Available online at [Link].

com
Construction
and Building

Construction and Building Materials 22 (2008) 2124–2129

MATERIALS
[Link]/locate/conbuildmat

Performance of lightweight concrete with silica fume

after high temperature
Harun Tanyildizi *, Ahmet Coskun
Department of Construction Education, Firat University, Elazig, Turkey

Received 28 May 2006; received in revised form 18 July 2007; accepted 21 July 2007
Available online 11 September 2007

Abstract

In this study, the effect of silica fume on compressive and splitting tensile strength of lightweight concrete after high temperature was
investigated experimentally and statistically. The mixes incorporating 0%, 10%, 20% and 30% silica fumes were prepared. After being
heated to temperatures of 200, 400 and 800 C, respectively, the compressive and splitting tensile strength of lightweight concrete was
tested. This article adopts Taguchi approach with an L16 (45) to reduce the numbers of experiment. Two control factors (percentage
of silica fume and heating degree) for this study were used. The level of importance of these parameters on compressive and splitting
tensile strength was determined by using analysis of variance (ANOVA) method.
 2007 Elsevier Ltd. All rights reserved.

Keywords: Lightweight concrete; Silica fume; High temperature; Mechanical properties; Analysis of variance (ANOVA)

1. Introduction in replace of cement in the concrete industry. The most

worldwide available supplementary cementing materials
The demand for structural lightweight concrete in many are silica fume (SF), a by-product of silicon metal, and
applications of modern construction is increasing, owing to fly ash (FA), a by-product of thermal power stations [5].
the advantage that lower density results in a significant Air and environment pollution has became a problem,
benefit in terms of load-bearing elements of smaller cross- thus, the idea of using waste material has gained popular-
sections and a corresponding reduction in the size of the ity. FA and SF are two of the most common concrete
foundation [1]. Lightweight aggregates are broadly classi- ingredients due to their pozzolanic properties [6,7].
fied into two types – natural (pumice, diatomite, volcanic The fire resistance capacity of concrete is complicated
cinders, etc.) and artificial (perlite, expanded shale, clay, because not only is concrete a composite material with
slate, sintered PFA, etc.). Lightweight concrete can easily components having different thermal characteristics, it
be produced by utilizing natural lightweight aggregate, also has properties that depend on moisture and porosity
i.e., pumice or perlite aggregate [2]. Structural lightweight [8]. A number of research studies indicated that the addi-
concrete has its obvious advantages of higher strength/ tion of silica fume highly densities the pore structure of
weight ratio, better tensile strain capacity, lower coefficient concrete, which can result in explosive spelling due to the
of thermal expansion due to air voids in the lightweight build-up of pore pressure by steam [9]. Since the evapora-
aggregate [3,4]. tion of physically absorbed water starts at 80 C which
In view of the global sustainable development, it is induces thermal cracks, such concretes may show inferior
imperative that supplementary cementing materials be used performance as compared to pure concretes at elevated
temperatures [10]. With the growth of engineering
*
Corresponding author. Fax: +90 424 2367064.
structures, applications of structural lightweight concrete
E-mail address: htanyildizi@fi[Link] (H. Tanyildizi). are increased day by day. Therefore, questions about the

0950-0618/$ - see front matter  2007 Elsevier Ltd. All rights reserved.
doi:10.1016/[Link].2007.07.017
 H. Tanyildizi, A. Coskun / Construction and Building Materials 22 (2008) 2124–2129 2125

performance of structural lightweight concrete at high tem- Table 2

perature need to examined. Levels of the variables used in the experiments
One of the main purposes of this study is to investigate Variable Level 1 Level 2 Level 3 Level 4
the effect of experimental parameters on compressive and Heating degree, H (C) 20 200 400 800
splitting tensile strength of lightweight concrete after high Silica fume, A (%) 0 10 20 30
temperature. The effect of the experimental parameters
on compressive and splitting tensile strength was evaluated
statistically and the level of significance of the parameters Table 3
affecting compressive and splitting tensile strength was Mixture proportion of concretes
determined by using analysis of variance (ANOVA) Designation Cement Silica W/C Pumice Super
method. of mixture (kg/m3) fume aggregates plasticizer
(kg/m3) (kg/m3) (kg/m3)

