Cement & Concrete Composites 26 (2004) 347–357

www.elsevier.com/locate/cemconcomp

Effect of silica fume on mechanical properties

of high-strength concrete
a,*
M. Mazloom , A.A. Ramezanianpour b, J.J. Brooks c

a
Department of Civil Engineering, Shahid Rajaee University, Post Code 16788, P.O. Box 16785-163, Tehran, Iran
b
Civil Engineering Department, Amirkabir University, P.O. Box 15875-4413, Tehran, Iran
c
School of Civil Engineering, University of Leeds, LS2 9JT, UK

Abstract
This paper presents the results of experimental work on short- and long-term mechanical properties of high-strength concrete
containing different levels of silica fume. The aim of the study was to investigate the effects of binder systems containing different
levels of silica fume on fresh and mechanical properties of concrete. The work focused on concrete mixes having a fixed water/binder
ratio of 0.35 and a constant total binder content of 500 kg/m3 . The percentages of silica fume that replaced cement in this research
were: 0%, 6%, 10% and 15%. Apart from measuring the workability of fresh concrete, the mechanical properties evaluated were:
development of compressive strength; secant modulus of elasticity; strain due to creep, shrinkage, swelling and moisture movement.
The results of this research indicate that as the proportion of silica fume increased, the workability of concrete decreased but its
short-term mechanical properties such as 28-day compressive strength and secant modulus improved. Also the percentages of silica
fume replacement did not have a significant influence on total shrinkage; however, the autogenous shrinkage of concrete increased as
the amount of silica fume increased. Moreover, the basic creep of concrete decreased at higher silica fume replacement levels. Drying
creep (total creep
basic creep) of specimens was negligible in this investigation. The results of swelling tests after shrinkage and
creep indicate that increasing the proportion of silica fume lowered the amount of expansion. Because the existing models for
predicting creep and shrinkage were inaccurate for high-strength concrete containing silica fume, alternative prediction models are
presented here.
Ó 2003 Elsevier Ltd. All rights reserved.

Keywords: High-strength concrete; Silica fume; Compressive strength; Elastic modulus; Creep; Shrinkage; Swelling

1. Introduction which are produced by the hydration of ordinary Port-

land cement. The main results of pozzolanic reactions
Nowadays high-strength and high-performance con- are: lower heat liberation and strength development;
crete are widely used throughout the world and to lime-consuming activity; smaller pore size distribution.
produce them it is necessary to reduce the water/binder In high-performance concrete, which contains high
ratio and increase the binder content. Superplasti- quality and expensive materials, cracking provides the
cisers are used in these concretes to achieve the required greatest concerns for the designers because harmful
workability; moreover, different kinds of cement materials can penetrate from them to the concrete easily
replacement materials are usually added to them be- and start to destroy it and also corrode reinforcement.
cause a low porosity and permeability are desirable. Some of these cracks are related to drying and auto-
Silica fume is the one of the most popular pozzolanas, genous shrinkage of concrete. Therefore, to improve the
whose addition to concrete mixtures results in lower durability of high-strength concrete, its autogenous and
porosity, permeability and bleeding because their oxides drying shrinkage should be addressed and necessary
(SiO2 ) react with and consume calcium hydroxides, work on its mix design should be done to minimize
them. It is worth noting that autogenous shrinkage of
concrete is because of chemical reactions during the
*
Corresponding author. Tel.: +98-21-7844025; fax: +98-21- hydration of cementitious materials; nevertheless, dry-
2935040. ing shrinkage occurs as a result of moisture movement
E-mail address: moospoon@yahoo.com (M. Mazloom). from concrete to the atmosphere.

0958-9465/$ - see front matter Ó 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0958-9465(03)00017-9
348 M. Mazloom et al. / Cement & Concrete Composites 26 (2004) 347–357

