Building and Environment 37 (2002) 607 – 614

www.elsevier.com/locate/buildenv

Factors a!ecting the thermal properties of concrete and

applicability of its prediction models
M.I. Khan ∗
School of Civil Engineering, University of Nottingham, Nottingham NG7 2RD, UK

Received 22 February 2001; received in revised form 12 June 2001; accepted 19 June 2001

Abstract

In this paper an investigation on thermal conductivity of mortar, concrete and its major constituent aggregate is presented. Four di!erent
types of rocks were tested for their conductivities at dry and fully saturated states, which were further crushed and used as coarse
aggregates in concrete. Thermal conductivity of mortar and concrete have been measured at various moisture contents. The applicability
of Campbell-Allen and Thorne’s model in assessing the in5uence of aggregate type has been investigated. Using a few advanced models
for thermal conductivity of porous material, it has been demonstrated that there exists a possibility to gain insight into the nature of the
pore that is likely to be encountered in the concrete. The relationship between concrete conductivity and aggregate conductivity at various
degrees of saturation including at dry and fully saturated conditions have been developed. Further, equivalent conductivity of solid, which
is a necessary input parameter for porous material models, has been evaluated for all the types of concrete under consideration.  c 2002
Elsevier Science Ltd. All rights reserved.

Keywords: Aggregate type; Concrete; Moisture content; Models; Nature of the pores; Porosity; Thermal conductivity

1. Introduction and contribution of the component heat transfer to the overall

thermal behaviour of the concrete.
An analysis of conduction heat transfer through structure A review of earlier investigations reveal that the type of
is of great importance in civil engineering problems such aggregate, porosity and moisture content have the maximum
as heat 5ow into a building in energy e=cient building de- in5uence on thermal conductivity of concrete [1–5], unlike
sign, thermal loading of structures due to diurnal variations the cement hydrates for which thermal conductivity is not
of temperature, planning and design of building for thermal subjected to large variations. Thus, thermal conductivity of
comfort, design of radiation shield in nuclear power sta- aggregates primarily determines the insulating quality of
tions, analysis of bridge deck and other exposed structures the concrete [2]. The thermal conductivity of rocks, com-
for solar thermal loading, etc. The knowledge of thermal monly used as aggregates in concrete, ranges from 1.163 to
conductivity and other thermal transport properties of con- 8:6 W=mK [3– 6].
struction materials involved in the process of heat transfer is Quartzite, sandstone and other quartzose rock have the
essential in predicting the temperature pro>le and heat 5ow highest thermal conductivity; granite, gneiss, limestone
through the material. Concrete, being one of the most com- and dolomite have intermediate, whilst basalt and dolerite
monly used construction material, its thermal conductivity demonstrate lowest conductivities [3,5]. The thermal con-
draws importance. The problem of heat transfer in concrete ductivity of rock depends not only on its composition but
is complicated. In order to facilitate the better understanding also on its degree of crystallization. Rocks with crystalline
of conduction phenomena, one must subdivide the concrete structure show higher heat conduction than amorphous and
into its constituents and facets and study the heat transfer vitreous rocks of the same composition [2,6]. Mineralogical
through each of these to understand the relative importance character of the aggregate would greatly in5uence the ther-
mal conductivity of ordinary dense concrete [7]. Aggregate
with less thermal conductivity produce the less conductive
∗Tel.: +44-115-9513903; fax: +44-115-9513898. concrete whereas the more conductive aggregates produce
E-mail address: [email protected] (M.I. Khan). more conductive concrete [1,5,7]. Thus, aggregate type

