Construction and Building Materials 101 (2015) 932–941

Contents lists available at ScienceDirect

Construction and Building Materials

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

An experimental study on thermal conductivity of iron ore sand cement

mortar
Kai Liu a,⇑, Zhi Wang a, Can Jin a, Fang Wang b, Xueyuan Lu c
a
School of Transportation Engineering, Hefei University of Technology, Hefei 230009, China
b
School of Civil Engineering, Anhui Jianzhu University, Hefei 230601, China
c
Anhui Expressway Holding Group Co., Ltd., Hefei 230088, China

h i g h l i g h t s

IOS was used as aggregate at different replacement levels to make IOS cement mortar.

Hot-wire method and laser flash method were used to test the thermal conductivity.

The effects of five major factors on thermal conductivity were evaluated.

SPSS partial correlation analysis was conducted to determine the precedence factors.

A prediction formulation between thermal conductivity and the factors was developed.

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

Article history: The objective of this paper is to determine the effect of water–cement ratio, total sand–cement ratio, river
Received 5 June 2015 sand–iron ore sand ratio, temperature and density on the thermal conductivity of iron ore sand (IOS)
Received in revised form 15 September cement mortar. The IOS was filtered by wet magnetic separator and incorporated in a cement mortar
2015
mixture. The hot-wire method and the laser flash method were used to test the thermal conductivity
Accepted 17 October 2015
of the IOS cement mortar. Meanwhile, the SPSS partial correlation analysis was conducted to determine
the precedence of the five factors according to their effects on thermal conductivity. After that, a predic-
tion formulation of thermal conductivity was developed with MATLAB. Experimental results show that
Keywords:
Thermal conductivity
among the five factors, river sand–iron ore sand ratio ranks the best effect on thermal conductivity,
Cement mortar followed by water–cement ratio, total sand–cement ratio, temperature and density in sequence.
Iron ore sand Ó 2015 Elsevier Ltd. All rights reserved.
Hot-wire method
Laser flash method

1. Introduction concrete without glass bubble shows the mean conductivity of

In recent years, the application of electric technology such as
deicing and snow melting rapidly increases [1–4]. The key problem
in deicing and snow melting is the way to heat up. However, con-
ventional materials such as the ordinary Portland cement concrete
and conventional cement mortar often hold low thermal conduc-
tivity in heat conduction and result in significant energy loss.
Meanwhile, inorganic heat transfer cement is too expensive to be
widely applied. The present focus in pavement industry is to find
a material with high thermal conductivity as well as low price.
The thermal conductivity of concrete and mortar can be easily
reduced or raised by replacing one or more of its constituents
with thermally insulating or thermally conductive materials. The
⇑ Corresponding author.
E-mail address: liukai@hfut.edu.cn (K. Liu).
2.25 W/mK and as the amount of glass bubble increases, the
conductivity values decreases down to 1.32 W/mK [5]. Foamed
concrete can be used as thermal insulation materials in many
application cases, the thermal conductivity at different tempera-
tures can be changed from 0.131 W/mK to 0.484 W/mK [6]. The
non-structural grade oil palm shell foamed concrete with a density
of 1.10 g/cm3 holds the lowest thermal conductivity of 0.40 W/mK
[7]. Wood-based materials have been incorporated in concrete
products with the aim to reduce the thermal conductivity and
the lightening of concrete by wood shavings increases the thermal
insulation capacity by decreasing the thermal conductivity and dif-
fusivity, the lowest thermal conductivity can be about 0.20 W/mK
[8]. Rice husk concrete can compete with hemp concrete in terms
of thermal insulation with a dry thermal conductivity ranging from
about 0.10 W/mK to 0.14 W/mK depending on the mix proportion
[9]. Cement paste with natural sand content and the effect of silica

