tmax=27.6 ºC 35

1000 30 1000 30

Temperature, °C
Temperature, °C
800 25 800 25

Strain, µm/m

Strain, µm/m
600 20 600 20
400 15 400 15
200 10 200 10
0 5 0 5
0 10 20 30 40 0 10 20 30 40
-200 0 -200 0
Clock time, day Clock time, day
Strain (035-0S-0NA-1) Strain (035-0S-20NA-1)
Strain (035-0S-0NA-2) Strain (035-0S-20NA-2)
Temp. (035-0S-0NA) Temp. (035-0S-20NA)
(a) 035-0S-0NA (b) 035-0S-20NA

1200 Tmax=25.3 ºC 35 1200 Tmax=24.3 ºC 35

1000 30 1000 30

Temperature, °C
Temperature, °C

Strain, µm/m
800 25
Strain, µm/m

800 25
600 20 600 20
400 15 400 15
200 10 200 10
0 5 0 5
0 10 20 30 40 0 10 20 30 40
-200 0 -200 0
Clock time, day Clock time, day
Strain (035-0S-40NA-1) Strain (035-0S-60NA-1)
Strain (035-0S-40NA-2) Strain (035-0S-60NA-2)
Temp. (035-0S-40NA) Temp. (035-0S-60NA)
(c) 035-0S-40NA (d) 035-0S-60NA
1200 Tmax=22.2 ºC 35
1000 30
Temperature, °C

800 25
Strain, µm/m

600 20
400 15
200 10
0 5
0 10 20 30 40
-200 0
Clock time, hour
Strain (035-0S-78NA-1)
Strain (035-0S-78NA-2)
Temp. (035-0S-78NA)
(e) 035-0S-78NA
Fig. 4. Measured total deformation and temperature in 0.35 w/b ratio cementitious systems of different aggregate contents.

mal strain is consistent with the temperature change at different

aggregate contents and levels off after around two days when the
temperature is maintained constant.
1200
Xn  
035-0S-0NA
CTDi þ CTDiþ1
1000 eth ¼ ðT iþ1  T i Þ  ð3Þ
i¼1
2
800 035-0S-20NA
Strain, µm/m

600 0:75t

035-0S-40NA
CTDðtÞ ¼ 77  e 2:5 þ CTD0 ð4Þ
400
where CTD0 is the asymptotic value after the early age, t (hour) is
035-0S-60NA
200 the age of the specimen.
035-0S-78NA
At the same time, a zero time is needed for the autogenous
0
0 10 20 30 40 50 shrinkage characterization. There exists a lot of ambiguity regard-
-200 M20 time, day ing the definition of this point [13,29,36–38]. A qualitative guide-
line typically used is the change of the mix from a fluid state to a
035-0S-0NA 035-0S-20NA 035-0S-40NA
035-0S-60NA 035-0S-78NA
solid state, after which stress generation occurs [13,37]. This corre-
sponds to the transition point on the CTD curves [29] at around
Fig. 5. Measured total deformation against the maturity time at 20 °C. 10 h (Fig. 7), which is consistent with the criterion gauged by the
 Z. Liu, W. Hansen / Construction and Building Materials 121 (2016) 429–436 433

500 40 200

400 35 150
30
300

Temperature, °C

Strain, µm/m
100
Strain, µm/m

25
200 50
20
100 0
15
0.1 1 10 100
0
-50
0 2 4 6 8 10 10
-100 5 -100
Time, day
-200 Time, hour 0
035-0S-0NA 035-0S-20NA 035-0S-40NA
strain (035-0S-0NA) strain (035-0S-60NA)
035-0S-60NA 035-0S-78NA
temp. (035-0S-0NA) temp. (035-0S-60NA)

(a) Typical length and temperature change profiles. Fig. 8. Calculated thermal strain for different aggregate contents.

20 1000
Cooling
Heating 900 035-0S-0NA

15 800
035-0S-20NA
CTD, µm/m/ºC

Strain, µm/m
700
600 035-0S-40NA
10
500
400 035-0S-60NA

5 300 035-0S-78NA
200
100
0
0 0.2 0.4 0.6 0.8 1 0
0 5 10 15 20 25 30 35 40 45 50
Aggregate content by volume
Time, day
(b) Calculated CTD for different aggregate
contents at 28 days. Fig. 9. Calculated autogenous shrinkage by removing the thermal effects (solid
lines) and the simulation by the F-T model (dashed lines).
Fig. 6. CTD calculation from length and temperature measurement.

tation of self-desiccation in the capillary pore system of cement

100
paste as a result of cement hydration [40,41].
s m
e ¼ e0  eð t Þ ð5Þ
80
where e0 is the ultimate autogenous shrinkage strain, s is the time
CTD, µm/m/ºC

60 characteristic and m is the curvature parameter.