0360-1323/02/$ - see front matter
c 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 3 6 0 - 1 3 2 3 ( 0 1 ) 0 0 0 6 1 - 0
608 M.I. Khan / Building and Environment 37 (2002) 607–614

can cause nearly twice an increase in thermal conductivity This facilitated the evaluation of an equivalent shape of
of concrete [8]. Thermal conductivity increases with an the pores in concrete, together with a value for equivalent
increase in moisture content and it has been reported that fraction of entrapped pores.
thermal conductivity of concrete in moist state is 70% more
than that in dry state [1].
The information pertaining to experimental data and the- 2. Models for concrete
oretical prediction of porosity related to thermal conductiv-
ity of concrete is scarce. Boulder Canyon Project report [5] 2.1. Campbell-Allen and Thorne’s model
was the earliest in this regard and presented the empirical
method for prediction of thermal conductivity and heat ca- The theoretical model given by Campbell-Allen and
pacity of concrete. This method is based on the knowledge Thorne [1] for the prediction of thermal conductivity of
of mix proportions and petrographic composition of the ag- concrete assumes it to consist of a continuous mortar phase,
gregate and is purely empirical in nature. Campbell-Allen in which the discontinuous coarse aggregate phase remains
and Thorne [1] postulated a theoretical model for the pre- dispersed. Assuming two parallel paths, one through the
diction of thermal conductivity of concrete. This model is connecting mortar layers and the other through a path con-
based on the principle of Ohm’s law where the input pa- sisting of aggregate and mortar in series; and the geometrical
rameters are volume fractions and the thermal conductivi- features of uniformly dispersed cubical aggregate particle
ties of the coarse aggregate and mortar. Harmathy [2], on in a regular arrangement in the matrix. The expression for
the other hand, considered porosity of concrete, conductiv- thermal conductivity of the concrete suggested is given by
ity of equivalent solid and moisture content as the factors on km ka (1 − M )2
which conductivity of concrete is dependent and hence are k = km (2M − M 2 ) ; (1)
ka M + km (1 − M )
the input parameters for the model cited by Harmathy [2].
where
A model of this form is more acceptable from the funda-
mental nature of heat transfer in concrete and other building M = 1 − (1 − p)1=3 (2)
materials [8].
where p is the volume of mortar per unit volume of concrete,
However, experimental determination of equivalent con-
k the thermal conductivity and su=xes ‘m’ and ‘a’ refer to
ductivity of solid is considered to be di=cult. In this con-
mortar and aggregate, respectively.
text, concrete can be considered to be a two-phase material
made up of solid phase and a 5uid phase either air or water.
The solid phase, however, is a composite with its constituent 2.2. Models for porous material
component being aggregate and solid portion of the mortar.
Usually, the ratio of the aggregate conductivity to that of There are two types of models namely e!ective medium
the solid portion of mortar may be of the order of 2–3 [1,4] approximation (EMA) and Ohm’s law models for porous
whereas that for the equivalent solid and air or water is of materials. Various Ohm’s law models are similar to
the order of 100 or 10. Thus, it is justi>ed to assume con- Campbell-Allen and Thorne’s model but the volume frac-
crete as a two-phase material with solid and 5uid phases as tions of mortar and the aggregate are replaced by volume
stated above. There are large numbers of models available fraction of solid and pores, respectively. However, no such
for such two-phase mixtures [8]. However, all such models Ohm’s law model for porous material has been applied
require equivalent conductivity of solid and porosity as im- to concrete. One of the EMA models applied to concrete
portant input parameters. Methods for evaluating solid con- is Hamilton and Crosser’s model cited by Harmathy [2].
ductivity are suggested in the literature [8,9] and the use of This is a two-phase model, can be further extended to a
such methods enable one to estimate the equivalent solid multiphase material model, and is given below:
conductivity of concrete. However, invariably these meth- (1 − P)ka + EPkf
k= ; (3)
ods are based on models for thermal conductivity, which (1 − P) + EP
also take into account the shape or nature of the pores. where
In this paper, a review of the models for thermal con-
nka
ductivity of concrete including those models which take E= (4)
into account the nature=shape of pores have been pre- (n − 1)ks + kf
sented. Experimentally determined thermal conductivity of with ks standing for the conductivity and volume fraction of
rocks, mortar and concrete are reported. The applicability phase 1, which is the principal continuous phase. Phase 2 is
of Campbell-Allen and Thorne’s model [1] in assessing the a dispersed phase and the value of n is the function of phase
in5uence of aggregate type has been investigated. Using the distribution geometry and has to be determined experimen-
porous material model cited by Harmathy [2] for concrete tally. According to Harmathy [2] when both the phases are
and that suggested for rocks by Zimmerman [9], thermal essentially continuous, from a survey of available experi-
conductivity of concrete has been predicted after estimating mental data, a value of n = 1:5 seemed to be reasonable. For
solid conductivity through Bhattacharjee et al.’s model [8]. application to cement paste, mortar and concrete, this model
 M.I. Khan / Building and Environment 37 (2002) 607–614 609