http://dx.doi.org/10.1016/j.conbuildmat.2015.10.108
0950-0618/Ó 2015 Elsevier Ltd. All rights reserved.
 K. Liu et al. / Construction and Building Materials 101 (2015) 932–941
fume, class C fly ash and blast furnace slag on the thermal conduc-
tivity of mortar is investigated, the thermal conductivities change
from 0.72 W/mK to 1.02 W/mK [10]. Encapsulated phase change
materials are used in concrete to improve thermal properties of
the concrete, called PCM-concrete, the biggest predictions of the
effective thermal conductivity of PCM-concrete can beyond
2.0 W/mK [11]. In order to improve the efficiency of melting snow
and deicing, pavement concrete material such as common con-
crete, steel fiber concrete and carbon fiber concrete are studied
with the thermal conductivity no less than 1.5 W/mK, the biggest
can be 3.02 W/mK [12]. Another study on steel fiber concrete
shows that the biggest thermal conductivity is 2.47 W/mK [13].
The thermophysical characteristics of reactive sand concrete with
different fiber volumes under different temperatures are analyzed,
and the upper limit of the thermal conductivity of that concrete is
2.28 W/mK [14]. The thermal conductivity of four types self-
consolidating concrete ranges between 2.8 W/mK and 3.6 W/mK
at room temperature [15].
As is known to us, if we want to lighten the concrete, it can be
just obtained by introducing gas or foam into the concrete paste, or
by totally or partially replacing standard aggregate with low
weight and preferentially low cost components [16]. Similarly,
the effective thermal conductivity of composites is strongly depen-
dent on the volume fraction and corresponding properties of each
of the constituents [17,18]. Metal sand with high thermal conduc-
tivity can improve the thermal conductivity of mortar or concrete.
Compared with metal sands such as aluminum sand and copper
sand, iron ore sand is more suitable for its low cost in engineering
fields. Researches on mechanical properties of concrete or mortar
that contains iron as aggregate have been conducted [19,20].
Besides, the possibility of using iron ore tailings to replace natural
aggregate to prepare ultra-high performance concrete under two
different curing regimes is investigated [21]. Concrete and mortar
are heterogeneous and porous solid materials, and the heat propa-
gation in concrete at normal operating temperatures is primarily
by conduction. So many factors, such as aggregate type, cement
paste proportion, aggregate gradation, porosity, moisture and tem-
perature, will affect the thermal conductivity of concrete [22–25].
Therefore, the work of this paper mainly consists of six steps. First,
the IOS was used as aggregate at various replacement levels to
make IOS cement mortar. Second, several specimens were modeled
with different mix proportions. Thirdly, based on these specimens,
the thermal conductivities were detected by hot-wire method and
laser flash method. Fourthly, each relationship between each factor
and the thermal conductivity can be obtained with regression anal-
ysis. Fifthly, SPSS partial correlation analysis was conducted to
determine the precedence factors. At last, a prediction formulation
between thermal conductivity and the five factors was developed
then. The new thermal conductive material presented in this paper
facilitates applications in fields such as chemical industries,
petroleum pipeline, deicing and snow melting, and heating.
2. Materials and methods
2.1. Source materials
The ordinary Portland cement with a 28-day compressive strength of more than
42.5 MPa is used as a basic cementitious material for all the mixtures in accordance
with the Chinese Standard GB 175-2007. The mixing water is tap water. River sand
and IOS are used as aggregates in this mixture. The river sand is excavated from the
lower reaches of a river and it has an apparent density of 2.68 g/cm3. Its fineness
modulus is 3.0 and average diameter is 0.35 mm. The river sand meets the require-
ment of Chinese Standard GB/T 14684-2011. IOS is filtered by wet magnetic sepa-
rator. The apparent density of IOS is on the order of 4.37 g/cm3. Its natural water
proportion is 12.4%. The results of sieving test are shown in Table 1. Besides, the
Fe-grade in IOS is 81% while the rest in IOS is mineral powder and impurity. The
weight of each adopted material is calculated according to different mix proportion
and then these materials are well stirred for molding.
933
Table 1
Sieving test of IOS.
Mesh size/ Accumulated Accumulated Accumulated
mm sieve residue (by sieve residue (by sieve residue (by
mass)/% results of mass)/% results of mass)/% results of
sieving test one sieving test two sieving test three
0.6 7.1 7.4 7.9
0.3 15.1 15.7 15.4
0.15 23.8 24.4 23.6
0.075 55.2 56.3 54.7
Less than 98.2 99.1 98.7
0.075
2.2. The test parameters and mix proportions
Thermal conductivity measurements are performed with particular reference to
their dependence on some other contributing factors such as water–cement (W/C)
ratio, total sand–cement ratio (S/C), river sand–iron ore sand ratio (R/I) and temper-
ature (T). To analyze the sensitivity of each test parameter, a series of different val-
ues is selected. W/C ratio, S/C ratio and R/I ratio are all mass ratio. Each specimen
with a specific mix proportion is shown in Table 2. If the specimen does not contain
river sand, the total sand only contains IOS; otherwise, the total sand contains river
sand and IOS.
2.3. Preparation of specimens
Two types of specimens are prepared according to the type of tests to be real-
ized: one is cuboid specimen with the size of 4  4  16 cm3, the other is a cylinder
specimen with the diameter of 12.5 mm and 3 mm in thickness. These different
sizes are related to measuring devices whose dimensions are imposed. Pure IOS
specimen is made of cement, water and IOS. The mixed sand specimen is formed
by cement, water, river sand and IOS. Cuboid specimen is mixed by mortar mixer,
vibrated and molded by plain bumper. After that, they should be maintained
28 days in curing box in constant temperature (20 ± 1 °C) and humidity (not less
than 90%). The durations of the different phases of mixing must be sufficiently long
to allow a good homogenization and short enough to avoid leaving too much water
to evaporate in ambient air. The ages of cuboid specimens are all specified as
28 days by experimental control. All specimens are dried enough before the thermal
conductivity test.
Cylinder specimen is used to test the thermal conductivity at different temper-
atures. A PVC test tube with the diameter of 13 mm is used as a mold because the
cylinder specimen is difficult to prepare. A bamboo sign is used to tamp the test
tube until the mold is filled up with mortar. The specimens in the PVC test tubes
will also be maintained in 20 ± 1 °C and the humidity is no less than 90%. After
28 days, the mold will be cut into a cylinder with the thickness of 3 mm. After that,
the cylinder specimen will be taken out and polished with a file carefully until the
specimen meets standard size. Similar with cuboid specimens, all cylinder speci-
mens are specified as 28 days and dried enough before the thermal conductivity
test. The IOS cement mortar can be used in heat conduction layer for road deicing
and snow melting. The position of the heat conduction layer is under the water-
proof tack coat. Therefore, the IOS cement mortar would be waived of the effect
of water from road deicing and snow melting.
2.4. Measurement method
There are various testing devices and their interpreting methods for measure-
ment of thermal conductivity. They can be broken into four distinct methods. First,
the two-linear-parallel-probe method [26,27] has been widely accepted to deter-
mine thermal conductivity, Two probes are inserted into two parallel holes drilled
in the sample, where one probe is used as a heating source and the other as a tem-
perature sensor. Also used are the plane-heat-source method [28] and the hot-
guarded plate method [29]. They are similar to the two-linear-parallel-probe
method in their basic principle but require additional efforts to cut the sample thin,
and then firmly place the thermal probe on with epoxy [23]. Laser flash method
[30–32] is a kind of thermal diffusion method, which also be used these years for
a better accuracy.
In this study, the QTM-300 device manufactured from KEM in Japan (Fig. 1(a))
and NETZSCH LFA457 laser thermal conductivity meter made in Germany (Fig. 2)
are used to measure the thermal conductivity of these specimens.
The basic principle of QTM-300 is the same as that of the two-linear-parallel-
probe method. However the thermal conductivity is measured by the probe
method, which has been modified from the transient-hot-wire method. The QTM-
300 device makes it possible to measure the thermal conductivity of the sample
within 60 s by using probes containing temperature sensors. The measurement
934 K. Liu et al. / Construction and Building Materials 101 (2015) 932–941

range of QTM-300 is 0.023–12 W/mK with an error of about ±3%, and the thermal
conductivity of IOS cement mortar is also under the range. The cuboid specimens

0.60
3.0
1.0
27
are tested at room temperature (25 °C) and under ambient moisture (65% relative
humidity) (Fig. 1(c)). The instrument should be calibrated before the test (Fig. 1(b)).