High-performance concrete should be controlled Table 2

during its early ages. For instance, fresh concrete may Mix proportions of concrete containing different levels of silica fume
bleed or coarse aggregates may separate from the paste. Mix components Concrete mixes
Also its volume changes at this age are very important. OPC SF6 SF10 SF15
Of course before initial set, concrete has plastic prop- Cement (kg/m3 ) 500 470 450 425
erties and has high-tension strain capacity; conse- Silica fume (kg/m3 ) – 30 50 75
quently, the possibility of cracking in it is low. However, Superplasticiser (kg/m3 ) 8.17 9.78 11.71 13.34
shrinkage at this age may weaken the transition zone Gravel: 1203 kg/m3 , sand: 647 kg/m3 , water: 175 kg/m3 , W =b ¼ 0:35.
between aggregates and paste, thus drying shrinkage
cracking may increase in future [1]. Another point is
initial set and the creation of the internal skeleton of containing different levels of silica fume are given in
concrete. At this age tension strain capacity of concrete Table 2. Crushed granite sand and gravel with a nominal
is too low and therefore it is very sensitive to shrinkage maximum size of 10 mm were used as the aggregates.
cracks [2]. Moreover, Holt [1] shows the maximum The control mix was cast using OPC, while the other
autogenous shrinkage often occurs at this age. mixes were prepared by replacing part of the cement
Current methods of estimating movements in con- with silica fume at four different replacement levels on
crete, such as ACI [3] and CEB [4], do not apply for mass-for-mass basis. The water/cement ratio and the
high-performance concrete. They were developed for slump of control high-strength concrete were 0.35 and
plain concrete before the extensive use of mineral and 100  10 mm, respectively. The same water/binder ratio
chemical admixtures. In addition, shrinkage of high- of 0.35 was used for the other concrete mixes with the
performance concrete has a significant contribution same amount of slump. Consequently, the dosage of
from autogenous shrinkage and less contribution from superplasticiser changed due to the effect of the different
drying shrinkage; only the latter is considered by pre- levels of silica fume. The superplasticiser used is based
diction methods. In order to develop existing or new on melamine formaldehyde and lignosulfonate.
prediction models, experimental data are needed.
This paper compares strength, elasticity, shrinkage,
swelling, creep and moisture movement of high-strength 3. Test procedure
concretes containing different levels of silica fume.
For each mix, the following specimens were made: 24
2. Materials and mix proportions 100 mm cubes for compressive strength; eight 80  270
ðdiameter  lengthÞ mm cylinders for creep; four
The cementitious materials used were ordinary Port- 80  270 mm and four 150  300 mm cylinders for
land cement (OPC) and silica fume (SF), their chemical shrinkage; two 80  270 mm and two 150  300 mm
compositions and physical properties being given in cylinders for swelling. After being de-moulded at the age
Table 1. Details of the mix proportions for the concrete of one day, all creep and shrinkage specimens were
cured in water at 20  2 °C until the age of 7 days.
Table 1 Subsequently, half of them were sealed and then all
Chemical composition and physical properties of cementitious mate- creep and shrinkage specimens were kept in a controlled
rials environment of 20  2 °C and 50  5% relative humidity
Item Cementitious materials, % throughout the test duration. Compressive strength and
Ordinary Portland cement Silica fume swelling specimens remained in 20  2 °C water all the
SiO2 21.46 91.70 time.
Al2 O3 5.55 1 The specimens of autogenous shrinkage and basic
Fe2 O3 3.46 0.9 creep were sealed with aluminium waterproofing tape,
CaO 63.95 1.68 which was found to be very effective since the specimens
MgO 1.86 1.8
showed very minimal weight loss. The OPC control mix
Cl – 0.08
SO3 1.42 0.87 recorded a weight loss of 0.03% after 240 days. This is
K2 O 0.54 – well within the limit of 0.05% as recommended by the
Na2 O 0.26 – JCI [5]. The measurement of shrinkage and creep was
LOI – 2 performed using a mechanical Demec gauge of 100 mm
Compounds
gauge length at four circumferential positions of the
C3 S 50.96 –
C2 S 23.1 – specimens. This measurement commenced at about 7
C3 A 8.85 – days after casting, i.e. after specimens were removed
C4 AF 10.53 – from the water tank. The measurement of swelling
Fineness started at about 24 h after casting, i.e. after specimens
SSA (m2 /kg) 330 14,000
were de-moulded and stored in water.
 M. Mazloom et al. / Cement & Concrete Composites 26 (2004) 347–357 349

Compressive strength was determined according to product preventing further reaction of SF with calcium
BS 1881: Part 116: 1983 at various ages over a period of hydroxide beyond 90 days. In the case of the control
400 days. The apparatus and method of strain mea- concrete, hydration is at a less advanced stage and
surement used for shrinkage, swelling and creep were strength still shows significant increases. According to
those as described previously [6]. For the determination Neville and Brooks [9] the 365-day/28-day strength ratio
of total and basic creep, total and autogenous shrinkage is about 1.25 in concrete without mineral admixtures,
were treated as being additive respectively, i.e. creep is which is quite near the results of this experiment. At the
defined as the change in deformation since application age of 400 days, the compressive strength of control
of load, corrected for shrinkage of companion load-free concrete and concrete mixes containing different pro-
specimen. At the ages of 7 and 28 days, creep specimens portions of silica fume were the same. However, at the
were subjected to a sustained compressive stress of 10 age of 28 days, the strength of concrete containing 15%
MPa, which corresponded to a range of initial stress/ silica fume was about 21% more than that of control
strength ratio of 0.14 to 0.22, until the age of 400 days. concrete. Therefore, the inclusion of silica fume in con-
crete mixture, mainly affects short-term strength of
concrete. It is of interest to compare the strength of dry-
4. Results and discussion stored concrete with that of the wet-stored. The casting
programme included three specimens of each mix that
4.1. Workability of fresh concrete were cured in water for 7 days and then kept in the
control room of 20  2 °C and 50  5% relative humidity
Table 2 shows the dosage of superplasticiser which until the age of 400 days. The results of their 400-day
was necessary for mixes containing different levels of compressive strength and moist cured specimens for 400
silica fume to have a constant slump of 100  10 mm, days are compared in Table 4. It can be seen that moist
measured according to BS 1881: Part 102: 1983. It can curing has improved the compressive strength of control
be observed that the mixes incorporating higher silica concrete about 10% but it did not have any significant
fume content tended to require higher dosages of su- effects on specimens containing silica fume. It is worth
perplasticiser. The higher demand of superplasticiser adding that utilizing silica fume contributed to the
with the concrete containing silica fume can be attrib- compressive strength of dry-stored specimens by about
uted to the very fine particle size of silica fume that 9% at the age of 400 days. However, it had no influence
causes some of the superplasticiser being adsorbed on its on long-term strength of wet-stored concrete. According
surface [7]. It is worth adding that mixes incorporating to the Concrete Society [10] concrete containing silica
more silica fume were more cohesive and this is in fume should be moist cured at least for 7 days.
agreement with the findings of Khatri and Sirivivat-
nanon [7].
4.3. Elastic modulus