40
3.2. Modelling of autogenous shrinkage using Pickett’s model

The presence of discrete aggregate particles restrains concrete

20
shrinkage by reducing the relative content of the shrinking

0
1 10 100 1000 2.5
Time, hour
035-0S-0NA 035-0S-20NA 035-0S-40NA
2.0
035-0S-60NA 035-0S-78NA

Fig. 7. Simulated CTD for different aggregate contents. 1.5

n value

ultrasonic pulse velocity (UPV) value at 1500 m/s [13]. As a result, 1.0
the zero time was selected to be 10 h since the contact of water Va=0.2 y = 2.2228x-0.229 (R² = 0.973)
with cement (or the mixing time) in this study. Subtraction of y = 2.297x-0.162 (R² = 0.999)
0.5 Va=0.4
the thermal strain (Fig. 8) from the total deformation (Fig. 5) then
yields the autogenous shrinkage strain, as shown in Fig. 9. It clears Va=0.78 y = 2.6376x-0.12 (R² = 0.972)
illustrates the restraining effect of aggregate particles on the 0.0
1 10 1 00
shrinkage of the cementitious matrix that is gradually reduced
with an increasing aggregate content. Furthermore, it is noted M20 time, day
the development of autogenous shrinkage can be modelled by Fig. 10. Back-calculated aggregate restraining factor as a function of time for the
the F-H model (Eq. (5)), originally for fitting cement hydration 0.35-0S mixes of three different aggregate contents (90-day shrinkage strain is
curves [39], which validates autogenous shrinkage as the manifes- extrapolated using the F-H model).
434 Z. Liu, W. Hansen / Construction and Building Materials 121 (2016) 429–436

3.0 -0.3

2.5

2.0 -0.2

Parameter
Parameter
1.5 y = 0.739x + 2.0459
R² = 0.9687
1.0 -0.1
y = 0.1778x - 0.2521
0.5 R² = 0.9075

0.0 0.0
0 0.5 1 0 0.2 0.4 0.6 0.8 1

Aggregate content by volume Aggregate content by volume

Fig. 11. Correlation between parameters a and b in Eq. (7) with aggregate content.

cementitious matrix and also by creating internal resistance to the 80

shrinkage as a non-shrinkage phase. Pickett derived a composite 035-0S-78NA
70 035-25S-78NA

Compressive strength, MPa

model by simplifying concrete into a single spherical non- 035-50S-78NA
shrinking aggregate particle coated by a homogeneous spherical 60
shrinking body [13]. The relative shrinkage ratio eePc is governed by
50
the aggregate content V c and a restraining factor n (Eq. (6)).
40
ec
¼ ð1  V c Þn ð6Þ
eP 30

20
where ec is the autogenous shrinkage strain of concrete or mortar,
eP is the autogenous shrinkage strain of paste. 10
As shown in Fig. 10, the restraining factor n calculated at a ser- 0
ies of ages and aggregate contents is found to be time-dependent in 1 3 7 14 28
the form of Eq. (7). The 035-0S-60NA mix is intentionally omitted Age, day
in Fig. 10 and will be reserved for independent validation.
Fig. 13. Compressive strength of concrete mixes at 78% aggregate volume and
n ¼ a  tb ð7Þ different slag cement contents.

It can be seen in Fig. 11 that there is a fairly good linear corre-

lation in the two parameters a and b when plotted against the
aggregate content. A similar finding is reported in [38] as well.
When the interpolated parameters are inserted back into the
improved Pickett’s model for the 035-0S-60NA mix, the predicted
shrinkage fits well the measured one (Fig. 12).
3.3. Slag cement effect
Hydration of slag cement alone is a very slow process requiring
the break-down and dissolution of the glassy structure. Its reaction
rate can be accelerated by the hydroxyl ions liberated during the
hydration of portland cement, which involves a two-stage reaction
[42]. The initial reaction is with alkali hydroxide while the subse-
quent reaction is primarily with calcium hydroxide, the latter also
being known as the pozzolanic reaction. This process consists of
the consumption of a porous product (calcium hydroxide) and
the precipitation of a dense product (calcium silicate hydrate),
which leads to a more refined and condensed pore structure
[4,43]. This can be seen from a higher compressive strength in
the later ages for the concrete mixes with slag cement (Fig. 13).
The finer capillary pore system in turn creates a lower relative
humidity, accompanied by increased self-desiccation, as evidenced
by the larger long-term autogenous shrinkage in the binary cemen-
titious systems in Fig. 14.