0.60
The LFA457 is a device for conveniently measuring the thermal conductivity

0.5
3.0
26
through the method stipulated in the ‘‘KS L 1604 Fine ceramics—determination of
thermal diffusivity, specific heat capacity, and thermal conductivity of monolithic

0.60
3.0
ceramics by laser flash method” [31]. The temperature range of LFA457 laser ther-

25

0
mal conductivity meter is from 125 °C to 1100 °C, and the thermal conductivity
range is from 0.1 W/mK to 2000 W/mK. The diameter of the sample should be

0.60
2.7
1.0
24
12.5 mm and less than 3 mm in thickness. As for the measurement principle of
LFA, the thermal diffusivity is obtained by using the temperature disparity curve

0.60
for the laser pulse detected at the detector placed on the opposition side of the sam-
2.7
0.5
23

ple at which the laser is transmitted. The thermal conductivity is obtained by mul-
tiplying the measured thermal diffusivity and the specific heat obtained by using
0.60
2.7

differential scanning calorimetry and the density of the sample. It can be calculated
22

by the following standard expression:

0.60

kðTÞ ¼ aðTÞ  C p ðTÞ  qðTÞ ð1Þ

2.5
1.0
21

where kðTÞ is the thermal conductivity, aðTÞ is the thermal diffusivity, C P ðTÞ is the
0.60

specific heat, qðTÞ is the density.

2.5
0.5
20

0.60
2.5

3. Results and discussion

19

0
0.60

3.1. Test results

2.2
1.0
18

In this experiment, each cuboid specimen should be tested

0.60
0.5
2.2
17

three times in three different rectangular surfaces. Similarly, each

cylinder specimen should be tested two times in the two circular
0.60
2.2
16

surfaces. The average value is the thermal conductivity of these

0

specimens (Table 3). The thermal conductivity of single material

0.60
2.0
1.0
15

and inorganic heat transfer cement on market in different temper-

atures are shown in Table 4 [12,33–35] and the data is partially
0.60
0.5
2.0

derived from additional experiments.

14

The thermal conductivity of IOS is lower than pure iron sand

0.60

because that some mineral powder and impurity will be kept in

2.0
13

IOS after magnetic separation. As a matter of fact, the metal sands

such as aluminum sand and copper sand are better than iron ore
0.58
2.0
1.0
12

sand for their higher thermal conductivity. However, these metal

sands are not suitable for popularization and application for high
0.58
0.5
2.0
11

cost.
0.58
2.0
10

3.2. Analysis of the results

0
0.55
2.0
1.0

3.2.1. Dependence of thermal conductivity on water–cement ratio

9

Experimental outputs indicate that the mixture would be

0.55

heterogeneous and porous if W/C ratio is 0.50 while S/C ratio

0.5
2.0
8

exceeds 2.0. Meanwhile, multiple the test results may be inaccu-

0.55

rate in this mixing condition. Therefore, S/C ratio will be kept 2.0
2.0
7

in analyzing the effect of W/C ratio on thermal conductivity. The

evolution of the thermal conductivity as function of the various
0.53
2.0
1.0

samples in different W/C ratios is illustrated on Fig. 3.

6

From Fig. 3(a) and (b), thermal conductivity decreases as W/C

0.53
0.5
2.0

ratio increases under different R/I ratios and temperatures. The

5

reason of this phenomenon may be that thermal conductivity of

0.53
2.0

IOS cement mortar changes considerably due to its porosity since

4

the pores are filled up with air when W/C ratio increases and air
0.50

is the poorest conductor compared to solid and liquid. The thermal

2.0
1.0
3
The mix proportion of each specimen.

conductivity of air approximately is 0.03 W/mK. Therefore, the

0.50

thermal conductivity of mortar will decrease when the water in

0.5
2.0
2

the pores be displaced by some air. Fig. 3(c) shows that when R/I
ratio is 1.0 and W/C ratio rises from 0.50 to 0.60, the thermal con-
0.50
2.0