4.2. Compressive strength The results of secant modulus of elasticity of concrete

specimens containing different levels of silica fume,
For concrete stored in water, the development of which were obtained in the creep tests, are shown in
compressive strength with age is shown in Table 3. It can Table 5. In fact, the cylindrical specimens of 80  270
be seen that the compressive strength development of mm height were loaded at the ages of 7 and 28 days.
concrete mixtures containing silica fume was negligible Because secant modulus is related to the level of applied
after the age of 90 days; however, there were 26% and stress and also loading rate, all the specimens of this
14% strength increases in the control concrete after one research were subjected to a stress of 10 MPa and the
year compared to its 28 and 90 days strength, respec- time taken to apply it was about 10 min; the stress/
tively. The difference in strength development in OPC strength ratio was between 0.14 and 0.22. As shown in
concrete and SF concrete can be attributed [8] to Table 5, increasing the silica fume replacement level in-
the rapid formation of an inhibiting layer of reaction creases the secant modulus of concrete. Also ACI 318-95
[11] has predicted the static modulus of the investigated
Table 3
Development of compressive strength with age (MPa) Table 4
Compressive strength of dry and moist cured specimens after 400 days
Concrete Age (days) (MPa)
mixes
7 14 28 42 90 365 400
Method of Concrete mixes
OPC 46 52 58 62 64 73 74 curing
SF6 50.5 58 65 69 71 73 73 OPC SF6 SF10 SF15
SF10 52 61 67.5 71 74 74 73 Moist 74 73 73 76
SF15 53 63 70 73 76 75 76 Dry 67 72 73 74
350 M. Mazloom et al. / Cement & Concrete Composites 26 (2004) 347–357

Table 5 Table 6
7- and 28-day compressive strength and secant modulus of elasticity 400-day compressive strength and elastic modulus recovery
Kind and age of Compressive Measured Predicted Kind of concrete and Measured Compressive Predicted
concrete strength modulus modulus by age of loading modulus strength modulus by
(MPa) (GPa) Eq. (1) (GPa) (GPa) (MPa) Eq. (1) (GPa)
OPC 7 days 46 28.8 30.2 OPC 7 days 38 74 38.4
28 days 58 34.4 34 28 days 38.12
SF6 7 days 50.5 31 31.7 SF6 7 days 38.23 73 38.1
28 days 65 35.5 36 28 days 38.1
SF10 7 days 52 31.1 32.2 SF10 7 days 38.21 73 38.1
28 days 67.5 37 36.6 28 days 38.3
SF15 7 days 53 31.5 32.5 SF15 7 days 38.64 76 38.9
28 days 70 38.1 37.3 28 days 38.75

specimens properly. This is in agreement with the pre- ulus recovery. Also the ACI 318-95 [11] (Eq. (1)) could
vious findings [12]. It is worth noting that sealed and predict this modulus with acceptable accuracy.
drying specimens had similar values of secant modulus.
ACI 318-95 [11] presents the following equation to cal- 4.4. Swelling
culate elastic modulus:
Table 7 shows the swelling characteristics for the
0:5
Ec ¼ 4:7ðfc Þ ð1Þ various types of concrete with different proportions of
silica fume. These results are related to the cylindrical
where fc is the compressive strength of standard cylinder specimens of 80  270 mm and 150  300 mm height. It
specimen of 150  300 mm height in MPa and Ec is static is worth noting that the latter specimens exhibit very
modulus in GPa. little decrease in swelling. This could be explained sim-
Because 100 mm cube specimens were utilized to ply in terms of the larger diameter of the 150  300 mm
measure the compressive strength, a factor of 0.9 has height specimens, which cause more difficulties for water
been used to estimate the equivalent cylinder strength. to penetrate their internal layers and also self-desication.
This factor is greater than the usual factor of 0.8 that Increasing the silica fume replacement level decreased
is generally used for lower strength concrete, and was the swelling of concrete at all ages. Rao [14] and also
chosen after considering recent data presented by Imam Fattuhi and Al-Khaiat [15] agree with this finding. It has
et al. [13]. also been reported that increasing the level of silica fume
Elastic modulus recovery was calculated by dividing lowers the permeability of concrete [16] and, conse-
the sustained 10 MPa stress to instantaneous strain re- quently, the penetration of water decreases.
covery at the time of unloading. It should be mentioned
that the specimens loaded at the ages of 7 and 28 days, 4.5. Shrinkage
remained under the load for 396 and 376 days, respec-
tively, so that they were unloaded at the ages of 403 and Figs. 1 and 2 show the shrinkage of cylindrical
404 days, respectively. The compressive strength of specimens of 80  270 mm and 150  300 mm height,
concrete at these ages was assumed to be equal to their respectively. Also shown are the predictions by the ACI
400-day strength. Table 6 shows the measured elastic 209 [3] and the CEB-FIP 1990 [4] methods, which were
modulus recovery, compressive strength and estimated developed for drying shrinkage only, i.e. they do not
elastic modulus recovery according to the ACI 318-95 really apply to autogenous (sealed) shrinkage. Figs. 1
[11]. As shown in this table, loading age and silica fume and 2 show that at early ages both methods tend to
replacement level had no influence on the elastic mod- underestimate the shrinkage of drying specimens but, at

Table 7
Swelling of 80  270 mm and 150  300 mm high specimens in microstrain
Age (days) Concrete mixes
OPC SF6 SF10 SF15
80  270 mm 150  300 mm 80  270 mm 150  300 mm 80  270 mm 150  300 mm 80  270 mm 150  300 mm
25 128 124 112 106 96 93 64 61
100 144 140 112 108 128 124 80 78
425 168 165 152 148 152 150 128 125
 M. Mazloom et al. / Cement & Concrete Composites 26 (2004) 347–357 351