250
800
Measured shrinkage strian, µm/m

200 700 25% slag cement

Strain, µm/m

600
150
500
50% slag cement
100 400
Control mix
300
50
200
0 100
0 10 20 30 40
0
M20 time, day 0.1 1 10 100
035-0S-60NA (measured) 035-0S-60NA (modelled) M20 time, day
035-0S-40NA 035-25S-40NA 035-50S-40NA
Fig. 12. Measured and predicted shrinkage strain development for the 035-0S-
60NA mix by the improved Pickett’s model using a time-dependent restraining Fig. 14. Autogenous shrinkage strain in 0.35 w/b ratio mortar systems of different
factor. slag cement contents.
 Z. Liu, W. Hansen / Construction and Building Materials 121 (2016) 429–436 435

800 scaled shrinkage of a pure portland cement system by assuming

035-0S-78NA the slag cement is non-reactive [38], as shown in the hatched area
700 035-25S-78NA
035-50S-78NA of Fig. 16. The contribution from pozzolanic reaction increases with
600 time and a higher slag cement content yields more contribution. In
addition, the initial overlap of the two curves further confirms the
Strain, µm/m

500
‘‘diluting” effect of slag cement.
400 The negative effect of slag cement on the autogenous shrinkage
300
can be nullified by the incorporation of LWA in concrete mix, in
which LWA acts as a water storage agent that replenishes the con-
200 sumption of pore water via the continuous release of retained
100 moisture [41,44,45]. This is clearly shown in Fig. 17 that the auto-
genous shrinkage of a 0.35 w/b mortar with LWA is substantially
0 reduced. It is worth noting that internal curing by LWA leads to a
0 10 20 30 40 50 60 70 80
significant initial expansion happening mostly within two days,
Compressive strength, MPa
which may be attributed to the cement paste expansion cured
Fig. 15. Correlation between autogenous shrinkage strain and compressive under a saturated condition [41]. As for the blended mix with
strength. 50% slag cement replacement, an addition of 25% LWA is not suffi-
cient to mitigate the development of autogenous shrinkage.

However, the pozzolanic reaction lags behind cement hydration 4. Conclusions

at the early hydration stage. Slag cement particles act more like
inert fillers and thus the amount of reactive cementitious materials This study covers an extensive experimental program on the
is reduced. This is demonstrated by their lower compressive measurement of autogenous shrinkage in a wide range of cemen-
strength (Fig. 13) and the reduced the autogenous shrinkage in titious systems. The following major conclusions are obtained.
the very early age (Fig. 14). This is called the ‘‘diluting” effect of
slag cement [38]. The resemblance in slag cement effect on the (1) Thermal effects is accounted for by the maturity concept and
shrinkage and strength developments is further illustrated by a the coefficient of thermal deformation (CTD), from which
good correlation between the autogenous shrinkage strain and autogenous shrinkage is separated from the measured total
compressive strength (Fig. 15). deformation. Zero time for autogenous shrinkage measure-
The contribution of slag cement towards the total autogenous ment is determined by the transition point on the modelled
shrinkage development can be quantified by the difference CTD development with time, corresponding to a fluid-solid
between the measured shrinkage of a blended system and the transition of the cementitious mix.

800 800

Contribution Contribution
600 600
Strain, µm/m

Strain, µm/m

from slag cement from slag cement

400 400

200 200

0 0
0.1 1 10 100 0.1 1 10 100
M20 time, day M20 time, day
25S 0.75×control 50S 0.5×control

Fig. 16. Added shrinkage strain due to slag cement.