ductivity will drop from 1.81 W/mK to 1.56 W/mK. The average
1

reduction is 0.25 W/mK while the variation ratio is 13.8%. When

Mix proportion

R/I ratio is 0.5, the reduction is 0.41 W/mK while the variation is
W/C ratio

17.5%. Similarly, When R/I ratio is 0, the reduction is 0.39 W/mK

S/C ratio
R/I ratio

while variation ratio is 14.4%. This is attributed to IOS having a

Table 2

higher water–adsorption capacity than sand. It has more internal

pores with the decrease proportion of IOS if the water proportion
 K. Liu et al. / Construction and Building Materials 101 (2015) 932–941
Fig. 1. The testing process with QTM-300. (a) QTM-300 thermal conductivity meter. (b) Calibration. (c) Testing the cuboid specimen.
Fig. 2. LFA457 laser thermal conductivity meter.
is appropriate. The mixture would also be heterogeneous and
porous for water shortage if the proportion of IOS is too large.
When R/I ratio is 1.0, the thermal conductivity of this IOS cement
mortar is the least variation ratio among the three variation ratios.
It shows that the mixture has a good workability in that mixing
condition. From Fig. 3(c), a linear decreasing relationship between
W/C ratio and thermal conductivity can be observed when T is
25 °C and S/C ratio is 2.0. Meanwhile, the similar relationship can
also be found in different R/I ratios (Fig. 3(a) and temperatures
(Fig. 3(b)). In conclusion, W/C ratio and thermal conductivity will
always maintain a linear decreasing relationship regardless of the
change of other variables.
3.2.2. Dependence of thermal conductivity on total sand–cement ratio
Excessive proportion of cement and iron ore sand will result in
incomplete hydration when W/C ratio is 0.50 and S/C ratio is 3.0.
Specimen is difficult to mold under this condition, either. W/C ratio
is kept 0.60 when analyzing the effect of the S/C ratio on thermal
conductivity. The relationships between S/C ratio and thermal con-
ductivity can be observed in Fig. 4.
As shown in Fig. 4(a) and (b), thermal conductivity increases as
S/C ratio increases under different R/I ratios and temperatures. The
proportion of IOS in the specimens increases as S/C ratio increases.
The thermal conductivity of this IOS cement mortar will increase as
the proportion of IOS increases because IOS is a good material for
heat conduction in the mixture. A maximal thermal conductivity
can be obtained if cement is directly mixed with IOS in this test.
However, the framework structure of specimen will not strong
enough without sand. As a result, the flexural and the compressive
strengths of the specimens are not desirable. The sand proportion
in specimen should be carefully decided according to the specific
applications. As shown in Fig. 4(c), the variation ratio of thermal
conductivity is obvious when the S/C ratio is within the range from
2.0 to 2.2 and the range from 2.5 to 2.7. The IOS and cement are
935
well mixed in the mixing process and the specimens will be with
smooth surface under this condition. It shows that the S/C ratio
within the range from 2.0 to 2.2 and the range from 2.5 to 2.7
are typical to be used as construction mix proportion. What can
also be drawn from Fig. 4(c) is a linear increasing relationship
between S/C ratio and thermal conductivity under T is 25 °C, W/C
ratio is 0.60. This trend will not change even other variables change
(e.g. R/I ratios and temperatures). In conclusion, a linear increasing
relationship between S/C ratio and thermal conductivity will not
vary regardless of the change of other variables.
3.2.3. Dependence of thermal conductivity on river sand–iron ore sand
ratio
The thermal conductivity of this material significantly depends
on the proportion of IOS. To determine the change of thermal con-
ductivity affected by the R/I ratio, the thermal conductivity of the
specimens (Table 3) under different mix proportion (Table 2) such
as 1–3, 7–9, 13–15, 19–21, 25–27 are selected. The results are
shown in Fig. 5.
From Fig. 5(a), it is evident that the thermal conductivity of IOS
cement mortar gradually reduces as R/I ratio increases. When R/I
ratio reaches 0.5, the mass of river sand is the half of the IOS in
the specimen. The thermal conductivity has an average reduction
of 0.40 W/mK compared to R/I ratio is 0. The mean variation rate
is 15.4%. When R/I ratio reaches 1.0, the mass of river sand is as
much as IOS and the mean reduction of thermal conductivity is
0.43 W/mK compared to R/I ratio is 0.5 while the variation ratio
is 19.5%. The reason for that phenomenon is the proportion of
IOS decrease as R/I ratio increases. The thermal conductivity of
river sand is much smaller than IOS. Therefore, the thermal con-
ductivity of the material decreases observably as R/I ratio
increases.
In this experiment, the proportion of IOS in the specimen
reaches the minimal value when R/I ratio is 1.0. The thermal con-
936 K. Liu et al. / Construction and Building Materials 101 (2015) 932–941

ductivity is still bigger than conventional cement mortar. It can be

1.99
1.93
1.87
proved that incorporating the IOS in cement mortar can further

27
improve its thermal conductivity. Besides, a flexural test as well
as a compressive test on cuboid specimens is used to determine

2.32
2.29
2.29
26
the mechanical property of this IOS cement mortar. The results
show that the IOS cement mortar has a better mechanical property

2.79
2.76
2.71
25
compared with conventional cement mortar. The maximal com-
pressive strength reaches 39.75 MPa. Fig. 5(b) shows the linear fit-

1.83
24
ting result of relationship between R/I ratio and thermal
/
/
conductivity when T is 25 °C, W/C ratio is 0.60 and S/C ratio is
2.27 2.0. Based on this analysis method, a linear decreasing relationship
23
/
/

between R/I ratio and thermal conductivity will be kept regardless

2.69

of the change other variables.

22
/
/
1.76
1.74
1.71

3.2.4. Dependence of thermal conductivity on temperature

21

Thermal conductivity is affected by temperature. Tests at three

2.21
2.17
2.13

temperatures (20 °C, 0 °C, 25 °C) are carried out in this test. The
20

relationship between temperature and thermal conductivity is

2.64
2.61
2.56

shown in Fig. 6.
19

As shown in Fig. 6(a) and (b), the thermal conductivity reduces

as temperature increases all the time regardless of the variation of
1.63
18

W/C ratio, S/C ratio and R/I ratio. However, the dependence of ther-
/
/

mal conductivity on temperature is far less important than its

2.11
17

dependence on W/C ratio, S/C ratio or R/I ratio. The thermal con-
/

ductivity of IOS cement mortar significantly depends on the pro-

2.48

portion of IOS. Therefore, this phenomenon can be explained by

16
/
/

metal heat conduction theory. The iron ion and electronic con-
1.62

1.56
1.60

tained in composite structure after hydration. The higher the tem-

15

perature becomes, the faster the rate of electron motion will be.
This phenomenon can increase the probability collision of iron ions
1.93
2.04
2.02
14