OPC OPC
600
700
ACI ACI
600 500 Eq. 2 (drying)
Eq. 2 (drying)
Shrinkage

500

Shrinkage
400
400 CEB CEB
300
300
Eq. 2 (sealed) 200
200
100 100
0 0
0 200 400 600 800
0 200 400 600 800
Time (days) Time (days)

SF6 SF6
700 600
ACI ACI
600 500 Eq. 2 (drying)
Shrinkage

500

Shrinkage
Eq. 2 (drying) 400
400
CEB 300 CEB
300 Eq. 2 (sealed)
200 200
100 100
0
0 200 400 600 800 0
0 200 400 600 800
Time (days)
Time (days)
SF1 0
SF10
700 600
ACI ACI
600
Eq. 2 (drying) 500 Eq. 2 (drying)
500
Shrinkage

CEB
Shrinkage

400 400 CEB

300 300
200 Eq. 2 (sealed) 200
100
100
0
0 200 400 600 800 0
0 200 400 600 800
Time (days)
Time (days)
SF1 5
700 SF15
ACI 600
600
Eq. 2 (drying) 500
500 ACI Eq. 2 (drying)
Shrinkage

Shrinkage

400 CEB 400

CEB
300 300

200 Eq. 2 (sealed) 200

100 100
0 0
0 200 400 600 800 0 200 400 600 800
Time (days) Time (days)

Fig. 1. Shrinkage of 80  270 mm high specimens (microstrain). Fig. 2. Shrinkage of 150  300 mm high specimens (microstrain).

later ages, the CEB and ACI underestimate and over- fume percentage, esh ðtÞ ¼ shrinkage after t days of dry-
estimate total shrinkage, respectively. For the shrinkage ing.
determined in this investigation the following expression Figs. 1 and 2 demonstrate the improvements in pre-
was developed using regression analysis: dictions using Eq. (2) compared with those by the ACI
ðtÞ and CEB methods. Tables 8 and 9 compare the final
esh ðtÞ ¼ 516 y 10
6 ð2Þ measured shrinkage of 80  270 mm and 150  300 mm
ð0:3SF þ 12:6Þ þ ðtÞ
high specimens with the results of different predic-
where y ¼ yd ¼ 1:14
0:007ðV =SÞ P ys (for drying tion methods. According to Neville et al. [17] the error
specimens), or y ¼ ys ¼ 0:014SF þ 0:39 (for sealed coefficients of the above-mentioned models can be
specimens), V =S ¼ volume to surface ratio, SF ¼ silica compared by the following equation:
352 M. Mazloom et al. / Cement & Concrete Composites 26 (2004) 347–357

Table 8 Table 10
Shrinkage of 80  270 mm high specimens after 587 days of drying Values of total, autogenous and drying shrinkage of 80  270 mm high
(microstrain) specimens on completion of the tests (microstrain)
Concrete mixes Measured Predicted value Kind of Concrete mixes
value shrinkage
Eq. (2) ACI [3] CEB [4] OPC SF6 SF10 SF15
OPC 532 505 604 452 Total 532 528 523 512
SF6 528 504 604 412 Autogenous 198 231 264 297
SF10 523 503 604 397 Drying 334 297 259 215
SF15 512 501 604 382
Error 4.0 15.4 22.5
coefficient (%) genous shrinkage. This is in agreement with the results
of other researchers [19–23].
Generally, the levels of shrinkage of drying concrete
Table 9 seemed quite large for a water/binder ratio of 0.35. This
Shrinkage of 150  300 mm high specimens after 406 days of drying fearure can be attributed to a lower volume of aggregate
(microstrain) in the high-strength concrete, which acts as a restraint to
Concrete mixes Measured Predicted value shrinkage of the cement paste.
value
Eq. (2) ACI [3] CEB [4] Table 10 shows total shrinkage, autogenous shrink-
age and also the difference between them, which is
OPC 468 445 554 409
SF6 462 443 554 373 drying shrinkage, for the 80  270 mm high specimens
SF10 446 442 554 359 after 587 days of drying. As it can be seen, there is sig-
SF15 435 441 554 346 nificant increase in autogenous shrinkage at high levels
Error 3.1 23.3 18.2 of silica fume. In fact, inclusion of 10% and 15% silica
coefficient (%)
fume increases the autogenous shrinkage of concrete by
33% and 50%, respectively. The effect of silica fume on
" #0:5 autogenous shrinkage can be explained by its influence
1 X ½CmðtÞ
CpðtÞ2 on the pore structure and pore size distribution of
M¼ 100 ð3Þ concrete as well as its pozzolanic reaction. According to
CavðtÞ n
Sellevold [24] the inclusion of silica fume at high re-
where M ¼ error coeficient, CmðtÞ ¼ observed shrinkage placement levels significantly increases the autogenous
or creep after time t, CpðtÞ ¼ predicted shrinkage or shrinkage of concrete due to the refinement of pore size
creep after time t, CavðtÞ ¼ mean observed shrinkage or distribution that leads to a further increase in capillary
creep for a number of observations n. tension and more contraction of the cement paste. Pre-
As shown in these tables, the error coefficients of Eq. vious experimental results [25] on the pore structure of
(2) are much less than those of the ACI and CEB pre- mortars using mercury porosimetry technique showed
diction methods. The ACI and CEB prediction equa- that as silica fume content increased, the pore size dis-
tions are: tribution was shifted toward a finer distribution, the
average pore size reduced and the porosity decreased. It
esh ðtÞ ¼ t=ð35 þ tÞ esh ðuÞ ðACI209-92Þ
was found that the addition of silica fume and also the
where esh ðuÞ ¼ ultimate shrinkage and dosage of silica fume greatly influence the self-desicca-
0:5 tion and autogenous shrinkage of cement paste. In ad-
esh ðtÞ ¼ ½t=ðbsh þ tÞ esh ðuÞ ðCEB-FIP1990Þ
dition, the pozzolanic reaction of silica fume, which was
where bsh ¼ shape function of the specimen. found to be less sensitive to self-desiccation, also leads to
Also Figs. 1 and 2 show that silica fume did not have an increase in autogenous shrinkage [26].
considerable influence on drying specimens (total The high level of autogenous shrinkage of the high-
shrinkage). The average amount of total shrinkage after strengh concrete mixes reported here has practical im-
587 days of drying for the 80  270 mm and 150  300 plications. Autogenous shrinkage of concrete occurs as a
mm specimens was 524 and 450 microstrain, respec- result of chemical reactions during the hydration of ce-
tively, with corresponding standard deviations of 7.5 mentitious materials and is not related to moisture
and 14.2, respectively. It should be noted that Bisson- movement from concrete to the atmosphere. This means
nette et al. [18] disagree with this finding and believe the large sizes of structural elements or painting the surface
ultimate deformation does not differ much from one of them do not reduce this kind of shrinkage. Conse-
specimen size to the other and only the rate of drying is quently, any restraint to the deformation can induce
affected by the size of the specimen. Figs. 1 and 2 also tension stress and cracking in concrete members. For
show silica fume to considerably affect the shrinkage of instance, the reinforcement bars of structural elements
sealed specimens. It is clear that the general effect of [27] or stiff structural supports or even adjacent struc-
increasing the silica fume inclusion is to increase auto- tural members can resist autogenous shrinkage and
 M. Mazloom et al. / Cement & Concrete Composites 26 (2004) 347–357 353