had been extended to three-phase system. The concrete was namely, enclosed pores and enclosing pores. The expres-
assumed to be consisting of three phases namely: the solid sions describing the same models are given below:
consisting of solid particles of hydrated cement and aggre-
k
gates, the air voids and the water when moisture is present. = k1(1−f) × k2(f) ; (9)
However, when material is dry, it will be a two-phase sys- ks
tem and it is true in the fully saturated state. Therefore, the where k1 and k2 are the conductivities of an idealized enclos-
equivalent conductivity of solid becomes an essential input ing and enclosed pores, respectively, and f stands for frac-
to such type of a model. tion of enclosed pores. The enclosed pores are nearly sur-
An advanced model on similar EMA principle for 5uid rounded by solid in all directions, whereas enclosing pores
saturated rocks has been suggested by Zimmerman [9], nearly surround the solid from all directions. This system of
which in addition to porosity takes pore shapes also into pore structure takes into account, the interconnectivity for
account in terms of aspect ratio of the pores. The expres- the pore system and the types of pores such as ‘ink bottle’
sions relating to the equivalent conductivity of the porous pores which may be either due to capillarity of concrete or
material with porosity, conductivity of solid, and aspect due to shrinkage cracks. An indication of the fraction of
ratio, etc. are given below: enclosed pores can be obtained through mercury intrusion
k (1 − P)(1 − r) + rP porosimetry (MIP) and represent the fraction of entrapped
= ; (5) pores. Further k1 can be obtained as
ks (1 − P)(1 − r) + P
where 1
k1 = : (10)

 (Aks + B)
(1 − r) 4 1
= + (6)
3 2 + (r − 1)N 1 + (r − 1)(1 − N ) Similarly k2 in dry and saturated states are given by

with r = kf =ks and N being a factor that depends on the k2 (dry) = C − Dks ; (11)
aspect ratio of the pore. The aspect ratio can be de>ned
considering the spheroid as a degenerate ellipsoid which has
two axes of equal length. The parameter aspect ratio denoted log k2 (sat) = E − Fks + Gks2 ; (12)
by characterizes the shape, and is the ratio of the length
where A–G are functions of porosity in dry and saturated
of the unequal axis to that of length of one of the equal
states.
axes. Further → ∞ signi>es needle-like tubular pore, for
a spherical pore → 1, and that for a coin shaped crack
→ 0. The oblate pores will have ¡ 1, and for prolate
pores ¿ 1. Thus, N is related to in the following manner: 3. Experimental programme

For oblate pores: 3.1. Materials

(2 − sin 2)
N= ; (7) In order to eliminate the in5uence of cement, ordinary
2 tan  sin 
portland cement taken from the same batch was used
where  = cos−1 ( ). throughout the investigation. Four di!erent rock samples
were chosen for thermal conductivity measurement tests.
For prolate pores: These rocks were from di!erent geological sites having


