Construction and Building Materials 70 (2014) 351–361

Contents lists available at ScienceDirect

Construction and Building Materials

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

Review

Measurement of thermal properties of brick materials based

on clay mixtures
N. Laaroussi a,⇑, G. Lauriat b, M. Garoum a, A. Cherki a, Y. Jannot c
a
Ecole Supérieure de Technologie de Salé, Université Mohammed V Rabat-Agdal, Laboratoire d’Energétique, Matériaux et Environnement (LEME), Avenue Prince Héritier, BP. 227
Salé Medina, Morocco
b
Laboratoire MSME, UMR CNRS 8208, Université Paris-Est, 77454 Marne-la-Vallée Cedex 2, France
c
LEMTA, Nancy-Université, CNRS, 2, Avenue de la Forêt de Haye, BP 160-54504 Vandoeuvre Cedex, France

h i g h l i g h t s

 Transient, steady state hot-plate and flash methods used for clay thermal properties.
 A comparative study based on the different experimental results is performed.
 The specific heat, thermal conductivity and diffusivity measurements are discussed.
 Numerical simulations of 3D-transient heat conduction conducted for validations.

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

Article history: Thermal properties of building materials such as clay used in construction are measured using transient
Received 29 January 2014 and steady state hot-plate and flash methods. The experimental methods are applied to measure the
Received in revised form 20 June 2014 thermal properties of clay samples. The estimation of these thermal characteristics is based on a one
Accepted 23 July 2014
dimensional model. A tridimensional model of the heat transfer in the system is considered to determine
the validity conditions of the one dimensional model used to represent the center temperature. The use of
a brick as a thermal insulating material requires prior knowledge of all its thermal properties. For that
Keywords:
purpose, we have conducted an experimental study in order to characterize the thermal properties of
Thermal insulating
Building materials
brick samples, coming from the Moroccan Slaoui’s factory. The thermal diffusivity ‘‘a’’, specific heat ‘‘c’’
Thermal properties and thermal conductivity ‘‘k’’ of this clay are experimentally obtained and their values are reported for
Brick design purposes. Finally, a comparative study based on different experimental methods is performed
Clay material and the measurements are compared. The error estimations are found less than 3%.
Ó 2014 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352
2. Chemical analysis of clay sample . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352
3. Description of the experimental approaches used for thermal property measurements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
3.1. Specific heat (c) and heat capacity (qc) determinations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
3.2. Thermal conductivity determination by the centered hot plate method (k) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
3.3. Diffusivity coefficient (a) determination by the flash method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
3.4. Thermal effusivity (E) determination by the hot plate transient method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
4. Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
4.1. Unidimensional model validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
4.2. Thermal conductivity (k) measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356

⇑ Corresponding author. Tel.: +212 661663710.

E-mail address: n.laaroussi@est.um5a.ac.ma (N. Laaroussi).

http://dx.doi.org/10.1016/j.conbuildmat.2014.07.104
0950-0618/Ó 2014 Elsevier Ltd. All rights reserved.
352 N. Laaroussi et al. / Construction and Building Materials 70 (2014) 351–361

4.3. Thermal effusivity (E) and thermal capacity (qc) measurements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
4.4. Thermal diffusivity (a) measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 358
4.5. Comparisons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360
5. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361

Nomenclature

a thermal diffusivity (m2 s1) Greek symbols

c specific heat (J kg1 K1) W quadratic error between experimental and theoretical
Ch thermal capacity of the heating element per area unit curves
(J m2 K1) k thermal conductivity (Wm1 K1)
e thickness (mm) q density (kg m3)
eh thickness of heating element (mm) qc volumetric thermal capacity (J m3 K1)
E thermal effusivity (Wm2 K1 s1/2) /0 heat flux density (Wm2)
hc convective heat transfer coefficient (Wm2 K1) / laplace transform of the heat flux density
I current intensity (A) h laplace transform of the temperature
L sample and heating element length and width (m)
p laplace parameter Subscripts
Re electrical resistance of the heating element (X) M mean value
Rc thermal contact resistance between the heating element al aluminum
and the sample (K m2 W1) exp experimental
S heat exchange surface between the heating element and mod model
the sample (m2) h heating element
t time (s) p polystyrene
T temperature (°C) s sample
T0 initial temperature of the system (°C) HPS Hot plate method in steady state regime
U voltage HPT asymmetrical transient Hot Plate method
DSC differential scanning calorimeter