cause microcracks. Cracking can increase the perme- OPC

90
ability of concrete and therefore, especially in severe 80 CEB
environments, its durability decreases [28]. It should be 70 ACI

Specific creep
mentioned that Loukili et al. [29] believe this shrinkage 60
in very high-strength concrete stops after 10 days. 50 Eq. 4 Drying
This investigation shows one of the ways to minimize 40 Sealed
30
autogenous shrinkage and also the cracking probability
20
of high-strength concrete is to add not more than 10% 10
silica fume to the mix. Some other researches believe 0
that fiber reinforced concrete is very useful in this field 0 200 400 600
Time (days)
[30–34]. Of course, new recommendations of RILEM
[35] should be considered in steel fibers. Some investi- SF6
80
gators recommend utilizing expansive admixtures to CEB
70 ACI
compensate autogenous shrinkage [36,37]. Another

Specific creep
60 Eq. 4
method is to use lightweight aggregates in concrete [38] 50
Drying
because their water adsorption is high and the internal 40
Sealed
water lost by self-desiccation of cement paste is imme- 30
diately replaced by moisture from the lightweight ag- 20

gregate. Also shrinkage-reducing admixtures are useful 10

to decrease autogenous shrinkage [39]. These chemical 0

0 200 400 600
materials reduce the surface tension of capillary water. Time (days)
Montani [40] has suggested applying expansive and
SF10
shrinkage-reducing admixtures together as a method to 80
CEB
control concrete shrinkage. Specific creep 70
60 ACI

4.6. Creep 50 Eq. 4 Drying

40
Sealed
30
Figs. 3 and 4 show the results of specific creep, i.e.
20
creep per unit stress, for cylindrical specimens of
10
80  270 mm height, which were loaded at the ages of 7 0
and 28 days, respectively. Specimens were subjected to a 0 200 400 600
sustained compressive stress of 10 MPa during all the Time (days )

creep tests and the stress/strength ratios were less than SF15
0.3. The basic creep of sealed specimens can be com- 80
70 CEB
pared to the total creep of drying specimens. It can be
Specific creep

60
observed that the difference in creep of unsealed and AC
50
sealed specimens is small in the control concrete and Drying
40
almost zero in specimens containing silica fume. This Eq. 4 Sealed
30
implies that there is no interaction between creep and
20
shrinkage and also factors affecting the rate of drying,
10
which are specimen size and the relative humidity of
0
atmosphere, had no influence on the creep of the high- 0 200 400 600
strength concrete specimens investigated here. Time (days)
The predictions by the ACI [3] and CEB [4] methods
Fig. 3. Specific creep of specimens loaded at the age of 7 days
are compared to the results of this investigation in Figs. (microstrain/MPa).
3 and 4, which in fact show that both methods overes-
timate the creep. Consequently, for this investigation, a
satisfactory model is expressed using regression analysis As shown in Figs. 3 and 4, Eq. (4) presents the creep
as follows: of concrete containing different levels of silica fume
ðtÞ
0:6 much better than the ACI and CEB models. The error
CðtÞ ¼ 0:6
 ð103
3:65SFÞ  y  10
6 coefficients of Eq. (4), ACI [3] and CEB [4] prediction
ð26:5
SFÞ þ ðtÞ
methods for specimens loaded at the age of 7 days were
ð4Þ 6%, 36.6% and 52.6%, and also for specimens loaded at
where y ¼ 1:08
0:0114t0 (correction factor according the age of 28 days were 8.9%, 27.2% and 47.8%,
to the loading age t0 ), SF ¼ silica fume percentage, respectively. The ACI and CEB prediction equations
CðtÞ ¼ specific creep after t days of loading. are:
354 M. Mazloom et al. / Cement & Concrete Composites 26 (2004) 347–357

