Thermal Conductivity of Clay Bricks (Dondi-Italy-2004)
Thermal Conductivity of Clay Bricks (Dondi-Italy-2004)
Thermal Conductivity of Clay Bricks (Dondi-Italy-2004)
Michele Dondi1; Francesca Mazzanti2; Paolo Principi3; Mariarosa Raimondo4; and Giorgio Zanarini5
Abstract: In the present work the thermal conductivity of 29 samples of clay bricks was measured and the correlations of the thermal
performance with the compositional, physical, and microstructural features of products were investigated. The results obtained directed
our attention toward a better understanding of the role played by some parameters 共i.e., mineralogical components and pore size
Downloaded from ascelibrary.org by Technische Universitat Munchen on 07/01/15. Copyright ASCE. For personal use only; all rights reserved.
distribution兲, other than bulk density, in improving or depressing the insulating properties of bricks. Among them, the unfavorable role of
quartz, Ca-rich silicates, and amorphous phase came out, while the role of pore size and specific surface should be more accurately
evaluated in the structural design of materials.
DOI: 10.1061/共ASCE兲0899-1561共2004兲16:1共8兲
CE Database subject headings: Bricks; Thermal insulation; Bulk density; Microstructures; Mineralogy.
Introduction features of bricks play a very important role 共Schulle and Kutzen-
dorfer 1988; Jungk and Krcmar 1996; Rimpel and El Ghazzali
Due to the ever increasing requirements for energy saving and a 1998; Schlegek et al. 1999; Dondi et al. 2000兲.
pressing competition with alternative building materials, the ther-
This work is aimed at outlining the thermal conductivity of
mal insulating properties of clay bricks have recently become
clay bricks trying to single out the compositional, physical, or
more and more important 共Krahl 1989兲. Many studies have been
microstructural parameters which affect their thermal behavior
devoted to better understand the way of improving the thermal
most significantly. Moreover, a statistical treatment of data was
performance of clay bricks, acting on both the physical properties
performed in order to quantify the influence of the above men-
of terracotta 共porosity, etc.兲 and the geometrical design of prod-
ucts 共Krahl 1989; Schmidt-Reinholtz 1990; Anton 1993; Jungk tioned characteristics on thermal conductivity.
and Krcmar 1996; Rimpel and Schmedders 1996; Jungk et al.
1997; Hauck et al. 1998兲.
These studies point out that the thermal conductivity of bricks Materials and Methods
is mainly related to their bulk density, so that increasing the ther-
mal insulating properties implies the production of materials with Twenty-nine samples of clays, collected in 21 different brick-
a higher porosity 共Schmidt-Reinholtz 1990; Jungk and Krcmar works, were selected in order to represent the wide range of raw
1996; Jungk et al. 1997兲. However, the correlation between ther- materials utilized by the Italian brick industry 共Fabbri and Dondi
mal conductivity and bulk density is not statistically significant 1995兲. The sampling procedure and the manufacturing of clay
since data exhibit on the whole a considerable scattering 共Fig. 1兲. brick elements suitable for the conductivity measurements were
As a matter of fact, bulk density alone is not able to describe and described in detail in previous work 共Dondi et al. 2000兲. Sam-
accurately reflect the thermal behavior of clay bricks. The differ- pling was carried out in every brickwork, after clay treatment and
ent analytical methods used to measure the thermal conductivity extrusion, while drying and firing were performed in a single
关Albenque 1992; Anton 1993; Ente Nazionale Italiano di industrial plant. After firing, all products were ground and pol-
Unificazione 共UNI兲 1994兴 probably account for some discrepan- ished to get the suitable geometry 共disks of 200⫾1 mm diameter,
cies but, in most cases, the compositional and microstructural 20⫾2 mm of thickness, 0.05% planarity兲 for thermal conductivity
measurements.