1 cos2  1 + sin  di!erent mineralogical composition. These rocks cover a
N= 2 − ln ; (8) wide range of thermal conductivities and are being mostly
sin  2 sin3  1 − sin 
commonly used as aggregates in concrete. The rock sam-
where  = cos−1 (1= ). ples were tested for their thermal conductivities and then
In case of concrete, pores may be capillary pores formed crushed to obtain the coarse aggregates for concrete. Two
due to hydration process or may be shrinkage cracks formed types of >ne aggregates, sand type I (reddish land quarried
at the mortar aggregate interface. The nature of the pores quartz sand) and sand type II (greyish river sand), which
whether oblate or prolate will depend upon the relative ori- contains certain amount of mica, were used. The physical
entation of these pores with respect to heat 5ow direction. properties of materials used are presented in Table 1. The
Thus, the single value of suggested in this model might water absorption and speci>c gravity of aggregates were
be considered to that corresponding to an equivalent shape determined in accordance with ASTM C127: 1977 and
with respect to the heat conduction. ASTM C 128: 1973, respectively. The grading of both >ne
A third approach considering the nature of the pores has and coarse aggregates were adjusted to obtain the standard
been suggested by Bhattacharjee et al. [8] for thermal con- grading of zone B of Road Note No. 4. The >nal grading
ductivity of building materials, where two types of pores used for casting the mortar and the concrete specimens are
have been postulated for the purpose of heat conduction given in Table 2.
610 M.I. Khan / Building and Environment 37 (2002) 607–614

Table 1 Table 3
Physical properties of materials Measured porosity of mortar and concrete

Material Speci>c gravity Absorption (%) Type of concrete Porosity (%)

Ordinary portland cement 3.15a — Sand type I Sand type II

Sand type I 2.69 5.20
Sand type II 2.70 4.70 Mortar 22.99 24.10
Basalt 2.70 0.30 Basalt concrete 14.93 16.63
Limestone 2.69 0.80 Limestone concrete 15.58 17.01
Siltstone 2.66 1.83 Siltstone concrete 16.59 19.27
Quartzite 2.67 0.25 Quartzite concrete 16.50 18.52

a Assumed value.

3.4. Conditioning of specimens

Table 2
Grading of >ne and coarse aggregates
For thermal conductivity measurement in dry state, all the
Sieve size Percentage passing ◦
specimens were dried in an oven at 105 ± 5 C for 24 h and
Coarse aggregate Fine aggregate then cooled in a dessicator. For the measurement in satu-
rated state, the rock samples were kept in water and then
10:0 mm 100 100
4:75 mm 3.8 100
boiled vigorously in water for 5 h so as to achieve the max-
2:36 mm 15.6 100 imum saturated condition. Whereas for mortar and concrete
1:18 mm 3.1 94.5 specimens, three di!erent degrees of saturation were used
600
m — 66.7 for the measurement of thermal conductivity.
300
m — 2.8
150
m — 1.4
75
m — — 3.5. Test procedure
Pan — —
A high temperature thermal conductivity tester (model
HC-60, make-M=S EKO instrument, Japan), working on
3.2. Mix proportions non-steady line source hot wire method was used with a
range of 0.233–11:630 W=mK and an accuracy of ±5%. The
The volumetric mix proportions of all types of mortar and working method is based on application of constant heat
concrete were kept same as; cement : >ne aggregate : coarse 5ux through an in>nitely long thin source embedded in an
aggregate :: 1 : 2:33 : 4:66 at constant water–cement ratio of in>nitely homogeneous material, initially in equilibrium at
0.60 throughout the investigation. These proportions were constant temperature. The detailed principle of this method
similar to the work enumerated by Campbell-Allen and is available in the review given by Pratt [10]. For an applied
Thorne [1]. suitable constant current, for the expected range of thermal
conductivity of the material, to the line source hot wire, the
3.3. Specimen preparation and curing time versus temperature plot was obtained through a plot-
ter and the conductivity was calculated using the relevant
The thermal conductivity was measured for 120 × 120 × equation as described in the literature [10]. The test was per-
40 mm specimens, rectangular in cross-section, for rocks, formed on all samples of rocks, mortar and concrete in dry
mortar and concrete as per the requirement of the exper- as well as in saturated conditions at ambient temperature.
imental set-up. Wooden moulds were made for casting of For each condition, three samples were tested eight times
mortar and concrete. The mixing of mortar and concrete was each and the average is reported as the result.
done in an electrically operated mixer and all specimens The porosity of mortar and concrete has been determined
were fully compacted using platform vibrator. Six mortar in accordance with ASTM C 642: 1975 and the results
specimens were cast; three with sand type I and three with obtained are reported in Table 3.
sand type II. In total, 24 concrete specimens were cast; six
specimens with each type of coarse aggregate (three with
each type of >ne aggregate). After 24 h of casting all mortar 4. Results and discussions
and concrete specimens were demoulded and kept for cur-
ing in an open water tank, at ambient temperature (approx- 4.1. Thermal conductivity of rocks
◦
imately 24 ± 4 C). After 27 days of curing, the surfaces of
all the specimens were >nished properly to achieve smooth The thermal conductivity values of the rocks measured
surfaces in order to maintain the proper contact between the in dry and fully saturated conditions have been shown in
sensor and the specimen for the measurement of thermal Table 4. These rocks were crushed and used as aggregates
conductivity as per the requirement of the method. in the concrete. During crushing the aggregates are likely to
 M.I. Khan / Building and Environment 37 (2002) 607–614 611