OPC Table 11
70
Values of creep of 80  270 mm high specimens on completion of the
BCE
60 tests (microstrain)
ACI
Specific creep
50 Age of
Eq. 4 Concrete mixes
40 Drying loading
OPC SF6 SF10 SF15
30 Sealed (days)
20 7 595 510 459 417
28 413 407 381 328
10

0
0 100 200 300 400
Time (days) values for specimens loaded at the ages of 7 and 28 days
SF6
are shown in Table 11. As the proportion of silica fume
60 increased to 15%, the creep of concrete decreased by
CEB
50 ACI 20–30%. This finding is in agreement with the results
Specific creep

40
of other investigators [7,41].
Eq. 4 Drying
30
Sealed
20 4.7. Moisture movement
10
Moisture movement under alternating wetting and
0 drying cycles is a common occurrence in practice. The
0 100 200 300 400
Time (days) magnitude of this cyclic moisture movement clearly
depends upon the duration of the wetting and drying
SF10
60 periods but since drying is very much slower than wet-
CE
50
ting, the consequence of prolonged dry weather can be
ACI
reversed by a short period of rain. The movement also
Specific creep

40
Eq. 4 depends upon the range of relative humidity and the
Drying
30 composition of the concrete, as well as the degree of
Sealed
20 hydration at the onset of initial drying. Experiments
were conducted to investigate the influence of silica fume
10
content on the moisture movement of high-strength
0
0 100 200 300 400
concrete.
Time (days)

SF15 4.7.1. Shrinkage specimens

60
The shrinkage specimens of 80  270 mm height were
CE
50 submerged in water after 587 days of drying for a period
Specific creep

40
ACI of 60 days. Afterwards, they were removed and stored in
Drying a controlled environment of 20  2 °C and 50  5%
30 Eq. 4
Sealed relative humidity for 60 days. The results of these tests
20
are shown in Table 12. The general effect of silica fume
10 inclusion is to decrease moisture movement on the first
0 wetting cycle, e.g. the SF15 mix recovered 34.2% of the
0 200 400
587 day-shrinkage compared with 37.6% for the OPC
Time (days)
mix. The shrinkage of all specimens in the second drying
Fig. 4. Specific creep of specimens loaded at the age of 28 days cycle was similar for all mixes, and lower than their
(microstrain/MPa). swelling in the second stage of the experiment. This
could have been due to further hydration occurring

CðtÞ ¼ 1=EðtÞ ðt0:6 Þ=ð10 þ t0:6 Þ/ðuÞ ðACI209-92Þ

Table 12
Moisture movement of shrinkage specimens under alternating wetting
where /ðuÞ ¼ ultimate creep coefficient and and drying (microstrain)
0:3 Method of storage Concrete mixes
CðtÞ ¼ 1=Eðt ¼ 28Þ ½t=ðbh þ tÞ /ðuÞ ðCEB-FIP1990Þ
OPC SF6 SF10 SF15
where bh ¼ shape function of the specimen. 587 days in air 532 528 523 512
Figs. 3 and 4 demonstrate that silica fume had a 60 days in water 200 192 185 175
significant influence on the long-term creep. The final 60 days in air 155 150 144 140
 M. Mazloom et al. / Cement & Concrete Composites 26 (2004) 347–357 355

during wet storage leading to increased physical and Table 15

chemical bonds. Water penetration into drying concrete (mm)
Kind of specimen Concrete mixes

4.7.2. Creep specimens OPC SF6 SF10 SF15

The creep specimens were unloaded at the age of 403 Creep (t0 ¼ 7 days) 6 5 7 7
Creep (t0 ¼ 28 days) 8 7 8 8
days and their creep recovery was measured for 130
Shrinkage 11 7 7 7
days. Almost all the creep recovery happened during the Swelling 8 3 4 4
first month after unloading. After the recovery tests
finished, all specimens were submerged in water for 60
days. Tables 13 and 14 show the results of the experi-
included in Table 15. The maximum standard deviation
ments. The creep recovery or reversible creep measured
of these readings was 0.79 mm. It is worth noting that a
after unloading the unsealed specimens was 30% irre-
new more exact method has been developed recently for
spective of SF content. It was apparent that the swelling
determining the water permeability of concrete [45].
caused by re-wetting of creep specimens stored for 60
It can be seen that as the proportion of silica fume
days in water was more than that of shrinkage speci-
increased, the 24-h water penetration decreased. This
mens. LÕHermite [42,43] agrees with this result and
suggest an agreement with the previous experimental
considers this additional recovery as an increase of
results [25] on the pore structure of mortars using mer-
swelling due to load. This finding establishes that water
cury porosimetry technique which showed as silica fume
does play a significant role in the mechanism of creep.
content increased, the pore size distribution was shifted
toward a finer distribution, the average pore size re-
4.7.3. Effect of loading on permeability duced and the porosity decreased. Also it is worth not-
In order to investigate the effect of creep and ing that the water penetration in the cover of swelling
shrinkage on the microstructure of concrete, it was de- specimens was lower than that of shrinkage specimens,
cided to measure water penetration into the specimens probably because of much longer period of wet curing in
after drying them. The water penetration test was per- swelling specimens, resulting in a higher degree of hy-
formed by immersing the dry concrete cylinders for 24 h dration and lower porosity. However, in the creep
in water. The specimens were split in half by the indirect specimens silica fume did not decrease the water pene-
tensile test, and readings were taken to determine as to tration of the skin concrete but increased it in some
how far the water had penetrated into the concrete. circumstances. The 24-h water penetration of the skin
Haque [44] has utilized this method to compare the concrete in all the specimens loaded at the age of 7 days
water permeability of different concrete mixes previ- was lower than that of the ones loaded at the age of 28
ously. At least five readings were taken on each speci- days.
men and the average values of water penetration are