1. Introduction hot plate. The consequences in terms of thermal conductivity and

Clay is commonly defined as an earthy material that is plastic at
certain water contents and hardens when heated. Most products
obtained from clay have been investigated for building applica-
tions. It is widely used in making bricks which are made out of
clay-bearing subsoil. Hollow bricks of several widths are used in
single and double walls construction. Hollow bricks are lighter
and easier to handle, and have different thermal properties accord-
ing to the shape and to the number of holes. Some brick designs
have very high thermal insulation properties, making them suit-
able for zero-energy building.
Several methods for measuring the thermal properties of
various solid materials are well known. Many studies have been
published concerning the characterization of thermal properties
of materials. Toppi and Mazzarella [1] have carried out experimen-
tal correlations for thermal properties estimation of composite
materials. Khabbazi et al. [2] conducted an experimental study of
thermal and mechanical properties of new insulating material
based on granular cork embedded in cement mortar. Jannot et al.
[3] used the symmetrical transient hot plate method which
requires two similar samples to characterize thin insulating mate-
rials. Recently, Bal et al. [4] adopted the asymmetrical device to
characterize the laterite based bricks with milled waste additive
materials by using only one sample. André et al. [5] presented an
experimental set-up based on the hot wire method for the thermal
characterization of materials. Coquard et al. [6,7] investigated an
experimental and theoretical study of the hot-ring method applied
to low density thermal insulators. An application to the thermal
resistance and thermal conductivity measurement of a heteroge-
neous material is presented by Jannot et al. [8], who used a tiny
specific heat have been studied in [9], using the laser flash
technique. Hernández-Olivares et al. [10] suggested the use of
cork-gypsum composites for building applications. They reported
the mechanical properties, microstructure, acoustic properties,
but only the thermal conductivity of cork-gypsum composite.
Many experimental methods for thermal materials properties
characterization have been performed from research efforts all
over the world. The aim of this paper is for investigating experi-
mental methods for estimating the thermal properties of clay sam-
ples (coming from the Moroccan Slaoui’s factory). The thermal
properties of clay samples are reported, in order to provide indus-
trial designers with values based on high levels of confidence.
2. Chemical analysis of clay sample
The hollow Slaoui’s bricks, made of clay mixed with 22% water
are hardened by drying for three hours at 110 °C before being fired.
The bricks are burnt at a temperature of 780 °C for 24 h. The ther-
mal properties of clay bricks are influenced by the chemical and
mineral content of the raw materials, the firing temperature, and
the atmosphere in the kiln. A rational analysis is necessary to
determine the relative proportions of substances which may be
present in the clay bricks. For the determination of substance com-
positions, a complete chemical analysis is necessary [11].
The brick produced is a complicated mixture and it is necessary
to study its chemical and physical properties. The physical proper-
ties are affected by the chemical composition of the clay and the
impurities present in pure clay. The Slaoui bricks are made from
several types of clay. Typical compositions of the two main clays
used are reported in Tables 1 and 2.
 N. Laaroussi et al. / Construction and Building Materials 70 (2014) 351–361 353

Table 1
Chemical composition of used clays [11].

Percent composition of mass Red clay (Rommanie) Gray clay (Rommanie) Yellow clay (Khmisset) Clay cellars
SiO2 49.45 51.27 34.19 65.76
Al2O3 14.24 14.55 9.75 14.56
Fe2O3 5.03 9.66 3.73 8.82
CaO 3.76 9.01 23.97 1.08
MgO 12.97 9.39 2.70 0.37
K2O 2.77 0.33 1.65 1.01
TiO2 0.78 1.42 0.56 0.89
P2O5 0.10 0.15 0.17 0.09
Carbonates 5.10 0.90 39.71 1.11

Table 2
Percentage of composition of Slaoui’s brick [11].
In order to obtain the thermal properties of the clay composites,
Type of Red clay Gray clay Yellow clay Clay two kind of clay samples have been manufactured (Table 3). Sam-
clays (Rommanie) (Rommanie) (Khmisset) cellars ples manufactured with different thicknesses and similar composi-
(%) (%) (%) (%)
tions, have been tested with different experimental methods. The
Composition of 57.14 14.28 14.28 14.28 samples represent a material having some heterogeneity.
mass

Table 3
3. Description of the experimental approaches used for thermal
Dimensions and densities of the studied samples. property measurements

Samples Sample 1 Sample 2

3.1. Specific heat (c) and heat capacity (qc) determinations
Dimensions (mm3) 100  100  26 100  100  21
Density (kg m3) 1777 1652
The measurements of the specific heat clays were based on the
use of two differential scanning calorimeters, (lDSC3) and (lDSC7
Evo), designed for the study of samples in isothermal or/and scan-
ning modes over a wide temperature range.
Clay samples were crushed and reduced to powder before being
analysed. The powder was inserted into a standard cell, which was
inserted into the micro-calorimeter (lDSC7 Evo), initially main-
tained at 25 °C and providing an operating temperature range
[45 °C, 120 °C]. With the (lDSC3) operating in the temperature
range [20 °C, 120 °C], the sample was in solid form of mass
244.6 mg. The Micro DSC’s specifications provide for various
modes of operation depending on the applications to be carried
out. The scanning by temperature increment is especially attrac-
tive for determining the heat capacity of solids because the mea-
surements have a better accuracy.
The specific heat variation of the clay (Fig. 1) clearly shows the
temperature dependence of the specific heat. The differences for
both measurements do not exceed 1% at 20 °C and 2% at 60 °C. Since
thermal properties tests are conducted at ambient conditions, an
average value of specific heat capacity close to 821 J kg1 K1 is
retained. The thermal capacities, qc, of the two samples at ambient
temperature were deduced from the densities reported in Table 3.