1
Dr., Institute of Science and Technology for Ceramics, via Granarolo The thermal conductivity of bricks was measured by the hot
64, 48018 Faenza, Italy. E-mail: [email protected]
2
plate method, according to UNI 7745 共UNI 1977兲 standard with a
Dr., Institute of Science and Technology for Ceramics, via Granarolo Dynatech TCFGM apparatus. For each typology of product, six
64, 48018 Faenza, Italy. E-mail: [email protected]
3 specimens were tested with an experimental uncertainty lower
Professor, Dept. of Energetic, Univ. of Ancona, via Brecce Bianche,
100 Ancona, Italy. E-mail: [email protected] than 0.5%.
4
Dr., Institute of Science and Technology for Ceramics, via Granarolo Phase composition, open, closed, and total porosity, bulk den-
64, 48018 Faenza, Italy. E-mail: [email protected] sity, pore size distribution, and pore specific surface were deter-
5
Dr., Consorzio Aveolater, Viale A. Moro 16, 40127 Bologna, Italy. mined on fired products.
E-mail: [email protected] The phase composition was quantitatively determined by x-ray
Note. Associate Editor: Christopher K. Y. Leung. Discussion open powder diffraction 共Rigaku Miniflex, Cu K␣ radiation兲 with the
until July 1, 2004. Separate discussions must be submitted for individual reference intensity ratio method (Al2 O3 as internal standard兲. The
papers. To extend the closing date by one month, a written request must
experimental error is within 5% relative.
be filed with the ASCE Managing Editor. The manuscript for this paper
was submitted for review and possible publication on May 30, 2002; Open porosity 共OP兲 and bulk density were quantified by mea-
approved on March 19, 2003. This paper is part of the Journal of Mate- suring dry weight, water-saturated weight, and the weight sus-
rials in Civil Engineering, Vol. 16, No. 1, February 1, 2004. ©ASCE, pended in water, according to ASTM C 373 共ASTM 1994b兲. Total
ISSN 0899-1561/2004/1-8 –14/$18.00. porosity 共TP兲 was calculated as the ratio between bulk density
LM 0.47
D 0.49
LS 0.49
Fig. 1. Thermal conductivity versus bulk density values WSP 0.60
WSN 0.39
WPP 0.54
and specific weight according to ASTM C 329 共ASTM 1994a兲; WPN 0.42
CEP 0.54
the amount of closed porosity 共CP兲 was estimated by the differ-
CEN 0.50
ence: CP⫽TP⫺OP.
CAP 0.44
The pore size distribution 共in the 0.01–100 m range兲 was CAN 0.46
determined by mercury intrusion porosimetry 共Carlo Erba Poro- ATP 0.44
simeter 2000兲 with an experimental uncertainty of about 1% rela- ATN 0.42
tive. The pore specific surface analysis was performed by nitrogen RIP 0.46
absorption 共Micromeritics FlowSorb II 2300兲 following the RIN 0.48
Brunauer, Elmet, Teller single point method according to ASTM RDB 0.46
C 1069 共ASTM 1997兲. RIX 0.41
A statistical elaboration of data was performed by simple 共lin- ILS 0.52
ear binary correlation兲 and multivariate analysis techniques 共fac- ILP 0.63
tor analysis, multiple linear regression analysis and structural
equation modeling兲 using the StatSoft Statistica 5.0 software. Fac- well as the presence of a certain mineralogical component
tor analysis was carried out on the main physical, compositional 共Rimpel and El Ghazzali 1998兲, can play a very important role.
and microstructural variables extracting principal components Moreover, this thesis is supported by more detailed studies of the
共four factors according to the screen test for eigenvalues兲. Mul- heat transfer mechanism, which can be enhanced or depressed, for
tiple linear regression was executed by the forward stepwise example, by the different free mean path of air molecules en-
method, including intercept in the model and setting F⫽1.00 to trapped into the pores and, consequently, by the different pore
enter and F⫽0.00 to remove. size 共Schlegek et al. 1999兲.
Fig. 3. Thermal conductivity versus: 共a兲 open porosity, 共b兲 total porosity, 共c兲 bulk density, 共d兲 mean pore size, 共e兲 pore specific surface, and
共f兲 pore size selection
In order to shed light on the complex relationship between by K-feldspar and wollastonite in improving the thermal insulat-
thermal conductivity and bulk density, as well as on the role of ing properties of bricks, in contrast to the opposite tendency ex-
microstructural features and mineralogical composition, we per- hibited by plagioclase, pyroxene and illite-mica 共Fig. 4兲.