5. Conclusions
Table 13
Moisture movement of creep specimens loaded at the age of 7 days
From the results presented in this paper, using con-
under alternating wetting and drying (microstrain)
crete containing 0–15% silica fume, the main conclu-
Method of storage Concrete mixes sions are:
OPC SF6 SF10 SF15
396 days under load 595 510 459 417 1. In concrete mixtures with a constant slump of
130 days after 180 160 138 130 100  10 mm, those incorporating higher silica fume
unloading replacement levels tended to require more dosages
60 days in water 230 224 224 220
of superplasticiser.
2. The compressive strength of concrete mixtures con-
taining silica fume did not increase after the age of
Table 14 90 days.
Moisture movement of creep specimens loaded at the age of 28 days 3. The modulus of elasticity–compressive strength rela-
under alternating wetting and drying (microstrain)
tionship was similar to that of the ACI method.
Method of storage Concrete mixes The modulus of elasticity at unloading the creep spec-
OPC SF6 SF10 SF15 imens was independent of silica fume content.
376 days under load 413 407 381 328 4. Silica fume did not affect the total shrinkage; how-
130 days after 125 122 114 100 ever, as the proportion of silica fume increased, the
unloading autogenous shrinkage of high-strength concrete in-
60 days in water 210 200 190 185
creased and its drying shrinkage decreased.
356 M. Mazloom et al. / Cement & Concrete Composites 26 (2004) 347–357