3.2. Thermal conductivity determination by the centered hot plate

method (k)
Fig. 1. Temperature variation of the clay specific heat capacity.

We apply the steady state, hot plate method [8] to estimate the
thermal conductivity. The method is based on temperature mea-
surements at the center of heating element inserted between the
sample and a polyethylene foam. The heating element, foam and
sample have the same cross-section area S.
Fig. 2 describes the principle of the experiment. The heating
element is composed of a (100  100  0.22 mm3) plane resistance
inserted between the sample and the insulating polyethylene foam
having the dimensions (100  100  10 mm3), thermal conductiv-
ity k2 = 0.04 W m1 K1 and a thickness e2 = 10 mm. Most of the
heat dissipated into the heating element having an electric
resistance Re  40 X, passes through the upper part of the heating
Fig. 2. Experimental device of the centered hot plate method in steady state element. All the temperatures are measured by K-type thermocou-
regime. ples with a 0.005 K resolution.
354 N. Laaroussi et al. / Construction and Building Materials 70 (2014) 351–361
The set is then placed between two aluminum blocks of dimen-
sions (100  100  50 mm3), which ensure a constant temperature
on the unheated faces of the sample and of the polyethylene foam.
As shown in Fig. 2, a thermocouple is stuck on the center of the
lower face of the heating element to measure the temperature T0,
another one measures the temperature T1 of the unheated face of
the sample and a third measures the temperature T2 of the
unheated face of the insulating foam. With this configuration, we
can write at steady state and assuming one-dimensional heat
conduction:
/0 ¼ /1 þ /2
k1
/1 ¼ ðT 0  T 1 Þ
e1
k2
/2 ¼ ðT 0  T 2 Þ
e2
/1 is the heat flux density through the sample, /2 the heat flux den-
sity through the insulation foam and, /0 the flux through both faces
of the heating element. e1 is the thickness of the sample and k1 the
thermal conductivity of the sample to be determined. The tension U
applied through the heating element is measured with a 0.01 V
accuracy. Therefore:
U2
/0 ¼
2Re S
By combining Eqs. (1)–(4), we obtain:
" #
e1 U2 k2
k1 ¼  ðT 0  T 2 Þ
T 0  T 1 2Re S e2
Eq. (5) allows the determination of the sample thermal conductivity
once the system reaches the steady state regime.
The measurement is performed relative to a known reference
material, the foam. The estimation of the thermal conductivity
being validated by an experimental study on an insulating sample
such as the foam, whose conductivity is known, enabled us to cal-
ibrate the experimental device.
3.3. Diffusivity coefficient (a) determination by the flash method
The thermal diffusivity measurement is based on the flash
method [16,17]. An energy pulse heats one side of a plane-parallel
sample and the transient temperature rise T(t) on the backside due
to the energy input is recorded. The flash method is applied to
measure the thermal diffusivity of samples [12] from solving the
heat equation in Laplace space using the quadrupole model [13].
The heat losses from both sides of the sample to its surrounding
are represented by heat convective coefficients hc, assumed equal
(Fig. 3). The lateral surface of the sample is insulated. The sample
Fig. 3. Sample with heat losses on both faces and insulated lateral surface.
ð1Þ
ð2Þ
Fig. 4. Scheme of the experimental hot plate device.
ð3Þ
is crossed by a heat flux /0 (W m2) emitted by a flash lamp during
a short time s. The temperature rise is measured by a thermocou-
ple placed in the center of the rear face (Fig. 3). In the flash method,
the front surface is often coated with a thin, opaque and black
layer, in order to increase the absorptivity of the flash radiative
heat flux at the front surface, so that the rear surface temperature
may be easily detected by the thermocouple. Since the black layer
is usually very thin, its thermal resistance is negligible provided
ð4Þ that the sample has an adequate thickness.
According to the above assumptions, the transfer matrix may be
expressed by the product of the following three matrices:
 " sinhðqeÞ
# 
1 0 coshðqeÞ kqS
1 0
ð5Þ ð6Þ
hc 1 kq:S sinhðqeÞ coshðqeÞ hc 1
qffiffi
p
where q ¼ a
, with p the Laplace parameter and a the thermal
diffusivity of the sample. k and e are the thermal conductivity and
thickness of the sample.
The expression of temperature rise in the Laplace space reads:
/0
p
ð1  esp Þ
hðpÞ ¼ 2 sinhðqeÞ
ð7Þ
hc þ 2hc coshðqeÞ þ kqS sinhðqeÞ
kqS
s is the elapsed time from the flash pulse heating.
The numerical inversion of this expression, carried out by the
De Hoog algorithm [14], leads to the theoretical expression of the
temperature. The Levenberg–Marquardt algorithm [15] is used
for the estimations of the parameters a, /0 and hc that reduce the
quadratic error between the temperatures recorded at the center
of the rear sample side and theoretical expression of the tempera-
ture evolution.
3.4. Thermal effusivity (E) determination by the hot plate transient
method
The thermal conductivity of the sample can be also determined
from the measurement of the thermal effusivity, E = (kqc)1/2, pro-
vided that the thermal capacity qc is known, or simultaneously
measured. The hot plate transient method described in this subsec-
tion allows measurements of E and qc.
The asymmetrical experimental device used in this work is rep-
resented in Fig. 4. A plane heating element having the same section
(100  100 mm2) as the sample is placed under the sample. A type
K thermocouple made of two wires with a 0.005 mm diameter is
stuck on the lower face of the heating element.
The sample is placed between two extruded polystyrene blocks
(Fig. 4) having a thickness of 50 mm. The sample and the polysty-
rene blocks are set between two aluminum blocks with a thickness
of 40 mm. A heat flux step is sent into the heating element and the
 N. Laaroussi et al. / Construction and Building Materials 70 (2014) 351–361 355