and also increase the hindrance role of electron drift. As a result,

the thermal resistance increases. The macroscopical embodiment
2.42

2.31
2.40
13

of this phenomenon is that the thermal conductivity decrease

when the temperature increases. The variation rate of thermal con-
1.64
12

ductivity at negative temperatures is less than that at positive tem-

/
/

peratures. This is also due to the different intensity of the thermal

2.05

motion. Fig. 6(c) shows a fitting relationship between temperature

11
/
/

and thermal conductivity when W/C ratio is 0.60, S/C ratio is 2.0
2.43

and R/I ratio is 1.0. A linear decreasing relationship between tem-

10
/
/

perature and thermal conductivity will be kept regardless of the

change other variables.
1.75
1.72
1.68
9
2.22
2.18
2.08

3.2.5. Dependence of thermal conductivity on density

8

There is almost no change on the volume of these specimens

from 20 °C to 25 °C and the densities are shown in Table 5. The
2.63
2.61
2.49
7

density of conventional cement mortar is about 2.00 g/cm3, the

density of the cement mortar that mixed with IOS (4.37 g/cm3) will
1.79

increases from 2.34 g/cm3 to 2.78 g/cm3. Therefore, the apparent

6
/
/

density at 25 °C is chosen in test and the relationship is shown in

2.27

Fig. 7.
The thermal conductivities (W/mK) of the specimens.

5
/
/

In these tests, the density of specimen can be varied by many

2.68

factors (e.g. W/C ratio, S/C ratio and R/I ratio). Therefore, the rela-
4
/
/

tionship between thermal conductivity and density is not linear

1.81
2.05
2.01

and fitted with a second order polynomial regression (Fig. 7). The
3

air contained in specimen is considered to no convection from

20 °C to 25 °C and its thermal conductivity is far less than that
2.49
2.44
2.34
2

of solid. The number of pores in the specimens will decrease and

so does the air stored in the pores decrease if the density of spec-
2.78
2.76
2.7

imen increases. Based on above reasons, thermal conductivity

1

increases when density of specimen increases. Actually, the den-

Mix proportion

sity of specimen reaches about 2.78 g/cm3 when the thermal con-
ductivity reaches its peak value either. If the density of the IOS
20 °C

cement mortar exceeds 2.78 g/cm3, that means a large number of

25 °C
Table 3

0 °C

IOS and little cement contained in the specimen. The mixture

would also be incomplete hydration product because of cement
 K. Liu et al. / Construction and Building Materials 101 (2015) 932–941 937

Table 4
The thermal conductivity (W/mK) of each material in different temperatures.

The type of material Water 42.5R River Pure iron Iron ore Air Portland cement Conventional Inorganic heat
cement sand sand sand concrete cement mortar transfer cement
Thermal conductivity at 20 °C 1.59 (ICE) 0.17–0.61 (0.39) 3.0 38–42 (40) 32.3 0.018 1.37–2.05 (1.71) 0.77 4.5–5.5 (5.0)
Thermal conductivity at 0 °C 0.49 0.19–0.65 (0.42) 3.3 40–45 (42.5) 31.4 0.022 1.33–2.13 (1.73) 0.83 5.0–6.0 (5.5)
Thermal conductivity at 25 °C 0.53 0.19–0.65 (0.42) 3.5 40–50 (45) 30.2 0.030 1.35–2.18 (1.77) 0.93 5.0–7.0 (6.0)

2.10

λ (W/m⋅K)
1.95 -20

1.80
0

1.65

T
25
1.50
0.50 0.52 0.54 0.56 0.58 0.60
W/C

(a) (b)

3.0
R/I=0
R/I=0.5
2.7 R/I=1.0

λ = -4.05(W/C) + 4.76
2.4 R=0.90
λ (W/m⋅K)

2.1 λ = -4.1(W/C) + 4.4

R=0.92

1.8
λ = -2.56(W/C) + 3.11
R=0.92
1.5
0.50 0.52 0.54 0.56 0.58 0.60
W/C
(c)
Fig. 3. Relationship between W/C ratio and the thermal conductivity under different conditions. (a) Condition 1: different R/I with T = 25 °C, S/C = 2.0. (b) Condition 2:
different T with R/I = 1.0, S/C = 2.0. (c) Fitted line with T = 25 °C, S/C = 2.0.

shortage. This condition result in the mixture has not formed a
composite structure and makes the thermal conductivity decrease.
Besides, the size of IOS is smaller than the sand. When IOS and
sand are mixed together, the mixture would have better perfor-
mance in framework structure. Furthermore, the IOS with smaller
size can fill up some voids in the specimen and can make the sam-
ple texture more compact. Based on the above reasons, the IOS
cement mortar has a better performance in thermal conductivity
than the conventional cement mortar.
can be divided into linear correlation analysis, partial correlation
analysis and distance analysis. All the five factors in this test—W/
C ratio, S/C ratio, R/I ratio, temperature and density not only have
effects on thermal conductivity, but also have correlated relation-
ships among themselves. Accordingly, the partial correlation is
conducted to analyze the correlation coefficient between five fac-
tors and thermal conductivity.
The correlation coefficient between x and y could be calculated
as the following standard expression:

X
n
3.3. Correlation analysis on the influencing factors of thermal ðxi 
xÞðyi  y

Þ
conductivity i¼1
rij ¼ sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð2Þ
X n
P
Correlation analysis is a commonly statistical method which is ðxi 
xÞ2 ni¼1 ðyi  y
Þ2
i¼1
used to test the effect of each variable to the dependent object. It
938 K. Liu et al. / Construction and Building Materials 101 (2015) 932–941

2.0

1.9

λ (W/m⋅K)
-20

1.8

0
1.7

T
1.6
25
1.5
2.0 2.2 2.4 2.6 2.8 3.0
S/C

(a) (b)
3.0
R/I=1.0
2.8 R/I=0.5
R/I=0
2.6
λ =0.4(S/C) + 1.57
2.4 R=0.89
λ(W/m⋅K)

2.2
λ =0.34(S/C)+ 1.3
2.0
R=0.84
1.8
λ = 0.32(S/C) + 0.91
1.6 R= 0.96
2.0 2.2 2.4 2.6 2.8 3.0
S/C
(c)
Fig. 4. Relationship between S/C ratio and the thermal conductivity under different conditions. (a) Condition 1: different R/I with T = 25 °C, W/C = 0.60. (b) Condition 2:
different T with R/I = 1.0, W/C = 0.60. (c) Fitted line with T = 25 °C, W/C = 0.60.