800 800
Shrinkage strian, µm/m
Shrinkage strian, µm/m

600 600

400 400

200 200

0 0
0 10 20 30 40 0 10 20 30 40
-200 -200

-400 M20 time, day -400 M20 time, day

035-0S-40NA 035-50S-40NA
035-0S-30NA-10LWA 035-50S-30NA-10LWA
035-0S-20NA-20LWA 035-50S-20NA-20LWA

Fig. 17. Autogenous shrinkage strain in mortar mix of 40% sand volume with and without LWA.
436 Z. Liu, W. Hansen / Construction and Building Materials 121 (2016) 429–436
(2) A wide selection of aggregate contents (0%, 20%, 40%, 60%
and 78%) in the autogenous shrinkage measurement yields
an improved Pickett’s model for evaluating the aggregate
effect, in which a time-dependent aggregate retraining fac-
tor is used.
(3) Partial replacement with slag cement initially leads to a sup-
pressed autogenous shrinkage due to its ‘‘diluting” effect,
but ultimately causes an increase in the autogenous shrink-
age because of the pozzolanic reaction. This is found to
resemble the development of compressive strength in con-
crete mixes. The long-term increased autogenous shrinkage
by slag cement can be mitigated by the introduction of LWA
as a partial replacement of normal sand.
Acknowledgements
The authors would like to thank Jan Pantolin (Technical Services
Supervisor) and Bo Meng (Graduate Student Research Assistant),
Department of Civil and Environmental Engineering, for their assis-
tance in laboratory work.
References
[1] J. Kropp, Relations between transport characteristics and durability, in:
Proceedings Performance Criteria for Concrete Durability, Chapman & Hall,
London, 1995, pp. 65–89.
[2] M. Pigeon, R. Pleau, Durability of concrete in cold climates, in: Modern
Concrete Technology Series 4, E & FN SPON, 1995.
[3] P.-C. Aïtcin, High-Performance Concrete, E & FN SPON, New York, 1998, p. 568.
[4] D. Manmohan, P.K. Mehta, Influence of pozzolanic, slag, and chemical
admixtures on pore size distribution and permeability of hardened cement
pastes, Cem. Concr. Agg. 3 (1) (1981) 63–67.
[5] X.D. Chen, S.X. Wu, Influence of water-to-cement ratio and curing period on
pore structure of cement mortar, Constr. Build. Mater. 38 (2013) 804–812.
[6] B. Persson, G. Fagerlund, Self-desiccation and its importance in concrete
technology, in: Proceedings of the Third International Research Seminar in
Lund, 2001, p. 257.
[7] L.E. Copeland, R.H. Bragg, Self-desiccation in Portland cement pastes, in:
Research Bulletin RX052, Portland Cement Association, Skokie, Illinois, 1955, p.
112.
[8] E. Tazawa, S. Miyazawa, Autogenous shrinkage of concrete and its importance
in concrete technology, in: Proceedings of the Fifth International RILEM
Symposium on Creep and Shrinkage of Concrete, Barcelona, Spain, 1993, pp.
159–168.
[9] B. Persson, Self-desiccation and its importance in concrete technology, Mater.
Struct. 30 (1997) 293–305.
[10] E. Tazawa, S. Miyazawa, Influence of cement and admixture on autogenous
shrinkage of cement paste, Cem. Concr. Res. 25 (1995) 281–287.
[11] S.N. Lim, T.H. Wee, Autogenous shrinkage of ground-granulated blast-furnace
slag concrete, ACI Mater. J. 97 (2000) 587–593.
[12] P. Lura, K. Van Breugel, I. Maruyama, Effect of curing temperature and type of
cement on early-age shrinkage of high-performance concrete, Cem. Concr. Res.
31 (2001) 1867–1872.
[13] 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.
[14] 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 (2011) 13–20.
[15] A.M. Pailere, M. Buil, J.J. Serrano, Effect of fiber addition on the autogenous
shrinkage of silica fume concrete, ACI Mater. J. 86 (2) (1989) 139–144.
[16] S. Miyazawa, E. Tazawa, T. Sato, K. Sato, Autogenous shrinkage of ultra high
strength concrete caused by restraint of reinforcement, Trans. Jpn. Concr. Inst.
15 (1993) 115–122.
[17] O.M. Jensen, P.F. Hansen, Autogenous deformation and change of the relative
humidity in silica fume-modified cement paste, ACI Mater. J. 93 (6) (1996)
1238–1249.
[18] G. Pickett, Effect of aggregate on shrinkage of concrete and a hypothesis
concerning shrinkage, ACI J. Proc. 51 (1) (1956) 581–590.
[19] T.C. Hansen, K.E.C. Nielsen, Influence of aggregate properties on concrete
shrinkage, ACI J. 62 (7) (1965) 783–793.
[20] D.W. Hobbs, The dependence of the bulk modulus, Young’s modulus, creep,
shrinkage and thermal expansion of concrete upon aggregate volume
concentration, Mater. Struct. 4 (2) (1971) 107–114.