transient temperature T(t) is recorded. Since the thermocouple is in

contact with polystyrene that is a deformable material, the pres-
ence of the thermocouple does not increase the thermal contact
resistance between the heating element and the polystyrene. Fur-
thermore, since polystyrene is an insulating material, this contact
thermal resistance will be neglected in comparison the thermal
resistance of the polystyrene block. The theoretical model origi-
nates from the integral transform of the heat equation [13]. The
hypothesis is that the heat transfer remains unidirectional 1D at
the center of the sample during the experiment. The value given
by relations (10) is affected by convective effects through the lat-
eral faces of the sample that may affect the flux through the sam-
ple, which is different from the /0 value used in Eq. (7) for the
estimation of the thermal effusivity E. So, the recording processing
of T(t) is carried out by supposing that the heat transfer at the cen-
ter of the heating element is 1D.
By considering the very low value of the heat flux reaching the
aluminum blocks through the polystyrene and their high thermal
capacity, their temperatures are supposed equal and constant.
These assumptions lead to:
       
h 1 0 1 Rc A B Ai Bi 0
¼
U01 ch 1 0 1 C D Ci Di U1 Fig. 5. Scheme of the model system.
   ð8Þ
A1 B1 0
¼ The principle of the method is to estimate the value of the
C 1 D1 U1
sample effusivity that minimizes the sum of the quadratic error
P
     w ¼ Nj¼0 ½T exp ðt j Þ  T mod ðtj Þ2 between the experimental and the
h Ai Bi 0
¼ ð9Þ theoretical curves calculated with relation (13). The inverse
U02 Ci Di U2
Laplace transformation is performed using the De Hoog algorithm
with: [14].
The sensitivity analysis is based on the interpretation of the
/0 reduced sensitivity of T(t) to X: X @X @T
ðtÞ providing information on
U0 ¼ ¼ U01 þ U02 ð10Þ
p the influence of the parameter X on the temperature T according
h is the Laplace transform of the temperature T(t), /01 the Laplace to Beck [19]. The sensitivity analysis has been conducted for the
transform of the heat flux density living the heating element set of parameters (E, qc, Rc, Ch).
(upstream), /02 the Laplace transform of the heat flux density living
the heating element (downstream), /0 the Laplace transform of the 4. Results and discussion
total heat flux density produced in the heating element, /0, Ch the
thermal capacity of the heating element per area unit: Ch = qhcheh; 4.1. Unidimensional model validation
Rc the thermal contact resistance between the heating element
and the sample, /1 and /2 the Laplace transforms of heat flux den- A tridimensional heat conduction model has been numerically
sities input on the upper and lower aluminum blocks, respectively. solved in order to determine the accuracy of the measurement
 pffiffiffi techniques of the thermophysical properties. These measurements
qc pffiffiffi sinh qEc e p
A ¼ D ¼ cosh e p ; B¼ pffiffiffi ; are indeed based on analytical solutions of the transient, one-
E E p dimensional, heat conduction equation. The heat transfer by con-
pffiffiffi  qc pffiffiffi duction in the experimental set-up (Fig. 5) is more or less 3D, espe-
C ¼ E: p sinh e p ð11Þ
E cially after a long period of time depending on the thermal
qffiffiffi  diffusivity of the sample. The measurements are extracted from
rffiffiffiffi
sinh p
e the transient temperature changes at the center of the heating ele-
p ai i
Ai ¼ Di ¼ cosh ei ; Bi ¼ qffiffiffi ; ment. Therefore, we conducted some numerical 3D simulations in
ai p
ki ai order to check the maximum time at which the heat conduction
rffiffiffiffi rffiffiffiffi
differs at this point from the 1D assumption. When using an exper-
p p
C i ¼ ki sinh ei ð12Þ imental device as shown in Fig. 5, it is obviously required to carry
ai ai
out at first the following numerical computations for a sample
where qc is the sample thermal capacity, e the sample thickness, ki whose thermophysical properties are expected to be located
the polystyrene thermal conductivity, ai the polystyrene thermal within presumed ranges of values.
diffusivity and ei the polystyrene thickness. The 3D model is based on the assumption that the end horizon-
Combining the five Eqs. (8)–(12), leads to: tal surfaces of the device (y = 0 and y = H) are maintained at the
uniform and constant temperature T0, and that the vertical sides
U0 ðpÞ are in contact with ambient air at T0 with a uniform surface heat
hðpÞ ¼ D ð13Þ
1
B1
þ DB i transfer coefficient by natural convection and radiation, hc.
i
The transient, tridimensional heat conduction equations read:
The contact resistances have been estimated by applying a ten-
sion of 10 V through the heating element and by measuring the tem- @T
qal cal ¼ kal r2 T for 0 6 y 6 e4 and H  e4 6 y 6 H; 8x; z
peratures without any sample. Considering kal = 200 W m1 K1, we @t
obtain Rc = 104 k W1 m2. ð14Þ
356 N. Laaroussi et al. / Construction and Building Materials 70 (2014) 351–361