Table 4
Thermal conductivity of the rocks in dry and saturated states

Type of concrete Thermal conductivity (W=mK)

Dry state Saturated state

Basalt 4.03 4.30

Limestone 3.15 3.49
Siltstone 3.52 5.22
Quartzite 8.58 8.63

Table 5
Measured thermal conductivity of mortar and concrete in dry and fully
saturated states
Fig. 1. Thermal conductivity of the mortar at various moisture contents.
Type of concrete Thermal conductivity (W=mK)

Sand type I Sand type II

content present in it, as compared to mortar made with sand
Dry Fully saturated Dry Fully saturated type II, which contains mica.
Mortar 1.90 2.65 1.37 1.95 It is evident from these results that the aggregate with less
Basalt concrete 2.26 3.52 1.97 3.24 thermal conductivity produced the less conductive concrete
Limestone concrete 2.03 2.92 1.60 2.71 whereas the more conductive aggregates produced more
Siltstone concrete 2.21 3.61 1.91 2.90
conductive concrete. For example, quartzite aggregate which
Quartzite concrete 2.77 4.18 2.29 3.49
has conductivity of 8:58 W=mK produced the concrete with
conductivity of 2.77 and 2:29 W=mK with sand type I and
sand type II, respectively. Whilst limestone which has the
break in random planes, however, it has been assumed that conductivity of 3:15 W=mK produced the concrete with con-
the conductivity of such dispersed aggregates present in the ductivity of 2:03 and 1:60 W=mK with sand type I and sand
matrix of the mortar would demonstrate equivalent conduc- type II, respectively. It may be noted here that the quartzite
tion as measured in the parent rock. To ensure this, conduc- concrete with sand type I demonstrated much higher value
tivity of rocks was measured in three orthogonal directions. of conductivity compared to that of limestone concrete made
The experimentally obtained values of thermal conductiv- with the same sand. But in case of quartzite concrete with
ity of rocks (Table 4) are in well agreement with the earlier sand type II, the value of conductivity was not so high when
>ndings [1–3] except for the basalt. The basalt used in this compared with limestone concrete made with the same type
investigation was dense, >ne textured and crystalline com- of sand. This variation is due to the fact that quartzite con-
pared to other basalts which resulted in higher conductivity. crete made with sand type I have more quartz content present
compared to that made with sand type II. This variation
4.1.1. In8uence of moisture content was also observed in mortar made with sand type I and sand
It can be seen that the rocks with higher absorption at type II.
maximum saturation (Table 1) has the higher rate of increase
in conductivity as compared to dry conductivity (Table 4). 4.2.2. In8uence of moisture content
For example, siltstone which has absorption of 1.99% by The variation of thermal conductivity of mortar with mois-
weight at maximum saturation demonstrated an increase of ture content is shown in Fig. 1. This >gure demonstrates that
about 48% with respect to its dry conductivity. The rocks the highest value of moisture content in each case represents
which have less absorption at maximum saturation demon- complete saturation and the linear variations of conductivity
strated least increase with respect to their dry conductiv- with moisture content are obtained for both the types of sand.
ity; e.g. basalt and quartzite, with absorptions of 0.40% and Further the graph obtained for the mortar made with sand
0.30%, demonstrated an increase of only about 6.50% and type I and sand type II are almost parallel; this is quite ex-
0.60%, respectively, with respect to their dry conductivities. pected, as the grading of both sands is similar as well as the
mix proportions are same in both the mortars. Thus porosity,
4.2. Thermal conductivity of mortar and concrete pore structure and the process of absorption=deabsorption
moisture in both of the mortars on their thermal conduction
4.2.1. In8uence of aggregate type is also similar. It may be interesting to note here that, in dry
The thermal conductivity values of mortar and concrete state, the mortar made with sand type I has a thermal con-
in dry and fully saturated conditions are reported in Table 5. ductivity about 38% more as compared to that of sand type
It can be seen from these results that thermal conductivity II, whilst in saturated state the former has a conductivity
of mortar made with sand type I is higher, because of quartz which is about 36% more compared to that made with the
612 M.I. Khan / Building and Environment 37 (2002) 607–614