where rij is the correlation coefficient. x as well as y respects the (0.239). The result also proves that the proportion of IOS can
mean value of x and y. xi and yi are the i observations of x and y. directly affect the thermal conductivity of the IOS cement mortar.
Assume the number of variables is m. The partial correlation
coefficient stipulates that the relationship of two variables out of
3.4. Thermal performance prediction model
m is separately researched while the other variables are fixed.
Instead of constructing the multiple regression models in each
Based on the experimental data, a prediction formulation
computation, the following formula was used to obtain the partial
between thermal conductivity and the five factors was developed
correlation coefficient.
with MATLAB. The thermal conductivity of IOS cement mortar
ðR1 Þij can be predicted by the relationship as functions of W/C ratio
r 0ij ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð3Þ (W/C), R/I ratio (R/I), S/C ratio (S/C), temperature (T) and density
ðR1 Þii ðR1 Þjj (q) as shown in Eq. (4). The comparison of prediction and the lab
test value of thermal conductivity of specimen is shown in Fig. 8.
where r 0ij is the
partial correlation coefficient.
2 3
r 11 r12 . . . r 1m k ¼ 3:56ðW=CÞ  0:77ðR=IÞ þ 0:3ðS=CÞ  0:002T  0:6q2
6 r 21 r22 . . . r 2m 7
6 7
R ¼ ðr ij Þmm ¼ 6 .. .. .. 7 is the correlation coefficient þ 3:35q  0:6 ð4Þ
4 . . ... . 5
r m1 rm2 . . . r mm The sample correlation coefficient R of 0.99 indicates that Eq.
matrix and the superscript 1 indicates the inverse of the matrix. (4) is quite reasonable. The thermal conductivity can be calculated
The SPSS partial correlation analysis is used to obtain the effect from Eq. (4) under different W/C ratios, S/C ratios, R/I ratios and
precedence of the five factors on the thermal conductivity of the temperatures in the case that the specimen can be shaped. With
IOS cement mortar. The results of the partial correlation analysis the condition S/C = 2.0, W/C = 0.50 and T = 25 °C, The comparison
are shown in Table 6. of theoretical model (Maxwell’s model, Cubic dispersed phase
The seriation of precedence factors on the thermal conductivity parallel-serial model) [8,36] with the empirical formula in this
in sequence in Table 6 is that: R/I ratio (0.940), W/C ratio paper is shown in Table 7. The iron ore sand cement mortar can
(0.916), S/C ratio (0.761), Temperature (0.616), Density be treated as a two-phase composite material that is constituted
 K. Liu et al. / Construction and Building Materials 101 (2015) 932–941 939

2.4
Average Line W/C=0.60
2.6 2.6 S/C=2.0
2.2
T=25
Linear Fit
2.4

λ(W/m⋅k)
λ(W/m⋅K)
2.0
2.2 2.2

1.8 λ = -0.75(R/I) + 2.31

2.0 R=0.99

1.8 1.6
1.77

0.0 0.5 1.0 0.0 0.5 1.0

R/I R/I

(a) (b)
Fig. 5. Effect of R/I ratio on thermal conductivity. (a) The thermal conductivities under different R/I ratios. (b) Fitted line with T = 25 °C, W/C = 0.60, S/C = 2.0.

S/C=3.0
W/C=0.50 2.5 S/C=2.5
2.5 W/C=0.55 S/C=2.0
2.0
2.0 W/C=0.60 λ (W/m⋅K)
λ (W/m⋅K)

1.5
1.5

1.0 1.0

0.5 0.5
25 25
0.0 0.0
0 0 0 0
)

)
0.5 0.5
T(

T(
R/I -20 R/I -20
1 1

(a) (b)

1.62

λ = -0.0013T+1.6
1.60 R=0.97
λ(W/m⋅K)

1.58
W/C=0.60
S/C=2.0
R/I=1.0
1.56
Linear Fit

-30 -20 -10 0 10 20 30

T( )
(c)
Fig. 6. Effect of temperature on thermal conductivity under different conditions. (a) Condition 1: different R/I, W/C with S/C = 2.0. (b) Condition 2: different R/I, S/C with W/
C = 0.60. (c) Fitted line with W/C = 0.60, S/C = 2.0, R/I = 1.0.

of cement mortar (continuous phase) and iron ore sand (dispersed
phase). The deviation percentage between calculated value and
measured value are given in brackets.
It seems that the data from empirical formula is closer to the
measured value. Therefore, the empirical formula may be more
suitable for calculating the thermal conductivity of the iron ore
sand cement mortar and simplifies the calculation.
4. Conclusions
An experimental study was conducted to determine the con-
tributing factors on the thermal conductivity through the use of
the QTM-300 as well as LFA457 laser thermal conductivity meter.
From the lab test results, a partial correlation analysis was con-
ducted to find the precedence relationship among five factors in
940 K. Liu et al. / Construction and Building Materials 101 (2015) 932–941

3.0

2.45
2.45
2.45
25

27
Polynomial Fit

2.61
2.61
2.61
26
2.5

2.78
2.78
2.78

λ (W/m⋅K)
25
2.41
2.41
2.41
24

2.0
2.58
2.58
2.58
2
λ = -6.02ρ + 33.49ρ - 43.92
23

R=0.86
2.72
2.72
2.72
22

1.5
2.39
2.39
2.40

2.3 2.4 2.5 2.6 2.7 2.8

21

ρ (g/cm3)
2.57
2.56
2.55
20

Fig. 7. Relationship between density and thermal conductivity at 25 °C.