@T according to the material layers (a total of 2.55  106 control vol-

qp cp ¼ kp r2 T for e4 6 y
@t umes). Computations were first performed by using the thermo-
6 e3 þ e4 and H  ðe3 þ e4 Þ 6 y 6 H  e4 ; 8x; z ð15Þ physical properties reported in Table 4, those of the sample
being assumed to be accurate.
@T Two values of the heat transfer coefficient were considered:
qh ch ¼ kh r2 T þ S0 for e3 þ e4 6 y hc = 0 and hc = 10 W m2 K1, The first one allows simulation of
@t
6 H  ðe1 þ e3 þ e4 Þ; 8x; ð16Þ the 1D model. Figs. 6 and 7 show 3D-temperature maps at different
times along two side surfaces of the device for hc = 0 and hc = 10
The source term S0 introduced in the energy equation is for the vol- W m2 K1. The temperature field for hc = 0 is obviously 1D: it is
umetric heat flux produced in the heating element of volume Vh, the reference case for analyzing of the other case because all these
S0 = ReI2/Vh. solutions are numerical (i.e. approximate). As expected, Fig. 6
shows 1D temperature fields. On the other hand Fig. 7 shows that
@T
qs cs ¼ ks r2 T for H  ðe3 þ e4 þ e1 Þ 6 y the temperature distribution is almost 1D at the beginning of the
@t heating while noticeable deviations from the 1D solution are seen
6 H  ðe3 þ e4 Þ; 8x; z ð17Þ as the heating time increases. The transient temperature distribu-
tions at the middle plane of the device, i.e. at the center of which
The above system of equations is subjected to the following initial
the thermocouple is located, are shown in Figs. 8 and 9. Compari-
and boundary conditions:
sons between Figs. 7 and 9 show that the 3D effects are magnified
T ¼ T0 at t ¼ 0 8x; y; z ð18Þ around the thermocouple location.
Tridimensional simulations of the experimental set-up were
T ¼ T0 at y ¼ 0 and y ¼ H; 8t > 0; 8x; z ð19Þ performed in order to define the limits of validity of the 1D model.
The two samples depicted in Table 3 are considered (100  100 
@T 26 mm3 and 100  100  21 mm3). Three possible values of the
k ¼ hC ðT  T 0 Þ at x ¼ 0 and x ¼ L; 8t > 0; 8y; z ð20Þ
@x thermal conductivity of the clay mixture were retained for deter-
mining the influence of k on the unsteady temperature field within
@T the experimental set-up. Therefore, the thermal conductivity of the
k ¼ hC ðT  T 0 Þ at z ¼ 0 and z ¼ L 8t > 0; 8x; y ð21Þ
@x sample 2 was varied in the range [0.28 6 k 6 0.42 W m1 K1]
where hc (W m2 K1) is the surface heat transfer coefficient while qc = 1.458  106 J m3 K1. For each k-values, the tempera-
assumed to be uniform along the walls. ture evolution at the center of the heated face was computed either
Along the contact surfaces of the solid layers, the thermal con- with hc = 10 W m2 K1 or with hc = 0 (1D heat conduction). The
tact thermal resistances are neglected because polystyrene is a soft time tmax corresponding to a relative difference of 1% between
material. We considered a heating element with a thickness the two simulated temperatures was then estimated. The results
eh = 0.22 mm and a thermal capacity qhch = 1.5  106 J m3 K1. are represented in Fig. 10.
Since the thermal contact resistances and thermal resistance of Fig. 11 shows the temperature rise at the center of the heated
the heating element are supposed negligible, the temperature of side of the two samples of thicknesses 21 mm and 26 mm for
the heating element Th is equal to the sample temperature TS at hc = 0 and hc = 10 W m2 K1. The objective of these simulations
the interface [3]. is to determine the time (tmax) corresponding to a 1% difference
Therefore, heat flux and temperature continuities are assumed between the 1D and the 3D models. Fig. 11 shows that the 1D
at the interfaces: model is valid for times lower than tmax = 1200 s for a clay sample
with the thickness range [21 6 e 6 26 mm]. The conclusion is
@T @T @T @T @T @T that the duration of the measurements must be limited to times
kal ¼ kp ; kp ¼ kh ; kh ¼ ks 8x; z ð22Þ
@y @y @y @y @y @y less than about 1200 s.
9
T al ¼ T p at y ¼ e4 and y ¼ H  e4 >
>
> 4.2. Thermal conductivity (k) measurements
T p ¼ T h at y ¼ e3 þ e4 =
8x; z ð23Þ
T h ¼ T s at y ¼ H  ðe1 þ e3 þ e4 Þ >
> The experiments carried out using the steady-state hot plate
>
;
T s ¼ T p at y ¼ H  e3  e4 method (HPS) were repeated three times in order to estimate the
measurement errors. The three measured values of the thermal
The temperature field T(x, y, z, t) is obtained by solving Eqs.
conductivity and the average values are reported in Table 5 for
(14)–(23). Calculations were carried out using the finite-volume
the samples 1 and 2.
based code ANSYS 14.0 [18]. A second order scheme was used for
The average value of the thermal conductivity of the clay is thus
the spatial discretization and the time integration was performed
kHPS = (k1 + k2)/2  0.35 W m1 K1. The error on the estimated
with a second-order implicit scheme with a time-step t = 20 s. Uni-
value of kHPS caused by the experience uncertainties is defined
form 100  100 meshes were used in the (x, z) planes while the
from the differences |k1,2–kHPS|/kHPS which gives quite satisfactory
numbers of grid points in the y-direction are given in Table 3
results (maximum error of about 1%).