Fig. 2. In5uence of moisture content on thermal conductivity of the Fig. 3. Relationship between the thermal conductivity of aggregate and
concrete made with (a) sand type I (b) sand type II. the thermal conductivity of concrete in (a) dry state (b) saturated state.

latter. The minor di!erence may be attributed to the rela- ductivity produce concrete with higher conductivity. Fur-
tively more porous nature of the sand type I itself compared ther, it can be seen that the rate of increase of thermal con-
to the sand type II, as is evident from their water absorption ductivity of concrete with respect to aggregate is more in
value given in Table 1. Thus, resulting in a slightly di!erent saturated condition as compared to the dry condition. This
moisture con>guration in moist mortars under consideration. rate of increase was found similar for both sand type I and
The variation of thermal conductivity of concrete with sand type II. However, concrete made with sand type I dis-
moisture content is shown in Fig. 2. This >gure demon- played more conduction as compared to concrete made with
strates that there is an increase in the thermal conductivity sand type II, because the quartz content in sand type I is
of concrete due to the increase in the moisture content. The more as observed from visual inspection and the colour of
rate of the increase in the thermal conductivity of concrete the sand.
is more up to about 4.5% of moisture content by weight, be-
yond which the rate of increase is less. It has been observed
that from dry state to 50% degree of saturation, the rate of 5. Prediction of concrete conductivity
increase was more, beyond which it was not so signi>cant.
The detailed discussion on this topic has been published The thermal conductivity of concrete have been calculated
elsewhere [11]. both in dry and saturated states by substituting the experi-
mental values of thermal conductivity of aggregate and mor-
4.3. Relationship between aggregate conductivity and tar in Campbell-Allen and Thorne’s model (Eq. (1)). These
concrete conductivity computed values of thermal conductivity of concrete are
plotted against their corresponding experimentally obtained
The variation of thermal conductivity of concrete with values in Fig. 4 for dry and saturated states. It is evident
that of aggregate in dry and saturated states is presented from this >gure that Campbell-Allen and Thorne’s model
in Fig. 3. This >gure reveals that the variations in thermal overpredicts thermal conductivity of concrete by about 13–
conductivities are linear in both dry as well as in saturated 32% in dry state and the inaccuracy increases with increase
conditions. It is evident from these >gures that the aggre- in aggregate conductivity. In case of saturated state, the
gates with less thermal conductivity produce concrete with prediction is within −13% to +8%. Therefore, this model
less conductivity whereas the aggregates with higher con- can be still made applicable for calculating the thermal
 M.I. Khan / Building and Environment 37 (2002) 607–614 613