2.69
2.69
2.69
19

Table 6
The results of partial correlation analysis of each factor and the coefficient thermal
2.37
2.37
2.37

conductivity.
18

Analysis of the Control variables Partial correlation

2.52
2.52
2.52

variable coefficient
17

S/C ratio W/C ratio & R/I ratio & q & 0.761
2.63
2.63
2.63

temperature
16

W/C ratio R/I ratio & q & temperature & S/C 0.916
ratio
2.34
2.34
2.34

R/I ratio & q & S/C ratio & W/C 0.616

15

Temperature
ratio
R/I ratio S/C ratio & W/C ratio & 0.940
2.41
2.41
2.41
14

temperature & q
Density R/I ratio & S/C ratio & W/C ratio & 0.239
2.54
2.53
2.53

temperature
13
2.37
2.37
2.37
12

3.00
2.42
2.42
2.42
11

2.75
Calculated Value (W/m⋅K)
2.52
2.52
2.52
10

2.50
2.38
2.38
2.38
9

2.25
2.48
2.47
2.47
8

2.00
2.56
2.56
2.56
Apparent density (g/cm3) of every specimen in different temperature.

1.75
Test Point
2.42
2.42
2.41

R=0.99
6

1.50
2.49
2.49
2.49

1.50 1.75 2.00 2.25 2.50 2.75 3.00

5

Measured Value (W/m⋅K)

2.57
2.57
2.57
4

Fig. 8. Comparison of the measured and calculated values.

2.46
2.46
2.45
3

Table 7
2.51
2.51
2.51

The contrast of theoretical model and empirical formula (W/mK).

2

Mix R/I Measured Empirical Maxwell’s Cubic dispersed

2.58
2.58
2.58

proportion value formula model phase parallel–

1

serial model
W/C = 0.50, 0 2.70 2.73 2.17 2.71 (0.4%)
Mix proportion

S/C = 2.0, (1.1%) (19.6%)

T = 25 °C 0.5 2.34 2.33 1.62 2.07 (11.5%)
(0.4%) (30.8%)
20 °C

25 °C
Table 5

1.0 1.81 1.87 1.41 1.74 (3.9%)