Table 4 4.3. Thermal effusivity (E) and thermal capacity (qc) measurements
Parameters of simulation for sample 1.

Materials e (mm) k (W m1 K1) qc (J m3 K1) My We conducted the asymmetrical hot plate transient method
a
Sample1 26 0.35 1.458  106 50 (HPT) four times on the same sample. The parameter sensitivity
a
Sample2 21 0.35 1.356  106 50 analysis described in Section 3.4 was used to determine the rela-
Polyethylene 50 0.032 4.80  104 50 tive importance of factors influencing the characterization of the
Aluminum 40 200 2.368  106 50
sample thermal properties. The Levenberg–Marquard algorithm
Heating element 0.22 0.2 1.50  106 5
[15] was used to identify the parameters (E, qc, Rc, Ch) from the
a
Data is from experiment, My: mesh number in the y-direction. temperature evolution during each experiment. This method
 N. Laaroussi et al. / Construction and Building Materials 70 (2014) 351–361 357

Fig. 6. Isotherms at different times for hc = 0 W m2 K1.

Fig. 7. Isotherms at different times for hc = 10 W m2 K1.

Fig. 8. Isotherms at different times in the center plane (z = L/2) for hc = 0 W m2 K1.

allows the minimization of the sum of the quadratic error between Fig. 12 shows the time-evolution of the experimental and theo-
the experimental and the theoretical data. retical temperature with their residues defined as the difference
358 N. Laaroussi et al. / Construction and Building Materials 70 (2014) 351–361
Fig. 9. Isotherms at different times in the center plane (z = L/2) for hc = 10 W m2 K1.
Fig. 10. Simulation of Tc at the center of the heating element for the sample 2 by
considering different conductivities.
between the experimental and simulated curves. Fig. 13 illustrates
 
the reduced sensitivity curves E @T
@E
; qc @@T @T @T
qc ; Rc @Rc ; C h @C related to
h
the parameter identifications.
As expected, the theoretical model is very close to the measured
temperatures at the beginning of the experiment, while the dis-
crepancies increase for times larger than about 1200 s. This devia-
tion is due to the lateral heat losses that start to affect the
temperature in the vicinity of the thermocouple, located at the
center of the sample. The time step between two measurements
was then varied to ensure that the thermal effusivity was accu-
rately predicted, especially in the experiment starting.
It is worth mentioning that the sensitivities of E and qc vary
with time and become high enough for t > 800 s to use the identi-
fication algorithm. Thus, the parameters E and qc can be identified
as shown in Fig. 13.
This sensitivity analysis shows that the thermal effusivity E and
the thermal capacity qc can be estimated, simultaneously by con-
sidering tmax = 1200 s, in agreement with the findings of the
numerical simulations of the experimental set-up, as discussed in
Section 4.1.
Fig. 11. Simulation of Tc at the center of the heating element for the samples 1 and
2.
It can be readily shown in Fig. 13 that the thermal contact resis-
tance Rc and the heat capacity of the heating element Ch have
almost no influence on the thermogram because these parameters
are insensitive to the variation of the temperature during the
experiments. The thermal effusivity and thermal capacity values
obtained from the identification method for five trials are reported
in Table 6. The percentage deviation of each trial is compared with
the mean value, leading to a maximum deviation of 3.4% for qc of
sample 2. It should be underlined that the qc-values are close to
those measured by using DSC (see Section 3.1).
4.4. Thermal diffusivity (a) measurements
Another measurement done with the flash method allowed the
determination of the thermal conductivity kFlash, combining the
diffusivity results and the thermal capacity values calculated as
described in Section 3.3.
The thermal diffusivity was measured by the flash method and
the sensitivity analysis was carried out for the two samples with
thicknesses e = 21 mm and e = 26 mm. Fig. 14 represents the
 N. Laaroussi et al. / Construction and Building Materials 70 (2014) 351–361 359

Table 5
Values of the measured thermal conductivities of the two samples.