2. Experimental study M 400 – 0.77 1038 4.8

H 360 40 0.77 1028 4.8
A 320 80 0.77 1015 4.8
2.1. Materials E 280 120 0.77 1005 4.8

ASTM Type I Portland cement (PC) which produced as

TS EN 197–1–CEM I 42.5 in Turkey was used in this 2.3. Mix proportions
study. Silica fume from Antalya Electro Metallurgy Enter-
prise was selected for this work. The chemical analysis The mix proportions of binders are presented in Table 3,
properties of the cement and silica fume are presented in which also shows the use of a Rheobuild 1000 superplasti-
Table 1. Crushed pumice (scoria) was used as the aggregate cizer (Rheobuild 1000 is a ready-to-use, chloride free,
in the production of lightweight concrete. Scoria aggregate liquid admixture which meets ASTM C 494 requirements
was obtained from natural deposits in Elazig city (Turkey). for types A and F admixtures, AS 1478 Type HWR) to
Lightweight aggregate was scoria aggregate with a specific improve the workability. In mixtures containing silica
gravity of 2, a maximum grain size 16 mm and water fume, 0%, 10%, 20% and 30% of Portland cement by
absorption by mass of 23%. weight was replaced with silica fume. Cubic specimens of
dimensions 100 · 100 · 100 mm were prepared to deter-
2.2. Design of experiments mine the effect of different temperatures on compressive
strength. Cylinders of 100 mm in diameter and 200 mm in
Design of experiments is a power analysis tool for mod- height were prepared to determine the effect of different
eling and analyzing the influence of process variables over temperatures on splitting tensile strength. For each temper-
some specific variable [11]. The most important stage in the ature, three specimens were prepared.
design of experiment lies in the selection of the control
factors. As many as possible should be included, so that 2.4. Curing and heating regimes
it would be possible to identify non-significant variables
at the earliest opportunity [12]. In general, the compressive The specimens were demolded 24 h after the casting and
and splitting tensile strength experimental parameters will placed in a water tank at 20 ± 2 C. After 28 days water
mainly depend on the manufacturing conditions employed. curing, they were heated in an electric furnace up to 200,
Table 2 shows the detail of the variables used in the exper- 400 and 800 C. Each temperature was maintained for
iment. It is noted that there is parameters at four levels. 1 h to achieve the thermal steady state [13]. The heating
Only 16 experiments are needed to study the entire experi- rate was at 2.5 C per min. The specimens were allowed
mental parameters using the L16 (45) orthogonal array. to cool naturally to room temperature.

3. Results
Table 1
The chemical properties of cement and silica fume
3.1. Experimental findings
Bulk oxide % by mass
Portland Cement Silica fume The change in mechanical properties of concrete
SiO2 21.12 91 exposed to high temperature is dependent on materials as
Al2O3 5.62 0.58 well as environmental factors (such as initial strength
Fe2O3 3.24 0.24
before exposure to high temperature, moisture content)
CaO 62.94 0.71
MgO 2.73 0.33 [14]. As the cement paste is exposed to increasing tempera-
LOI 1.42 1.84 tures the following effects can be distinguished: the expul-
Specific surface area (cm2/g) 3430 – sion of evaporable water (100 C), the beginning of the
Particle size – 96.5% < 45 lm dehydration of the hydrate calcium silicate (180 C), the
Specific gravity (g/cm3) 3.10 2.2
decomposition of calcium hydroxide (500 C) and of the
2126 H. Tanyildizi, A. Coskun / Construction and Building Materials 22 (2008) 2124–2129