Table 6
Estimated solid conductivity and fraction of enclosed pores

Type of concrete Sand type I Sand type II

ks (W=mK) f ks (W=mK) f

Mortar 3.43 0.875 2.48 0.870

Basalt concrete 4.26 0.809 4.03 0.782
Limestone concrete 3.26 0.838 3.31 0.759
Siltstone concrete 4.52 0.791 3.69 0.821
Quartzite concrete 5.22 0.843 4.47 0.825

Table 7
Estimated solid conductivity

Type of concrete ks (W=mK) f

Dolerite 2.43 0.786

Barytes 2.50 0.809
Haematite 5.44 0.821

the above mentioned solid conductivity agrees well with the

experimentally determined value, the evaluated solid con-
ductivity is acceptable. A trial on this line using Harmathy’s
model [2] failed to provide satisfactory results, and con-
ductivity at the saturated state estimated using the evalu-
ated solid conductivity, deviated from experimental values
by as much as +17% to +49%. Thus other advanced model
Fig. 4. Predicted thermal conductivity values of concrete against experi- namely, the one suggested by Bhattacharjee et al. [7] and
mentally obtained values in (a) dry state (b) saturated state. Zimmerman [8] was thought to be more suitable for evalu-
ation of solid conductivity. The method suggested by Bhat-
tacharjee et al. seems to be simpler, although involves a trial
conductivity of concrete in saturated state while as, in case and error procedure. The fraction of enclosed pores f is
of dry state the values are overpredicted. Thus, approxi- estimated for a trial value of conductivity of solid ks from
mate estimates of thermal conductivity can be made through the ratio of experimentally determined dry to saturated con-
this model especially when the aggregate conductivity is ductivities. This step is followed by validation of ks and f
low. This is because simple Ohm’s law models are accu- through a comparison of experimentally determined conduc-
rate enough when the ratio of conductivities of constituent tivities with those estimated using the above ks and f. When
phases are low. In dry state, the same ratio of aggregate to disagreement is observed from the comparison, a new trial
mortar conductivity becomes high. Thus, resultant inaccu- value of ks is selected using the following equation, till the
racy introduced is higher; however, this model is amply able results converge simultaneously.
to explain the in5uence of aggregate type on conductivity
ks (new) = ks (old) × kdry =kdry (estimated); (13)
of concrete.

ks (new) = ks (old) × ksat =ksat (estimated): (14)

6. Evaluation of pores through models
A simple computer programme has been used to facilitate
6.1. Solid conductivity and enclosed pores the above trial and error procedure. The values of solid
conductivity ks and the fraction of enclosed pores f thus
In order to apply the porous models, evaluation of equiv- obtained are reported in Table 6. Although the experimental
alent solid conductivity of concrete becomes essential. This porosity values were not available for Campbell-Allen and
estimation can be carried out from the models if both dry and Thorne’s data, approximate porosity values were estimated
saturated conductivities are known. The simplest method using the procedure as suggested in the literature [12]. The
is to evaluate the solid conductivity from the model using equivalent conductivities of the solid have been calculated
known value of thermal conductivity in dry state. If the and are given in Table 7, together with their fraction of en-
estimated thermal conductivity in the saturated state using closed pores.
614 M.I. Khan / Building and Environment 37 (2002) 607–614