Experiment number T0 (°C) T1 (°C) T2 (°C) e1 (mm) e2 (mm) k2 (W m1 K1) /0 (W/m2) k1 (W m1 K1)
(a) Sample 1
1 37 24.20 20.80 26 10.2 0.04 240 0.344
2 35 23.30 19.80 0.345
3 36.40 23.80 20.20 0.350
Mean value 0.346
Standard deviation 0.63%
k2 (W m1 K1)
(b) Sample 2
1 41 32.70 27.32 21 10.2 0.04 240 0.358
2 40.41 31.69 27.54 0.352
3 40.17 31.44 27.36 0.349
Mean value 0.353
Standard deviation 1%

Fig. 13. Time-evolution of the reduced sensitivity curves of fitting parameters for
Fig. 12. Time-evolution of the experimental and simulated hot plate temperatures,
the two samples, (a) /0 = 90 W/m2, e = 26 mm, (b) /0 = 90 W/m2, e = 21 mm.
(a) /0 = 90 W/m2, e = 26 mm, and (b) /0 = 90 W/m2, e = 21 mm.

based on the interpretation of the reduced sensitivity of Tc(t) to,

experimental curve and the simulated curve with the estimated  
parameters; the residues defined as the difference between a @T
@a
c @T c
ðtÞ; /0 @/ 0
ðtÞ; h @T c
C @h ðtÞ providing information on the influence
C

the experimental points and the model are also represented in of the parameters (a, /0, hc) on the temperature Tc(t). The sensitiv-
Fig. 14. ity analysis predicts that the parameters may be estimated
The reduced sensitivities of the center temperature Tc(t) to the accurately with this method because the sensitivities to the
different parameters were calculated. The sensitivity analysis is parameters are uncorrelated for all times and will be easily to
360 N. Laaroussi et al. / Construction and Building Materials 70 (2014) 351–361

Table 6
Thermal effusivity and volumetric heat capacity from transient centered hot plate
measurements.

Experiment number E (W m2 K1 s1/2) qc (106 (J m3 K1)

Sample 1
1 705 1.453
2 712 1.427
3 700 1.417
4 679 1.413
5 682 1.381
Mean value 696 1.418
Standard deviation 2.1% 1.8%
Sample 2
1 695 1.374
2 679 1.403
3 680 1.296
4 691 1.336
5 668 1.405
Mean value 683 1.363
Standard deviation 1.6% 3.4%

Fig. 15. Time-evolution of the reduced sensitivity curves of fitting parameters for
samples, /0  1000 W, (a) e = 26 mm, s = 10 s, and (b) e = 21 mm, s = 10 s.

Table 7
Thermal diffusivity from Flash method.

Experiment number Sample 1 Sample 2

a  107 (m2 s1) a  107 (m2 s1)
1 2.428 2.572
2 2.410 2.574
3 2.472 2.600
4 2.434 2.614
5 2.455 2.595
Mean value 2.439 2.591
Standard deviation 1.0% 0.7%