hydrate calcium silicate (that begins around 700 C). The was 4.15 (3.11)%, 14.14 (11.46)% and 70.15 (80.15)% at the
alterations produced by high temperatures are more evi- 200, 400 and 800 C, respectively. The reduction occurred
dent when the temperature surpasses 500 C. At this tem- at compressive and splitting tensile strength of lightweight
perature level, most changes experienced by concrete can concrete containing 20% silica fume was 6.92 (4.69)%,
be considered irreversible [15]. 14.43 (12.91)% and 68.06 (78.87)% at the 200, 400 and
In this study, the specimens were used to investigate the 800 C, respectively. The reduction occurred at compres-
most unfavorable situation (high temperatures or fire) for sive and splitting tensile strength of lightweight concrete
concrete mixes containing silica fume. After they were containing 30% silica fume was 3.55 (5.8)%, 24.38
heated in an electric furnace up to 200, 400 and 800 C, (40.62)% and 77.14 (75.08)% at the 200, 400 and 800 C,
the specimens were tested for compressive and splitting ten- respectively. It can be seen from Figs. 1 and 2 that light-
sile strength. The results obtained from compressive and weight concretes containing 20% silica fume (A) showed
splitting tensile strength tests on concrete mixes for temper- slightly better performance. The decrease in compressive
atures of 20, 200, 400 and 800 C are given in Figs. 1 and 2, and splitting tensile strength of lightweight concrete was
respectively. The test results indicated that each tempera- prevented with silica fume admixture. The reduction in
ture range had a distinct pattern of strength loss. It can compressive strength and splitting tensile strength can be
be seen from Figs. 1 and 2 that the compressive and split- attributed to the driving out of free water and fraction
ting tensile strength of lightweight concrete drops with tem- water of hydration of concrete due to high temperatures
perature starting from 200 C in this study. The reduction or fire. Dehydration of concrete causes a decrease in its
occurred at compressive and splitting tensile strength of strength, elastic modulus, coefficient of thermal expansion
lightweight concrete containing 0% silica fume was 8.78 and thermal conductivity [16]. Poor microstructure is asso-
(4.98)%, 12.86 (13.51)% and 83.64 (80.13)% at the 200, ciated with generation of undesirable configuration of C–
400 and 800 C, respectively. The values given in parenthe- S–H crystals, and increased cracking at high temperature
ses represent the splitting tensile strength reductions. The curing. Generally, the C–S–H crystals grow long and
reduction occurred at compressive and splitting tensile thin/narrow and occupy less space in the matrix at high
strength of lightweight concrete containing 10% silica fume temperatures, resulting in decreased density of the micro-
structure. The increased micro-cracking is the result of high
40
thermal stresses that are generated due to the induced tem-
M perature gradients. Thus, homogeneity and density of con-
35
Compressive strength (MPa)

crete microstructure are adversely affected due to the

H
30 aforementioned factors [17]. Furthermore, color of light-
A weight concrete exposed to 200 C does not change, while
25
E straw yellow, off white when the lightweight concrete are
20 exposed to temperature of 400 and 800 C, respectively.
15
3.2. Statistical analysis of the results
10

5 In this study, ANOVA and F-test were performed to see

0 statistically significant process parameters and percent
0 20 200 400 800
Temperature (ºC) Table 4
S/N ratio of compressive strengtha
Fig. 1. Compressive strength results after high temperature.
Control factor Mean S/N ratios (dB)

3.5 Level 1 Level 2 Level 3 Level 4

b
Heating degree, H (C) 30.424581 29.900149 28.850693 18.185609
Splitting tensile strength (MPa)

3 M
Silica fume, A (%) 25.929469 27.691504 28.554078b 25.185981
H a
2.5 Overall mean S/N ratio = 26.840258 dB.
A b
Optimum level.
2 E

1.5
Table 5
S/N ratio of splitting tensile strengtha
1
Control factor Mean S/N ratios (dB)
0.5
Level 1 Level 2 Level 3 Level 4
b
0 Heating degree, H (C) 8.597381 8.184065 6.517198 4.814925
0 20 200 400 800 Silica fume, A (%) 4.860953 5.220097 5.664445b 2.738224
Temperature (ºC) a
Overall mean S/N ratio = 4.173832 dB.
b
Fig. 2. Splitting tensile strength results after high temperature. Optimum level.
 H. Tanyildizi, A. Coskun / Construction and Building Materials 22 (2008) 2124–2129 2127