[21] W. Hansen, Constitutive model for predicting ultimate drying shrinkage of
concrete, J. Am. Ceram. Soc. 70 (5) (1987) 329–332.
[22] Z.C. Grasley, Measuring and modeling the time-dependent response of
cementitious materials to internal stresses (Ph.D. thesis), University of
Illinois at Urbana-Champaign, Urbana, IL, 2006.
[23] D.P. Bentz, O.M. Jensen, Mitigation strategies for autogenous shrinkage
cracking, Cem. Concr. Compos. 26 (2004) 677–685.
[24] A. Radlinska, F. Rajabipour, B. Bucher, R. Henkensiefken, G. Sant, J. Weiss,
Shrinkage mitigation strategies in cementitious systems: a closer look at
differences in sealed and unsealed behavior, Transp. Res. Rec. J. Transp. Res.
Board 2070 (2008) 59–67.
[25] M. Jose Oliveira, A.B. Ribeiro, F.G. Branco, Combined effect of expansive and
shrinkage reducing admixtures to control autogenous shrinkage in self-
compacting concrete, Constr. Build. Mater. 52 (2014) 267–275.
[26] D.P. Bentz, W.J. Weiss, Internal Curing : A 2010 State-of-the – Art Review,
National Institute of Standards and Technology (NIST) Report 7765, 2011,
p. 82.
[27] J.S.M. Shilstone, Concrete mixture optimization, Concr. Inter. 12 (6) (1990)
33–39.
[28] O.M. Jensen, P.F. Hansen, Influence of temperature on autogenous deformation
and relative humidity change in hardening cement paste, Cem. Concr. Res. 29
(1999) 567–575.
[29] A. Loukili, D. Chopin, A. Khelidj, J.Y. Le Touzo, A New approach to determine
autogenous shrinkage of mortar at an early age considering temperature
history, Cem. Concr. Res. 30 (2000) 915–922.
[30] E.J. Bjøntegård, Ø. Sellevold, Interaction between thermal dilation and
autogenous deformation in high performance concrete, Mater. Struct. 34
(2001) 266–272.
[31] P. Turcry, A. Loukili, L. Barcelo, J.M. Casabonne, Can the maturity concept be
used to separate the autogenous shrinkage and thermal deformation of a
cement paste at early age?, Cem Concr. Res. 32 (2002) 1443–1450.
[32] ASTM C 39, Standard Test Method for Compressive Strength of Cylindrical
Concrete Specimens, ASTM International, West Conshohocken, PA, 2009.
[33] T.A. Hammer, Ø. Bjontegaard, E.J. Sellevold, Measurement methods for testing
of early age autogenous strain, in: Proceedings of Early Age Cracking in
Cementitious Systems, Cachan, France, 2002, pp. 207–215.
[34] ASTM C 1074, Standard Practice for Estimating Concrete Strength by the
Maturity Method, ASTM International, West Conshohocken, PA, 2009.
[35] R.C. Tank, N.J. Carino, Rate constant functions for strength development of
concrete, ACI Mater. J. 88 (1991) 74–83.
[36] P.-C. Aitcin, Dymystifying autogenous shrinkage, Concr. Int. 21 (1999) 54–
56.
[37] M.S. Meddah, P.C. Aitcin, N. Petrov, A new approach for the determination of
the starting point of autogenous shrinkage strains measurement (ASS), ACI
Mater. J. 234 (2006) 473–484.
[38] Y. Wei, Modelling of autogenous deformation in cementitious materials,
restraining effect from aggregate, and moisture warping in slabs on grade (Ph.
D. thesis), University of Michigan, Ann Arbor, USA, 2008.
[39] P.F. Hansen, E.J. Pedersen, Maturity computer for controlled curing and
hardening of concrete, Nordisk Betong 1 (1977) 21–25 (In Danish).
[40] E.A.B. Koenders, Simulation of volume changes in hardening cement-based
materials (Ph.D. thesis), Delft University of Technology, Delft, The Netherlands,
1997.
[41] P. Lura, Autogenous deformation and internal curing of concrete (Ph.D. thesis),
Delft University of Technology, Delft, The Netherlands, 2003.
[42] D.M. Roy, G.M. Idorn, Hydration, structure and properties of blast-furnace slag
cements, mortars and concrete, ACI J. Proc. 79 (6) (1982) 445–457.
[43] R.F.M., Bakker, Permeability of blended cement concretes, ACI S.P. (Fly ash,
Silica fume, Slag and Other Mineral By-products in Concrete) 79 (1983) 589–
605.
[44] D.P. Bentz, E.A.B. Koenders, S. Monnig, H.-W. Reinhardt, K. van Breugel, G. Ye,
Materials science-based models in support of internal water curing, in: K.
Kovler, O.M. Jensen (Eds.), RILEM State-of-the-Art Report ‘‘Internal Curing of
Concrete”, RILEM Publications S.A.R.L, France, 2007, pp. 29–43.
[45] A. Bentur, S.I. Igarashi, K. Kovler, Prevention of autogenous shrinkage in high-
strength concrete by internal curing using wet lightweight aggregates, Cem.
Concr. Res. 31 (11) (2001) 1587–1591.