formed a statistical study of the results through different ap- As far as the influence of the microstructural variables 共bulk
proaches 共binary and multiple regression analysis, factorial analy- density, open, closed, and total porosity, mean pore size, pore
sis兲. selection, and specific surface兲, the total porosity, rather than bulk
The mutual correlations between thermal conductivity and the density, appears to be the only one with a statistically significant
microstructural 共Fig. 3兲 or mineralogical parameters 共Fig. 4兲 were relation with thermal conductivity 共Fig. 3兲. Some samples with a
first evaluated. The results obtained provide just some trends, higher total porosity actually have poorer thermal insulating prop-
such as the vague positive correlation of thermal conductivity erties, probably because of the contrasting influence exerted by
with bulk density or the negative correlations with open porosity pores having different size.
and mean pore size. In the other cases, no significant relationship The factorial analysis, performed through the analysis of the
is detectable. Plotting the thermal conductivity against the content main components, confirms that the thermal conductivity is influ-
of single mineralogical components reveals a different role played enced by several variables and that it is quite difficult to point out
Fig. 4. Thermal conductivity versus: 共a兲 K-feldspar, 共b兲 wollastonite, 共c兲 plagioclase, 共d兲 pyroxene, and 共e兲 illite-mica content
the more significant parameters, together with porosity, on bulk tural variables 共pore size selection兲; factor 2 explains the variance
density. of bulk density versus open and total porosity 共Fig. 5兲. Factors 3
From a strictly statistical point of view, the analysis of the and 4 account for the variance of some phases 共mica, pyroxene,
main components allows us to classify all the parameters into hematite兲 as well as some microstructural parameters 共mean pore
different groups, according to the correlation existing among size and pore specific surface兲 共Fig. 6兲.
them, and to reduce the number of significant variables 共Cooley The thermal conductivity is influenced in a complex way by
and Lohnes 1971兲. many variables and the four factors are able to explain no more
In our case, the principal components analysis extracted four than half of its variance. However, a certain effect of bulk density,
significant factors accounting for 67% of the total variance. Plot- open, and total porosity can be claimed, as the role of pore size
ting the factors it is possible to single out the relative position of and phase composition also stands out.
each parameter and, based on their mutual distance, the most A further statistical analysis was performed by a stepwise mul-
significant chemical and physical analogies stand out 共Figs. 5– 6兲. tiple regression, taking the thermal conductivity as a dependent
As shown in the square of Fig. 5, where the generic variables A, variable and the main compositional and physical parameters as
B, C, and D are represented, A is positively correlated with D, independent ones; this procedure selects progressively the inde-
negatively correlated with C, while there is no correlation be- pendent parameters with the greater significance. Once two outli-
tween A and B 共Cooley and Lohnes 1971兲. ers were eliminated 共sample X and WPS兲, the multiple regression
Factor 1 explains most variance of some compositional 共cal- provided a forecast of the thermal conductivity value with a fair
cium silicates, quartz兲, physical 共closed porosity兲 and microstruc- multiple correlation coefficient (R⫽0.780, R 2 ⫽0.608) with a
Fig. 6. Weight plot of factors 3 and 4 obtained by extraction of main Standard Standard Probability
components: 共PSE兲 pore size selection; 共MPS兲 mean pore size; 共PSS兲 N⫽27  error B error level p
pore specific surface; 共P50兲 fraction of pores smaller than 50 m; Intercept — — 0.603 0.047 0.000
共OP兲 open porosity; 共TP兲 total porosity; 共CP兲 closed porosity; 共BD兲 Open porosity ⫺0.882 0.157 ⫺0.008 0.001 0.000
bulk density; 共Qz兲 quartz; 共AM兲 amorphous; 共KF兲 K-feldspar; 共HE兲 Wollastonite 0.423 0.166 0.006 0.002 0.018
hematite; 共PE兲 periclase; 共IM兲 illite-mica; 共CC兲 calcium carbonate; Quartz 0.600 0.195 0.004 0.001 0.005
共CS兲 calcium silicates; 共PX兲 pyroxene; 共PL兲 plagioclase; 共ME兲 meli- Melilite 0.330 0.178 0.002 0.001 0.077
lite; 共WO兲 wollastonite; and 共TC兲 thermal conductivity
Note: Multiple correlation coefficients: R⫽0.780; R 2 ⫽0.608.