0 °C

(3.3%) (22.1%)
 K. Liu et al. / Construction and Building Materials 101 (2015) 932–941
affecting thermal conductivity of the specimen. A prediction for-
mulation between thermal conductivity and the five factors was
developed with MATLAB. Several conclusions can be drawn as
follows:
(1) In this paper, a laboratory study on five parameters which
mostly affect thermal conductivities of this IOS cement mor-
tar was considered. The SPSS partial correlation analysis was
conducted to indicate the precedence relationship among
five factors in affecting thermal conductivity. The prece-
dence sequence is that: R/I ratio, W/C ratio, S/C ratio,
temperature and density.
(2) The increase of W/C ratio leads to a rise of pore proportion in
mortar and a drop of thermal conductivity. The increase of S/
C ratio as well as the decrease of R/I ratio leads to a rise of
IOS proportion and thermal conductivity of specimen.
The increase of temperature leads a rise of thermal resis-
tance and a drop of the thermal conductivity. Density has
little effect on the thermal conductivity of specimen.
(3) Based on the test results from the five parameters, a pro-
posed formulation correlates test results well by indicating
the sample correlation coefficient of 0.99. It seems that the
data from empirical formula is closer to the measured value.
Therefore, the empirical formula may be more suitable for
calculating the thermal conductivity of the iron ore sand
cement mortar and simplifies the calculation.
(4) The IOS cement mortar not only has high performance on
thermal conductivity, but also has low-cost feature. The
new thermal conductive material presented in this paper
facilitates applications in fields such as chemical industries,
petroleum pipeline, deicing and snow melting, and heating.
Acknowledgements
This work was supported by National Natural Science Founda-
tion of China (Grant Nos. 51108150, 51408005 and 51508147),
Technological Project of Anhui Expressway Holding Group Co.,
Ltd. (Grant No. AHGS 2013-9), and Science Foundation for Excellent
Youth Scholars of Anhui Province (Grant No. 2013SQRL046ZD).
References
[1] I. Yoshitake, N. Yasumura, M. Syobuzako, A. Scanlon, Pipe heating system with
underground water tank for snow thawing and ice prevention on roads and
bridge decks, J. Cold Reg. Eng. 25 (2) (2011) 71–86.
[2] H.M. Zhao, S.G. Wang, Z.M. Wu, G.J. Che, Concrete slab installed with carbon
fiber heating wire for bridge deck deicing, J. Transp. Eng.-ASCE 136 (6) (2010)
500–509.
[3] N. Sugawara, K. Hokari, T. Watanabe, H. Sugawara, Energy saving
characteristics of a new type of road-heating system, Atmos. Res. 46 (1998)
113–122.
[4] Huajun Wang, Linbo Liu, Zhihao Chen, Experimental investigation of hydronic
snow melting process on the inclined pavement, Cold Reg. Sci. Technol. 63
(2010) 44–49.
[5] T.S. Yun, Y.J. Jeong, T.-S. Han, K.-S. Youm, Evaluation of thermal conductivity
for thermally insulated concretes, Energy Build. 61 (2013) 125–132.
[6] M.A.O. Mydin, Y.C. Wang, Elevated-temperature thermal properties of
lightweight foamed concrete, Constr. Build. Mater. 25 (2011) 705–716.
[7] U.J. Alengaram, B.A. Al Muhit, M.Z. bin Jumaat, M.L.Y. Jing, A comparison of the
thermal conductivity of oil palm shell foamed concrete with conventional
materials, Mater. Des. 51 (2013) 522–529.
941
[8] D. Taoukil, A. El Bouardi, F. Sick, A. Mimet, H. Ezbakhe, T. Ajzoul, Moisture
content influence on the thermal conductivity and diffusivity of wood-
concrete composite, Constr. Build. Mater. 48 (2013) 104–115.
[9] M. Chabannes, J.C. Benezet, L. Clerc, E. Garcia-Diaz, Use of raw rice husk as
natural aggregate in a lightweight insulating concrete: an innovative
application, Constr. Build. Mater. 70 (2014) 428–438.
[10] Ramazan. Demirbog, Influence of mineral admixtures on thermal conductivity
and compressive strength of mortar, Energy Build. 35 (2003) 189–192.
[11] P. Meshgin, Y.P. Xi, Multi-scale composite models for the effective thermal
conductivity of PCM-concrete, Constr. Build. Mater. 48 (2013) 371–378.
[12] Liu. Kai, Influence factor of thermal conductivity of cement concrete and its
prediction model, J. Build. Mater. 15 (6) (2012) 771–777.
[13] N.N. Htun, S. Sukchai, S. Hemavibool, Properties of concrete material for
thermal energy storage, in: 2014 International Conference and Utility
Exhibition on Green Energy for Sustainable.
[14] Y. Ju, H.B. Liu, J.H. Liu, K.P. Tian, S. Wei, S. Hao, Investigation on thermophysical
properties of reactive powder concrete, Sci. China Technol. Sci. 54 (12) (2011)
3382–3403.
[15] W. Khaliq, V. Kodur, Thermal and mechanical properties of fiber reinforced
high performance self-consolidating concrete at elevated temperatures, Cem.
Concr. Res. 41 (11) (2011) 1112–1122.
[16] B. Chen, N. Liu, A novel lightweight concrete-fabrication and its thermal and
mechanical properties, Constr. Build. Mater. 44 (2013) 691–698.
[17] J.K. Lee, Prediction of thermal conductivity of composites with spherical fillers
by successive embedding, Arch. Appl. Mech. 77 (7) (2007) 453–460.
[18] R. Osiroff, D.P.H. Hasselman, Effect of interfacial thermal barrier on the
thermal stresses near spherical inclusion in matrix subjected to linear heat
flow, J. Compos. Mater. 25 (12) (1991) 1588–1598.
[19] C. Czaderski, M. Shahverdi, R. Bronnimann, C. Leinenbach, M. Motavalli,
Feasibility of iron-based shape memory alloy strips for prestressed
strengthening of concrete structures, Constr. Build. Mater. 56 (2014) 94–105.
[20] M. Alwaeli, J. Nadziakiewicz, Recycling of scale and steel chips waste as a
partial replacement of sand in concrete, Constr. Build. Mater. 28 (1) (2012)
157–163.
[21] S.J. Zhao, J.J. Fan, W. Sun, Utilization of iron ore tailings as fine aggregate in
ultra-high performance concrete, Constr. Build. Mater. 50 (2014) 540–548.
[22] A.H.C. Shin, U. Kodide, Thermal conductivity of ternary mixtures for concrete
pavements, Cement Concr. Compos. 34 (4) (2012) 575–582.
[23] K.-H. Kim, S.-E. Jeon, J.-K. Kim, S. Yang, An experimental study on thermal
conductivity of concrete, Cem. Concr. Res. 33 (2003) 363–371.
[24] H. Uysal, R. Demirboga, R. Sahin, R. Gul, The effects of different cement
dosages, slumps, and pumice aggregate ratios on the thermal conductivity and
density of concrete, Cem. Concr. Res. 34 (2004) 845–848.
[25] R. Demirboga, R. Gul, The effects of expanded perlite aggregate, silica fume and
fly ash on the thermal conductivity of lightweight concrete, Cem. Concr. Res.
33 (2003) 723–727.
[26] P. Morabito, Measurement of the thermal properties of different concretes,
High Temp. High Pressures 21 (1989) 51–59.
[27] A. Donati, A. Lanciani, P. Morabito, et al., Evaluation of thermal conductivity
and diffusivity of concrete by means of the two-linear-parallel-probe method,
High Temp. High Pressures 19 (1987) 371–380.
[28] J.-M. Zhou, Z.-X. Zhu, D.-J. Xie, H. Li, W. Chen, Y.-X. Zhang, Thermal physical
property of refractory material measured by plane heat source method with
constant heat rate, J. Central South Univ. (Sci. Technol.) 42 (5) (2011) 1467–
1472.
[29] C.S. Sanjaya, T.-H. Wee, T. Tamilselvan, Regression analysis estimation of
thermal conductivity using guarded-hot-plate apparatus, Appl. Therm. Eng. 31
(10) (2011) 1566–1575.
[30] D.C. Harris, L.F. Johnson, R. Seaver, et al., Optical and thermal properties of
spinel with revised (increased) absorption at 4 to 5 mu m wavelengths and
comparison with sapphire, Opt. Eng. 52 (8) (2013) 12.
[31] J. Cha, J. Seo, S. Kim, Building materials thermal conductivity measurement and
correlation with heat flow meter, laser flash analysis and TCi, J. Therm. Anal.
Calorim. 109 (1) (2012) 295–300.
[32] S. Yamanaka, T. Maekawa, H. Muta, T. Matsuda, S. Kobayashi, K. Kurosaki,
Thermal and mechanical properties of SrHfO3, J. Alloy. Compd. 381 (1–2)
(2004) 295–300.
[33] Y. Yen, Review of intrinsic thermophysical properties of snow, ice, sea ice and
frost, North. Eng. 23 (1) (1991) 187–218.
[34] Liu Weidong, Tian Bo, Hou Ziyi, Experimental study on thermal conductivity of
concrete, J. China Foreign Highway 32 (1) (2012) 226–229.
[35] Xiao Jianzhaung, Song Zhiwen, Zhang Feng, An experimental study on thermal
conductivity of concrete, J. Build. Mater. 13 (1) (2010) 17–21.
[36] Weiping Zhang, Hongguang Min, Gu Xianglin, Yunping Xi, Yishan Xing,
Mesoscale model for thermal conductivity of concrete, Constr. Build. Mater. 98
(2015) 8–16.