5. The total creep and basic creep were similar, and de- [18] Bissonnette B, Pierre P, Pigeon M. Influence of key parameters on
creased as the level of silica fume replacement in- drying shrinkage of cementitious materials. Cement Concrete Res
1999;29:1655–62.
creased. Consequently, drying creep was negligible
[19] De Larrard F, Ithurralde G, Acker P, Chauvel D. High
in the high-strength concrete specimens investigated performance concrete for a nuclear containment. Proceedings,
here; hence, there was no interaction between creep Second International Symposium on Utilization of High-Strength
and shrinkage, also specimen size and the relative hu- Concrete, Berkley, ACI SP; 1990. p. 121–7.
midity of the atmosphere had no effect on creep. The [20] Persson BSM. Shrinkage of high performance concrete. Proceed-
general effect of silica fume inclusion was to decrease ings, International Workshop on Autogenous Shrinkage of
Concrete. Hiroshima: Japan Concrete Institute; June 1998.
moisture movement. On re-wetting of dried concrete p. 101–18.
using specimens having undergone shrinkage and [21] Le Roy R, De Larrard F. Creep and shrinkage of high-
creep, re-wetting caused more swelling in creep spec- performance concrete: the LCPC experience. The Fifth Interna-
imens than in shrinkage specimens. This finding sug- tional RILEM Symposium on Creep and Shrinkage of Concrete.
gests that water does play a significant role in the London: E & FN Spon; 1993. p. 499–504.
[22] Jensen OM, Hansen PF. Influence of temperature on autogenous
mechanism of creep. deformation and relative humidity change in hardening cement
paste. Cement Concrete Res 1999;29:567–75.
[23] Mak SL, Ritchie D, Taylor A, Diggins R. Temperature effects on
References early age autogenous shrinkage in high performance concretes.
Proceedings, International Workshop on Autogenous Shrinkage
[1] Holt EE. Autogenous shrinkage at very early ages. Proceedings, of Concrete. Hiroshima: Japan Concrete Institute; June 1998.
International Workshop on Autogenous Shrinkage of Concrete, p. 153–62.
Japan Concrete Institute, Hiroshima, June 1998. p. 133–40. [24] Sellevold EJ. The function of silica fume in high strength concrete.
[2] Kasai Y, Yokoyama K, Matsui I. Tensile properties of early age Proceedings of International Conference on Utilization of High
concrete. In: Mechanical behavior of materials, vol 4. Japan: Strength Concrete, Stavanger, Norway; 1987. p. 39–50.
Society of Materials Science; 1972. p. 288–99. [25] Brooks JJ, Cabrera JG, Megat Johari MA. Factors affecting the
[3] ACI Committee 209. Prediction of creep, shrinkage and temper- autogenous shrinkage of silica fume high-strength concrete.
ature effects in concrete structures. ACI Manual of concrete Proceedings, International Workshop on Autogenous Shrinkage
practice, Part 1, 1997, 209R 1-92. of Concrete. Hiroshima: Japan Concrete Institute; June 1998.
[4] CEB-FIP Model Code for Concrete Structures 1990. Evaluation p. 185–92.
of the time dependent behavior of concrete. Bulletin dÕInforma- [26] Jensen OM, Hansen PF. Autogenous deformation and change of
tion No. 199. Lausanne: Comite European du Beton/Federation relative humidity in silica fume-modified cement paste. ACI Mater
Internationale de la Precontrainte; 1991. J 1996;93(6):539–43.
[5] JCI Committee Report, Technical Committee on Autogenous [27] Miyazawa S, Tazawa E, Sato T, Sato K. Autogenous shrinkage
Shrinkage of Concrete. In: Proceedings of International Work- stress of ultra-high-strength concrete caused by resistant of
shop on Autogenous Shrinkage of Concrete. Hiroshima: Japan reinforcement. Trans Jpn Concrete Institute 1993;15:115–22.
Concrete Institute; June 1998. p. 5–64. [28] Aitcin P-C. Autogenous shrinkage measurement. Proceedings,
[6] Neville AM, Liszka WZ. Accelerated determination of creep of International Workshop on Autogenous Shrinkage of Concrete.
lightweight aggregate concrete. Civil Eng Public Works Rev Hiroshima: Japan Concrete Institute; June 1998. p. 245–56.
1973;68(803):515–9. [29] Loukili A, Khelidj A, Richard P. Hydration kinetics, change
[7] Khatri RP, Sirivivatnanon V. Effect of different supplementary of relative humidity, and autogenous shrinkage of ultra-high-
cementitious materials on mechanical properties of high perfor- strength concrete. Cement Concrete Res 1999;29:577–84.
mance concrete. Cement Concrete Res 1995;25(1):209–20. [30] Soroushian P, Ravanbakhsh S. Control of plastic shrinkage
[8] Wild S, Sabir BB, Khatib JM. Factors influencing strength cracking with specialty cellulose fibers. ACI Mater J 1998;
development of concrete containing silica fume. Cement Concrete 95(4):429–35.
Res 1995;25(7):1567–80. [31] Wu HC, Lim YM, Li VC. Application of recycled tyre cord
[9] Neville AM, Brooks JJ. Concrete technology. London: Longman; in concrete for shrinkage crack control. J Mater Sci Lett
1990. 1996;15:1828–31.
[10] Concrete Society. Microsilica in concrete. Technical Report No. [32] Toledo Filho RD, Sanjuan MA. Effect of low modulus sisal and
41, TR.041. Wexham, Slough: The Concrete Society; 1993. polypropylene fibre on the free and restrained shrinkage of mortar
[11] ACI Committee 318. Building code requirements for reinforced at early age. Cement Concrete Res 1999;29:1597–604.
concrete. American Concrete Institute; 1995. [33] Paillere AM, Buil M, Serrano JJ. Effect of fiber addition on the
[12] Brooks JJ. How admixtures affect shrinkage and creep. Mag Am autogenous shrinkage of silica fume concrete. ACI Mater J
Concrete Institute––Int Tech Soc 1999;(April):35–8. 1989;86(2):139–44.
[13] Imam M, Vandewalle L, Mortelmans F. Are current concrete [34] Banthia N, Yan C, Mindess S. Restrained shrinkage cracking in
strength tests suitable for high strength concrete? Mater Struct fiber reinforced concrete: a novel test technique. Cement Concrete
1995;28:384–91. Res 1996;26(1):9–14.
[14] Rao GA. Influence of silica fume replacement of cement on [35] Vandewalle L, Leuven KU. RILEM TC 162-TDF: Test and
expansion and drying shrinkage. Cement Concrete Res 1998; design methods for steel fibre reinforced concrete. Mater Struct
28(10):1505–9. 2000;33:3–5.
[15] Fattuhi NI, Al-Khaiat H. Shrinkage of concrete exposed to hot [36] Hori A, Morioka M, Sakai E, Daimon M. Influence of expansive
and arid climate. J Mater Civil Eng 1999;(February):66–75. additives on autogenous shrinkage. Proceedings, International
[16] Shah SP, Ahmad SH. High performance concrete and applica- Workshop on Autogenous Shrinkage of Concrete. Hiroshima:
tions. London: Edvard Arnold; 1994. Japan Concrete Institute; June 1998. p. 177–84.
[17] Neville AM, Dilger WH, Brooks JJ. Creep of plain and structural [37] Nagataki S, Gomi H. Expansive admixtures (mainly ettringite).
concrete. London and New York: Construction Press; 1983. Cement Concrete Comp 1998;20:163–70.
 M. Mazloom et al. / Cement & Concrete Composites 26 (2004) 347–357 357

[38] Kohno K, Okamoto T, Isikawa Y, Sibata T, Mori H. Effects [42] LÕHermite R. What do we know about the plastic deformation
of artificial lightweight aggregate on autogenous shrinkage of and creep of concrete? Bull RILEM 1959;1:22–51.
concrete. Cement Concrete Res 1999;29:611–4. [43] LÕHermite R. Volume changes of concrete. Proceedings,
[39] Tazawa E, Miyazawa S. Influence of constituents and composi- Fourth International Symposium on the Chemistry of Ce-
tion on autogenous shrinkage of cementitious materials. Mag ment, NBS, Washington, DC; 1960. Paper V-3. p. 659–
Concrete Res 1997;49(178):15–22. 702.
[40] Montani S. Possibilities to control concrete shrinkage by means [44] Haque MN. Strength development and drying shrinkage of high-
of chemical admixtures. CHIMIA 1998;52(5):208–11. strength concretes. Cement Concrete Comp 1996;18:333–
[41] Persson B. Quasi-instantaneous and long-term deformations of 42.
high performance concrete with sealed curing. Adv Cement Based [45] Li Z, Chau CK. New water permeability test scheme for concrete.
Mater 1998;8:1–16. ACI Mater J 2000;97(1):84–90.