contribution of experimental parameters on the compres- The loss function is further transformed into a signal-to-
sive and splitting tensile strength of lightweight concrete. noise (S/N) ratio for determining the performance charac-
Larger F value indicates that the variation of the process teristic deviating from the desired value. The S/N ratio nij
parameter makes a big change on the performance charac- for the ith performance characteristic in the jth experiment
teristics [18–20]. can be expressed as
In the ANOVA, a loss function is used to calculate the nij ¼ 10 logðLij Þ; ð2Þ
deviation between the experimental value and the desired
value. Some characteristics, such as bond strength, com- The control factors and the levels given in Table 2 were
pressive strength and splitting tensile strength, also do used for calculating the S/N ratio. The S/N ratios of the
not negative values. Such characteristics are called larger compressive and splitting tensile strength were calculated
the better type quality characteristics. Larger the character- for each level of the experimental parameters given in Ta-
istics gives the better performance in compressive and split- ble 2. The S/N ratio results are shown in Tables 4 and 5.
ting tensile strength of lightweight concrete. Therefore, the Figs. 3 and 4 show the S/N response graph of each level
‘‘larger is better (LB)’’ loss function for this study was of the experimental parameters for the compressive and
selected to obtain the optimal conditions. The loss function splitting tensile strength.
Lij of LB performance characteristic can be expressed as Based on the S/N ratio, the maximum the compressive
and splitting tensile strength was obtained at 20 C and
1X n
1 20% silica fume admixture (Tables 4, 5 and Figs. 3, 4).
Lij ¼ ; ð1Þ
n k¼1 y 2ijk These findings were consistent with the experimental
where Lij is the loss function of the ith performance charac- results. The results of ANOVA for the compressive and
teristic in the jth experiment, n the number of tests, and yijk splitting tensile strength are presented in Tables 6 and 7.
the experimental value of the ith performance characteris- As shown in Tables 6 and 7, the heating degree has an
tics in the jth experiment at the kth test. utmost importance on compressive and splitting tensile
strength. The analysis indicated that the experimental error
was very low level (3.6% and 1.96%). Figs. 5 and 6 show the

33
31 Table 6
Mean S/N ratios (dB)

29 Results of ANOVA for compressive strength

27 Control Degrees of Sum of Variance FA0 Contribution
25 factor freedom square (VA) %
(f) (SSA)
23
21 Heating 3 404.620443 134.873481 74.975202a 89.971139
degree,
19
H (C)
17 Silica fume, 3 28.911873 9.637291 5.357301b 6.428825
15 A (%)
Error 9 16.190171 1.798908 – 3.600036
H1

H2

H3

H4

A1

A2

A3

A4

Total 15 449.722487 – – 100

Compressive strength parameters level
a
At least 99% confidence.
Fig. 3. S/N response graph of experimental parameters for compressive b
At least 95% confidence.
strength.

Table 7
9 Results of ANOVA for splitting tensile strength
Control Degrees Sum of Variance FA0 Contribution
Mean S/N ratios (dB)

7
factor of square (VA) %
5 freedom (SSA)
(f)
3
Heating 3 484.557130 161.519043 144.198477a 94.118252
1 degree,
H (C)
-1
6.011420b
H1

H2

H3

H4

A1

A2

A3

A4

Silica 3 20.200465 6.733488 3.923650

-3 fume,
A (%)
-5 Error 9 10.081045 1.120116 – 1.958098
Splitting tensile strength parameters level Total 15 514.838640 – – 100
a
Fig. 4. S/N response graph of experimental parameters for splitting tensile At least 99% confidence.
b
strength. At least 95% confidence.
2128 H. Tanyildizi, A. Coskun / Construction and Building Materials 22 (2008) 2124–2129

Silica splitting tensile strength may be the occurrence of micro-

Error
Fume, A and macro-cracks in concrete because of high tempera-
3.60 % 6.43 % ture. The decrease in compressive and splitting tensile
strength of lightweight concrete was prevented with sil-
ica fume admixture.
 Based on the S/N, the optimum parameters for the com-
pressive and splitting tensile strength were obtained
from specimens containing 20% silica fume and 20 C.
Based on ANOVA and F-test, the most effective param-
eters on the compressive and splitting tensile strength
were found as heating degree and silica fume,
respectively.