Table 8 (2) The type of aggregate has signi>cant in5uence on ther-

Estimated thermal conductivity of mortar and concrete mal conductivity of mortar and concrete.
Type of concrete k (W=mK) (3) The thermal conductivity of concrete increased with
the increase in moisture content. From dry state to 50%
Sand type I Sand type II
degree of saturation, the rate of increase was more
Dry Saturated Dry Saturated signi>cant, beyond which it was less signi>cant.
(4) When aggregate conductivity is low, Campbell-Allen
Mortar 1.85 2.33 1.29 1.73
Basalt concrete 2.80 3.20 2.54 2.95 and Thorne’s mortar–concrete approach is su=ciently
Limestone concrete 2.13 2.49 2.07 2.48 accurate in both dry and saturated states for predicting
Siltstone concrete 2.85 3.27 2.17 2.60 the in5uence of aggregate type on thermal conductivity
Quartzite concrete 3.29 3.72 2.68 3.17 of concrete. However, from fundamental principle of
heat transfer in porous media such as concrete, porous
media approach is better and can furnish additional
Table 9
Estimated saturated thermal conductivity and aspect ratio
information regarding the nature of the pore.
(5) Equivalent conductivity of solid also strongly depends
Type of Sand type I Sand type II upon the nature of the pores and about 80% of the
concrete pores in concrete are nearly of enclosed type. Further,
k (W=mK) k (W=mK)
the predominant pore shape in concrete con>rms to
Mortar 2.46 1:155 × 10−1 1.84 1:264 × 10−1 oblate nature. Although it has not been taken up in
Basalt 3.19 3:506 × 10−2 2.95 2:830 × 10−2 this work, the porous material approach can be easily
concrete
extended to partially moist concrete considering it to
Limestone 2.57 8:600 × 10−2 2.50 2:683 × 10−2
concrete be a three-phase material.
Siltstone 3.25 2:904 × 10−2 2.67 6:648 × 10−2
concrete
Quartzite 3.71 5:164 × 10−2 3.16 6:087 × 10−2 References
concrete
[1] Campbell-Allen D, Thorne CP. The thermal conductivity of concrete.
Magazine of Concrete Research 1963;15(43):39–48.
[2] Harmathy TZ. Thermal properties of concrete at elevated
6.2. Porosity and equivalent pore shape temperatures. Journal of Materials 1970;5:47–74.
[3] Marshall AL. The thermal properties of concrete. Building Science
The results obtained using both Harmathy [2] and Zim- 1972;7:167–74.
merman [8] models are given in Tables 8 and 9. In case of [4] Khan MI, Bhattacharjee B. Relationship between thermal
Hamilton and Crosser’s model cited by Harmathy [2] de- conductivities of aggregate and concrete. In: Reddy RR, editor. Civil
engineering materials and structures. Osmania University Hyderabad,
viation of dry conductivities are within −5:9% to +29% India: 1995. p. 162– 6.
whereas those for the saturated conductivities are within [5] Boulder Canyon Project Report. Thermal properties of concrete.
−8:5% to −14%. Similar variations for saturated conductiv- Final Report by USBR, Bulletin No. 1, Part VII, 1940.
ities (−5:9% to −13:3%) have been observed for Zimmer- [6] Zoldners NG. Thermal properties of concrete under sustained
man’s model. Thus, these solid conductivities and porosi- elevated temperatures. ACI Publication SP-25, 1971. p. 1–31.
[7] Neville AM. Properties of concrete. UK: Longman, 1995.
ties allow for better estimation of thermal conductivities of [8] Bhattacharjee B, Krishnamurthy S. A model for thermal conductivity
concrete irrespective of the aggregate type used. of porous building materials, under revision.
The value of N in Eqs. (7) and (8) can range from 0 to [9] Zimmerman RW. Thermal conductivity of 5uid-saturated rocks.
1; it ranges between 0 to 0.66 for oblate pores and between Journal of Petroleum Science and Engineering 1989;3:219–27.
0.66 and 0.9999 for prolate pores. For all the mortar and [10] Pratt AW. Heat transmission in low conductivity materials. Thermal
conductivity, Vol. 1. London: Academic Press, 1969.
concrete under consideration, the equivalent pore shape is [11] Khan MI, Bhattacharjee B. An experimental investigation on
oblate. This provides indirectly an idea about the dominant in5uence of aggregate type and moisture content on thermal
pore shape in concrete. The fraction of enclosed pores given conductivity of concrete. In: Choi E, editor. Building envelope
in Table 6 provides further an insight into the type of likely systems and technology. Centre for Continuing Education, Nanyang
pore structure that can exist in concrete. Technological University, Singapore: 1994. p. 522– 6.
[12] Neville AM, Brooks JJ. Concrete technology., 1st ed. UK: Longman,
1987.

7. Conclusions

(1) Thermal conductivity of rocks and mortar increases

with an increase in the moisture content.