4.5. Comparisons

Table 8 shows comparisons between the measured thermal

Fig. 14. Time-evolution of the experimental and simulated flash temperature
curves, /0  1000 W, (a) e = 26 mm, s = 10 s, and (b) e = 21 mm, s = 10 s.
estimate separately, especially the thermal diffusivity for which
the sensitivity is the highest. The results shown in Fig. 15 and
Table 7 present an overview of the experimental data.
properties of the two clay samples. The differences are based on
the HPT measurements, assumed to be the best accurate method.
First, there are discrepancies linked to the measurement tech-
niques, which are within about 3%. Second, the difference between
the samples composition and manufacturing process are negligible
for the thermal conductivity and remains quite small (3.5%) for the
thermal diffusivity and thermal capacity. The conclusion is that
the global uncertainties may reach about 7% for the transient
 N. Laaroussi et al. / Construction and Building Materials 70 (2014) 351–361
Table 8
Comparisons between the thermal properties of the two samples.
Methods Sample 1
2 6
k (W m 1
K 1
) a  10 7
(m s 1
) qc  10 (J m
DSC 1.458 (2.8%)
FLASH 2.439 (1.1%)
HPT 0.342 2.412 1.418
HPS 0.346 (1.17%)
predictions of thermal properties of the clay used for manufactur-
ing bricks.
For practical applications, the relevant thermal properties are
those of the final product (i.e. the hollow brick). These properties
involve the heat transfers by natural convection and surface
radiation within the holes as well as heat conduction through
the partitions. The apparatuses used for the measurements of
the thermal properties of the bricks show effective thermal prop-
erties. These effective properties include all the heat transfer
modes within the hollow bricks. Since the mass of the brick is that
of the clay partitions, it is relevant to accurately determine the
thermal properties of the clay, especially for the transient behav-
ior of the product.
5. Conclusion
This work is related to the setting up of new regulations by the
Moroccan government to increase the energy efficiency and sus-
tainability for new buildings. It is clear that the use of construc-
tion materials with the best thermal properties is the first
condition for greatly reducing the thermal heat losses of the final
product.
The thermal properties of clay samples, coming from the
Moroccan Slaoui’s factory [20], have been measured through the
use of different methods. The combination of data recorded using
various methods is indeed one of the most relevant way to improve
the reliability of the results. In this study, differential scanning cal-
orimeters, centered hot plate in steady state or transient regimes
and flash methods were applied to clay samples issued from a local
manufacturer. For the measurements of heat conductivity, diffusiv-
ity and volumetric capacity, the measurement techniques are
essentially based on the measure of the steady or transient temper-
ature at the center of a sample heated with a prescribed uniform
flux. Therefore, special care should be paid to check the validity
of the assumptions enabling the use of 1D heat conduction models
which are the basis for going through the thermal properties.
Numerical simulations of 3D, transient heat conduction within
the experimental device was thus carried out in order to determine
the validity of the experimental procedure according to the
expected ranges of thermal properties and thickness of a sample.
The comparisons between the different measurement techniques
showed fair accuracies.
361
Sample 2
3 1
K ) k (W m1 K1) a  107 (m2 s1) qc  106 (J m3 K1)
1.356 (0.5%)
2.591 (3.3%)
0.342 2.509 1.363
0.353 (3.2%)
References
[1] Toppi T, Mazzarella L. Gypsum based composite materials with micro-
encapsulated PCM: experimental correlations for thermal properties
estimation on the basis of the composition. Energy Build 2013;57:227–36.
[2] Khabbazi A, Garoum M, Terahmina O. Experimental study of thermal and
mechanical properties of new insulating material based on cork and cement
mortar. AMSE J Adv Modell Simulation 2005;74(7):73.
[3] Jannot Y, Felix V, Degiovanni A. A centered hot plate method for measurement
of thermal properties of thin Insulating materials. Measure Sci Technol
2010;21(3):1–8.
[4] Bal H, Jannot Y, Quenette N, Chenu A, Gaye S. Water content dependence of the
porosity, density and thermal capacity of laterite based bricks with millet
waste additive. Construct Build Mater 2012;31:144–50.
[5] André S, Rémy B, Pereira F, Cella N. Hot wire method for the thermal
characterization of materials: inverse problem application. Engenharia
Térmica 2003;4:55–64.
[6] Coquard R, Baillis D, Quenard D. Experimental and theoretical study of the hot-
ring method applied to low-density thermal insulators. Int J Thermal Sci
2008;47(3):324–38.
[7] Coquard R, Coment Ph, Flasquin G, Baillis D. Numerical and experimental
analysis of the hot-disk technique applied to low-density insulating materials
international journal of thermal sciences. IJTS 2013;65:242–53.
[8] Jannot Y, Remy B, Degiovanni A. Measurement of thermal conductivity and
thermal resistance with a tiny hot plate. High Temperatures-High Pressures
2009;39:11–31.
[9] Michot A, Smith D, Degot S, Gault C. Thermal conductivity and specific heat of
kaolinite: evolution with thermal treatment. J Eur Ceramic Soc
2008;28:2639–44.
[10] Hernández-Olivares F, Bollati MR, del Rio M, Parga-Landa B. Development
of cork–gypsum composites for building applications. Construct Build Mater
1999;13(4):179–86.
[11] Rapport d’essai N 0125PM/12, Briqueterie SLAOUI, Centre des Techniques &
Matériaux de Construction, CETEMCO, Casablanca, 04 Mai 2012.
[12] Degiovanni A, Laurent A. Une nouvelle technique d’identification de la
diffusivité thermique pour la méthode flash. Revue de Physique Appliquée
1986;21:229–37.
[13] Maillet D, André S, Batsale JC, Degiovanni A, Moyne C. Thermal quadrupoles:
solving the heat equation through integral transforms. Chichester (PA): John
Wiley & Sons Ltd; 2000.
[14] De Hoog F. An improved method for numerical inversion of Laplace
transforms. Soc Ind Appl Math 1982;3(3):357–66.
[15] Marquardt DW. An algorithm for least-squares estimation of nonlinear
parameters. J Appl Math 1963;11(2):431–41.
[16] Parker WJ, Jenkins RJ, Butler CP, Abbot GL. Flash method of determining
thermal diffusivity, heat capacity and thermal conductivity. J Appl Phys
1961;32(9):1679–84.
[17] Degiovanni A. Diffusivité et méthode flash. Revue Générale de Thermique
1977;185:420–41.
[18] Ansys Fluent 14 Users Guide (n.d), <www.fluentusers.com>.
[19] Beck JV, Arnold KJ. Parameter estimation in engineering and science. New
York: Wiley; 1977.
[20] Entreprise de Briqueterie Slaoui (Rabat-salé-zemmour-zaër, Salé), Ancienne
route de Méknès, 11000 Salé, Maroc.