Heating
Degree, H References
89.97 %
[1] Babu KG, Babu DS. Behaviour of lightweight expanded polystyrene
Fig. 5. The effect of experimental parameters on compressive strength. concrete containing silica fume. Cement Concrete Res 2003;33(5):
755–62.
[2] Kılıç A, Atisß CD, Yasßar E, Özcan F. High-strength lightweight
concrete made with scoria aggregate containing mineral admixtures.
Silica Cement Concrete Res 2003;33(10):1595–9.
Error Fume, A
1.96 % [3] Topcu IB. Semi-lightweight concretes produced by volcanic slags.
3.92 % Cement Concrete Res 1997;27(1):15–21.
[4] Al-Khaiat H, Haque MN. Effect of initial curing on early strength
and physical properties of lightweight concrete. Cement Concrete Res
1998;28(6):859–66.
[5] Demirboğa R, Orüng I, Gül R. Effects of expanded perlite aggregate
and mineral admixtures on the compressive strength of low-density
concretes. Cement Concrete Res 2001;31(11):1627–32.
[6] Bilodeau A, Malhotra VM. High-volume fly ash system: concrete
solution for sustainable development. ACI Mater J 2000;97(1):41–9.
[7] Yeğinobalı A, Sobolev KG, Soboleva SV, Kıyıcı B. Thermal
resistance of blast furnace slag high strength concrete cement. In:
1st international symposium on mineral admixtures in cement.
Turkey, Istanbul; 1997. p. 106–17.
[8] Shah SP, Ahmad SH. High performance concrete: properties and
Heating applications. McGraw-Hill; 1994. p. 341.
Degree, H [9] Poon CS, Azhar S, Anson M, Wong YL. Performance of metakaolin
94.12 % concrete at elevated temperatures. Cement Concrete Compos
2003;25(1):83–9.
Fig. 6. The effect of experimental parameters on splitting tensile strength. [10] Phan LT. Fire performance of high strength concrete: a report of the
state-of-the-art. Maryland: Building and Fire Research Laboratory,
National Institute of Standards and Technology; 1996.
relative importance of the experimental parameters used in [11] Davim PJ. A note on determination of optimal cutting conditions for
this study on the compressive and splitting tensile strength. surface finish obtained in turning using of experiments. J Mater
Process Technol 2001;116(2–3):305–8.
4. Conclusions [12] Nian CY, Yang WH, Tarng YS. Optimization of turning operation
with multiple performance characteristics. J Mater Process Technol
1999;95(1–3):90–6.
In this study, the effect of experimental parameters, [13] Mohamedbhai GTG. Effect of exposure time and rates of heating and
namely heating degree and silica fume, on the compressive cooling on residual strength of heated concrete. Mag Concrete Res
and splitting tensile strength is investigated experimentally 1986;38(136):151–8.
and statistically. Specific findings of this research include [14] Chan YN, Luo X, Sun W. Compressive strength and pore structure of
the following: high-performance concrete after exposure to high temperature up to
800 C. Cement Concrete Res 2000;30(2):247–51.
[15] Luccioni BM, Figueroa MI, Danesi RF. Thermo-mechanic model for
 It is found experimentally that the highest the compres- concrete exposed to elevated temperatures. Eng Struct 2003;25(6):
sive and splitting tensile strength results were obtained 729–42.
from the specimens containing 20% silica fume for all [16] Sakr K, Hakim EE. Effect of high temperature or fire on heavy weight
temperatures. concrete properties. Cement Concrete Res 2005;35(3):590–6.
[17] Tarun RN, Kraus RN. Temperature effects on high-performance
 The results of test show that the compressive and split- concrete. Report No. CBU-2002-07 REP-460, Department of Civil
ting tensile strength are decreased as the test tempera- Engineering and Mechanics College of Engineering and Applied
ture is increased. These decreases in compressive and Science, The University of Wisconsin, Milwaukee; 2002.
 H. Tanyildizi, A. Coskun / Construction and Building Materials 22 (2008) 2124–2129 2129

[18] Liao YS, Chu YY, Yan MT. Study of wire breaking process and [20] Lin JL, Wang KS, Yan BH, Tarng YS. An investigation into
monitoring of WEDM. Int J Mach Tools Manuf 1997;37(4):555–67. improving worn electrode reliability in the electrical
[19] Ross PJ. Taquchi tecniques for quality engineering. 2nd ed. New discharge machining process. Int J Adv Manuf Technol 2000;16(6):
york: McGraw-Hill; 1996. p. 102. 113–9.